CN116644663A - Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature - Google Patents
Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature Download PDFInfo
- Publication number
- CN116644663A CN116644663A CN202310574148.5A CN202310574148A CN116644663A CN 116644663 A CN116644663 A CN 116644663A CN 202310574148 A CN202310574148 A CN 202310574148A CN 116644663 A CN116644663 A CN 116644663A
- Authority
- CN
- China
- Prior art keywords
- temperature
- hot spot
- prediction
- characteristic measuring
- measuring point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000005259 measurement Methods 0.000 title claims abstract description 11
- 238000012544 monitoring process Methods 0.000 claims abstract description 25
- 239000013598 vector Substances 0.000 claims abstract description 15
- 238000004088 simulation Methods 0.000 claims abstract description 10
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims abstract description 7
- 238000005096 rolling process Methods 0.000 claims abstract description 4
- 239000002245 particle Substances 0.000 claims description 37
- 238000004590 computer program Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 8
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 238000012423 maintenance Methods 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 8
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000009413 insulation Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000004804 winding Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009529 body temperature measurement Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 229940086299 hot spot Drugs 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The application discloses a transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature, which comprises the following steps: acquiring the characteristic measuring point oil temperature of the oil immersed transformer and the monitoring information of the relevant state quantity of the characteristic measuring point for several days, and constructing multi-element time sequence sample data of a characteristic measuring point oil temperature prediction model; calculating a correlation coefficient matrix of the sample data, and extracting a characteristic vector of the oil temperature of the characteristic measuring point; then taking the characteristic vector as multidimensional input, taking the oil temperature of the characteristic measuring point as multidimensional output, establishing a multivariate time sequence prediction model, verifying the accuracy of the multivariate time sequence prediction model on sample data, and realizing single-step prediction (real-time rolling prediction) or multi-step prediction (short-term prediction) of the oil temperature of the characteristic measuring point; then, obtaining hot spot temperatures under different working conditions through simulation calculation of transformer temperature field distribution, and constructing inversion samples of the hot spot temperatures by combining monitoring data of the oil temperatures of the characteristic measuring points; and finally, establishing a hot spot temperature and characteristic measuring point oil temperature relation model, and verifying the effectiveness and accuracy of the model on an inversion sample, so that inversion prediction of a hot spot temperature future value is realized. The application can rapidly realize real-time or short-term prediction of the hot spot temperature by utilizing the relevant state quantity historical information which is easy to acquire in the transformer monitoring system, and has a certain reference value for the operation and maintenance management of the large-scale oil-immersed transformer.
Description
Technical Field
The application relates to the technical field of hot spot temperature monitoring of oil immersed transformers, in particular to a transformer hot spot temperature inversion prediction method based on characteristic measuring point oil temperature.
Background
The transformer is an important connection between the power grid and the generator set, and the safe and stable operation of the transformer is important for guaranteeing the long-term safety of the whole system. According to relevant domestic and foreign statistical analysis, transformer insulation aging or damage is an important cause of transformer accidents, and windings are one of main parts of transformer faults. Winding insulation and transformer temperature rise are closely related, and the temperature rise problem is more prominent along with the operation of a large-scale high-pressure oil immersed transformer. If the hot spot temperature of the transformer exceeds a threshold value, transformer accidents are easy to occur, and the safety of the transformer substation is directly affected; if the hot spot temperature is too low, the capacity of the transformer is not fully utilized, and the economy of the transformer substation needs to be improved. Therefore, monitoring the hot spot temperature of the large oil immersed transformer is beneficial to reducing the fault rate of the transformer, prolonging the service life of the transformer and ensuring the economic benefit of the transformer.
At present, for a large-sized oil immersed transformer, the difficulty of directly monitoring the hot spot temperature in batches by arranging a distributed optical fiber temperature sensor is high, so that the engineering site is used for estimating and predicting the hot spot temperature for the large-sized oil immersed transformer generally by using an indirect method, and the most commonly used method at the present stage is to estimate the hot spot temperature indirectly by the monitoring value of the top oil temperature according to an empirical model in the national standard GB/T1094.7 (corresponding to IEC 60076-7). However, the empirical model ignores some nonlinear characteristics of the transformer oil, and cannot reflect the influence of the change of the running state of the transformer on the hot spot temperature. With the wide application of the artificial intelligence technology in the power system, the operation history data which is easy to acquire except the hot spot temperature in the transformer monitoring system is utilized, and the intelligent algorithm is used for realizing continuous prediction of the hot spot temperature, so that the real-time monitoring of the operation condition of the transformer and the timely adjustment of the operation mode are facilitated. However, in the hot spot temperature prediction model established in the prior document, most of sample data are hot spot temperature test measurement data, and the acquisition of samples is limited by test conditions and actual running states of the transformer; in addition, the hot spot temperature prediction model of part of the literature can only calculate the hot spot temperature at the current moment according to the measured data of other states of the transformer at the same moment, but cannot reflect the change condition of the hot spot temperature in the time domain.
Disclosure of Invention
In order to solve the technical problems, the application provides a transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature, which combines finite element simulation and artificial intelligence algorithm, builds analysis samples by using multi-state quantity monitoring data of a transformer and hot spot temperature simulation data, can realize rapid continuous prediction of hot spot temperature, and has a certain reference value for operation and maintenance management of a large-scale oil immersed transformer.
In a first aspect, the application provides a transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature, which comprises the following steps:
step A, constructing multi-element time sequence sample data of a characteristic measuring point oil temperature prediction model according to the oil temperature of the characteristic measuring point of the oil-immersed transformer and the monitoring information of other relevant state quantities in a plurality of days;
step B, carrying out normalization processing on the multi-element time series sample data in the step A, calculating a correlation coefficient matrix of the multi-element time series sample data, and extracting a characteristic vector of the oil temperature of the characteristic measuring point;
c, taking the characteristic vector in the step B as multidimensional input, taking the characteristic measuring point oil temperature as multidimensional output, establishing a multielement time sequence prediction model, and verifying the validity and the accuracy of the model on the sample data in the step A so as to perform single-step prediction (real-time rolling prediction) or multi-step prediction (short-term prediction) on the characteristic measuring point oil temperature;
step D, performing simulation calculation on the distribution of the temperature field of the transformer, and extracting hot spot temperatures and characteristic measuring point oil temperatures under different working conditions to construct inversion samples of the hot spot temperatures;
and E, establishing a relation model of the hot spot temperature and the oil temperature of the characteristic measuring point, and verifying the validity and the accuracy of the inversion sample in the step D, so that inversion prediction of a hot spot temperature future value can be realized.
As a preferred technical scheme of the application: in the step a, a plurality of transformer oil temperature measuring points which are easy to obtain monitoring data are selected as characteristic measuring points based on the hot spot temperature inversion principle of the oil immersed transformer, such as top oil temperature, oil outlet temperature, oil inlet temperature and the like, and the multi-element time sequence sample data shown in the following formula is constructed by combining monitoring information of other relevant state quantities, such as voltage, current, active power, reactive power, cooler temperature and the like:
in the method, in the process of the application,the i-dimensional state quantity of the j-th group of data is that m is the total dimension of all state quantities including the oil temperature of the characteristic measuring point, and n is the total length of the time sequence, namely the total group number of the sample data.
As a preferred technical scheme of the application: in the step B, the linear correlation and the nonlinear correlation of the sequence are comprehensively considered, and a Pearson-Spearman mixed correlation coefficient matrix is calculated according to the following formula:
in the method, in the process of the application,average values of p-th and q-th dimension state quantities of n groups of data are respectively represented; to the p, q two-dimensional state quantity data x p 、x q Respectively forming new column vectors x 'in descending order' p 、x′ q ，/>And->At x' p 、x′ q The positions of (a) are respectively recorded asThe value is [ -1,1]Between (I)>Representing a complete negative correlation, +.>Indicating complete independence, ->Indicating a complete positive correlation.
As a preferred technical scheme of the application: in the step C, the multivariate time sequence prediction model is a CNN-GRU model, the basic structure of the multivariate time sequence prediction model comprises a CNN feature extraction module and a GRU prediction module, the CNN feature extraction module comprises a convolution layer, a pooling layer and a flattening layer, the GRU prediction module comprises a GRU circulation unit and a full connection layer, and the number of the convolution layers, the number of convolution kernels and the size of the convolution kernels in the CNN module and the number of the circulation units in the GRU module are further designed according to the prediction effect of sample data.
As a preferred technical scheme of the application: in the step D, multi-physical field coupling finite element simulation calculation is carried out on the temperature field distribution of the transformer under different working conditions, hot spot temperatures under various working conditions are extracted from the finite element simulation calculation, and inversion samples of the hot spot temperatures are constructed by combining monitoring data of the oil temperatures of characteristic measuring points under corresponding working conditions and normalized.
As a preferred technical scheme of the application: in the step E, a relation model of the hot spot temperature and the oil temperature of the characteristic measuring point is a PSO-BPNN model; the step E comprises the following steps E1 to E7:
e1, determining the topological structure of the BPNN, namely the unit numbers of an input layer, an hidden layer and an output layer of the BPNN;
e2, determining a PSO structure, taking a BPNN training error as a particle fitness, and initializing a particle swarm;
step E3, calculating the fitness value of the current position of each particle so as to measure the quality of the position, and recording the position with the largest fitness value, which is found by the ith particle in the d dimension under the current iteration number, as P id I.e., individual history optimal solutions; the position with the largest fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P gd I.e., the population history optimal solution.
E4, updating the speed and the position of each particle according to the individual history optimal solution and the group history optimal solution of the particle group by using the following formula to ensure that the particles move towards the direction of the more optimal solution:
v id (k+1)＝ωv id +c 1 r 1 (P id (k)-x id (k))+c 2 r 2 (P gd (k)-x id (k))
x id (k+1)＝x id (k)+v id (k+1)
wherein k is the iteration number; x is x id 、v id Respectively solving the position and the speed of the ith particle in the space for the d-dimensional target; omega is the inertial weight of the velocity; c 1 Individual acceleration coefficient, c, for individual particles 2 Population acceleration coefficients for particle populations; r is (r) 1 And r 2 Is [0,1]Random constants in the range;
e5, repeating the step E3 and the step E4 until a set termination condition is reached, such as the maximum iteration number is reached, the adaptability meets a preset threshold value or the running time exceeds a limit, and the like, and outputting a group history optimal solution calculated in the last iteration as a global optimal solution, namely PSO optimized BPNN weight and bias;
e6, initializing BPNN weight and bias according to the output result of the particle swarm, training and verifying a relation model of the hot spot temperature and the characteristic measuring point oil temperature on the hot spot temperature inversion sample in the step D;
and E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the step C and the relation model of the hot spot temperature and the oil temperature of the characteristic measuring point in the step E6, realizing real-time or short-term prediction of the future value of the hot spot temperature.
In a second aspect, the present application provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.
In a third aspect, the present application provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the above method.
Compared with the prior art, the embodiment of the application combines finite element simulation and artificial intelligence algorithm to establish a data-driven high-precision prediction model, and sample data in prediction modeling does not depend on a hot spot temperature measurement test, so that the method is not limited by test conditions and the actual running state of the transformer; based on the hot spot temperature inversion principle, the method utilizes relevant state quantity historical data which are easy to obtain in a transformer monitoring system to realize quick prediction of a hot spot temperature future value, and has a certain reference value for operation and maintenance management of a large-scale oil immersed transformer.
Drawings
FIG. 1 is a schematic flow chart of a transformer hot spot temperature inversion prediction method based on a characteristic measurement point oil temperature in an embodiment of the application;
FIG. 2 is a schematic diagram of a CNN-GRU network used for establishing a characteristic measuring point oil temperature prediction model in an embodiment of the application;
FIG. 3 is a schematic diagram of a BPNN structure used for establishing a relationship model between a hot spot temperature and a characteristic measurement point oil temperature in an embodiment of the application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
As shown in fig. 1, a transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature includes the following steps:
step A, aiming at the large oil immersed transformer, selecting top-layer oil temperature T easy to acquire monitoring data based on a hot spot temperature inversion principle top_oil Temperature T of oil outlet oil_out Oil inlet temperature T oil_in The characteristic measuring point oil temperature is used as a transformer; other relevant state quantities are high-voltage side phase voltage U A 、U B 、U C High side line voltage U AB 、U BC 、U CA High side current I A 、I B 、I C Active power P, reactive power Q, frequency f, power factor cos phi, water outlet temperature T of cooler water_out And cooler water inlet temperature T water_in The method comprises the steps of carrying out a first treatment on the surface of the The sampling time of the oil temperature and other monitoring data of related state quantities of the characteristic measuring point of the oil immersed transformer is 3 minutes, and the monitoring data of 5 days are extracted to construct multi-element time sequence sample data shown in the following formula:
in the method, in the process of the application,an ith dimension state quantity for the jth group of data, < >>Respectively represent T in the j-th group of data oil_out 、T oil_in 、T top_oil 、T water_in 、T water_out 、U A 、U B 、U C 、U AB 、U BC 、U CA 、I A 、I B 、I C 、P、Q、f、cosφ。
And B, carrying out normalization processing on the multi-element time sequence sample data in the step A, comprehensively considering the linear correlation and the nonlinear correlation of the sequence, and calculating a Pearson-Spearman mixed correlation coefficient matrix according to the following formula:
in the method, in the process of the application,average values of p-th and q-th dimension state quantities of n groups of data are respectively represented; to the p, q two-dimensional state quantity data x p 、x q Respectively forming new column vectors in descending order>And->At->The positions in (a) are marked as +.>The value is [ -1,1]Between (I)>Representing a complete negative correlation, +.>Indicating complete independence, ->Representing a complete positive correlation;
extracting according to the calculation result of the correlation coefficient matrixIs the characteristic vector of the oil temperature of the characteristic measuring point.
And C, taking the characteristic vector in the step B as multidimensional input, taking the oil temperature of the characteristic measuring point as multidimensional output, and establishing a CNN-GRU-based multivariate time sequence prediction model. The CNN-GRU network basic structure comprises a CNN feature extraction module and a GRU prediction module, wherein the CNN feature extraction module comprises a convolution layer, a pooling layer and a flattening layer, and the GRU prediction module comprises a GRU circulation unit and a full-connection layer. The specific structure of the CNN-GRU network is determined according to the prediction effect of the actual sample: the number and the size of the convolution kernels are 64, 2, 32 and 2 respectively, and the convolution step sizes are 1; the number of GRU circulating units is 1, and the number of units is 200; the activation functions are all ReLU functions.
And C, dividing the sample data in the step A into a training set and a testing set according to the ratio of 8:2. The single-step prediction model is set as follows: and predicting the oil temperature of the characteristic measuring point at the next sampling moment according to the transformer history monitoring information for 1 hour, namely, the input of the model single learning is 20 groups of characteristic vectors, and the output is 1 group of future target vectors. The multi-step prediction model is set as follows: and predicting a CNN-GRU model of the characteristic measuring point oil temperature of 1 hour in the future by using the history monitoring information of 5 hours, namely, outputting 20 sets of target vectors in the future by using 100 sets of characteristic vectors as input of single learning of the model. The results of the validation of this model on the test set are shown in the following table:
the validation result shown in the table can prove the validity and accuracy of the CNN-GRU model for the characteristic measuring point oil temperature prediction, so that the CNN-GRU model can be used for single-step prediction (real-time rolling prediction) or multi-step prediction (short-term prediction) of the characteristic measuring point oil temperature
And D, performing multi-physical field coupling finite element simulation calculation on the temperature field distribution of the transformer under 400 groups of different working conditions, extracting hot spot temperatures under various working conditions, constructing inversion samples of the hot spot temperatures by combining monitoring data of the oil temperatures of the characteristic measuring points under the corresponding working conditions, performing normalization processing on inversion, and dividing a training set and a testing set according to the proportion of 8:2.
And E, establishing a relation model of the hot spot temperature and the characteristic measuring point oil temperature based on the PSO-BPNN, and verifying the validity and accuracy of the inversion sample in the step D, so that inversion prediction of a hot spot temperature future value can be realized. The method specifically comprises the following steps E1 to E7:
step E1, determining the topological structure of the BPNN: by [ T ] oil_out ，T oil_in ，T top_oil ]As input, at a hot spot temperature T hot_spot For output, the hidden layer is 1 layer and 10 units;
e2, determining a PSO structure, setting the number of particles to be 80, setting the maximum iteration number to be 500, taking the BPNN training error as the particle fitness, and initializing a particle swarm.
Step E3, calculating the fitness value of the current position of each particle so as to measure the quality of the position, and recording the position with the largest fitness value, which is found by the ith particle in the d dimension under the current iteration number, as P id I.e., individual history optimal solutions; the position with the largest fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P gd I.e., the population history optimal solution.
E4, updating the speed and the position of each particle according to the individual history optimal solution and the group history optimal solution of the particle group by using the following formula to ensure that the particles move towards the direction of the more optimal solution:
v id (k+1)＝ωv id +c 1 r 1 (P id (k)-x id (k))+c 2 r 2 (P gd (k)-x id (k))
x id (k+1)＝x id (k)+v id (k+1)
wherein k is the iteration number; x is x id 、v id Respectively solving the position and the speed of the ith particle in the space for the d-dimensional target; omega is the inertial weight of the velocity, here set to linearly time varying decreasing inertial weight; c 1 Individual acceleration coefficient, c, for individual particles 2 Taking c as a group acceleration coefficient of the particle swarm 1 ＝c 2 ＝2；r 1 And r 2 Is [0,1]Random constants in the range;
e5, repeating the step E3 and the step E4 until a set termination condition is reached, such as the maximum iteration number is reached, the adaptability meets a preset threshold value or the running time exceeds a limit, and the like, and outputting a group history optimal solution calculated in the last iteration as a global optimal solution, namely PSO optimized BPNN weight and bias;
and E6, initializing BPNN weight and bias according to the output result of the particle swarm, inverting a relation model of the hot spot temperature and the characteristic measuring point oil temperature on the sample training set by using the hot spot temperature in the step D, and verifying on the testing set, wherein the result shows that the error of the hot spot temperature sample observation value and the PSO-BPNN model fitting value is not more than 0.8 ℃, and the maximum absolute value of the relative error is not more than 1.5%, so that the engineering requirements can be completely met.
And E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the step C and the relation model of the hot spot temperature and the oil temperature of the characteristic measuring point in the step E6, realizing real-time or short-term prediction of the future value of the hot spot temperature.
In addition, the embodiment of the application also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the steps of the method when executing the computer program.
Furthermore, the embodiment of the application also provides a computer readable storage medium, on which a computer program is stored, which when being executed by a processor, implements the steps of the method.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The examples described above represent only a few preferred embodiments of the application, but the application is not limited to the specific embodiments described above, which are merely illustrative and not limiting. Many modifications and substitutions may be made by one of ordinary skill in the art without departing from the spirit and scope of the application as defined by the appended claims, which are to be considered as within the scope of the application.
Claims (8)
1. The transformer hot spot temperature inversion prediction method based on the characteristic measuring point oil temperature is characterized by comprising the following steps of:
step A, constructing multi-element time sequence sample data of a characteristic measuring point oil temperature prediction model according to the oil temperature of the characteristic measuring point of the oil-immersed transformer and the monitoring information of other relevant state quantities in a plurality of days;
step B, carrying out normalization processing on the multi-element time series sample data in the step A, calculating a correlation coefficient matrix of the multi-element time series sample data, and extracting a characteristic vector of the oil temperature of the characteristic measuring point;
c, taking the characteristic vector in the step B as multidimensional input, taking the characteristic measuring point oil temperature as multidimensional output, establishing a multielement time sequence prediction model, and verifying the validity and the accuracy of the model on the sample data in the step A so as to perform single-step prediction (real-time rolling prediction) or multi-step prediction (short-term prediction) on the characteristic measuring point oil temperature;
step D, performing simulation calculation on the distribution of the temperature field of the transformer, and extracting hot spot temperatures and characteristic measuring point oil temperatures under different working conditions to construct inversion samples of the hot spot temperatures;
and E, establishing a relation model of the hot spot temperature and the oil temperature of the characteristic measuring point, and verifying the validity and the accuracy of the inversion sample in the step D, so that inversion prediction of a hot spot temperature future value can be realized.
2. The method according to claim 1, wherein in the step a, a plurality of transformer oil temperature measuring points which are easy to obtain monitoring data are selected as characteristic measuring points based on the principle of inversion of hot spot temperature of the oil immersed transformer, such as top oil temperature, oil outlet temperature, oil inlet temperature, etc., and the following multi-element time series sample data are constructed by combining monitoring information of other relevant state quantities, such as voltage, current, active power, reactive power, cooler temperature, etc.:
in the method, in the process of the application,the i-dimensional state quantity of the j-th group of data is that m is the total dimension of all state quantities including the oil temperature of the characteristic measuring point, and n is the total length of the time sequence, namely the total group number of the sample data.
3. The method according to claim 1, wherein in the step B, the Pearson-Spearman mixed correlation coefficient matrix is calculated by taking into consideration the linear correlation and the nonlinear correlation of the sequences, as follows:
in the method, in the process of the application,average values of p-th and q-th dimension state quantities of n groups of data are respectively represented; to the p, q two-dimensional state quantity data x p 、x q Respectively forming new column vectors x 'in descending order' p 、x′ q ，/>And->At x' p 、x′ q The positions in (a) are marked as +.>The value is [ -1,1]Between (I)>Representing a complete negative correlation, +.>Indicating complete independence, ->Indicating a complete positive correlation.
4. The method according to claim 1, wherein the multivariate time series prediction model in the step C is a CNN-GRU model, the basic structure of which includes a CNN feature extraction module and a GRU prediction module, the CNN feature extraction module includes a convolution layer, a pooling layer and a flattening layer, the GRU prediction module includes a GRU circulation unit and a full connection layer, and the number of convolution layers, the number of convolution kernels, the size of the convolution kernels, and the number of circulation units in the GRU module are further designed according to the prediction effect of the sample data.
5. The method of claim 1, wherein in the step D, the multi-physical-field-coupled finite element simulation calculation is performed on the transformer temperature field distribution under different working conditions, the hot spot temperatures under various working conditions are extracted therefrom, the monitoring data of the characteristic measurement point oil temperature under the corresponding working conditions are combined, an inversion sample of the hot spot temperatures is constructed, and normalization processing is performed.
6. The method according to claim 1, wherein the relationship model between the hot spot temperature and the characteristic measurement point oil temperature in the step E is a PSO-BPNN model, and the step E includes the following steps E1 to E7:
e1, determining the topological structure of the BPNN, namely the unit numbers of an input layer, an hidden layer and an output layer of the BPNN;
e2, determining a PSO structure, taking a BPNN training error as a particle fitness, and initializing a particle swarm;
step E3, calculating the fitness value of the current position of each particle so as to measure the quality of the position, and recording the position with the largest fitness value, which is found by the ith particle in the d dimension under the current iteration number, as P id I.e., individual history optimal solutions; the position with the largest fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P gd I.e., the population history optimal solution.
E4, updating the speed and the position of each particle according to the individual history optimal solution and the group history optimal solution of the particle group by using the following formula to ensure that the particles move towards the direction of the more optimal solution:
v id (k+1)＝ωv id +c 1 r 1 (P id (k)-x id (k))+c 2 r 2 (P gd (k)-x id (k))
x id (k+1)＝x id (k)+v id (k+1)
wherein k is the iteration number; x is x id 、v id Respectively solving the position and the speed of the ith particle in the space for the d-dimensional target; omega is the inertial weight of the velocity; c 1 Individual acceleration coefficient, c, for individual particles 2 Population acceleration coefficients for particle populations; r is (r) 1 And r 2 Is [0,1]Random constants in the range;
e5, repeating the step E3 and the step E4 until a set termination condition is reached, such as the maximum iteration number is reached, the adaptability meets a preset threshold value or the running time exceeds a limit, and the like, and outputting a group history optimal solution calculated in the last iteration as a global optimal solution, namely PSO optimized BPNN weight and bias;
e6, initializing BPNN weight and bias according to the output result of the particle swarm, training and verifying a relation model of the hot spot temperature and the characteristic measuring point oil temperature on the hot spot temperature inversion sample in the step D;
and E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the step C and the relation model of the hot spot temperature and the oil temperature of the characteristic measuring point in the step E6, realizing real-time or short-term prediction of the future value of the hot spot temperature.
7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 6 when the computer program is executed by the processor.
8. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310574148.5A CN116644663A (en) | 2023-05-19 | 2023-05-19 | Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310574148.5A CN116644663A (en) | 2023-05-19 | 2023-05-19 | Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116644663A true CN116644663A (en) | 2023-08-25 |
Family
ID=87639221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310574148.5A Pending CN116644663A (en) | 2023-05-19 | 2023-05-19 | Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116644663A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118032160A (en) * | 2023-12-07 | 2024-05-14 | 广东科源电气股份有限公司 | Transformer wireless temperature measurement method, system, electronic equipment and storage medium |
-
2023
- 2023-05-19 CN CN202310574148.5A patent/CN116644663A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118032160A (en) * | 2023-12-07 | 2024-05-14 | 广东科源电气股份有限公司 | Transformer wireless temperature measurement method, system, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108717579B (en) | Short-term wind power interval prediction method | |
| CN106295899B (en) | Wind power probability density Forecasting Methodology based on genetic algorithm Yu supporting vector quantile estimate | |
| CN113126019B (en) | Remote estimation method, system, terminal and storage medium for error of intelligent ammeter | |
| CN110222897A (en) | A kind of distribution network reliability analysis method | |
| CN111525587B (en) | Reactive load situation-based power grid reactive voltage control method and system | |
| Yang et al. | Multi-dimensional scenario forecast for generation of multiple wind farms | |
| CN106529731A (en) | Regional power grid photovoltaic power station cluster division method | |
| CN113592144B (en) | Medium-long term runoff probability forecasting method and system | |
| CN112115636B (en) | Advanced prediction method and system for insulation aging life of power cable | |
| CN112100911A (en) | Solar radiation prediction method based on deep BISLTM | |
| CN116644663A (en) | Transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature | |
| CN113344288A (en) | Method and device for predicting water level of cascade hydropower station group and computer readable storage medium | |
| CN114046903A (en) | Real-time temperature prediction method and system for high-voltage cable core | |
| CN117828413A (en) | Transformer oil temperature prediction method and system based on LSTM neural network | |
| CN104834816A (en) | Short-term wind speed prediction method | |
| CN110276478B (en) | Short-term wind power prediction method based on segmented ant colony algorithm optimization SVM | |
| CN112363012A (en) | Power grid fault early warning device and method | |
| CN111914384A (en) | Chemical process state monitoring method based on autocorrelation latent variable model | |
| CN116307139A (en) | Wind power ultra-short-term prediction method for optimizing and improving extreme learning machine | |
| CN115713255A (en) | Low-voltage transformer area line loss estimation method based on deep learning | |
| CN114139809A (en) | Dynamic frequency response curve prediction method and system based on XGboost power system | |
| CN112581311A (en) | Method and system for predicting long-term output fluctuation characteristics of aggregated multiple wind power plants | |
| CN113256018B (en) | Wind power ultra-short term probability prediction method based on conditional quantile regression model | |
| CN111695229A (en) | Novel distributed non-Gaussian process monitoring method based on GA-ICA | |
| CN111382895A (en) | Seawater desalination load prediction method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |