CN111127235B

CN111127235B - Fracturing sand blocking early warning method and device and related products

Info

Publication number: CN111127235B
Application number: CN201911193283.5A
Authority: CN
Inventors: 刘颜鹏
Original assignee: Neusoft Corp
Current assignee: Neusoft Corp
Priority date: 2019-11-28
Filing date: 2019-11-28
Publication date: 2022-05-10
Anticipated expiration: 2039-11-28
Also published as: CN111127235A

Abstract

The application discloses a fracturing sand blocking early warning method, a fracturing sand blocking early warning device and related products, and cycle characteristics are extracted by acquiring original data of each fracturing construction cycle in a fracturing construction process. Because the historical construction process is a construction process with sand blockage, and the cycle reference characteristics extracted from the original data before sand blockage occurs in the historical construction process reflect the characteristics of fracturing construction before sand blockage occurs, the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are matched with the cycle reference characteristics, and when the matching is successful, the sand blockage occurring in the fracturing construction process can be predicted. The application provides a technical scheme can effectively judge that the stifled problem of sand is about to take place before this fracturing work progress takes place sand stifled. Therefore, when the matching is successful, the early warning of the fracturing sand plugging is carried out, so that constructors can conveniently take corresponding measures to avoid the sand plugging, and the adverse effect caused by the sand plugging problem is avoided or reduced.

Description

Fracturing sand blocking early warning method and device and related products

Technical Field

The application relates to the technical field of oil and gas exploitation, in particular to a fracturing sand blocking early warning method and device and a related product.

Background

In the field of oil and gas exploitation, a construction technology for artificially forming cracks and fractures in a stratum under the action of water power is called a fracturing technology. The fracturing can improve the flowing environment of oil in the underground, so that the yield and the recovery rate of an oil well are improved, and the fracturing plays an important role in improving the flowing condition of the bottom of the oil well.

The fracturing construction generally comprises the working procedures of circulation, pressure test, trial extrusion, fracturing, sand adding, alternate extrusion, viewing diffusion pressure, moving pipe columns and the like. Due to the complexity of the formation, fracture construction is often subject to risks and anomalies. Among the various risks and anomalies, sand plugging is a very common type of problem. When sand blocking occurs, high pressure is formed to suppress a damaged pipeline, equipment is damaged, construction materials such as fracturing fluid are wasted, the cost of oil and gas exploitation operation is greatly increased, casualties can be caused, stratum seepage is damaged, and fracturing construction wells are scrapped.

In the prior art, sand blockage can be judged according to the condition of formation cracks and fractures or the analysis of a construction oil pressure curve. But sand blockage has often occurred when one can judge that a sand blockage problem exists. It can be seen that the prior art lacks a technical scheme capable of effectively early warning sand blockage to avoid or reduce adverse effects caused by the sand blockage problem.

Disclosure of Invention

Based on the problems, the application provides a fracturing sand blocking early warning method, a fracturing sand blocking early warning device and related products, so that effective early warning is carried out before sand blocking occurs, and adverse effects caused by the sand blocking problem are avoided or reduced.

The embodiment of the application discloses the following technical scheme:

in a first aspect, the application provides a method for early warning of fracturing sand plugging, comprising:

obtaining original data of one fracturing construction period in the fracturing construction process; the raw data comprises at least the following parameters: oil pressure, discharge capacity, sand adding amount and water adding amount;

extracting periodic features from the raw data;

matching the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process with the cycle reference characteristics, and performing fracturing sand plugging early warning when the matching is successful; the periodic reference features are extracted from original data before sand blockage occurs in the historical construction process in advance.

Optionally, extracting periodic features from the raw data specifically includes:

and providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a period characteristic corresponding to the fracturing construction period.

providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a first type of characteristic corresponding to the fracturing construction period;

obtaining a second type of characteristics by using parameters included in the original data of each acquisition time of the fracturing construction period, wherein the second type of characteristics at least comprise: the variance, the average value and the maximum value corresponding to each parameter;

and obtaining the periodic characteristics according to the first type of characteristics and the second type of characteristics.

Optionally, the method further comprises:

acquiring a plurality of pieces of training data; each piece of training data comprises original data of one fracturing construction period in the historical construction process; the plurality of pieces of training data include at least two lengths; the original data of one fracturing construction period in the historical construction process at least comprises the following parameters: oil pressure, discharge capacity, sand adding amount and water adding amount;

training the self-encoder to be trained by using the plurality of pieces of training data, and obtaining the trained self-encoder when any one of a first condition and a second condition is met;

the first condition is that the value of the loss function of the self-encoder is smaller than a first preset threshold value, and the difference between the output and the input of the self-encoder to be trained is positively correlated with the value of the loss function;

the second condition is that the iteration number of training the self-encoder to be trained by using a gradient descent optimization method reaches a second preset threshold value.

Optionally, matching the cycle characteristics and the cycle reference characteristics corresponding to each fracturing construction cycle in the fracturing construction process specifically includes:

constructing a first feature vector corresponding to each construction period in the fracturing construction process according to the period features corresponding to each fracturing construction period in the fracturing construction process;

judging whether the first feature vector falling into a feature vector matching range exists or not; if yes, determining that the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are successfully matched with the cycle reference characteristics; the feature vector matching range is determined according to the periodic reference features.

Optionally, before determining whether there is the first feature vector that falls within a feature vector matching range, the method further includes:

acquiring original data corresponding to each construction period before sand blockage occurs in the historical construction process, and extracting historical characteristics from the original data;

constructing a second characteristic vector corresponding to each construction period before sand blockage occurs in the historical construction process according to the historical characteristics;

clustering each second feature vector, and determining the second feature vector of which the category is the category to be pre-warned from each second feature vector according to the clustering result; the second feature vector of the category to be early-warned comprises the periodic reference feature;

and determining the feature vector matching range according to the second feature vector of the category to be pre-warned.

Optionally, determining the feature vector matching range according to the second feature vector of the category to be early warned specifically includes:

obtaining an average feature vector by using the second feature vector of the category to be pre-warned;

obtaining the distance between each second feature vector of the category to be pre-warned and the average feature vector, and determining the maximum distance;

and determining the range with the maximum distance as the radius as the feature vector matching range by taking the average feature vector as an origin.

obtaining each first node by using the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process, and connecting the first nodes end to end according to the time sequence to obtain a fracturing construction track;

comparing the track similarity of the fracturing construction track with the historical construction track, and when the similarity meets a preset condition, determining that the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are successfully matched with the cycle reference characteristics; the historical construction track is obtained in advance according to the periodic reference characteristics.

Optionally, the method further comprises:

obtaining a second node corresponding to each construction period before sand blockage occurs in the historical construction process by using the historical characteristics;

determining second nodes to be early-warned from the second nodes; the second node to be early-warned comprises the periodic reference feature;

and connecting the second nodes to be pre-warned end to end according to a time sequence to obtain the historical construction track.

In a second aspect, the application provides a device of stifled early warning of fracturing sand, includes:

the initial data acquisition module is used for acquiring initial data of one fracturing construction period in the fracturing construction process; the raw data comprises at least the following parameters: oil pressure, discharge capacity, sand adding amount and water adding amount;

the periodic feature extraction module is used for extracting periodic features from the original data;

the characteristic matching module is used for matching the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process with the cycle reference characteristics; the periodic reference characteristic is extracted from original data before sand blockage occurs in the historical construction process in advance;

and the early warning module is used for carrying out early warning on the fracturing sand plugging when the characteristic matching module is successfully matched.

Optionally, the periodic feature extraction module specifically includes:

and the first extraction unit is used for providing parameters included by the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included by the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder as a period characteristic corresponding to the fracturing construction period.

Optionally, the periodic feature extraction module specifically includes:

the second extraction unit is used for providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a first type of characteristic corresponding to the fracturing construction period;

the third extraction unit is used for obtaining a second type of characteristics by using parameters included in the original data of each acquisition time of the fracturing construction cycle, and the second type of characteristics at least comprise: the variance, the average value and the maximum value corresponding to each parameter;

and the fourth extraction unit is used for obtaining the periodic characteristics according to the first type of characteristics and the second type of characteristics.

Optionally, the apparatus further comprises:

the training data acquisition module is used for acquiring a plurality of pieces of training data; each piece of training data comprises original data of one fracturing construction period in the historical construction process; the plurality of pieces of training data include at least two lengths; the original data of one fracturing construction period in the historical construction process at least comprises the following parameters: oil pressure, discharge capacity, sand adding amount and water adding amount;

the self-encoder training module is used for training a self-encoder to be trained by utilizing the plurality of pieces of training data, and when any one of a first condition and a second condition is met, the trained self-encoder is obtained; the first condition is that the value of the loss function of the self-encoder is smaller than a first preset threshold value, and the difference between the output and the input of the self-encoder to be trained is positively correlated with the value of the loss function; the second condition is that the iteration number of training the self-encoder to be trained by using a gradient descent optimization method reaches a second preset threshold value.

Optionally, the feature matching module specifically includes:

the characteristic vector first construction unit is used for constructing a first characteristic vector corresponding to each construction period in the fracturing construction process according to the period characteristics corresponding to each fracturing construction period in the fracturing construction process;

the first matching unit is used for judging whether the first characteristic vector falling into a characteristic vector matching range exists or not; if so, determining that the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are successfully matched with the cycle reference characteristics; the feature vector matching range is determined according to the periodic reference features.

Optionally, the apparatus further comprises:

the historical characteristic extraction unit is used for acquiring original data corresponding to each construction period before sand blockage occurs in the historical construction process and extracting historical characteristics from the original data;

the characteristic vector second construction unit is used for constructing a second characteristic vector corresponding to each construction period before sand blockage occurs in the historical construction process according to the historical characteristics;

a clustering unit, configured to cluster the second feature vectors;

the vector type determining unit is used for determining second characteristic vectors of which the types are to-be-early-warning types from the second characteristic vectors according to clustering results; the second feature vector of the category to be early-warned comprises the periodic reference feature;

and the matching range determining unit is used for determining the matching range of the feature vector according to the second feature vector of the category to be pre-warned.

Optionally, the matching range determining unit specifically includes:

the average characteristic vector obtaining subunit is configured to obtain an average characteristic vector by using the second characteristic vector of the category to be pre-warned;

the maximum distance obtaining subunit is configured to obtain distances between the second feature vectors of the categories to be pre-warned and the average feature vector, and determine a maximum distance therebetween;

and the matching range determining subunit is configured to determine, as the feature vector matching range, a range in which the average feature vector is used as an origin and the maximum distance is used as a radius.

Optionally, the feature matching module specifically includes:

the first track acquisition unit is used for acquiring each first node by using the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process, and connecting the first nodes end to end according to a time sequence to acquire a fracturing construction track;

the second matching unit is used for comparing the track similarity of the fracturing construction track with the historical construction track, and when the similarity meets a preset condition, determining that the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are successfully matched with the cycle reference characteristics; the historical construction track is obtained in advance according to the periodic reference characteristics.

Optionally, the apparatus further comprises:

the historical characteristic extraction unit is used for obtaining the original data corresponding to each construction period before sand blockage occurs in the historical construction process and extracting historical characteristics from the original data;

the node acquisition unit is used for acquiring second nodes corresponding to the construction periods before sand blockage occurs in the historical construction process by using the historical characteristics;

the node determining unit is used for determining second nodes to be pre-warned from the second nodes; the second node to be early warned comprises the periodic reference feature;

and the track second acquisition unit is used for connecting the second nodes to be pre-warned end to end according to a time sequence to acquire the historical construction track.

In a third aspect, the present application provides a computer readable storage medium, in which a computer program is stored, and when the program is executed by a processor, the method for early warning of fracturing sand plugging as provided in the first aspect is implemented.

In a fourth aspect, the present application provides a processor for executing a computer program, which when executed performs the method for fracture sand blockage warning as provided in the first aspect.

Compared with the prior art, the method has the following beneficial effects:

the method and the device extract the cycle characteristics by acquiring the original data of each fracturing construction cycle in the fracturing construction process. Because the historical construction process is a construction process with sand blockage, and the cycle reference characteristics extracted from the original data before sand blockage occurs in the historical construction process reflect the characteristics of fracturing construction before sand blockage occurs, the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are matched with the cycle reference characteristics, and when the matching is successful, the sand blockage occurring in the fracturing construction process can be predicted. The application provides a technical scheme can effectively judge that the stifled problem of sand is about to take place before this fracturing work progress takes place sand stifled. Therefore, when the matching is successful, the early warning of the fracturing sand plugging is carried out, so that constructors can conveniently take corresponding measures to avoid the sand plugging, and the adverse effect caused by the sand plugging problem is avoided or reduced.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive exercise.

Fig. 1 is a flowchart of a method for early warning of a fractured sand plug according to an embodiment of the present disclosure;

fig. 2 is a flowchart of a method for training a self-encoder according to an embodiment of the present disclosure;

FIG. 3 is a diagram of a neural network architecture for a self-encoder;

FIG. 4 is a flow diagram of another implementation of extracting periodic features from raw data;

FIG. 5 is a flow chart of another method for early warning of frac sand plugging provided by an embodiment of the present application;

fig. 6 is a flowchart of another method for early warning of a fractured sand plug according to an embodiment of the present disclosure;

FIG. 7a is a schematic illustration of a historical construction track;

FIG. 7b is a schematic illustration of a fracture construction trajectory;

fig. 8 is a schematic structural diagram of the fracturing sand plugging early warning device provided in this embodiment;

fig. 9 is a hardware structure diagram of the equipment for early warning of fractured sand plugging provided by this embodiment.

Detailed Description

At present, in the field of oil and gas exploitation, sand blockage can only be determined to exist when or after the sand blockage occurs, and effective early warning cannot be achieved, so that adverse effects caused by the sand blockage are difficult to avoid, for example, construction material waste, exploitation cost rise, casualties, stratum seepage damage, construction well abandonment and the like are caused.

