CN113514739B - IWOA-BP algorithm-based oil paper insulation aging evaluation method - Google Patents
IWOA-BP algorithm-based oil paper insulation aging evaluation method Download PDFInfo
- Publication number
- CN113514739B CN113514739B CN202110664278.9A CN202110664278A CN113514739B CN 113514739 B CN113514739 B CN 113514739B CN 202110664278 A CN202110664278 A CN 202110664278A CN 113514739 B CN113514739 B CN 113514739B
- Authority
- CN
- China
- Prior art keywords
- aging
- model
- algorithm
- whale
- iwoa
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
-
- 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
Abstract
The invention relates to an oil paper insulation aging evaluation method based on an IWOA-BP algorithm, belonging to the technical field of electricity; comprises the following steps of 1: manufacturing an oiled paper insulation model; step 2: carrying out PDC experiments on the oil paper insulation models in different aging states; and 3, step 3: performing oiled paper insulation aging evaluation; and 4, step 4: and (4) dividing the aging state by combining the comprehensive conductivity by taking different aging time as a standard. The invention provides a transformer oil paper insulation aging state evaluation model based on a PDC method and an IWOA-BP model for the first time.
Description
Technical Field
The invention belongs to the technical field of electricity, and particularly relates to an oil paper insulation aging evaluation method based on an IWOA-BP algorithm.
Background
With the progress of times, the demand of the national people on the electric power resources in China is continuously improved, and the current situation to be faced not only is the contradiction between the supply quantity and the supply cost of the electric power and the demand quantity of the public and enterprises, but also the contradiction between the increase of the demand of the supply quantity of the electric power and the increase of the demand scale of the electric power resources and the non-reduction of the supply cost of the electric power. To solve these contradictions fundamentally, it is necessary to start from a basic aspect. The method improves the existing traditional means, such as power facility maintenance, power consumption in the process of power transmission and transformation, state evaluation of power equipment and the like, and further reduces the cost of power resource supply.
In the power system, the oil-paper insulation power transformer is widely used in each energy conversion critical position, so the safe and stable operation of the power transformer is important for the safety and reliability of the power system operation. However, in the process of long-term operation, the oil-paper insulation transformer is influenced by a plurality of factors to generate an aging phenomenon. The factors comprise electrical aging, thermal aging, mechanical aging and aging under the action of chemical factors, and whatever aging phenomenon occurs, the performance of the transformer is directly reduced and the service life of the transformer is shortened; even more, the whole power system is broken down, which causes great loss of national economy. Therefore, for the state evaluation of the transformer oilpaper insulation aging, certain necessity exists for the stable operation of the power system.
For the insulation aging state evaluation of the power transformer, two discrimination methods are currently adopted: a discrimination method using physical elements as characteristic quantities, namely an electrical diagnosis method; a method for discriminating chemical elements as characteristic quantities, namely a chemical diagnostic method. The chemical diagnosis method mainly comprises a furfural content detection method in oil, a dissolved gas analysis method in oil, a polymerization degree measurement method of transformer insulating oil paperboard and the like. However, since chemical diagnostics are affected by various factors during implementation, such as sampling problems: inconvenient sampling, etc.; sample problem: problems with transportation and storage of the sample; problems in the test: a plurality of random errors exist in the test process, the influence on the test result is large, and the like; the polymerization degree measurement method of the oil paper board has the problems that a cover is required to be hung for coring, an insulating structure is damaged, field operation is inconvenient and the like, so that an electrical diagnosis method which is simpler and more convenient to operate and has more reliable test results is usually selected for evaluating the aging state of the transformer insulating oil paper.
The electric diagnostic methods mainly include a return voltage method, a frequency domain dielectric method, a polarization-depolarization current method, and the like. The return voltage method is sensitive in response and strong in interference resistance, but only can identify the overall insulation condition and cannot distinguish the respective aging states of oil and paper; although the frequency domain dielectric method can realize the test by using a normal-pressure power supply and can effectively evaluate the insulation aging state, the frequency domain dielectric method is greatly influenced by noise and needs high-precision noise elimination; the polarization depolarization current rule has the advantages of nondestructive detection, simplicity in operation, small interference and the like, and therefore, the polarization depolarization current rule is widely applied to transformer insulation aging state evaluation.
Disclosure of Invention
In order to better evaluate the aging state of the oil paper insulation of the power transformer, the invention provides an IWOA-BP algorithm-based oil paper insulation aging evaluation method. In the traditional PDC method, the polarization depolarized current obtained through testing is fitted to current data so as to judge the aging state of the oil paper insulation of the power transformer, and the aging state of the oil paper insulation cannot be accurately evaluated. According to the method, the evaluation accuracy is improved by combining the BP neural network optimized by the improved whale algorithm and the 'comprehensive conductivity' of the aging characteristic quantity, and the evaluation and classification of the oil paper insulation aging state are realized, so that the oil paper insulation aging state is better reflected.
