CN116363137B - Cleaning effect evaluation method and system for guiding automatic cleaning of oil pipe - Google Patents

Abstract

The invention discloses a cleaning effect evaluation method and a system for guiding automatic cleaning of an oil pipe, and relates to the technical field of data processing, wherein the method comprises the following steps: acquiring automatic cleaning information of a target oil pipe set; collecting an in-pipe picture of the target oil pipe collection by using a picture collecting module; obtaining a dirt residual type and a pipe surface damage type; obtaining M dirt feature sets based on a dirt feature analysis model; m pipe surface damage characteristic sets are obtained based on the pipe surface damage characteristic analysis model, the M pipe surface damage characteristic sets are combined and input into the cleaning effect evaluation model for evaluation, the M cleaning effect evaluation results are weighted in stages, and the target oil pipe automatic cleaning effect evaluation result is obtained according to the processing result. The invention solves the technical problems of low intelligent degree and poor reliability of the evaluation result of the oil pipe cleaning effect evaluation in the prior art, and achieves the technical effects of improving the evaluation efficiency and the evaluation quality.

Description

Cleaning effect evaluation method and system for guiding automatic cleaning of oil pipe

Technical Field

The invention relates to the technical field of data processing, in particular to a cleaning effect evaluation method and system for guiding automatic cleaning of an oil pipe.

Background

In the process of petroleum transportation by using a petroleum transportation pipeline, impurities in crude oil can be deposited on a pipe wall after long-time use, so that the transportation efficiency is affected. Therefore, the oil pipe needs to be cleaned, and with the use of an automatic cleaning technology, the workload of cleaning the oil pipe is greatly reduced, and the cleaning speed is improved. However, the cleaning effect of the oil pipe after automatic cleaning cannot be reliably evaluated, the existing manual sampling inspection mode is low in effect evaluation efficiency, the requirement cannot be met, the evaluation result is limited by the capability of an inspector, and reliable guidance cannot be provided for automatic cleaning of the oil pipe.

Disclosure of Invention

The application provides a cleaning effect evaluation method and system for guiding automatic cleaning of an oil pipe, which are used for solving the technical problems of low intelligent degree of oil pipe cleaning effect evaluation and poor reliability of an evaluation result in the prior art.

In view of the above problems, the present application provides a cleaning effect evaluation method and system for guiding automatic cleaning of oil pipes.

In a first aspect of the present application, a cleaning effect evaluation method for guiding automatic cleaning of an oil pipe is provided, where the method is applied to an automatic cleaning guiding system of an oil pipe, and the automatic cleaning guiding system of an oil pipe is communicatively connected to a picture acquisition module, and the method includes:

Acquiring automatic cleaning information of a target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

taking the automatic cleaning track as a picture collecting route, and collecting pictures in pipes of the target oil pipe set by using a picture collecting module to obtain M oil pipe picture sets, wherein M is the number of oil pipes in the target oil pipe set and is an integer greater than or equal to 1, and the M oil pipe picture sets are provided with cleaning sequence number identifiers;

inputting the automatic cleaning mode into a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, and obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type;

constructing a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a SLOWFAST model, capturing pictures of M oil tube inner picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube inner picture sets according to a first inner picture capturing step length to obtain M first oil tube background picture sets, and respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model to obtain M dirt feature sets;

Constructing a tube surface damage capturing channel, a tube surface capturing channel and a feature analysis layer of a tube surface damage feature analysis model based on the SLOWFAST model, capturing pictures of M tube inner picture sets according to a second capturing step length to obtain M tube surface damage capturing picture sets, capturing pictures of M tube inner picture sets according to the second tube inner picture capturing step length to obtain M second tube background picture sets, and respectively inputting the M tube surface damage capturing picture sets and the M second tube background picture sets into a dirt capturing channel and a tube surface capturing channel of the tube surface damage feature analysis model to obtain M tube surface damage feature sets;

inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, and obtaining M cleaning effect evaluation results;

and carrying out staged weighting treatment on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the treatment result.

In a second aspect of the present application, there is provided a cleaning effect evaluation system for guiding automatic cleaning of an oil pipe, the system comprising:

The automatic cleaning information acquisition module is used for acquiring automatic cleaning information of the target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

the oil pipe in-pipe picture obtaining module is used for collecting pictures in pipes of the target oil pipe set by taking the automatic cleaning track as a picture collecting route and obtaining M oil pipe in-pipe picture sets, wherein M is the number of oil pipes in the target oil pipe set and is an integer greater than or equal to 1, and the M oil pipe in-pipe picture sets are provided with cleaning sequence number identifiers;

the pipe surface damage type obtaining module is used for inputting the automatic cleaning mode into the cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

the first capturing step length obtaining module is used for obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, and obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type;

The dirt feature set obtaining module is used for constructing a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a SLOWFAST model, capturing pictures of M oil tube in-tube picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube in-tube picture sets according to a first in-tube picture capturing step length to obtain M first oil tube background picture sets, and respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model to obtain M dirt feature sets;

the pipe surface damage characteristic obtaining module is used for constructing a pipe surface damage capturing channel, a pipe surface capturing channel and a characteristic analysis layer of a pipe surface damage characteristic analysis model based on the SLOWFAST model, capturing pictures of M pipe inner picture sets according to a second capturing step length to obtain M pipe surface damage capturing picture sets, capturing pictures of M pipe inner picture sets according to a second pipe inner picture capturing step length to obtain M second pipe background picture sets, and respectively inputting the M pipe surface damage capturing picture sets and the M second pipe background picture sets into a dirt capturing channel and a pipe surface capturing channel of the pipe surface damage characteristic analysis model to obtain M pipe surface damage characteristic sets;

The evaluation result obtaining module is used for inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, so as to obtain M cleaning effect evaluation results;

the target evaluation result obtaining module is used for carrying out staged weighting processing on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the processing result.