Based on the above problems, the embodiments of the present application provide a method and an apparatus for early warning of fractured sand plugging, and related products. For the requirement of early warning of sand blockage in the fracturing construction process, extracting the period characteristics from the original data of each fracturing construction period in the process, matching the period characteristics with the period reference characteristics extracted in the historical construction process before sand blockage occurs, and finally early warning of sand blockage according to the matching result. Therefore, the problems of material waste and casualties caused by untimely early warning of sand blockage are avoided.

In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Method embodiment

Referring to fig. 1, the figure is a flowchart of a method for early warning of a fractured sand plug provided in an embodiment of the present application.

As shown in fig. 1, the method for early warning of fractured sand plugging provided by this embodiment includes:

step 101: and obtaining the original data of one fracturing construction period in the fracturing construction process.

The fracturing construction process mentioned in the step refers to any fracturing construction process needing sand plugging early warning. For example, the construction site has N fracturing construction wells, namely a first construction well, a second construction well, … and an Nth construction well. The fracturing construction process corresponding to each construction well is respectively called a first fracturing construction process, a second fracturing construction process, …, an Nth fracturing construction process and the like. If sand blocking occurs during fracturing construction in the first construction well, the first fracturing construction process may be specifically referred to as a historical construction process in this embodiment, and the rest fracturing construction processes that do not occur sand blocking and are corresponding to the construction wells that continue fracturing construction may be respectively used as the fracturing construction processes mentioned in this step.

In practical application, the fracturing construction process usually lasts for tens of minutes or even several hours. Each fracturing job comprises a plurality of fracturing job cycles. The fracturing construction period can be set according to specific needs, for example, the fracturing construction period is set to be 20 minutes. Each fracture construction cycle may also include multiple acquisition times, for example, four acquisition times being t1, t2, t3, and t4, respectively. During the concrete implementation of the step, the parameters of the fracturing construction can be collected at each collecting moment of the fracturing construction period and used as the original data of the fracturing construction period.

For ease of understanding, the raw data for one fracture construction cycle is described below by way of example in connection with table 1.

TABLE 1 parameters collected at each collection time of the fracturing construction cycle

The initial data of a fracturing construction cycle is not limited to oil pressure, displacement, sand addition and water addition, and may further comprise other types of parameters. The type of parameter in the raw data is not limited herein.

As shown in table 1, the specific values of the parameters acquired at each acquisition time may be the same or different. For example, the displacement acquired at time t2 is the same as the displacement acquired at time t 3; the oil pressure collected at time t1 is different from the oil pressure collected at time t 2.

When the step is implemented specifically, the raw data can be obtained through some sensing devices. For example, oil pressure is collected for each fracture cycle using an oil pressure sensor.

Step 102: periodic features are extracted from the raw data.

The method comprises the step of extracting cycle characteristics corresponding to the fracturing construction cycle by using original data obtained in the previous step. The cycle characteristics reflect characteristics of the fracture construction operation during the fracture construction cycle. As an example, the periodic characteristics may include: maximum oil pressure, average sand addition, etc. Various types of periodic characteristics can be extracted according to actual requirements.

Extracting periodic features from raw data includes a number of possible implementations.

As an exemplary implementation, the periodic features may be extracted through machine learning. As another exemplary implementation, the periodic features may also be extracted by machine learning in combination with preset statistical parameters. The preset statistical parameters may include any one or more of the following:

maximum, minimum, variance, standard deviation, or mean, and the like.

Step 103: and matching the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process with the cycle reference characteristics, judging whether the matching is successful, and executing the step 104 if the matching is successful.

The fracturing construction process is a continuous process, the periodic characteristics corresponding to each adjacent fracturing construction period possibly have correlation, and the periodic characteristics corresponding to each fracturing construction period can reflect the trend of whether sand blocking is about to occur or not. Therefore, in order to accurately warn sand blockage, in the step, the periodic characteristics corresponding to one fracturing construction period are not matched with the periodic reference characteristics, but the periodic characteristics corresponding to each fracturing construction period in the fracturing construction process are taken as a whole, and the whole is matched with the periodic reference characteristics.

The periodic reference features are extracted from original data before sand blockage occurs in the historical construction process in advance. For example, the historical construction process comprises a plurality of construction cycles, sand blockage occurs in the Y-th construction cycle (Y is an integer larger than 1), the original data of the Y-1 construction cycle can be obtained, and the cycle reference characteristics corresponding to the Y-1 construction cycle are extracted by utilizing the original data; or acquiring original data (Y is more than X, and X is an integer) from the Y-X construction period to the Y-1 construction period, and extracting cycle reference characteristics respectively corresponding to the Y-X construction period to the Y-1 construction period by using the original data.

If the periodic reference feature corresponds to one construction period in the historical construction process, taking the periodic feature corresponding to each fracturing construction period in the fracturing construction process as a whole, and matching the whole with the periodic reference feature; if the periodic reference characteristics correspond to a plurality of construction periods in the historical construction process, the step specifically comprises the steps of taking the periodic characteristics corresponding to each fracturing construction period in the fracturing construction process as a first whole, taking the periodic reference characteristics corresponding to the plurality of construction periods as a second whole, and matching the two whole.

It can be understood that the periodic reference features reflect the features of fracturing construction before sand blockage occurs in the historical construction process, and therefore when matching is successful, the periodic reference features indicate that the fracturing construction process has the features matched with the fracturing construction process before sand blockage occurs in the historical construction process, and sand blockage risks exist. And when the matching is unsuccessful, the fracture construction process does not have the characteristics matched with the characteristics before sand blockage occurs in the historical construction process, the sand blockage risk does not exist, and the original data needs to be continuously acquired and analyzed, and then further judgment is carried out.

Step 104: and when the matching is successful, performing fracturing sand blocking early warning.

The early warning of the fracturing sand blockage comprises a plurality of possible implementation modes, such as controlling an alarm on a construction site to sound so as to prompt that sand blockage is about to occur; and a prompt lamp on a construction site can be controlled to flicker to prompt that sand blockage is about to occur. The warning method is not particularly limited.

The method for early warning of the fracturing sand blocking is provided by the embodiment of the application. The method extracts cycle characteristics by acquiring the original data of each fracturing construction cycle in the fracturing construction process. Because the historical construction process is a construction process with sand blockage, and the cycle reference characteristics extracted from the original data before sand blockage occurs in the historical construction process reflect the characteristics of fracturing construction before sand blockage occurs, the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are matched with the cycle reference characteristics, and when the matching is successful, the sand blockage occurring in the fracturing construction process can be predicted. The early warning method provided by the application can effectively judge that the sand blocking problem is about to occur before sand blocking occurs in the fracturing construction process. Therefore, when the matching is successful, the early warning of the fracturing sand plugging is carried out, so that constructors can conveniently take corresponding measures to avoid the sand plugging, and the adverse effect caused by the sand plugging problem is avoided or reduced.

In the embodiment of the application, a self-encoder can be applied to extract the cycle characteristics corresponding to each fracturing construction cycle. Before describing in detail the implementation of step 102 using an autoencoder, the training process of the autoencoder is first introduced.