In order to achieve the purpose, the specific technical scheme of the oil paper insulation aging evaluation method based on the IWOA-BP algorithm is as follows:
an IWOA-BP algorithm-based oiled paper insulation aging assessment method comprises the following steps, and the following steps are sequentially carried out:
step 1: manufacturing an oiled paper insulation model;
step 2: carrying out PDC experiments on the oil paper insulation models in different aging states;
PDC: the PDC method is one of electrical diagnostic methods in an insulation aging evaluation method for a power transformer, and applies a dc high voltage to an oil-paper insulation of the transformer to test a Polarization depolarization current flowing through a circuit, thereby evaluating an aging state of the oil-paper insulation of the transformer.
The PDC method test principle is as follows:
the operation of Polarization and Depolarization Current (PDC) is simple, and the testing process is mainly divided into two stages: a polarization phase and a depolarization phase. The basic schematic diagram of the PDC process is shown in fig. 1. The polarization stage is that when the experiment is started, a switch is switched to a contact 1, and a DC high voltage U with a certain amplitude is applied to an oiled paper insulation model 0 At this stage, the data of the current flowing through the circuit recorded by the picoampere meter is the polarization current i p . The duration of this phase is denoted t p . After this phase, a depolarization phase is entered. The switch is turned to the contact 2 to short circuit and ground the two poles of the oil-paper insulation model, the current value flowing through the circuit jumps to a negative value instantly in the phase, and the current data recorded by the Pian ammeter is depolarized current i d Duration of this phase is denoted t d . The experimental results of the polarization depolarizing current under ideal conditions are shown in fig. 2.
The PDC test circuit principle is shown in FIG. 3, and the experimental device comprises an upper computer, a Keithley6517A high-resistance meter, a vacuum constant-temperature aging box, an oil paper insulation model, a GPIB-USB-HS data acquisition card, a coaxial cable and the like.
And step 3: performing oiled paper insulation aging evaluation;
step 31: electric powerThe conductivity is one of the most direct and dominant features for evaluating the insulation performance of an insulating medium, and the insulation performance is evaluated by the conductivity and is influenced by the insulation geometry and the dimensions of the tested sample. Therefore, selecting the electrical conductivity as a characteristic quantity for evaluating the insulation aging state enables a judgment of the insulation aging state as a whole. Calculating the comprehensive conductivity
When the polarization time is long enough, the following expression can be used to calculate:
in formula (9), U 0 Denotes the polarization voltage, C 0 Is the power frequency insulation geometric capacitance of the tested sample 0 Is a vacuum dielectric constant (. epsilon.) 0 ＝8.85×10 -12 F/m)，i p (t end ) Representing the average value of the current at the end of the polarization current, i d (t end ) Represents the average value of the depolarization current terminal current; in the test process, the conductivity is unstable, so that the conductivity when the current tends to be stable is selected as the comprehensive conductivity.
Step 32: optimizing a BP neural network by an IWOA algorithm;
whale optimization algorithm: whale Optimization Algorithm (WOA) is a meta-heuristic intelligent optimization algorithm, and the calculation of the algorithm is realized by simulating a spiral bubble net predation mode of a whale at the head. The whale optimization algorithm selects a random whale position or a prey position, updates the current whale position through the position, further carries out global optimization, and finally converges to obtain an optimal solution.
Improving whale optimization algorithm: whale Optimization Algorithm (WOA) is a meta-heuristic intelligent optimization algorithm, and the calculation of the algorithm is realized by simulating a spiral bubble net predation mode of a whale at the head. The whale optimization algorithm selects a random whale position or a prey position, updates the current whale position through the position, further carries out global optimization, and finally converges to obtain an optimal solution.
Step 33: based on the optimized BP neural network model, PDC current data is used as input, and comprehensive conductivity is used as output, and the aging state is evaluated and predicted;
the IWOA algorithm optimizes a BP neural network model:
the BP neural network is a feedforward neural network which is widely applied at present, and the network structure of the BP neural network is mainly divided into three layers: the structure of the input layer, the hidden layer and the output layer is shown in fig. 4. The BP neural network is mainly characterized in that: the input signal is propagated forward, but the error is propagated backward, in contrast. Although the advantages of the BP neural network are numerous, the BP neural network is easy to fall into a local optimal condition due to the existing phenomenon of 'over training', and the convergence is poor. Therefore, in order to achieve better optimization effect, the BP neural network is optimized by using an IWOA algorithm. The method utilizes the advantages of strong global optimization capability, high convergence speed and the like of IWOA to complement the defects of the BP neural network, and finally obtains a more perfect algorithm model.
The basic idea of the IWOA algorithm for optimizing the BP neural network is as follows: extracting the weight and the threshold of the BP neural network, taking the two factors as the position information of the whale with the IWOA, and continuously updating the position of the whale with the optimal solution position by the approaching of the whale with the optimal solution position, so that the weight and the threshold of the network can be continuously updated, the optimal BP neural network model structure parameters are obtained, and finally the optimization of the IWOA to the BP neural network is realized.