One or more technical solutions provided in the present application have at least the following technical effects or advantages:

the method comprises acquiring automatic cleaning information of a target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode, then taking the automatic cleaning track as a picture acquisition route, carrying out picture acquisition on the inside of a pipe of the target oil pipe set by using a picture acquisition module to obtain M oil pipe picture sets, wherein M is the number of the oil pipes in the target oil pipe set and is an integer greater than or equal to 1, the M oil pipe picture sets are provided with cleaning sequence number identifiers, inputting the automatic cleaning mode into a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type, then obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type, constructing a dirt capturing channel, a tube surface capturing channel and a characteristic analysis layer of a dirt characteristic analysis model based on a SLOWFAST model, capturing pictures of M oil tube inner picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube inner picture sets according to the first inner picture capturing step length to obtain M first oil tube background picture sets, respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt characteristic analysis model to obtain M dirt characteristic sets, further constructing a tube surface damage capturing channel, the tube surface capturing channel and the characteristic analysis layer of a tube surface damage characteristic analysis model based on the SLOWFAST model, capturing pictures of the M oil tube inner picture sets according to a second capturing step length to obtain M tube surface damage capturing picture sets, and carrying out picture capturing on M oil pipe inner picture sets according to the second in-pipe picture capturing step length to obtain M second oil pipe background picture sets, respectively inputting the M pipe surface damage capturing picture sets and the M second oil pipe background picture sets into a dirt capturing channel and a pipe surface capturing channel of a pipe surface damage characteristic analysis model to obtain M pipe surface damage characteristic sets, inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model to evaluate, obtaining M cleaning effect evaluation results, carrying out staged weighting treatment on the M cleaning effect evaluation results based on cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the treatment result. The technical effect of improving the reliability of the evaluation result of the automatic cleaning effect and the evaluation efficiency is achieved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic flow chart of a cleaning effect evaluation method for guiding automatic cleaning of an oil pipe according to an embodiment of the present application;

fig. 2 is a schematic flow chart of obtaining a first capturing step and a first in-pipe picture capturing step in a cleaning effect evaluation method for guiding automatic cleaning of an oil pipe according to an embodiment of the present application;

fig. 3 is a schematic flow chart of obtaining an evaluation result of the automatic cleaning effect of the target oil pipe in the cleaning effect evaluation method for guiding the automatic cleaning of the oil pipe according to the embodiment of the present application;

fig. 4 is a schematic structural diagram of a cleaning effect evaluation system for guiding automatic cleaning of an oil pipe according to an embodiment of the present application.

Reference numerals illustrate: the device comprises an automatic cleaning information obtaining module 11, an oil pipe internal picture obtaining module 12, a pipe surface damage type obtaining module 13, a first capturing step length obtaining module 14, a dirt characteristic set obtaining module 15, a pipe surface damage characteristic obtaining module 16, an evaluation result obtaining module 17 and a target evaluation result obtaining module 18.

Detailed Description

The application provides a cleaning effect evaluation method and a cleaning effect evaluation system for guiding automatic cleaning of an oil pipe, which are used for solving the technical problems of low intelligent degree of oil pipe cleaning effect evaluation and poor reliability of an evaluation result in the prior art.

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present application based on the embodiments herein.

It should be noted that the terms "comprises" and "comprising," along with any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus, but may include other steps or modules not expressly listed or inherent to such process, method, article, or apparatus.

Example 1

As shown in fig. 1, the present application provides a cleaning effect evaluation method for guiding automatic cleaning of an oil pipe, wherein the method is applied to an automatic cleaning guiding system of the oil pipe, and the automatic cleaning guiding system of the oil pipe is in communication connection with a picture acquisition module, and the method includes:

step S100: acquiring automatic cleaning information of a target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

step S200: taking the automatic cleaning track as a picture collecting route, and collecting pictures in pipes of the target oil pipe set by using a picture collecting module to obtain M oil pipe picture sets, wherein M is the number of oil pipes in the target oil pipe set and is an integer greater than or equal to 1, and the M oil pipe picture sets are provided with cleaning sequence number identifiers;

in one possible embodiment, the image acquisition module is a functional module for acquiring images of the inner wall of the oil pipe, and is formed by an image acquisition device, preferably, the image acquisition device can be a camera, an infrared camera or the like. The target oil pipe set is any oil pipe which is automatically cleaned and is required to evaluate the cleaning effect of the inner wall of the pipe. And acquiring the automatic cleaning information of the target oil pipe set by calling the automatic cleaning record of the target oil pipe set. Wherein, the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode. The automatic cleaning track is a route for automatically cleaning all the oil pipes in the target oil pipe set according to an automatic cleaning scheme, and in the actual cleaning process, the distance of the automatic cleaning oil pipes is limited and is generally 10 km, so that the automatic cleaning is carried out in stages according to the transmission direction of the pipeline. The automatic cleaning mode is an automatic cleaning mode for gathering and using the target oil pipe and comprises the modes of mechanical cleaning, physical cleaning, chemical cleaning and the like. The automatic cleaning information is acquired, so that an analysis direction is provided for analyzing the cleaning effect.

In one embodiment, the cleaning route in the automatic cleaning track is used as a picture collecting route, and picture collection is performed on the inner wall of the pipeline of the target oil pipe set by using the picture collecting module according to the positions of the oil pipes in the route, so that M oil pipe in-pipe picture sets are obtained. The M oil pipe inner picture sets are pictures for visually representing the appearance states of the inner walls of the pipelines in the target oil pipe sets, each oil pipe picture set corresponds to the corresponding target oil pipe one by one, and the cleaning sequence number identification is a sequence number determined according to the sequence of the target oil pipe in the picture acquisition route. If an oil pipe is cleaned by the 10 th oil pipe, the corresponding cleaning serial number is identified as 10. And providing basic analysis data for subsequent cleaning effect evaluation by obtaining an image set in the oil pipe.

Step S300: inputting the automatic cleaning mode into a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

further, the step S300 of the embodiment of the present application further includes:

Step S310: taking the cleaning defect as an index, and calling a cleaning defect data set from an oil pipe cleaning database, wherein the cleaning defect data set comprises a dirt residue type data set and a pipe surface damage type set;

step S320: taking the automatic cleaning mode as an index, and calling an automatic cleaning mode set from an oil pipe cleaning database;

step S330: obtaining an automatic cleaning mode-cleaning defect data mapping relation according to the corresponding relation between the cleaning defect data set and the automatic cleaning mode set;

step S340: and constructing the cleaning defect analysis module based on the mapping relation between the automatic cleaning mode and the cleaning defect data.