Referring to fig. 2, a flowchart of a method for training an auto-encoder according to an embodiment of the present application is shown.

As shown in fig. 2, training the self-encoder for periodic feature extraction includes:

step 201: a plurality of pieces of training data are acquired.

The present embodiment trains an auto-encoder that can adapt to input data of different lengths.

In practical applications, the length of the raw data of one fracturing construction cycle obtained in step 101 is related to the type number of the parameters and the acquisition time, so that the length of the acquired raw data may be non-fixed. Taking table 1 as an example, since the parameters in the raw data of the fracturing construction cycle are 4 types and there are 4 acquisition times, the length of the raw data shown in table 1 is 16. It can be understood that if the acquisition time is changed to 3, the length of the original data becomes 12; if the parameters are of 5 types in total, the length of the original data becomes 20, and so on.

After the self-encoder is trained, the raw data obtained in step 101 is used as the input data of the self-encoder. In order to adapt the trained self-encoder to input data of different lengths and prevent the application of erroneous features extracted from the encoder, a plurality of pieces of training data are acquired in this step 201 and the acquired training data are required to include at least two lengths. As an example, the training data has 200 pieces, wherein the length of 100 pieces of training data is 20, the length of 50 pieces of training data is 12, and the length of 50 pieces of training data is 16.

Each piece of training data comprises raw data of one fracturing construction period in a historical construction process. Here, the raw data includes at least the following parameters: oil pressure, discharge capacity, sand addition amount and water addition amount. The form of a piece of training data can be referred to table 1.

Step 202: and training the self-encoder to be trained by utilizing a plurality of pieces of training data.

An auto-encoder can generally provide a recurring function, the output and input of which are very consistent. When training the self-Encoder to be trained, the coefficients of the encoding structure Encoder and the decoding structure Decoder in the self-Encoder are adjusted according to the difference between the output and the input in each iteration.

In a specific implementation, a loss function may be set, which is a function on the input and output from the encoder, which may be simply expressed as loss. The difference between the input and output of the self-encoder to be trained is positively correlated with the value of the penalty function. It will be appreciated that the smaller the value of the loss function, the smaller the difference between the output and the input, and the closer the training from the encoder is to the ideal.

In particular implementations, the self-encoder to be trained may also be trained using a gradient descent optimization method. As an example, the gradient descent optimization method may be any one of:

random Gradient descent (SGD), root mean square transfer algorithm (RMSprop), or adaptive moment estimation algorithm (Adam).

Step 203: judging whether any one of the first condition and the second condition is satisfied, and executing step 204 when any one of the conditions is satisfied; and when the first condition and the second condition are not met, returning to execute the step 202.

The first condition is that a value of a loss function of the self-encoder is less than a first preset threshold. When the first condition is satisfied, the value of the loss function is small enough, that is, the difference between the output and the input of the self-encoder is small enough, and at this time, the self-encoder satisfies the working requirement, and the training may be stopped, i.e., step 204 is performed.

And the second condition is that the iteration number of training the self-encoder to be trained by using a gradient descent optimization method reaches a second preset threshold value. When the second condition is met, which indicates that the number of iterations is sufficient, the optimization effect of the gradient descent optimization method can enable the self-encoder to meet the working requirement, and the training can be stopped, i.e. the step 204 is entered.

Step 204: a trained self-encoder is obtained.

In practical application, the trained self-encoder obtained in step 204 can be directly applied in step 102 for extracting the periodic features; in addition, test data can be divided from a plurality of pieces of training data, the trained self-encoder is tested by using the test data, coefficients of an encoding structure and a decoding structure are further optimized and adjusted, and the self-encoder obtained after testing is applied to the step 102 for extracting the periodic characteristics.

Alternative implementations of step 102 are described and illustrated, respectively.

An alternative implementation of extracting periodic features from raw data will first be described with reference to fig. 3. This implementation mainly extracts periodic features through machine learning.

Referring to fig. 3, a diagram of a neural network architecture for a self-encoder is shown.

As shown in fig. 3, the self-encoder includes an input layer including a plurality of input nodes a1, an output layer including a plurality of output nodes a2, and a hidden layer. Based on the above description of the trained self-encoder, the trained self-encoder can adapt to inputs with different lengths, and the number of input nodes a1 in the self-encoder is variable. The number of input nodes a1 corresponds to the number of output nodes a 2.

The hidden layer includes a plurality of intermediate nodes a 3. Each input node a1 is associated with a respective intermediate node a 3; each intermediate node a3 is associated with a respective output node a 2.

In this implementation manner, the parameters included in the raw data at each acquisition time of the fracturing construction cycle obtained in step 101 may be provided to the input node a1 of the self-encoder, the parameters included in the raw data at each acquisition time of the fracturing construction cycle may be provided to the output node a2 of the self-encoder, and the result of the intermediate node a3 of the self-encoder is extracted as the cycle characteristic corresponding to the fracturing construction cycle.

Taking table 1 as an example, 16 data shown in table 1 are provided to the input node a1, respectively, and 16 data shown in table 1 are provided to the output node a2, respectively. The data of the middle node a3 of the trained self-encoder can reflect the period characteristics corresponding to the fracturing construction period obtained after the self-encoder performs machine learning by using the 16 data.

In practical applications, the self-encoder may learn the functional relationship between the data of the input node a1 and the data of the intermediate node a3 in a machine learning manner and also learn the functional relationship between the data of the intermediate node a3 and the data of the output node a2 by using the data of the input node a1 and the output node a 2. In this implementation, the result of the intermediate node a2 that meets the above functional relationship is used as the cycle characteristic corresponding to the fracturing construction cycle.

This implementation extracts periodic features through machine learning. Because the self-encoder is trained in advance before the periodic characteristics are extracted, when the periodic characteristics need to be extracted, the self-encoder can be used for extracting the periodic characteristics quickly and efficiently, and the early warning speed of the fracturing sand plugging early warning method provided by the embodiment on sand plugging can be promoted. In addition, the self-encoder can learn the characteristics which are difficult to capture through the universal statistical parameters, so the obtained periodic characteristics can more comprehensively reflect the characteristics of the fracturing construction.

An alternative implementation of extracting periodic features from raw data is described next with reference to fig. 4. The implementation mode mainly extracts the periodic characteristics through machine learning and combining preset statistical parameters.

Referring to fig. 4, a flow diagram of another implementation of extracting periodic features from raw data is shown.

Step 401: and providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of the self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a first type of characteristic corresponding to the fracturing construction period.

The detailed implementation of step 401 will not be described herein. The difference from the foregoing implementation is that in the foregoing implementation, the result of the intermediate node of the self-encoder is directly used as the period feature, whereas in the present implementation, the result of the intermediate node of the self-encoder is used as the first type of feature corresponding to the fracture construction period.

Step 402: obtaining a second type of characteristics by utilizing parameters included in the original data of each acquisition time of the fracturing construction period, wherein the second type of characteristics at least comprise: variance, mean and maximum values for each parameter.