And 4, step 4: and dividing the aging state by combining the comprehensive conductivity by taking different aging time as a standard.
Comprehensive conductivity: the conductivity is one of the most direct and dominant characteristic quantities for evaluating the insulation performance of an insulation medium, and the insulation performance is evaluated by the conductivity and is not affected by the insulation geometry and the size of the tested sample. And the comprehensive conductivity is selected as a characteristic quantity for evaluating the insulation aging state, so that the insulation aging state can be judged integrally.
Further, the step 1 specifically comprises the following steps, and the following steps are sequentially performed:
step 11: cutting the oiled paper into four pieces with the same shape, wherein a cavity is reserved between the two pieces;
step 12: connecting four pieces of oiled paper together by using a copper column;
step 13: drying the manufactured oiled paper model, and putting the oiled paper model into a vacuum ageing oven for vacuum heating for 24 hours;
step 14: and (3) putting the dried oil paper model into a container containing transformer oil which just submerges the model, and soaking for 24 hours to fully combine the transformer oil and the oil paper, so that the oil paper insulation model is manufactured.
Further, the step 2 specifically includes the following steps, and the following steps are sequentially performed:
step 21: putting the oiled paper model into an aging box, selecting temperature for aging at fixed time, and preparing a new oiled paper insulation model for an aging experiment at different temperatures;
step 22: carrying out an experiment on the aged oil paper insulation model every 24 hours;
step 23: connecting the instrument to three electrodes of the oil paper insulation model, and opening a Labview interface to measure PDC current data;
step 24: PDC current data were recorded.
Further, the whale optimization algorithm (IWOA) implementation algorithm used in the step 32 has the following implementation steps:
1) surrounding target prey
The whale can search and surround the prey, and when finding the position of the prey, the whale can continuously update the position of the whale by taking the position of the prey as a reference, so that the aim of surrounding the prey is fulfilled. In the process, the position of the whale with the closest position to the prey is taken as the optimal solution, the rest whales gradually update the self positions with the aim of approaching to the optimal solution position, and therefore the prey surrounding effect is achieved, and the mathematical model is as follows:
in the formula (1)
Representing the current optimal standing whale position, X (t) representing the current standing whale position, t representing the current iteration times, A and C are coefficient vectors, r 1 And r 2 Represents a [0,1]]A is a convergence factor that decreases linearly from 2 to 0 in the iterative process, and the expression of a is as follows:
t in formula (2) max Represents the maximum number of iterations;
2) spiral bubble net attack
The whale with the standing head can update the position of the whale with the shrinkage and the spiral movement, so that the whale gradually approaches to a prey and finally starts an attack. The mathematical model of the spiral movement of the whale at the head is as follows:
in the formula (3), D represents the distance between the current whale position and the optimal solution position, b is a spiral shape constant, and l represents a random number between [ -1,1 ];
judging whether the position is updated spirally or the position is shrunk and surrounded by the p value, wherein the mathematical model is as follows:
wherein p is a random number between [0,1 ];
3) target hunting search and locking
The whale with the standing head can not only track the prey through a spiral air bubble net mode, but also search the prey through the position of the whales with the standing head. When the coefficient vector | A | > is more than or equal to 1, changing the reference of the updating position of the standing whale from the optimal standing whale position to a randomly selected standing whale position, and further carrying out global optimization; when the coefficient vector | A | is less than 1, the whale still carries out position updating with the optimal solution position, and local optimization is carried out, wherein the mathematical model is as follows:
in the formula (5), X rand (t) indicates the position of a random whale;
the WOA algorithm principle is simple, parameters needing manual adjustment are few, the optimization capability of the algorithm is determined by a convergence factor a, and the smaller the value of a is, the more the optimization capability of the algorithm is biased to local optimization; the larger the value of a is, the closer the optimization capability of the algorithm is to the global optimization. The value of a varies in a linear manner, which may cause the algorithm to be difficult to find locally. Therefore, the nonlinear convergence factor a is introduced, and the local optimization capability and the global optimization capability of the algorithm can be ensured on the premise that the variation trend of the a is not changed. The mathematical model is as follows:
a＝(2-2t 2 /t max 2 )(1-t 3 /t max 3 ) (6)
by adding the nonlinear convergence factor, the early global optimization capability of the algorithm can be ensured, and the convergence speed is accelerated; but also can ensure the local optimizing ability in the later period.