In one embodiment, the cleaning defect analysis module is a functional module for analyzing damage to the inner wall of the oil pipe and the type of dirt residue which is easy to generate by using an automatic cleaning mode. The residual dirt type is determined according to the automatic cleaning mode used, and dirt such as chemical reagent, water drops, mechanical part fragments and the like remained on the surface of the pipe wall after cleaning. The damage type of the pipe surface is damage to the surface of the pipe wall through automatic cleaning, such as wall corrosion caused by cleaning agent, flushing mark caused by high-pressure water jet flushing, and the like. For example, if mechanical cleaning, i.e., mechanical decontamination of the inner wall of a pipe using various types of drills, cutters or wire brushes, is used, pieces of parts tend to remain on the inner wall surface and damage to the inner wall, such as scratches, grooves, raised edges, etc., tends to occur.

Specifically, a cleaning defect data set is obtained from a tubing cleaning database by indexing the cleaning defect. The cleaning defect data set is data describing the type of dirt left behind and the type of damage to the surface of the pipe, which are caused by cleaning the inner wall of the oil pipe. The automatic cleaning mode is used as a keyword, and an automatic cleaning mode set is obtained from an oil pipe cleaning database, wherein the automatic cleaning mode set comprises high-pressure water jet automatic cleaning, mechanical automatic cleaning, sand blasting automatic cleaning, chemical automatic cleaning and the like. And constructing an automatic cleaning mode-cleaning defect data mapping relation according to the corresponding relation between the cleaning defect data set and the automatic cleaning mode set, namely cleaning defects caused by the automatic cleaning mode, so as to construct a cleaning defect analysis module. Therefore, the automatic cleaning mode is input into the cleaning defect analysis module, and the corresponding cleaning defect data can be obtained through mapping matching, so that the dirt residual type and the pipe surface damage type are obtained.

Step S400: obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, and obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type;

Further, as shown in fig. 2, the step S400 further includes:

step S410: randomly selecting one dirt residual type from the dirt residual type without returning as a first dirt residual type;

step S420: a plurality of first historical oil pipe in-pipe picture sets and a plurality of first historical dirt characteristic sets are called from an oil pipe cleaning database based on the first dirt residual type, wherein the plurality of first historical oil pipe in-pipe picture sets and the plurality of first historical dirt characteristic sets are in one-to-one correspondence;

step S430: obtaining a plurality of first historical dirt picture frame sets of a plurality of first historical dirt picture sets in the oil pipe according to the plurality of first historical dirt feature sets;

step S440: calculating a first maximum number of frames to be separated and a first minimum number of frames to be separated based on the plurality of first historical dirt picture frame sets;

step S450: randomly selecting one dirt residual type from the dirt residual types without returning as a second dirt residual type, and obtaining a second maximum spacing frame number and a second minimum spacing frame number according to the second dirt residual type;

Step S460: randomly selecting a dirt residual type from the dirt residual type without returning as a P dirt residual type, and obtaining a P maximum spacing frame number and a P minimum spacing frame number according to the P dirt residual type;

step S470: obtaining a maximum dirt residual separation frame number set according to the first maximum separation frame number, the second maximum separation frame number and the Pmax separation frame number, and obtaining a minimum dirt residual separation frame number set according to the first minimum separation frame number, the second minimum separation frame number and the Pmin separation frame number;

step S480: a first capture step and a first in-tube picture capture step are obtained based on the maximum and minimum sets of fouling residual stand-off frames.

Further, the step S480 of the embodiment of the present application further includes:

step S481: taking the maximum dirt residual space frame number in the maximum dirt residual space frame number set as a first capturing step length;

step S482: the minimum dirt residual space frame number in the minimum dirt residual space frame number set is used as a first in-tube picture capturing step size.

In one embodiment, the first capturing step is a number of frames between two adjacent extractions of the image set when capturing and analyzing the larger dirt residues in the image set in the M oil pipe. The first in-pipe picture capturing step length is the number of interval frames extracted twice adjacently when the analysis of capturing all dirt residues in M oil pipe in-pipe picture sets is carried out. The first capture step size is larger than the first intra-tube picture capture step size.

In one embodiment, a dirt residual type is selected randomly from the dirt residual type without returning as a first dirt residual type, and the first dirt residual type is taken as a calling object, so that data when dirt residual analysis is carried out on histories, namely a plurality of first historical in-pipe picture sets and a plurality of first historical dirt characteristic sets, are called from an oil pipe cleaning database. And extracting and summarizing a plurality of corresponding picture frames in the picture set in the oil pipe according to analysis of the first historical dirt feature sets to obtain a plurality of corresponding picture frames of the dirt features in the picture set in the oil pipe, so as to obtain a first historical dirt picture frame set. Each first historical dirt picture frame set corresponds to a picture set in a historical oil pipe, namely, corresponds to one-time automatic cleaning of the historical oil pipe. And obtaining a plurality of first separated frame number sets according to the number of frames spaced between two adjacent frame numbers in each set of the plurality of first historical dirt frame number sets. Comparing the frame numbers from the first frame number sets, and selecting the maximum frame number and the minimum frame number from all the first frame number sets as the first maximum frame number and the first minimum frame number respectively. Thus, when the first dirt residual type exists on the inner wall of the oil pipe after the oil pipe is automatically cleaned, the situation of the number of frames separated in the acquired picture is obtained. Since the automatic cleaning is performed, the cleaning effect depends on the setting of the automation parameters during the cleaning process, and therefore, the occurrence of dirt is associated with the setting of the automation parameters, and therefore, the number of frames to be separated when the screen is extracted can be determined from the number of frames to be separated.