Taking the parameters shown in table 1 as an example, the second type of characteristics at least includes: oil pressure variance, oil pressure average value, oil pressure maximum value, displacement variance, displacement maximum value, displacement average value, sand addition variance, sand addition maximum value, sand addition average value, water addition variance, water addition maximum value and water addition average value.

For ease of understanding, the second category of characteristics of a fracture construction cycle is shown below by table 2. The second category of features shown in table 2 was obtained based on the raw data shown in table 1.

TABLE 2 second class of characteristics of fracture construction cycle

Statistical parameter	Oil pressure	Discharge capacity	Amount of sand added	Amount of added water
					Variance (variance)	0.38	2	0.07	0.19
Maximum value	94.2	24	3.5	2.2
					Mean value of	93.37	22	3.2	1.5

The present embodiment does not limit the execution order of

steps

401 and 402 in this implementation manner of step 102. For example, step 401 may be executed first and then step 402 may be executed, step 402 may be executed first and then step 401 may be executed, or

steps

401 and 402 may be executed at the same time. In fig. 4, the specific operations of the implementation are shown only in the execution sequence of executing step 401 first and then executing step 402 as an example.

Step 403: and obtaining the periodic characteristics according to the first type of characteristics and the second type of characteristics.

In the specific implementation of this step, the first type of features and the second type of features may be spliced, and the result after splicing is taken as a periodic feature. For example, the first class of features is K1, K2, …, K10; the second type of characteristics are L1, L2, … and L12, and a data set { K1, K2, …, K10, L1, L2, … and L12} which represents the cycle characteristics of the fracturing construction cycle can be obtained by splicing the two types of characteristics.

The implementation mode extracts the periodic characteristics through machine learning and by combining preset statistical parameters. The self-encoder can be used for rapidly and efficiently extracting the first type of characteristics, and further is beneficial to improving the early warning speed of the fracturing sand plugging early warning method provided by the embodiment on sand plugging. The second type of characteristics are captured through the preset statistical parameters and supplement the first type of characteristics obtained by the self-encoder, and finally, the periodic characteristics obtained by the implementation mode are more comprehensive, so that the accuracy of sand blockage early warning is improved.

The above provides two implementation manners for extracting the periodic features from the original data. In practical applications, the method for extracting the periodic features is not limited to the above two methods.

In practical applications, in the method for early warning of sand plugging in fracturing provided in the foregoing embodiment, the step 103 includes multiple possible implementation manners. The following provides another two fracturing sand plugging early warning methods by combining the embodiment and the drawings, and specifically describes a matching mode of the periodic characteristic and the periodic reference characteristic.

Referring to fig. 5, the figure is a flowchart of another method for early warning of sand plugging in fracturing provided by the embodiment of the present application.

As shown in fig. 5, the method for early warning of fractured sand plugging provided by this embodiment includes:

the following step 501-504 describes a specific implementation manner from obtaining the original data corresponding to each construction period before sand blockage occurs in the historical construction process to determining the feature vector matching range. In practical applications, steps 501-504 may be completed before matching. That is, it is not necessary to perform a round of steps 501-504 each time the method of the present embodiment is performed, and only a round of determining the matching range of the good feature vector is performed.

Step 501: and acquiring original data corresponding to each construction period before sand blockage occurs in the historical construction process, and extracting historical characteristics from the original data.

The implementation of extracting the historical features may refer to the related description of step 102 described above. As an exemplary implementation, the historical features may be extracted through machine learning. As another exemplary implementation, the historical features may also be extracted by machine learning in combination with preset statistical parameters.

Step 502: and constructing a second characteristic vector corresponding to each construction period before sand blockage occurs in the historical construction process by using the historical characteristics.

That is, one construction period corresponds to one second feature vector. The dimension of the second feature vector is related to the number of the historical features of the construction period. For example, if step 501 extracts two historical features from the original data corresponding to one construction period, the second feature vector corresponding to the construction period is a two-dimensional vector; if step 501 extracts three historical features from the original data corresponding to one construction period, the second feature vector corresponding to the construction period is a three-dimensional vector. Each dimension of the second feature vector corresponds to a historical feature.

Step 503: and clustering each second feature vector, and determining the second feature vector of which the category is the category to be pre-warned from each second feature vector according to the clustering result.

In practical application, each second feature vector may be clustered according to the value of the historical feature corresponding to each dimension of the second feature vector.

As an example, the respective second feature vectors are divided into three classes, a first class, a second class and a third class, respectively.

In the embodiment, historical characteristics corresponding to the construction period of sand blockage in the historical construction process can be extracted, and a second characteristic vector corresponding to the construction period of sand blockage is constructed by using the historical characteristics. If the second eigenvector corresponding to the construction period in which the sand blockage occurs can be classified into the first class according to the clustering mode, and the second eigenvector corresponding to the construction period before the sand blockage occurs is classified into the third class, the third class can be used as the class to be pre-warned.

That is, all the second feature vectors classified into the third class are used as the second feature vectors of the class to be warned. The second type of feature vectors corresponding to the construction period before the sand blockage occurs are divided into a third type, and historical features included in the second type of feature vectors corresponding to the construction period before the sand blockage occurs are specifically periodic reference features, so that the second feature vectors of the to-be-early-warning type include the periodic reference features.

Step 504: and determining a feature vector matching range according to the second feature vector of the category to be pre-warned.

An alternative implementation of this step is described below.

The above example is still used. It is understood that after clustering, there may be a plurality of second feature vectors that are classified into a third category, i.e., the second feature vectors that are classified into the category to be warned. The second feature vectors of the category to be early-warned are scattered in the distribution space of the second feature vectors. The purpose of determining the feature vector matching range in this embodiment is to subsequently use the feature vector matching range as a basis for judging whether sand blockage early warning needs to be performed.

In order to determine the feature vector matching range, firstly, obtaining an average feature vector by using a second feature vector of a category to be pre-warned; then, obtaining the distance between each second feature vector of the category to be pre-warned and the average feature vector, and determining the maximum distance in the distance; and finally, determining a range with the average characteristic vector as an origin and the maximum distance as a radius as a characteristic vector matching range.

To facilitate understanding of the above operation, the following description is made by way of example.

Assuming that there are w second eigenvectors (w is an integer greater than 1) of the category to be warned, each second eigenvector is a three-dimensional vector and is represented as [ u1, k1, i1], [ u2, k2, i2], …, [ uw, kw, iw ]. The formula for obtaining the average feature vector [ u ', k ', i ' ] is as follows:

u' ═ u1+ u2+ … + uw)/w formula (1)

k' ═ k1+ k2+ … + kw)/w equation (2)

i' ═ i (i1+ i2+ … + iw)/w equation (3)

It can be understood that the second feature vector of each category to be warned is respectively distant from the average feature vector by d1, d2, …, and dw. In practical application, the distance can be calculated according to European geometric axiom. The maximum distance D is determined, where D ═ max { D1, D2, …, dw }.

And finally, taking the average eigenvector [ u ', k ', i ' ] as an origin and D as a radius, so as to determine the spherical eigenvector matching range.

From the above steps 501-504, the feature vector matching range is actually determined according to the periodic reference feature.

The implementation manners of steps 505-506 in this embodiment are substantially the same as the implementation manners of steps 101-102 in the foregoing embodiments, and therefore, the descriptions of steps 505-506 may refer to the foregoing embodiments, which are not repeated herein.

Step 507: and constructing a first feature vector corresponding to each construction period in the fracturing construction process according to the period features corresponding to each construction period in the fracturing construction process.

In this step, the periodic characteristics refer to the periodic characteristics obtained by performing step 506.

One fracturing construction period corresponds to one first characteristic vector. The dimensionality of the first feature vector is related to the number of the periodic features corresponding to the fracturing construction period. For example, if step 506 extracts two period features from the original data corresponding to one fracturing construction period, the first feature vector corresponding to the fracturing construction period is a two-dimensional vector; if step 506 extracts three period features from the original data corresponding to one fracturing construction period, the first feature vector corresponding to the fracturing construction period is a three-dimensional vector. Each dimension of the first feature vector corresponds to a periodic feature.

Step 508: and judging whether a first feature vector falling into the feature vector matching range exists or not, if so, determining that the matching of the cycle features corresponding to each fracturing construction cycle and the cycle reference features is successful in the fracturing construction process, and executing a step 509.

Because the feature vector matching range is determined according to the periodic reference features, the feature vector matching range has correlation with the periodic reference features before sand blockage occurs in the historical construction process. In addition, the feature vector matching range is also determined according to each second feature vector classified as the category to be pre-warned, so that the feature vector matching range reflects the commonality of the second feature vectors of the category to be pre-warned. Through the analysis, if the first feature vector falls into the feature vector matching range, the first feature vector is matched with the periodic reference feature, and the first feature vector and the second feature vector of the category to be pre-warned have the commonalities. Therefore, it can be judged that sand clogging is likely to occur thereafter.

Step 509: and performing early warning on fracturing sand plugging.

The method for early warning of fracturing sand plugging provided by the embodiment utilizes the original data corresponding to each construction period before sand plugging occurs in the historical construction process, and gradually extracts the historical characteristics, constructs the second characteristic vector, clusters the second characteristic vector and determines the characteristic vector matching range. And a first characteristic vector is constructed by utilizing the periodic characteristics corresponding to the fracturing construction period in the fracturing construction process, and when the first characteristic vector falls into the characteristic vector matching range, the matching is judged to be successful and sand blockage needs to be warned. The method is characterized in that the commonality between the fracturing construction process and the historical construction process before sand blockage occurs is excavated by constructing vectors and determining the matching range of characteristic vectors, so that the trend of the sand blockage occurrence is judged, and early warning of the sand blockage is made in time. Because the feature vector matching range can be determined in advance, when the fracturing construction process needs to be judged whether sand blockage will occur or not, only the first feature vector needs to be constructed, and the first feature vector is matched with the feature vector matching range. The method is convenient and quick and can be automatically realized, so that the timeliness and effectiveness of sand blockage early warning are correspondingly improved.

Referring to fig. 6, this figure is a flowchart of another method for early warning of a fractured sand plug according to the embodiment of the present application.

As shown in fig. 6, the method for early warning of fractured sand plugging provided by this embodiment includes:

the following step 601-604 describes a specific implementation manner from obtaining the original data corresponding to each construction period before sand blockage occurs in the historical construction process to the historical construction track. In practical applications, step 601-604 may be completed before matching. That is, it is not necessary to execute step 601 and step 604 each time the method of the present embodiment is executed, and only one round of obtaining the historical construction track is executed.

Step 601: and acquiring original data corresponding to each construction period before sand blockage occurs in the historical construction process, and extracting historical characteristics from the original data.

The implementation manner of step 601 is the same as that of step 501 in the foregoing embodiment, and therefore details of the implementation of step 601 are not described herein.

Step 602: and obtaining a second node corresponding to each construction period before sand blockage occurs in the historical construction process by using the historical characteristics.

In concrete implementation, one construction period corresponds to one second node. The dimension of the space where the second node is located is related to the number of the historical features of the construction period. For example, if step 601 extracts two historical features from the raw data corresponding to one construction period, the second node corresponding to the construction period is located in the two-dimensional space; if step 601 extracts three historical features from the raw data corresponding to one construction period, the second node corresponding to the construction period is located in the three-dimensional space. Each dimension of the space in which the second node is located corresponds to a history feature.

Step 603: and determining the second nodes to be early-warned from the second nodes.

In practical application, the second node to be early-warned can be determined according to the construction period of sand blockage in the historical construction process. For example, if the construction period in which sand blockage occurs is the Y-th construction period of the historical construction process, the second nodes corresponding to the preset number of construction periods before the Y-th construction period may be used as the second nodes to be pre-warned. For example, the preset number is X, Y is larger than X, X is an integer, and the second nodes corresponding to the Y-X to the Y-1 construction periods are used as the second nodes to be pre-warned. Thus, X second nodes are determined.

According to the description of the implementation manner for extracting the periodic reference feature in the foregoing embodiment, it can be known that the historical feature included in the second node corresponding to the construction period before the sand blockage occurs is the periodic reference feature, and therefore, the second node to be early-warned includes the periodic reference feature. As an example, the coordinate of the second node corresponding to a certain construction period in the space is represented as J₁(b₁，c₁，e₁) Wherein b is₁，c₁And e₁And respectively referring the extracted three cycle reference characteristics of the construction cycle.

Step 604: and connecting the second nodes to be early-warned end to end according to the time sequence to obtain the historical construction track.

Still continuing with the foregoing example, the second nodes corresponding to the Y-X to Y-1 construction periods respectively are denoted as J as the second nodes to be pre-warned_Y-X(b_Y-X，c_Y-X，e_Y-X)，J_Y-X-1(b_Y-X-1，c_Y-X-1，e_Y-X-1)，…，J_Y-2(b_Y-2，c_Y-2，e_Y-2)，J_Y-1(b_Y-1，c_Y-1，e_Y-1). And connecting the second nodes to be early-warned end to end according to the time sequence, namely connecting the second nodes to be early-warned end to end according to the occurrence sequence of the construction period corresponding to each second node to be early-warned. That is, the starting point of the historical construction path is J_Y-XA mixture of J and_Y-Xand J_Y-X-1Joining, …, J_Y-2And J_Y-1Is connected with J_Y-1As the end point of the historical construction track.

The implementation manner of steps 605-606 is substantially the same as that of steps 101-102 in the foregoing embodiment, and therefore, the description of steps 605-606 can refer to the foregoing embodiment and will not be repeated herein.

Step 607: and obtaining each first node by utilizing the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process, and connecting the first nodes end to end according to the time sequence to obtain a fracturing construction track.

In this step, the periodic characteristics refer to the periodic characteristics obtained by performing step 606.

One fracturing construction cycle corresponds to one first node. The dimension of the space where the first node is located is related to the number of the periodic characteristics of the fracturing construction period. For example, if step 606 extracts two period features from the raw data corresponding to one construction period, the first node corresponding to the fracture construction period is located in the two-dimensional space; if step 606 extracts three cycle features from the raw data corresponding to one fracture construction cycle, the first node corresponding to the fracture construction cycle is located in the three-dimensional space. Each dimension of the space in which the first node is located corresponds to a periodic feature.

Step 608: and comparing the similarity of the fracturing construction track with the similarity of the historical construction track, judging whether the similarity of the two tracks meets a preset condition, if so, determining that the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are successfully matched with the cycle reference characteristics, and executing step 609.

Since the historical construction track is formed according to the second node to be early-warned, and the second node to be early-warned includes the periodic reference feature, the historical construction track is obtained in advance according to the periodic reference feature in this embodiment. The historical construction track reflects the characteristics of fracturing construction before sand plugging occurs in the historical construction process, so that if the track similarity of the fracturing construction track and the historical construction track meets a preset condition (for example, a certain preset similarity threshold value is reached), the fracturing construction track can be determined to be sufficiently similar to the historical construction track, and the fracturing construction track also reflects the characteristics of fracturing construction before sand plugging occurs. And further showing that the fracturing construction track is matched with the periodic reference characteristics. Therefore, it can be judged that sand clogging is likely to occur thereafter.

In the concrete implementation of the step, the similarity of the two tracks can be obtained by utilizing a relatively mature track similarity calculation method. As an example, a Dynamic Time Warping (DTW) algorithm may be applied to calculate the similarity of two trajectories. If the similarity of the two obtained tracks is the first similarity, and the first similarity is greater than or equal to the similarity threshold, the matching can be determined to be successful, and early warning needs to be performed on sand blockage which may occur.

See fig. 7a and 7 b. Fig. 7a is a schematic diagram of a historical construction track, and fig. 7b is a schematic diagram of a fracture construction track.

In fig. 7a, the starting point of the historical construction track is F71, and the track is from F71, F72, F73, F74 to the end point F75. F76 is a second node corresponding to the construction period of sand blockage in the historical construction process.

In fig. 7b, the starting point of the fracturing construction track is F81, and the track is along F81, F82, F83, F84 to F85, where F85 is the first node corresponding to the latest fracturing construction cycle in the current fracturing construction process.

According to the fig. 7a and 7b, it can be easily found that the two tracks have similarity, and if the early warning method provided by the embodiment is applied to compare the similarity of the two tracks before the next fracturing construction period is finished, it is possible to realize early warning in time before sand blockage occurs.

Step 609: and performing early warning on fracturing sand plugging.

The method for early warning of fracturing sand plugging provided by the embodiment gradually obtains the second node, determines the second node to be early warned and obtains the historical construction track by using the original data corresponding to each construction period before sand plugging occurs in the historical construction process. And obtaining first nodes by utilizing the cycle characteristics corresponding to the fracturing construction cycle of the fracturing construction process, obtaining fracturing construction tracks by utilizing the plurality of first nodes, and determining whether sand blockage needs to be early warned according to the similarity of the historical construction tracks and the fracturing construction tracks. According to the method, the historical construction track and the track similarity comparison are established to find the commonness between the fracturing construction process and the historical construction process before sand blockage occurs, so that the trend of sand blockage occurrence is judged, and sand blockage early warning is made in time. Since the historical construction track can be obtained in advance, when the sand blockage in the fracturing construction process needs to be judged, the fracturing construction track of the fracturing construction process only needs to be formed, and the similarity comparison between the two tracks is carried out. The method is convenient and quick and can be automatically realized, so that the timeliness and effectiveness of sand blockage early warning are correspondingly improved.

Based on the method for early warning of fracturing sand plugging provided by the embodiment, correspondingly, the application further provides a device for early warning of fracturing sand plugging. The following description is made with reference to the embodiments and the accompanying drawings.

Device embodiment

Referring to fig. 8, the diagram is a schematic structural diagram of a fractured sand plugging early warning provided by the embodiment of the application.

As shown in fig. 8, the device for early warning of sand plugging in fracturing provided by this embodiment includes:

an original data acquisition module 801, configured to acquire original data of a fracturing construction cycle in a fracturing construction process; the raw data comprises at least the following parameters: oil pressure, discharge capacity, sand adding amount and water adding amount;

a periodic feature extraction module 802, configured to extract periodic features from the raw data;

the characteristic matching module 803 is used for matching the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process with the cycle reference characteristics; the periodic reference characteristic is extracted from original data before sand blockage occurs in the historical construction process in advance;

and the early warning module 804 is used for carrying out early warning on the fracturing sand plugging when the characteristic matching module is successfully matched.

The device of stifled early warning of fracturing sand that above provides for this application embodiment. The device extracts cycle characteristics by acquiring the original data of each fracturing construction cycle in the fracturing construction process. Because the historical construction process is a construction process with sand blockage, and the cycle reference characteristics extracted from the original data before sand blockage occurs in the historical construction process reflect the characteristics of fracturing construction before sand blockage occurs, the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process are matched with the cycle reference characteristics, and when the matching is successful, the sand blockage occurring in the fracturing construction process can be predicted. The application provides an early warning device can effectively judge out the stifled problem of sand and be about to take place before this fracturing work progress takes place sand stifled. Therefore, when the matching is successful, the early warning of the fracturing sand plugging is carried out, so that constructors can conveniently take corresponding measures to avoid the sand plugging, and the adverse effect caused by the sand plugging problem is avoided or reduced.

The periodic feature extraction module 802 includes a variety of implementations, which are described in detail below.

Optionally, the periodic feature extraction module 802 specifically includes:

and the first extraction unit is used for providing parameters included in the original data at each acquisition time of the fracturing construction cycle to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction cycle to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a cycle characteristic corresponding to the fracturing construction cycle.

Optionally, the periodic feature extraction module 802 specifically includes:

The implementation mode extracts the periodic characteristics through machine learning and combining with preset statistical parameters. The self-encoder can be used for rapidly and efficiently extracting the first type of characteristics, and further is beneficial to improving the early warning speed of the fracturing sand plugging early warning method provided by the embodiment on sand plugging. The second type of characteristics are captured through the preset statistical parameters and supplement the first type of characteristics obtained by the self-encoder, and finally, the periodic characteristics obtained by the implementation mode are more comprehensive, so that the accuracy of sand blockage early warning is improved.

The above provides two implementation manners for extracting the periodic features from the original data. In practical applications, the manner of extracting the periodic features is not limited to the above two manners.

Optionally, the apparatus further comprises:

the self-encoder training module is used for training a self-encoder to be trained by utilizing the plurality of pieces of training data, and obtaining the trained self-encoder when a first condition or a second condition is met; the first condition is that the value of the loss function of the self-encoder is smaller than a first preset threshold value, and the difference between the output and the input of the self-encoder to be trained is positively correlated with the value of the loss function; the second condition is that the iteration number of training the self-encoder to be trained by using a gradient descent optimization method reaches a second preset threshold value.

Optionally, the feature matching module 803 specifically includes:

the characteristic vector first construction unit is used for constructing a first characteristic vector corresponding to each construction period in the fracturing construction process according to the period characteristic corresponding to each fracturing construction period in the fracturing construction process;

Optionally, the apparatus further comprises:

a clustering unit, configured to cluster the second feature vectors;

Optionally, the matching range determining unit specifically includes:

The device provided by the above embodiment performs historical feature extraction, constructs the second feature vector, clusters the second feature vector, and determines the feature vector matching range by using the original data corresponding to each construction cycle before sand blockage occurs in the historical construction process. And a first characteristic vector is constructed by utilizing the periodic characteristics corresponding to the fracturing construction period in the fracturing construction process, and when the first characteristic vector falls into the characteristic vector matching range, the matching is judged to be successful and sand blockage needs to be warned. The device excavates the commonality before sand blockage occurs in the fracturing construction process and the historical construction process by constructing vectors and determining the matching range of the characteristic vectors, thereby judging the trend of the sand blockage and giving early warning of the sand blockage in time. Because the feature vector matching range can be determined in advance, when the fracturing construction process needs to be judged whether sand blockage will occur or not, only the first feature vector needs to be constructed, and the first feature vector is matched with the feature vector matching range. The device is convenient and fast and can be automatically realized, so that the timeliness and effectiveness of sand blockage early warning are correspondingly improved.

Optionally, the feature matching module 803 specifically includes:

Optionally, the apparatus further comprises:

the node acquisition unit is used for acquiring a second node corresponding to each construction period before sand blockage occurs in the historical construction process by using the historical characteristics;

the node determining unit is used for determining second nodes to be pre-warned from the second nodes; the second node to be early-warned comprises the periodic reference feature;

The fracturing sand plugging early warning device provided by the embodiment obtains the second node, determines the second node to be early warned and obtains the historical construction track by utilizing the original data corresponding to each construction period before sand plugging occurs in the historical construction process. And obtaining first nodes by utilizing the cycle characteristics corresponding to the fracturing construction cycle of the fracturing construction process, obtaining fracturing construction tracks by utilizing the plurality of first nodes, and determining whether sand blockage needs to be early warned according to the similarity of the historical construction tracks and the fracturing construction tracks. The device is used for excavating the commonness between the fracturing construction process and the historical construction process before sand blockage occurs by constructing historical construction tracks and track similarity comparison, so that the trend of sand blockage occurrence is judged, and sand blockage early warning is timely made. Since the historical construction track can be obtained in advance, when the sand blockage in the fracturing construction process needs to be judged, the fracturing construction track of the fracturing construction process only needs to be formed, and the similarity comparison between the two tracks is carried out. The device is convenient and fast and can be automatically realized, so that the timeliness and effectiveness of sand blockage early warning are correspondingly improved.

Based on the method and the device for early warning of fractured sand plugging provided by the embodiment, the embodiment of the application further provides a computer-readable storage medium.

The storage medium stores a program, and the program is executed by a processor to realize part or all of the steps of the method for early warning of fractured sand plugging, which is protected by the method embodiment.

The storage medium may be a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and various media capable of storing program codes.

Based on the method, the device and the storage medium for early warning of the fractured sand plugging provided by the embodiment, the embodiment of the application provides a processor. The processor is used for running a program, wherein the program runs to execute part or all of the steps of the method for early warning of the fractured sand plugging protected by the method embodiment.

Based on the storage medium and the processor provided by the foregoing embodiments, the application further provides a fracturing sand blocking early warning device.

Referring to fig. 9, the figure is a hardware structure diagram of the equipment for early warning of fractured sand plugging provided in this embodiment.

As shown in fig. 9, the equipment for early warning of fractured sand clogging includes: memory 901, processor 902, communication bus 903, and communication interface 904.

The memory 901 stores a program capable of running on the processor, and when the program is executed, part or all of the steps of the method for early warning of fractured sand plugging provided by the foregoing method embodiments of the present application are implemented. The memory 901 may include high-speed random access memory and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device.

In this device, the processor 902 and the memory 901 transmit signaling, logic instructions, and the like through a communication bus. The device is capable of communicative interaction with other devices via the communication interface 904.

By executing the method through the program, the problem of sand blocking can be effectively judged to be about to occur before sand blocking occurs in the fracturing construction process. Therefore, when the matching is successful, the early warning of the fracturing sand plugging is carried out, so that constructors can conveniently take corresponding measures to avoid the sand plugging, and the adverse effect caused by the sand plugging problem is avoided or reduced.

It should be noted that, in the present specification, all the embodiments are described in a progressive manner, and the same and similar parts among the embodiments may be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus and system embodiments, since they are substantially similar to the method embodiments, they are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for related points. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts suggested as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

The above description is only one specific embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present application should be covered within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for early warning of fracturing sand blocking is characterized by comprising the following steps:

extracting periodic features from the raw data;

matching the cycle characteristics corresponding to each fracturing construction cycle in the fracturing construction process with the cycle reference characteristics, and performing fracturing sand plugging early warning when the matching is successful; the periodic reference characteristic is extracted from original data before sand blockage occurs in the historical construction process in advance;

the extracting of the periodic features from the raw data specifically includes:

providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a period characteristic corresponding to the fracturing construction period;

or, the extracting periodic features from the raw data specifically includes:

obtaining the periodic characteristics according to the first type of characteristics and the second type of characteristics;

wherein the self-encoder is used for feature extraction and has a three-layer structure.

2. The method of claim 1, further comprising:

3. The method according to claim 1, wherein the matching of the cycle characteristics and the cycle reference characteristics corresponding to each fracturing construction cycle in the fracturing construction process specifically comprises:

4. The method of claim 3, further comprising, prior to determining whether the first feature vector that falls within a feature vector match range exists:

5. The method according to claim 4, wherein the determining the feature vector matching range according to the second feature vector of the category to be early-warned specifically comprises:

6. The method according to claim 1, wherein the matching of the cycle characteristics and the cycle reference characteristics corresponding to each fracturing construction cycle in the fracturing construction process specifically comprises:

7. The method of claim 6, further comprising:

8. The utility model provides a device of stifled early warning of fracturing sand which characterized in that includes:

the periodic feature extraction module specifically comprises:

the first extraction unit is used for providing parameters included in the original data at each acquisition time of the fracturing construction period to an input node of a self-encoder, providing parameters included in the original data at each acquisition time of the fracturing construction period to an output node of the self-encoder, and extracting a result of an intermediate node of the self-encoder to serve as a period characteristic corresponding to the fracturing construction period;

or, the periodic feature extraction module specifically includes:

the fourth extraction unit is used for obtaining the periodic characteristics according to the first type of characteristics and the second type of characteristics;

wherein, the self-encoder is used for feature extraction and has a three-layer structure; and the early warning module is used for carrying out fracturing sand blocking early warning when the characteristic matching module is successfully matched.

9. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the method of frac sand blockage warning as claimed in any one of claims 1-7.

10. A processor for running a computer program which when run performs the method of frac sand blockage warning according to any one of claims 1 to 7.