When a nonlinear convergence factor is introduced, the linearly changing inertia weight cannot enable the algorithm to have a better optimization effect. In order to enhance the optimization capability of the algorithm, the change rule of the weight should be changed, and the inertia weight changing linearly is changed into the adaptive weight changing nonlinearly, and the mathematical model is as follows:
the mathematical model for updating the position of the whale with the self-adaptive weight factor is as follows:
further, the step 32: the specific implementation steps for optimizing the BP neural network by an IWOA algorithm are as follows:
step 32-1, initializing a BP neural network, determining a topological structure of the network, and initializing weights and thresholds in network parameters;
step 32-2, initializing an IWOA algorithm, setting the population number N of the whale with beluga and the maximum iteration time t max Taking the weight and the threshold of the BP neural network as the position vector of the IWOA algorithm when the iteration time t is 0;
step 32-3, calculating the fitness value of the individual of the whale and recording the current optimal fitness value and the position information of the current optimal fitness value;
step 32-4, updating the nonlinear convergence factor a according to the formula (6), updating the adaptive weight omega according to the formula (7), updating the coefficient vector A, C by a, and randomly selecting p, l;
step 32-5, updating the position information of the whale head according to the value of A, and updating the position information according to a formula when p is less than 0.5 in the formula (5) when | A | is more than or equal to 1; when the absolute value A < 1, updating the position information according to the formula (8);
step 32-6, judging whether the iteration number reaches the maximum iteration number, if so, finishing the algorithm; if not, adding 1 to the current iteration number t, and returning to the step 32-3;
and 32-7, outputting the optimal solution to obtain the optimal network parameters to obtain the optimal BP neural network structure, wherein the flow chart is shown in figure 5.
The invention provides an oil paper insulation aging evaluation method based on an IWOA-BP algorithm. The whale optimization algorithm is improved by introducing the nonlinear convergence factor and the self-adaptive weight, so that the convergence speed and the accuracy of the whale optimization algorithm are improved, and the local search and global optimization capability of the algorithm are improved; optimizing the weight and the threshold of the BP neural network through an improved whale optimization algorithm, and improving the problems that the BP neural network is easy to be over-trained, poor in convergence, easy to fall into local optimization and the like; an oil paper insulation model is built according to a PDC method, experiments are carried out, and polarized depolarized current data are measured; and introducing an oil paper insulation aging characteristic quantity of comprehensive conductivity, and performing oil paper insulation aging evaluation based on an IWOA-BP model. Finally, the evaluation result proves that the IWOA-BP evaluation model has certain practicability and certain development space in the aspect of transformer oil paper insulation life evaluation.
The oil paper insulation aging evaluation method based on the IWOA-BP algorithm has the following advantages:
1. the transformer oiled paper insulation aging state evaluation model based on the PDC method and the IWOA-BP model is firstly proposed in the system.
2. Introducing comprehensive conductivity as characteristic quantity for evaluating the insulation aging state of the oil paper;
3. the whale optimization algorithm is improved by adding the nonlinear convergence factor and the adaptive weight, the algorithm precision is improved, the convergence speed of the algorithm is improved, and the local search capability and the global optimization capability of the algorithm are enhanced; and optimizing the weight and the threshold of the BP neural network by using an improved whale optimization algorithm, and improving the convergence and the global optimization capability of the BP neural network.
4. Establishing an oil paper insulation model according to a PDC method, and measuring polarized depolarized current data; the innovation point of the method is that the comprehensive conductivity is introduced to be used as the characteristic quantity for evaluating the aging state of the oil paper insulation, and then the aging state of the oil paper insulation is evaluated by an IWOA-BP algorithm.
Drawings
FIG. 1 is a schematic diagram of the PDC method in the example.
FIG. 2 is a graph of the results of an ideal PDC experiment in the examples.
FIG. 3 is a schematic diagram of a PDC measurement circuit in an embodiment.
FIG. 4 shows the topology of the BP neural network in the embodiment.
FIG. 5 is a flow chart of optimizing BP neural network by IWOA algorithm in the embodiment.
Figure 6 is a graph comparing pdc current data after three days of thermal aging in the examples.
FIG. 7 is a graph comparing the evaluation results of the aging state of the oiled paper insulation in the examples.
Detailed Description
In order to better understand the purpose, structure and function of the present invention, the oil paper insulation aging evaluation method based on IWOA-BP algorithm of the present invention is described in further detail below with reference to the accompanying drawings.
As shown in fig. 1, the invention provides a Prony algorithm in the oil paper insulation Debye parameter identification method based on depolarization current Prony fitting, which utilizes the characteristic of the adaptive sampling interval to perform fitting operation on collected depolarization current data, and further identifies the parameters of a Debye equivalent circuit. The analysis of the low-frequency dielectric loss characteristic of the transformer oil paper insulation based on the polarization-depolarization current method uses the polarization depolarization current method to replace a frequency domain dielectric spectroscopy method to obtain the diagnosis information of the transformer oil paper insulation under the low-frequency condition, and makes insulation oil paper models under different thermal aging degrees and different water contents so as to research the oil paper insulation frequency domain diagnosis information under the polarization depolarization current method. The actual transformer oil paper insulation aging state field evaluation based on the polarization-depolarization current method evaluates the state of a plurality of transformers with different aging conditions under the field condition by the polarization depolarization current testing device, extracts three aging characteristic quantities of comprehensive conductivity, insulating oil conductivity and low-frequency dielectric loss factors to evaluate the transformer oil paper insulation aging, and finally proves the accuracy of the testing result.