In an embodiment of the present application, one of the soil residual types is randomly selected as the second soil residual type from the soil residual type non-return, and one of the soil residual types is randomly selected as the P-th soil residual type from the soil residual type non-return. According to the same method as the first dirt residual type for obtaining the first maximum and the first minimum frame numbers, obtaining the second maximum and the second minimum frame numbers according to the second dirt residual type, and obtaining the P maximum and the P minimum frame numbers according to the P dirt residual type.

In one possible embodiment, the set of maximum fouling residual stand-off frames is obtained by summing the first maximum stand-off frame number, the second maximum stand-off frame number, and the Pth maximum stand-off frame number. The first minimum spacing frame number, the second minimum spacing frame number and the Pth minimum spacing frame number are summarized to obtain a minimum dirt residual spacing frame number set. The maximum dirt residual interval frame number set describes the maximum frame number of all dirt residual types generated after the target oil pipe set is automatically cleaned in a picture. The minimum dirt residual interval frame number set describes the minimum frame number of all dirt residual types which are generated after the target oil pipe set is automatically cleaned and are separated in a picture.

Specifically, the maximum dirt residual space frame number set is subjected to space frame number screening, the maximum value is set as a first capturing step, then the space frame number in the minimum dirt residual space frame number set is screened, and the minimum value is set as a first in-pipe picture capturing step.

Step S500: constructing a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a SLOWFAST model, capturing pictures of M oil tube inner picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube inner picture sets according to a first inner picture capturing step length to obtain M first oil tube background picture sets, and respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model to obtain M dirt feature sets;

further, the step S500 in the embodiment of the present application further includes:

step S510: a plurality of sample oil pipe dirt capturing picture sets, a plurality of sample first oil pipe background picture sets and a plurality of sample dirt analysis characteristic sets are called from an oil pipe cleaning database to be used for constructing a sample data set;

Step S520: and performing supervision training on the dirt capturing channel, the pipe surface capturing channel and the characteristic analysis layer by using the constructed sample data set until output reaches convergence, so as to obtain the dirt characteristic analysis model.

In the embodiment of the present application, the slow model is applied to two parallel convolutional neural networks, one slow channel and one fast channel, for the same picture. The slow channel is used for extracting the picture set with higher resolution and more pictures, so that the occupied operation amount is larger, the allocated calculation resource is 70%, the fast channel is used for extracting the picture set with lower resolution and less pictures, the occupied operation amount is smaller, but the operation speed is high, and the allocated calculation resource is 30%. Then, the feature analysis layer is connected with the slow channel and the fast channel, is a full connection layer and is used for carrying out feature analysis on the picture input into the full connection layer to obtain the identified features. The dirt capturing channel is used for capturing the features of dirt with larger frame number, so that the dirt capturing channel corresponds to a fast channel, and the tube surface capturing channel is used for capturing the features of dirt with smaller frame number, and the dirt capturing channel corresponds to a slow channel.

In one possible embodiment, the image capturing is performed on the M sets of images in the oil pipe according to the first capturing step, that is, the image capturing is performed on the sets of images according to the number of frames separated in the first capturing step, for example, the number of frames separated is 16, and the images of the 1 st frame and the 17 th frame … … are extracted to obtain the M sets of images captured by the oil pipe dirt. And then, according to the number of the interval frames corresponding to the first in-pipe picture capturing step, picture extraction is carried out on the M oil pipe picture sets, so as to obtain M first oil pipe background picture sets. The M first oil pipe background picture sets comprise pictures corresponding to features with smaller frame numbers.

In one embodiment, the M dirt feature sets are obtained by inputting the obtained M oil pipe dirt capturing picture sets into a dirt capturing channel of a dirt feature analysis model to capture features, inputting the M first oil pipe background picture sets into a pipe face capturing channel to capture features, and inputting the captured features into a feature analysis layer to identify dirt features. The M dirt characteristic sets are sets for describing dirt residual types in M target oil pipes and dirt residual degrees corresponding to the dirt residual types. Illustratively, fouling in an oil pipe is characterized by the presence of water droplets, and the water droplets have a diameter of 2cm. Therefore, the target of intelligent feature recognition on dirt residues after automatic cleaning of the oil pipes in the target oil pipe set is realized, the efficiency of feature analysis is improved, and a reliable basis is provided for subsequent effect evaluation.

In one embodiment, the plurality of sample scale analysis feature sets in the constructed data set are identified as supervision data in the training process by retrieving the plurality of sample tubing scale capture frames, the plurality of sample first tubing background frames, the plurality of sample scale analysis feature sets from the tubing wash database as the constructed sample data set. And inputting the constructed sample data set into a dirt capturing channel, a pipe surface capturing channel and a characteristic analysis layer for training, and supervising the training process by using supervision data until the output result is converged, so as to obtain the dirt characteristic analysis model.

Step S600: constructing a tube surface damage capturing channel, a tube surface capturing channel and a feature analysis layer of a tube surface damage feature analysis model based on the SLOWFAST model, capturing pictures of M tube inner picture sets according to a second capturing step length to obtain M tube surface damage capturing picture sets, capturing pictures of M tube inner picture sets according to the second tube inner picture capturing step length to obtain M second tube background picture sets, and respectively inputting the M tube surface damage capturing picture sets and the M second tube background picture sets into a dirt capturing channel and a tube surface capturing channel of the tube surface damage feature analysis model to obtain M tube surface damage feature sets;

In one possible embodiment, the pipe face damage capture channel is used for feature capture of pipe face damage with larger frame numbers, so that the fast channel corresponds to the pipe face damage, and the pipe face capture channel is used for feature capture of pipe face damage with smaller frame numbers, and the slow channel corresponds to the pipe face damage.

In one possible embodiment, the image capturing is performed on the M sets of images in the oil pipe according to the second capturing step, that is, the image collection is performed on the images according to the number of frames separated in the second capturing step, so as to obtain the M sets of images captured by the damage of the oil pipe surface. And then, according to the number of the interval frames corresponding to the capturing step sizes of the pictures in the second pipe, the pictures in the M oil pipe sets are extracted, and M second oil pipe background picture sets are obtained. The M second oil pipe background picture sets comprise pictures corresponding to picture damage characteristics with smaller frame numbers.

In one embodiment, the M pipe surface damage feature sets are obtained by inputting the obtained M pipe surface damage capture screen sets into a pipe surface damage capture channel of a pipe surface damage feature analysis model to perform feature capture, inputting the M second oil pipe background screen sets into the pipe surface capture channel to perform feature capture, and inputting the captured features into a feature analysis layer to perform pipe surface damage feature recognition. The M pipe surface damage characteristic sets are sets for describing the pipe surface damage residual types in the M target oil pipes and the pipe surface damage residual degrees corresponding to the pipe surface damage residual types. Illustratively, a pipe face damage in an oil pipe is characterized by a scratch on the pipe wall, and the scratch is 10cm in length.

In one embodiment, a sample data set is constructed by retrieving a plurality of sample tubing face damage capture frames, a plurality of sample second tubing background frames, a plurality of sample tubing face damage analysis feature sets as tubing face damage from a tubing wash database. Based on the training method similar to the dirt characteristic analysis model, the pipe surface damage characteristic analysis model is obtained by performing supervision training on a pipe surface damage capturing channel, a pipe surface capturing channel and a characteristic analysis layer by utilizing a pipe surface damage construction sample data set.

In one embodiment, based on the same method steps as the first capturing step size and the first in-pipe picture capturing step size obtained by the dirt residual type, randomly selecting one pipe face damage residual type from the pipe face damage residual type without being put back as a first pipe face damage residual type, calling a plurality of first historical oil pipe in-pipe picture sets and a plurality of first historical pipe face damage characteristic sets from an oil pipe cleaning database based on the first pipe face damage residual type, obtaining a first damage maximum space frame number and a first damage minimum space frame number, randomly selecting one pipe face damage residual type from the pipe face damage residual type without being put back as a second pipe face damage residual type, obtaining a second damage maximum space frame number and a second damage minimum space frame number according to the second pipe face damage residual type, randomly selecting one pipe face damage residual type from the pipe face damage residual type without being put back as a P pipe face damage residual type, and obtaining a P damage maximum space frame number and a P damage minimum space frame number according to the P-th pipe face damage residual type. Further, a maximum pipe surface damage residual space frame number set and a minimum pipe surface damage residual space frame number set are obtained, the maximum pipe surface damage residual space frame number in the maximum pipe surface damage residual space frame number set is used as a second capturing step length, and the minimum pipe surface damage residual space frame number in the minimum pipe surface damage residual space frame number set is used as a second in-pipe picture capturing step length.

Step S700: inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, and obtaining M cleaning effect evaluation results;

in one embodiment, the historical data set is obtained by retrieving a plurality of historical soil feature sets, a plurality of historical pipe surface damage feature sets, and a plurality of historical cleaning effect evaluation results. And performing supervision training on the framework constructed on the basis of the BP neural network by using the evaluation historical data set until the output reaches convergence, so as to obtain the cleaning effect evaluation model. And inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model to obtain M cleaning effect evaluation results. The M cleaning effect evaluation results are the results of intelligent evaluation on the cleaning effect of the M target oil pipes after automatic cleaning according to the obtained feature set. By providing high-quality input data for the cleaning effect evaluation model according to the reliable feature set, the technical effect of obtaining a reliable evaluation result is achieved.

Step S800: and carrying out staged weighting treatment on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the treatment result.

Further, as shown in fig. 3, step S800 in the embodiment of the present application further includes:

step S810: according to the automatic cleaning track, K stages of automatic cleaning routes are called, and corresponding cleaning serial numbers are matched, so that K cleaning serial number sequences are obtained;

step S820: matching and dividing the M cleaning effect evaluation results according to the K cleaning sequence numbers to obtain K cleaning effect evaluation result sets;

step S830: weighting the K cleaning effect evaluation result sets according to the sequence of the sequence numbers in the K cleaning sequence number sequences to obtain K cleaning effect evaluation results;

step S840: and carrying out average value processing on the K cleaning effect evaluation results to obtain the target oil pipe automatic cleaning effect evaluation result.

Further, the step S850 in this embodiment of the present application further includes:

step S851: traversing the K cleaning sequence numbers to input weight value distribution models to obtain K weight distribution results;

step S852: and carrying out weighting treatment on the K cleaning effect evaluation result sets according to the K weight distribution results to obtain K cleaning effect evaluation results.

Further, step S851 in the embodiment of the present application further includes:

the weight distribution model comprises a weight calculation formula;

the weight value calculation formula is as follows:

，/>；

wherein ,for the weight value of the cleaning effect evaluation result corresponding to the ith cleaning sequence number in the cleaning sequence number,and n is the total number of the sequence numbers in the cleaning sequence number.

In one possible embodiment, the K phase auto-purge paths are invoked by following the phase paths in the auto-purge trajectory. Wherein the stage route is determined according to a single cleaning longest distance of an automatic cleaning mode. And obtaining K cleaning sequence numbers according to the cleaning sequence number identification matched with the corresponding cleaning sequence number, wherein the sequence order is determined according to the sequence order of the route passing through the target oil pipe, the corresponding sequence number value in the sequence is unchanged and is consistent with the sequence number value in the cleaning sequence number identification, and the sequence is not reordered.

In one embodiment, the M cleaning effect evaluation results are matched and divided according to K cleaning sequence number sequences, and the evaluation results belonging to one sequence are divided together to obtain K cleaning effect evaluation result sets. Because the impurity washed by the previous oil pipe in each stage is transferred to the next oil pipe when the automatic washing is carried out, the actual washing effect of the next oil pipe is lower than that of the previous oil pipe under the same washing parameters, and the corresponding weight value is higher and higher along with the backward sequential growth. And then inputting the K cleaning sequence numbers into a weight calculation formula in a weight distribution model, calculating to obtain weight values occupied by cleaning effect evaluation results corresponding to each cleaning sequence number in the sequence in the weighted calculation, and summarizing to obtain K weight distribution results. And then respectively carrying out weighting treatment on the K cleaning effect evaluation result sets according to the K weight distribution results, and obtaining K cleaning effect evaluation results according to the weighting treatment results. And averaging the obtained K cleaning effect evaluation results, and taking the obtained result as the automatic cleaning effect evaluation result of the target oil pipe.

In summary, the embodiments of the present application have at least the following technical effects:

according to the method, the automatic cleaning track of the oil pipe is used as a clue, the picture in the oil pipe is collected, so that the obtained picture data is orderly ensured, a reliable basis is provided for the subsequent evaluation result weighting treatment, then a dirt characteristic analysis model and a pipe surface damage analysis model are constructed by using a SLOWFAST model, efficient and high-quality picture capturing is performed, reliable characteristics are obtained, further the cleaning effect evaluation model is used for evaluation, the obtained evaluation result is subjected to staged weighting treatment based on the sequence in the automatic cleaning track, and the target oil pipe automatic cleaning effect evaluation result subjected to refined analysis is obtained. The technical effects of improving the reliability of the evaluation result and the intelligence degree and efficiency of the evaluation are achieved.

Example two

Based on the same inventive concept as the cleaning effect evaluation method for guiding the automatic cleaning of the oil pipe in the foregoing embodiments, as shown in fig. 4, the present application provides a cleaning effect evaluation system for guiding the automatic cleaning of the oil pipe, and the embodiments of the system and the method in the embodiments of the present application are based on the same inventive concept. Wherein the system comprises:

The automatic cleaning information obtaining module 11 is used for obtaining automatic cleaning information of the target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

the oil pipe in-pipe picture obtaining module 12 is configured to use the automatic cleaning track as a picture collecting route, and perform picture collection on the inside of the target oil pipe set by using the picture collecting module to obtain M oil pipe in-pipe picture sets, where M is the number of oil pipes in the target oil pipe set, is an integer greater than or equal to 1, and the M oil pipe in-pipe picture sets have cleaning sequence number identifiers;

the pipe surface damage type obtaining module 13 is used for inputting the automatic cleaning mode into a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

a first capturing step length obtaining module 14, where the first capturing step length obtaining module 14 is configured to obtain a first capturing step length and a first in-tube picture capturing step length based on the dirt residual type, and obtain a second capturing step length and a second in-tube picture capturing step length according to the pipe surface damage type;

The dirt feature set obtaining module 15 is configured to construct a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a slow feature analysis model, perform picture capturing on M oil tube inner picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, perform picture capturing on the M oil tube inner picture sets according to a first in-tube picture capturing step length to obtain M first oil tube background picture sets, and input the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model respectively to obtain M dirt feature sets;

the pipe surface damage characteristic obtaining module 16, where the pipe surface damage characteristic obtaining module 16 is configured to construct a pipe surface damage capturing channel, a pipe surface capturing channel and a characteristic analysis layer of a pipe surface damage characteristic analysis model based on the slow fast model, perform image capturing on M pipe inner image sets according to a second capturing step length to obtain M pipe surface damage capturing image sets, perform image capturing on M pipe inner image sets according to a second pipe inner image capturing step length to obtain M second pipe background image sets, and input the M pipe surface damage capturing image sets and the M second pipe background image sets into a dirt capturing channel and a pipe surface capturing channel of the pipe surface damage characteristic analysis model respectively to obtain M pipe surface damage characteristic sets;

The evaluation result obtaining module 17 is used for inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, so as to obtain M cleaning effect evaluation results;

the target evaluation result obtaining module 18, where the target evaluation result obtaining module 18 is configured to perform staged weighting processing on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M in-pipe images corresponding to the M cleaning effect evaluation results, and obtain a target oil pipe automatic cleaning effect evaluation result according to the processing result.

Further, the system further comprises:

the cleaning defect data acquisition unit is used for taking the cleaning defect as an index and calling a cleaning defect data set from the oil pipe cleaning database, wherein the cleaning defect data set comprises a dirt residual type data set and a pipe surface damage type set;

the automatic cleaning mode collection obtaining unit is used for taking the automatic cleaning mode as an index and calling the automatic cleaning mode collection from the oil pipe cleaning database;

the mapping relation obtaining unit is used for obtaining the mapping relation between the automatic cleaning mode and the cleaning defect data according to the corresponding relation between the cleaning defect data set and the automatic cleaning mode set;

And the defect analysis module construction unit is used for constructing the cleaning defect analysis module based on the automatic cleaning mode-cleaning defect data mapping relation.

Further, the system further comprises:

a first dirt-residue-type setting unit configured to randomly select one dirt-residue type from the dirt-residue types without being put back as a first dirt-residue type;

the first historical dirt characteristic set obtaining unit is used for calling a plurality of first historical oil pipe in-pipe picture sets and a plurality of first historical dirt characteristic sets from an oil pipe cleaning database based on the first dirt residual type, wherein the plurality of first historical oil pipe in-pipe picture sets and the plurality of first historical dirt characteristic sets are in one-to-one correspondence;

a first historical dirt picture frame set obtaining unit, configured to obtain a plurality of first historical dirt picture frame sets of a plurality of first historical oil pipe in-pipe picture sets according to the plurality of first historical dirt feature sets;

a first minimum frame number obtaining unit for calculating a frame number based on the plurality of first historical dirt picture frame sets to obtain a first maximum frame number and a first minimum frame number;

A second minimum frame number obtaining unit for randomly selecting one of the dirt residual types from the dirt residual types without returning as a second dirt residual type, and obtaining a second maximum frame number and a second minimum frame number according to the second dirt residual type;

the P minimum separation frame number obtaining unit is used for randomly selecting a dirt residual type from the dirt residual type without returning to be used as a P dirt residual type, and obtaining the P maximum separation frame number and the P minimum separation frame number according to the P dirt residual type;

the minimum separation frame number set obtaining unit is used for obtaining a maximum dirt residual separation frame number set according to the first maximum separation frame number, the second maximum separation frame number and the Pth maximum separation frame number, and obtaining a minimum dirt residual separation frame number set according to the first minimum separation frame number, the second minimum separation frame number and the Pth minimum separation frame number;

and a first intra-tube picture capturing step size obtaining unit for obtaining a first capturing step size and a first intra-tube picture capturing step size based on the maximum dirt residual separation frame number set and the minimum dirt residual separation frame number set.

Further, the system further comprises:

a first capturing step setting unit configured to set, as a first capturing step, a maximum dirt remaining space frame number in the maximum dirt remaining space frame number set;

and the first in-tube picture capturing step length setting unit is used for taking the minimum dirt residual spacing frame number in the minimum dirt residual spacing frame number set as a first in-tube picture capturing step length.

Further, the system further comprises:

the sample data set constructing unit is used for calling a plurality of sample oil pipe dirt capturing picture sets, a plurality of sample first oil pipe background picture sets and a plurality of sample dirt analysis characteristic sets from the oil pipe cleaning database to construct a sample data set;

and the supervision and training unit is used for performing supervision and training on the dirt capturing channel, the pipe surface capturing channel and the characteristic analysis layer by utilizing the constructed sample data set until the output reaches convergence, so as to obtain the dirt characteristic analysis model.

Further, the system further comprises:

k cleaning sequence obtaining units, which are used for calling K stages of automatic cleaning routes according to the automatic cleaning track and matching corresponding cleaning sequence to obtain K cleaning sequence;

K cleaning effect evaluation result obtaining units, wherein the K cleaning effect evaluation result obtaining units are used for carrying out matching division on the M cleaning effect evaluation results according to K cleaning sequence numbers to obtain K cleaning effect evaluation result sets;

the weighting processing unit is used for carrying out weighting processing on the K cleaning effect evaluation result sets according to the sequence of the sequence numbers in the K cleaning sequence number sequences to obtain K cleaning effect evaluation results;

and the target cleaning effect evaluation result obtaining unit is used for carrying out average processing on the K cleaning effect evaluation results to obtain the automatic cleaning effect evaluation result of the target oil pipe.

Further, the system further comprises:

k weight distribution result obtaining units, wherein the K weight distribution result obtaining units are used for traversing the K cleaning sequence numbers and inputting the K cleaning sequence numbers into a weight distribution model to obtain K weight distribution results;

and the K weight distribution result obtaining units are used for carrying out weighting processing on the K cleaning effect evaluation result sets according to the K weight distribution results to obtain K cleaning effect evaluation results.

Further, the system further comprises:

the weight distribution model comprises a weight calculation formula;

the weight value calculation formula is as follows:

，/>；

wherein ,for the weight value of the cleaning effect evaluation result corresponding to the ith cleaning sequence number in the cleaning sequence number,and n is the total number of the sequence numbers in the cleaning sequence number.

It should be noted that the sequence of the embodiments of the present application is merely for description, and does not represent the advantages and disadvantages of the embodiments. And the foregoing description has been directed to specific embodiments of this specification. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

The foregoing description of the preferred embodiments of the present application is not intended to limit the invention to the particular embodiments of the present application, but to limit the scope of the invention to the particular embodiments of the present application.

The specification and drawings are merely exemplary of the application and are to be regarded as covering any and all modifications, variations, combinations, or equivalents that are within the scope of the application. It will be apparent to those skilled in the art that various modifications and variations can be made in the present application without departing from the scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the present application and the equivalents thereof, the present application is intended to cover such modifications and variations.

Claims (9)

1. The method is applied to an automatic oil pipe cleaning guidance system which is in communication connection with a picture acquisition module, and the method comprises the following steps:

acquiring automatic cleaning information of a target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

taking the automatic cleaning track as a picture collecting route, and collecting pictures in pipes of the target oil pipe set by using a picture collecting module to obtain M oil pipe picture sets, wherein M is the number of oil pipes in the target oil pipe set and is an integer greater than or equal to 1, and the M oil pipe picture sets are provided with cleaning sequence number identifiers;

Inputting the automatic cleaning mode into a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, and obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type;

constructing a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a SLOWFAST model, capturing pictures of M oil tube inner picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube inner picture sets according to a first inner picture capturing step length to obtain M first oil tube background picture sets, and respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model to obtain M dirt feature sets;

constructing a tube surface damage capturing channel, a tube surface capturing channel and a feature analysis layer of a tube surface damage feature analysis model based on the SLOWFAST model, capturing pictures of M tube inner picture sets according to a second capturing step length to obtain M tube surface damage capturing picture sets, capturing pictures of M tube inner picture sets according to the second tube inner picture capturing step length to obtain M second tube background picture sets, and respectively inputting the M tube surface damage capturing picture sets and the M second tube background picture sets into a dirt capturing channel and a tube surface capturing channel of the tube surface damage feature analysis model to obtain M tube surface damage feature sets;

Inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, and obtaining M cleaning effect evaluation results;

and carrying out staged weighting treatment on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the treatment result.

2. The method of claim 1, wherein the automatic cleaning mode is input to a cleaning defect analysis module for cleaning defect analysis to obtain a dirt residue type and a pipe surface damage type, the method comprising:

taking the cleaning defect as an index, and calling a cleaning defect data set from an oil pipe cleaning database, wherein the cleaning defect data set comprises a dirt residue type data set and a pipe surface damage type set;

taking the automatic cleaning mode as an index, and calling an automatic cleaning mode set from an oil pipe cleaning database;

obtaining an automatic cleaning mode-cleaning defect data mapping relation according to the corresponding relation between the cleaning defect data set and the automatic cleaning mode set;

and constructing the cleaning defect analysis module based on the mapping relation between the automatic cleaning mode and the cleaning defect data.

3. The method of claim 1, wherein the obtaining a first capture step and a first in-tube picture capture step based on the scale residual type, the method comprising:

randomly selecting one dirt residual type from the dirt residual type without returning as a first dirt residual type;

a plurality of first historical oil pipe in-pipe picture sets and a plurality of first historical dirt characteristic sets are called from an oil pipe cleaning database based on the first dirt residual type, wherein the plurality of first historical oil pipe in-pipe picture sets and the plurality of first historical dirt characteristic sets are in one-to-one correspondence;

obtaining a plurality of first historical dirt picture frame sets of a plurality of first historical dirt picture sets in the oil pipe according to the plurality of first historical dirt feature sets;

calculating a first maximum number of frames to be separated and a first minimum number of frames to be separated based on the plurality of first historical dirt picture frame sets;

randomly selecting one dirt residual type from the dirt residual types without returning as a second dirt residual type, and obtaining a second maximum spacing frame number and a second minimum spacing frame number according to the second dirt residual type;

randomly selecting a dirt residual type from the dirt residual type without returning as a P dirt residual type, and obtaining a P maximum spacing frame number and a P minimum spacing frame number according to the P dirt residual type;

Obtaining a maximum dirt residual separation frame number set according to the first maximum separation frame number, the second maximum separation frame number and the Pmax separation frame number, and obtaining a minimum dirt residual separation frame number set according to the first minimum separation frame number, the second minimum separation frame number and the Pmin separation frame number;

a first capture step and a first in-tube picture capture step are obtained based on the maximum and minimum sets of fouling residual stand-off frames.

4. The method of claim 3, wherein the obtaining the first capture step and the first intra-tube picture capture step is based on a set of maximum and minimum fouling residual stand-off frame numbers, the method comprising:

taking the maximum dirt residual space frame number in the maximum dirt residual space frame number set as a first capturing step length;

the minimum dirt residual space frame number in the minimum dirt residual space frame number set is used as a first in-tube picture capturing step size.

5. The method of claim 1, wherein the constructing a fouling capture channel and a pipe surface capture channel of a fouling profile model based on the slow fast model comprises:

A plurality of sample oil pipe dirt capturing picture sets, a plurality of sample first oil pipe background picture sets and a plurality of sample dirt analysis characteristic sets are called from an oil pipe cleaning database to be used for constructing a sample data set;

and performing supervision training on the dirt capturing channel, the pipe surface capturing channel and the characteristic analysis layer by using the constructed sample data set until output reaches convergence, so as to obtain the dirt characteristic analysis model.

6. The method of claim 1, wherein the method comprises:

according to the automatic cleaning track, K stages of automatic cleaning routes are called, and corresponding cleaning serial numbers are matched, so that K cleaning serial number sequences are obtained;

matching and dividing the M cleaning effect evaluation results according to the K cleaning sequence numbers to obtain K cleaning effect evaluation result sets;

weighting the K cleaning effect evaluation result sets according to the sequence of the sequence numbers in the K cleaning sequence number sequences to obtain K cleaning effect evaluation results;

and carrying out average value processing on the K cleaning effect evaluation results to obtain the target oil pipe automatic cleaning effect evaluation result.

7. The method of claim 6, wherein the K cleaning effect evaluation result sets are weighted according to the order of sequence numbers in the K cleaning sequence number sequences, the method comprising:

Traversing the K cleaning sequence numbers to input weight value distribution models to obtain K weight distribution results;

and carrying out weighting treatment on the K cleaning effect evaluation result sets according to the K weight distribution results to obtain K cleaning effect evaluation results.

8. The method of claim 7, wherein the method comprises:

the weight distribution model comprises a weight calculation formula;

the weight value calculation formula is as follows:

；

wherein ,for the weight value of the cleaning effect evaluation result corresponding to the ith cleaning sequence number in the cleaning sequence number,/>And n is the total number of the sequence numbers in each cleaning sequence number.

9. A cleaning effect evaluation system for directing automatic cleaning of tubing, the system comprising:

the automatic cleaning information acquisition module is used for acquiring automatic cleaning information of the target oil pipe set, wherein the automatic cleaning information comprises an automatic cleaning track and an automatic cleaning mode;

the oil pipe in-pipe picture obtaining module is used for collecting pictures in pipes of the target oil pipe set by taking the automatic cleaning track as a picture collecting route and obtaining M oil pipe in-pipe picture sets, wherein M is the number of oil pipes in the target oil pipe set and is an integer greater than or equal to 1, and the M oil pipe in-pipe picture sets are provided with cleaning sequence number identifiers;

The pipe surface damage type obtaining module is used for inputting the automatic cleaning mode into the cleaning defect analysis module for cleaning defect analysis to obtain a dirt residual type and a pipe surface damage type;

the first capturing step length obtaining module is used for obtaining a first capturing step length and a first in-pipe picture capturing step length based on the dirt residual type, and obtaining a second capturing step length and a second in-pipe picture capturing step length according to the pipe surface damage type;

the dirt feature set obtaining module is used for constructing a dirt capturing channel, a tube surface capturing channel and a feature analysis layer of a dirt feature analysis model based on a SLOWFAST model, capturing pictures of M oil tube in-tube picture sets according to a first capturing step length to obtain M oil tube dirt capturing picture sets, capturing pictures of M oil tube in-tube picture sets according to a first in-tube picture capturing step length to obtain M first oil tube background picture sets, and respectively inputting the M oil tube dirt capturing picture sets and the M first oil tube background picture sets into the dirt capturing channel and the tube surface capturing channel of the dirt feature analysis model to obtain M dirt feature sets;

The pipe surface damage characteristic obtaining module is used for constructing a pipe surface damage capturing channel, a pipe surface capturing channel and a characteristic analysis layer of a pipe surface damage characteristic analysis model based on the SLOWFAST model, capturing pictures of M pipe inner picture sets according to a second capturing step length to obtain M pipe surface damage capturing picture sets, capturing pictures of M pipe inner picture sets according to a second pipe inner picture capturing step length to obtain M second pipe background picture sets, and respectively inputting the M pipe surface damage capturing picture sets and the M second pipe background picture sets into a dirt capturing channel and a pipe surface capturing channel of the pipe surface damage characteristic analysis model to obtain M pipe surface damage characteristic sets;

the evaluation result obtaining module is used for inputting the M dirt characteristic sets and the M pipe surface damage characteristic sets into a cleaning effect evaluation model for evaluation, so as to obtain M cleaning effect evaluation results;

the target evaluation result obtaining module is used for carrying out staged weighting processing on the M cleaning effect evaluation results based on the cleaning sequence number identifiers of the M oil pipe inner picture sets corresponding to the M cleaning effect evaluation results, and obtaining a target oil pipe automatic cleaning effect evaluation result according to the processing result.