Example 1:
step 1: making oiled paper insulation model
Step 11: cutting the oiled paper into four pieces with the same shape, wherein a cavity is reserved between the two pieces;
step 12: connecting four pieces of oiled paper together by using a copper column;
step 13: drying the oil paper model after the manufacture, putting the oil paper insulation model into a vacuum constant temperature aging oven for 24h, wherein the environmental conditions in the oven are as follows: 70 ℃ and 50 Pa;
step 14: placing the dried oil paper model into a container containing transformer oil just submerging the model, and soaking for 24 hours to fully combine the transformer oil with the oil paper, so that the oil paper insulation model is manufactured; carrying out oil immersion treatment on the oil paper insulation model, pouring 41# transformer oil to enable the transformer oil to just submerge the oil paper model, then placing the model in a box, and standing for 24 hours; after standing, the oiled paper insulation model can be put into an experiment;
step 2: and (3) carrying out PDC experiments on the oil paper insulation models in different aging states, and during the experiments, putting the pretreated oil paper insulation models into a vacuum constant-temperature aging box for thermal aging treatment. Accelerated heat aging experiments were carried out at 70 ℃, 90 ℃ and 110 ℃, with experiments at each temperature set being carried out for 7 days. The oiled paper insulation models used for the experiments at different temperatures were all reproduced and pre-treated. During the seven days of accelerated thermal aging of the oiled paper insulation model, one PDC test was performed on the oiled paper insulation model for 24 h. In the PDC experiment, the polarization voltage was set to 100V, the polarization time was set to 1500s, and the depolarization time was set to 3000 s.
Step 21: putting the oiled paper model into an aging box, selecting temperature for aging at regular time, and preparing a new oiled paper insulation model for an aging experiment at different temperatures;
step 22: every 24 hours, the aged oil paper insulation model is tested
Step 23: connecting the instrument to three electrodes of the oil paper insulation model, opening a Labview interface to measure PDC current data
Step 24: recording PDC current data
And step 3: evaluation of insulation aging of oiled paper
Step 31: calculating the comprehensive conductivity
When the polarization time is long enough, the following expression can be used to calculate:
in the formula (1), U 0 Denotes the polarization voltage, C 0 Is the power frequency insulation geometric capacitance of the tested sample 0 Is a vacuum dielectric constant (. epsilon.) 0 ＝8.85×10 -12 F/m)，i p (t end ) Representing the average value of the current at the end of the polarization current, i d (t end ) Represents the average value of the depolarization current end current. In the testing process, the conductivity is unstable, so that the conductivity when the current tends to be stable is selected as the comprehensive conductivity.
Step 32: optimizing BP neural network by IWOA algorithm
Step 33: and (3) based on the optimized BP neural network model, taking PDC current data as input and comprehensive conductivity as output, and performing evaluation prediction on the aging state.
And 4, step 4: and (4) dividing the aging state by combining the comprehensive conductivity by taking different aging time as a standard.
Analysis of experiments
In order to verify the effectiveness of the IWOA-BP algorithm in evaluating the transformer oil paper insulation aging state, firstly, a transformer oil paper insulation model is built by combining the PDC theoretical knowledge, and polarization-depolarization current data is obtained through experiments; secondly, introducing an aging characteristic quantity of comprehensive conductivity as a criterion for evaluating the insulation aging state; and finally, carrying out a simulation experiment by taking the measured polarization-depolarization current as input and the comprehensive conductivity as output through an IWOA-BP algorithm so as to verify the feasibility of the algorithm model.
PDC experiment
The PDC test circuit principle is shown in FIG. 3, and the experimental device comprises an upper computer, a Keithley6517A high-resistance meter, a vacuum constant-temperature aging box, an oil paper insulation model, a GPIB-USB-HS data acquisition card, a coaxial cable and the like.
The partial polarization depolarized current data measured in the PDC experiments are processed to obtain a polarization depolarized current curve as shown in fig. 6.
As can be seen from fig. 6, the higher the set temperature of the aging oven is, the larger the measured initial value of the polarization depolarization current is, and it can be seen that the polarization depolarization current at 110 ℃ is compared with the polarization depolarization current at 90 ℃ and 70 ℃, the initial slope of the polarization depolarization current is large, which indicates that the higher the temperature is, the more serious the oil paper insulation aging is in the aging time of the same time.
2. Oiled paper insulation aging evaluation model
At each set of temperatures, 5 sets of polarization depolarizing current data were measured at each time, for a total of 105 sets of data measured. The measured polarized depolarized current data is used as input data of the IWOA-BP model, and the measured comprehensive conductivity is used as output data of the evaluation model. Simulation experiments were performed with 85 groups as training sets and the remaining 20 groups as test sets. And comparing the WOA-BP model with the BP neural network model to verify the effectiveness and accuracy of the IWOA-BP model. The evaluation result of the aging state of the oil paper insulation is shown in fig. 7, and as can be seen from fig. 7, compared with the evaluation results of the BP model and the WOA-BP model, the evaluation result obtained by the IWOA-BP model is closer to an actual value, which shows that the model error is smaller and more accurate than the other two models, and the effectiveness of the model is fully reflected.
In addition, three evaluation indexes are selected to evaluate the prediction result in order to reflect the accuracy and the effectiveness of the prediction result. The three evaluation indexes are: mean relative error (MAPE), mean absolute error (MAD), and Root Mean Square Error (RMSE). The expression is as follows:
in the formula: in the above formula, y i Is the ith actual value;
is as followsi predicted values; and n is the total number of data.
The evaluation results are shown in table 1.
Table 1.
As can be seen from Table 1, the IWOA-BP model has only 9.0456% of average relative error control for the evaluation results, lower than 10%, and lower average absolute error and root mean square error compared to the WOA-BP and BP models. The evaluation result proves that the IWOA-BP model has more accurate evaluation effect and more applicability to the aging state of the oiled paper insulation. Therefore, the IWOA-BP model has certain feasibility for being used as a transformer oil paper insulation aging state evaluation model.
Table 2.
Finally, as shown in Table 2, the overall conductivity was higher than 3X 10 for the evaluation results -13 S/m, which can be evaluated as severe aging; for a combined conductivity higher than 1X 10 -13 S/m, less than 3X 10 -13 S/m, which can be evaluated as moderate aging; for overall conductivity below 1X 10 -13 S/m, which can be evaluated as light aging. Subsequent studies can focus on the severity of aging, and make a life assessment of the oiled paper insulation.
It is to be understood that the present invention has been described with reference to certain embodiments, and that various changes in the features and embodiments, or equivalent substitutions may be made therein by those skilled in the art without departing from the spirit and scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed, but that the invention will include all embodiments falling within the scope of the appended claims.
Claims (5)
1. An IWOA-BP algorithm-based oiled paper insulation aging assessment method is characterized by comprising the following steps which are sequentially carried out:
step 1: manufacturing an oiled paper insulation model;
step 2: carrying out PDC experiments on the oil paper insulation models in different aging states;
and step 3: performing oiled paper insulation aging evaluation;
step 31: calculating the comprehensive conductivity
When the polarization time is long enough, the following expression can be used to calculate:
in formula (9), U 0 Denotes the polarization voltage, C 0 Is the power frequency insulation geometric capacitance of the tested sample 0 Is a vacuum dielectric constant (. epsilon.) 0 ＝8.85×10 -12 F/m)，i p (t end ) Representing the average value of the current at the end of the polarization current, i d (t end ) Represents the average value of the depolarization current terminal current; in the testing process, the conductivity is unstable, so that the conductivity when the current tends to be stable is selected as the comprehensive conductivity;
step 32: optimizing the BP neural network by an IWOA algorithm;
step 33: based on the optimized BP neural network model, PDC current data is used as input, and comprehensive conductivity is used as output, and the aging state is evaluated and predicted;
and 4, step 4: and (4) dividing the aging state by combining the comprehensive conductivity by taking different aging time as a standard.
2. The IWOA-BP algorithm-based oiled paper insulation aging assessment method according to claim 1, wherein the step 1 specifically comprises the following steps, and the following steps are sequentially performed:
step 11: cutting the oiled paper into four pieces with the same shape, wherein a cavity is reserved between the two pieces;
step 12: connecting four pieces of oiled paper together by using a copper column;
step 13: drying the manufactured oiled paper model, and putting the oiled paper model into a vacuum ageing oven for vacuum heating for 24 hours;
step 14: and (3) putting the dried oil paper model into a container containing transformer oil which just submerges the model, and soaking for 24 hours to fully combine the transformer oil and the oil paper, so that the oil paper insulation model is manufactured.
3. The IWOA-BP algorithm-based oiled paper insulation aging assessment method according to claim 1, wherein the step 2 specifically comprises the following steps, and the following steps are sequentially performed:
step 21: putting the oiled paper model into an aging box, selecting temperature for aging at fixed time, and preparing a new oiled paper insulation model for an aging experiment at different temperatures;
step 22: carrying out an experiment on the aged oil paper insulation model every 24 hours;
step 23: connecting the instrument to three electrodes of the oilpaper insulation model, and opening a Labview interface to measure PDC current data;
step 24: PDC current data were recorded.
4. The method for evaluating the aging of the oiled paper insulation based on the IWOA-BP algorithm according to claim 1, wherein the IWOA algorithm used in the step 32 is implemented by the following steps:
1) surrounding target prey
The position of the whale with the closest position to the prey is taken as the optimal solution, the rest whales are gradually updated to the position of the optimal solution by taking the whales close to the optimal solution, and therefore the prey surrounding effect is achieved, and the applied mathematical model is as follows:
in the formula (1)
Representing the current optimal position of the whale with standing head, X (t) representing the current position of the whale with standing head, t representing the current iteration times, A and C being coefficient vectors, r 1 And r 2 Represents a [0,1]]A is a convergence factor that decreases linearly from 2 to 0 in the iterative process, and the expression of a is as follows:
t in formula (2) max Represents the maximum number of iterations;
2) spiral bubble net attack
The mathematical model of the spiral movement of the whale head is as follows:
in the formula (3), D represents the distance between the current whale position and the optimal solution position, b is a spiral shape constant, and l represents a random number between [ -1,1 ];
judging whether the position is updated spirally or the position is shrunk and surrounded by the p value, wherein the mathematical model is as follows:
wherein p is a random number between [0,1 ];
3) target hunting search and locking
When the coefficient vector | A | > is more than or equal to 1, changing the reference of the updating position of the standing whale from the optimal standing whale position to a randomly selected standing whale position, and further carrying out global optimization; when the coefficient vector | A | is less than 1, the whale still carries out position updating with the optimal solution position, and local optimization is carried out, wherein the mathematical model is as follows:
in formula (5), X rand (t) indicates the position of a random whale;
a nonlinear convergence factor a is introduced, and the mathematical model of the nonlinear convergence factor a is as follows:
a＝(2-2t 2 /t max 2 )(1-t 3 /t max 3 ) (6)
changing the inertia weight changing linearly into the adaptive weight changing nonlinearly, and the mathematical model is as follows:
the whale location updating mathematical model with the introduction of the adaptive weight factor is as follows:
5. the IWOA-BP algorithm-based oiled paper insulation aging assessment method according to claim 4, wherein the step 32: the specific implementation steps for optimizing the BP neural network by an IWOA algorithm are as follows:
step 32-1, initializing a BP neural network, determining a topological structure of the network, and initializing weights and thresholds in network parameters;
step 32-2, initializing an IWOA algorithm, setting the population number N of the whale with beluga and the maximum iteration time t max Taking the weight and the threshold of the BP neural network as the position vector of the IWOA algorithm when the iteration time t is 0;
step 32-3, calculating the fitness value of the individual of the whale and recording the current optimal fitness value and the position information of the current optimal fitness value;
step 32-4, updating the nonlinear convergence factor a according to the formula (6), updating the adaptive weight omega according to the formula (7), updating the coefficient vector A, C by a, and randomly selecting p, l;
step 32-5, updating the position information of the whale according to the value of A, and updating the position information according to a formula when p is less than 0.5 in the formula (5) when the absolute value of A is more than or equal to 1; when the absolute value A < 1, updating the position information according to the formula (8);
step 32-6, judging whether the iteration number reaches the maximum iteration number, if so, finishing the algorithm; if not, adding 1 to the current iteration number t, and returning to the step 32-3;
and 32-7, outputting the optimal solution to obtain optimal network parameters and obtain an optimal BP neural network structure.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110664278.9A CN113514739B (en) | 2021-06-16 | 2021-06-16 | IWOA-BP algorithm-based oil paper insulation aging evaluation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110664278.9A CN113514739B (en) | 2021-06-16 | 2021-06-16 | IWOA-BP algorithm-based oil paper insulation aging evaluation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113514739A CN113514739A (en) | 2021-10-19 |
| CN113514739B true CN113514739B (en) | 2022-09-06 |
Family
ID=78065637
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110664278.9A Active CN113514739B (en) | 2021-06-16 | 2021-06-16 | IWOA-BP algorithm-based oil paper insulation aging evaluation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113514739B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114184909B (en) * | 2021-11-23 | 2023-07-28 | 国网福建省电力有限公司莆田供电公司 | Oil paper insulation micro-water content assessment method based on particle swarm optimization confidence rule base |
| CN114264921B (en) * | 2021-11-23 | 2023-07-28 | 国网福建省电力有限公司莆田供电公司 | Oil paper insulation aging state assessment method based on extended confidence rule base |
| CN114236331A (en) * | 2021-12-03 | 2022-03-25 | 广西电网有限责任公司电力科学研究院 | Transformer insulation state identification method and system based on neural network and fingerprint database |
Family Cites Families (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9274150B2 (en) * | 2012-12-09 | 2016-03-01 | General Electric Company | Systems for synchrophasor data management |
| CN103197171B (en) * | 2013-03-01 | 2015-09-09 | 中国南方电网有限责任公司超高压输电公司贵阳局 | Based on the method for the characteristic parameter assessment oil paper insulation ageing state of time domain dielectric response |
| CN108431613A (en) * | 2015-12-01 | 2018-08-21 | 通用电器技术有限公司 | The intelligent assessment method of the paper insulated major insulation situation of transformer oil |
| CN107991584A (en) * | 2017-08-02 | 2018-05-04 | 云南电网有限责任公司昆明供电局 | A kind of transformer capacitor formula casing insulation ag(e)ing test method based on polarization/depolarization current |
| CN107860980A (en) * | 2017-10-17 | 2018-03-30 | 武汉泽电新材料有限公司 | A kind of time-frequency domain combines quick dielectric response method of testing |
| CN107831415B (en) * | 2017-10-20 | 2020-02-04 | 广东电网有限责任公司河源供电局 | Interval value fuzzy set method for transformer insulation paper aging state evaluation |
| CN107831300B (en) * | 2017-10-20 | 2020-02-04 | 广东电网有限责任公司河源供电局 | Transformer insulating oil degradation evaluation method based on three-dimensional trapezoidal probability fuzzy set |
| CN109001598B (en) * | 2018-05-31 | 2020-09-01 | 重庆大学 | Method for evaluating oil paper insulation aging state of transformer by applying depolarization electric quantity growth rate |
| CN108872820B (en) * | 2018-08-02 | 2020-10-16 | 国网吉林省电力有限公司电力科学研究院 | Method and system for evaluating insulation aging state of oil impregnated paper in high-voltage current transformer |
| US11436479B2 (en) * | 2018-08-09 | 2022-09-06 | International Business Machines Corporation | System and method for transfer of analog synaptic weight information onto neuromorphic arrays with non-ideal non-volatile memory device |
| CN110889228B (en) * | 2019-11-28 | 2023-04-18 | 国网吉林省电力有限公司电力科学研究院 | Transformer oil paper insulation aging prediction method based on chicken swarm optimization BP neural network |
| CN111680831A (en) * | 2020-05-25 | 2020-09-18 | 广东电网有限责任公司广州供电局 | Substation power utilization analysis system and method |
| CN112098469B (en) * | 2020-09-15 | 2023-06-30 | 北京安赛博技术有限公司 | Soil conductivity detection system and method |
| CN111948502B (en) * | 2020-09-15 | 2022-05-24 | 中国南方电网有限责任公司超高压输电公司广州局 | Method for evaluating oil paper insulation aging state of converter transformer |
| CN112505494B (en) * | 2020-10-30 | 2022-05-03 | 西安交通大学 | Method and device for evaluating insulation water content of oiled paper |
-
2021
- 2021-06-16 CN CN202110664278.9A patent/CN113514739B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113514739A (en) | 2021-10-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113514739B (en) | IWOA-BP algorithm-based oil paper insulation aging evaluation method | |
| CN110889228B (en) | Transformer oil paper insulation aging prediction method based on chicken swarm optimization BP neural network | |
| CN113049962B (en) | LSTM-based energy storage device operation situation deduction method | |
| CN108846227B (en) | Lithium ion battery capacity degradation prediction and evaluation method based on random forest and capacity self-recovery effect analysis | |
| CN114970332B (en) | Lithium battery model parameter identification method based on chaotic quantum sparrow search algorithm | |
| CN112580284B (en) | Hybrid capacitor equivalent circuit model and online parameter identification method | |
| CN113901707A (en) | SOH time sequence prediction method for vehicle lithium ion power battery | |
| CN109613446A (en) | A kind of aging of lithium battery detection method based on time series analysis | |
| CN112287520B (en) | Oil paper insulation expansion Debye model parameter identification method based on two-step optimization method | |
| CN112288147A (en) | Method for predicting insulation state of generator stator by BP-Adaboost strong predictor | |
| CN114371374A (en) | Cable insulation electric heating combined aging degree estimation method and system | |
| CN111896849B (en) | Oil paper insulation performance analysis method and device considering non-uniform medium | |
| CN116449219A (en) | Lithium battery model parameter identification method for improving VFFRLS | |
| Yang et al. | Health Status Prediction of Lithium Battery Based on LSTM Model with Optimization Algorithms | |
| Ding et al. | NSGA-II model-based dielectric frequency response parameters for aging and moisture analysis of transformer insulation | |
| CN116224109A (en) | Method for rapidly measuring capacity of lithium ion battery | |
| CN116127858A (en) | GIS equipment temperature rise prediction method and system based on improved sand cat algorithm optimization | |
| CN114280490B (en) | Lithium ion battery state of charge estimation method and system | |
| CN115329672A (en) | Converter transformer state reliability assessment method based on improved TOPSIS gray correlation | |
| CN115600066A (en) | Oil paper insulation moisture degree accurate evaluation method based on time domain dielectric spectrum | |
| Jiang et al. | Reinforcement Learning Based Genetic Algorithm for Aging State Analysis of Insulating Paper at Transformer Hotspot | |
| CN114543896A (en) | Capacitive equipment medium water content and aging evaluation method based on temperature drift electrical parameters | |
| Sun et al. | State of health estimation for lithium-ion batteries based on current interrupt method and genetic algorithm optimized back propagation neural network | |
| CN113283176B (en) | Optimization method and system for dielectric response extended debye model | |
| CN117669103A (en) | Novel intelligent calculation method for broadband oscillation damping coefficient of power 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |