A multi-factor integrated defect diagnosis method for dry-type iron core reactors

Abstract

The present invention provides a multi-factor integrated dry-type iron core reactor defect diagnosis method, which relates to the field of power system fault diagnosis, including: obtaining temperature field and structural field simulation models of dry-type iron core reactors to obtain temperature monitoring points and vibration monitoring points; obtaining temperature distribution characteristics after magnetic field, fluid and temperature multi-physics field coupling simulation based on temperature monitoring points and vibration monitoring points, and correcting parameters in vibration simulation according to temperature distribution characteristics to obtain vibration distribution characteristics, jointly establishing defect sample data sets with temperature distribution characteristics and vibration distribution characteristics, and dividing the defect sample data sets into training sets, verification sets and test sets; constructing an initial defect diagnosis model, and outputting defect types after training, adjusting parameters and testing the initial defect diagnosis model. The present invention comprehensively considers the influence of multiple indicators of temperature and vibration on defect diagnosis, and improves the accuracy of dry-type iron core reactor defect diagnosis.

Description

Technical Field

[0001] The invention relates to the field of power system fault diagnosis, and in particular to a multi-factor integrated defect diagnosis method for dry-type iron core reactors. Background Art

[0002] Dry-type iron core reactors are very important power transmission and transformation equipment in power systems, and are mainly used in low-voltage reactive power compensation cabinets. Because they use air cooling and have a relatively harsh operating environment, the external environment will cause insulation aging of dry-type iron core reactors, further affecting their stability. Fault diagnosis of dry-type iron core reactors is of great significance in helping to improve their reliability to meet the needs of high-quality power grids.

[0003] The temperature and vibration distribution during stable operation are important indicators of the operating status of dry-type air-core reactors. For example, inter-turn short-circuit defects can cause local temperature rise and local vibration intensification of dry-type reactors. At present, there are few studies on defect diagnosis and early warning algorithms for dry-type iron-core reactors, and they have certain limitations. Most of the defect diagnosis methods proposed by current research institutes are based on simple threshold judgment methods, relative comparison methods, and similar comparison methods. In terms of data acquisition, these methods often use simple two-dimensional simulation models to obtain defect diagnosis data, and do not consider the impact of temperature changes on material parameters in vibration simulation, and do not perform coupling simulation between the two, making the subsequent data samples used for defect diagnosis inaccurate; in terms of factors, these methods often only consider the impact of a single factor, and cannot timely warn of possible defects. It is difficult to detect early local defects in time and it is difficult to locate the specific defect location, resulting in low diagnostic accuracy of dry-type iron-core reactors.

[0004] Therefore, it is urgent to provide a solution to improve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a multi-factor integrated defect diagnosis method for dry-type iron core reactors, so as to improve the problem that the prior art cannot timely warn of possible defects, is difficult to timely discover early local defects and is difficult to locate the specific defect position, resulting in low accuracy of defect diagnosis of dry-type iron core reactors.

[0006] The present invention provides a multi-factor integrated dry-type iron core reactor defect diagnosis method, comprising the following steps:

[0007] Acquire a temperature field and a structural field simulation model of a dry-type iron core reactor based on physical characteristics, acquire a temperature distribution curve and a vibration distribution curve of the dry-type iron core reactor based on the simulation model, and acquire a temperature monitoring point and a vibration monitoring point of the dry-type iron core reactor based on the temperature distribution curve and the vibration distribution curve;

[0008] After performing a magnetic field, fluid and temperature multi-physics field coupling simulation based on the temperature monitoring points and the vibration monitoring points, the temperature distribution characteristics are obtained, and after correcting the material parameters in the vibration simulation according to the temperature distribution characteristics, the vibration distribution characteristics are obtained, and a defect sample data set is established by combining the temperature distribution characteristics and the vibration distribution characteristics, and the defect sample data set is divided into a training set, a verification set and a test set;

[0009] Construct an initial defect diagnosis model, train, adjust parameters and test the initial defect diagnosis model based on the training set, validation set and test set to obtain a trained prediction model, input the defect sample data set into the trained prediction model to output the defect type.

[0010] The beneficial effects of a multi-factor fusion dry-type iron core reactor defect diagnosis method provided by the present invention are: first, multi-physics coupling simulation is used to simulate the operating state of the dry-type iron core reactor when it is defective, and temperature and vibration defect data under these states are obtained as defect sample data sets. On the basis of the defect sample data sets, a multi-factor fusion dry-type iron core reactor defect diagnosis model is constructed; secondly, the parameters of the defect diagnosis model are optimized to form an optimal defect diagnosis model, and the weak position of the dry-type iron core reactor is determined by combining the temperature field and structural field simulation results, and then the vibration and temperature monitoring points are confirmed, and sensors are placed at the positions; finally, the vibration data and temperature data monitored by these sensors are used as input samples of the optimal defect diagnosis model, and the corresponding defect type is finally predicted more accurately.

[0011] Optionally, the process of obtaining the temperature monitoring point and the vibration monitoring point of the dry-type iron core reactor includes:

[0012] A temperature detection path is set in the axial direction and radial direction on the three-phase coil of the dry-type iron core reactor, a temperature distribution curve of the temperature change with displacement on the temperature detection path is obtained based on the temperature sensor, and a position corresponding to the highest temperature is obtained as a temperature monitoring point based on the temperature distribution curve;

[0013] The vibration displacement distribution curve of the core and the coil of the dry-type iron core reactor in the X, Y and Z axis directions is obtained based on the vibration sensor, and the position with the maximum vibration displacement is obtained as the vibration monitoring point based on the vibration displacement distribution curve.

[0014] Optionally, the temperature monitoring points are deployed at the lower, middle and upper parts of the three-phase coil, which are 13.3%, 49.7% and 84.4% of the length of the three-phase coil from bottom to top, respectively, and the number of temperature monitoring points and vibration monitoring points are both 9.

[0015] Optionally, when modifying material parameters in vibration simulation according to temperature distribution characteristics, Young's modulus and electrical conductivity of copper material are modified, where:

[0016] After correction, the expression of Young's modulus is:

[0017] = E 0 [ 1 − m α ( T − T 0 ) ] ;

[0018] Where E is the modified Young's modulus, is the initial temperature of the copper material, is Young's modulus, is the linear expansion coefficient, is the temperature of each structural component, is the ratio of the temperature coefficient of Young's modulus to the coefficient of linear expansion;

[0019] After correction, the expression of copper material conductivity is:

[0020] ;

[0021] in, is the conductivity of the corrected copper material, is the real-time temperature of the copper material.

[0022] Optionally, output the defect type process, including:

[0023] Randomly sampling from the defect sample data set to obtain N defect sample subsets, wherein the defect sample subsets include M feature quantities;

[0024] A corresponding decision tree is established for each defect sample subset, and m feature dimensions are randomly selected from the M feature quantities, and the feature dimensions are input into each decision tree to obtain N prediction results, where m≤M;

[0025] Voting is performed on the N prediction results to obtain the prediction result with the highest vote as the defect type.

[0026] Optionally, the basis for determining the defect type consists of a first determination condition and a second determination condition, the first determination condition being: when the hot spot temperature of the dry-type iron core reactor is lower than 155°C and the maximum vibration amplitude of its clamp is lower than the vibration peak value, the operating state of the dry-type iron core reactor is normal, otherwise there is overheating or mechanical defects, and the second determination condition being: when the hot spot temperature rise is lower than 95K and the maximum vibration amplitude of the reactor core or winding is lower than the vibration peak value, the operating state of the dry-type iron core reactor is normal, otherwise there is overheating or mechanical defects, wherein the first determination condition is equivalent to the second determination condition.

[0027] Optionally, the process of adjusting the parameters of the initial defect diagnosis model includes: obtaining an objective function based on the difference between the actual defect type value and the predicted value of the dry-type iron core reactor, and iteratively adjusting the parameters of the initial defect diagnosis model based on the verification set until the iterative adjustment is stopped when the objective function remains unchanged.

[0028] Optionally, the objective function is expressed as:

[0029] ;

[0030] in, is the objective function, is the actual defect type value of the reactor, is the predicted value of the random forest model, is the number of samples in the defect sample set, is an indicative function. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is an overall technical scheme diagram of the multi-factor integrated dry-type iron core reactor defect diagnosis method provided by the present invention;

[0032] Figure 2 It is a flow chart of the multi-factor integrated dry-type iron core reactor defect diagnosis method provided by the present invention;

[0033] Figure 3 It is a three-dimensional simulation model of a dry-type iron core reactor provided by the present invention;

[0034] Figure 4 It is a temperature distribution diagram of each component of the dry-type iron core reactor provided by the present invention;

[0035] Figure 5 It is a vibration displacement distribution diagram of each component of the dry-type iron core reactor provided by the present invention;

[0036] Figure 6 It is a diagram showing the arrangement of temperature and vibration monitoring points of a dry-type iron core reactor provided by the present invention;

[0037] Figure 7 It is a schematic diagram of the temperature and vibration displacement distribution of the inter-turn short-circuit defect of the dry-type iron core reactor provided by the present invention;

[0038] Figure 8 This is a defect diagnosis effect diagram of the dry-type iron core reactor provided by the present invention.

[0039] Reference numerals: core 1 , clamp 2 , coil 3 , base 4 . DETAILED DESCRIPTION

[0040] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Unless otherwise defined, the technical terms or scientific terms used herein should be understood by people with general skills in the field to which the present invention belongs. "Including" and similar words used in this article mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects.

[0041] In some embodiments, see Figure 1 The present invention provides a multi-factor integrated dry-type iron core reactor defect diagnosis method using the following technical solutions:

[0042] S1. Acquire a temperature field and a structural field simulation model of a dry-type iron core reactor based on physical characteristics, acquire a temperature distribution curve and a vibration distribution curve of the dry-type iron core reactor based on the simulation model, and acquire a temperature monitoring point and a vibration monitoring point of the dry-type iron core reactor based on the temperature distribution curve and the vibration distribution curve;

[0043] S2, obtaining temperature distribution characteristics after performing magnetic field, fluid and temperature multi-physical field coupling simulation based on the temperature monitoring points and vibration monitoring points, and obtaining vibration distribution characteristics after correcting material parameters in vibration simulation according to the temperature distribution characteristics, establishing a defect sample data set by combining the temperature distribution characteristics and the vibration distribution characteristics, and dividing the defect sample data set into a training set, a validation set and a test set;

[0044] S3. Construct an initial defect diagnosis model, train, adjust parameters and test the initial defect diagnosis model based on the training set, validation set and test set to obtain a trained prediction model, input the defect sample data set into the trained prediction model to output the defect type.

[0045] In some embodiments, see Figure 3 , represents the three-dimensional simulation model of the dry-type iron core reactor provided by the present invention, which specifically includes: an iron core 1, a clamp 2, a coil 3 and a base 4.

[0046] In some embodiments, when executing step S1 to obtain the temperature field and structural field simulation model of the dry-type iron core reactor, due to the different physical properties considered in the temperature-flow field coupling simulation and the structural field simulation of the dry-type iron core reactor, the simplified conditions considered in the temperature field and structural field simulation modeling are different. For example, in the temperature field simulation, structural components that have little effect on the thermal conduction of the dry-type iron core reactor, such as the core air gap and coil interlayer struts, are simplified and removed. However, for the structural field simulation of the dry-type iron core reactor, the influence of the core air gap on the magnetic field of the dry-type iron core reactor and the influence of the coil interlayer struts on the electromagnetic vibration of the dry-type iron core reactor cannot be ignored. In addition, in the present invention, the reactor specifically refers to a dry-type iron core reactor.

[0047] In some embodiments, the process of obtaining the temperature monitoring point and the vibration monitoring point of the dry-type iron core reactor in step S1 includes:

[0048] S1-1. A temperature detection path is set in the axial direction and radial direction on the three-phase coil of the dry-type iron core reactor, respectively. A temperature distribution curve of temperature variation with displacement on the temperature detection path is obtained based on a temperature sensor, and a position corresponding to the highest temperature is obtained based on the temperature distribution curve as a temperature monitoring point;

[0049] S1-2. Obtain the vibration displacement distribution curve of the core and coil of the dry-type iron core reactor in the X, Y, and Z axis directions based on the vibration sensor, and obtain the position with the maximum vibration displacement as the vibration monitoring point based on the vibration displacement distribution curve.

[0050] In some embodiments, when executing step S1-1, a temperature detection path is set on the three-phase coil of the dry iron core reactor in the axial direction and radial direction respectively, which are path 1 and path 2 respectively. According to the display of the temperature sensor, the temperature distribution curve of path 1 shows that the distribution of the three-phase coil is regular and symmetrical, and the inner coil is higher than the outer coil.

[0051] Furthermore, according to the display of the temperature sensor, the temperature distribution curve of path 2 shows that the coil temperature increases first and then slowly decreases from the bottom to the top, and the highest temperature is close to the top of the coil and lower than the top position. It can be seen that the temperature change trend of each layer of coil of the dry-type iron core reactor in the axial direction is basically the same, and the radial temperature change trend is larger.

[0052] For further information, see Figure 6 , Figure 6 The diagram shows the arrangement of the temperature monitoring points and vibration monitoring points. Each phase coil is divided into three areas for temperature monitoring and vibration monitoring. The temperature monitoring points and vibration monitoring points (represented by circles in the figure) are deployed at the bottom, middle, and top of the three-phase coil, which are 13.3%, 49.7%, and 84.4% of the length of the three-phase coil from bottom to top, respectively. There are 9 temperature monitoring points and 9 vibration monitoring points each.

[0053] Specifically, the same number of vibration sensors are placed on the other side of the dry-type iron core reactor coil temperature monitoring point and on the core column to perform vibration monitoring on the dry-type iron core reactor.

[0054] Furthermore, the three-phase coil is divided into A phase, B phase and C phase.

[0055] In some embodiments, when executing step S1-2 to obtain the vibration displacement distribution curve of the core and coil of the dry-type iron core inductor in the X, Y, and Z axis directions based on the vibration sensor, the vibration displacement distribution curve shows that the vibration displacement of the core and coil of the dry-type iron core inductor is relatively evenly distributed in the X and Y directions, and the vibration displacement distribution is mainly concentrated in the component in the Z direction.

[0056] See also Figure 4 , Figure 4 (a) shows the temperature distribution of the dry-type iron core reactor. Figure 4 (b) shows the temperature distribution of the core and the core clamp. Figure 4 (c) shows the temperature distribution of the inner coil. Figure 4 (d) shows the outer coil distribution, see Figure 5 , Figure 5 (a) shows the vibration displacement of the dry-type iron core reactor. Figure 5 (b) shows the core vibration displacement distribution. Figure 5 (c) in the figure shows the coil vibration displacement distribution.

[0057] according to Figure 4 and Figure 5 It can be seen that the temperature of the dry-type iron-core reactor is mainly concentrated in the inner coil, with the maximum temperature reaching 80.6℃; the vibration of the dry-type iron-core reactor is mainly concentrated on the iron core column corresponding to A, with the maximum vibration displacement of 4.6×10-3 mm. At the same time, the winding cannot be ignored in the study of the vibration distribution of the dry-type iron-core reactor. The vibration displacement of the coil corresponding to A is relatively large, reaching 1.33×10-4 mm.

[0058] In some embodiments, when executing step S2 to correct the material parameters in the vibration simulation according to the temperature distribution characteristics, it includes correcting the Young's modulus and the electrical conductivity of the copper material, wherein:

[0059] After correction, the expression of Young's modulus is:

[0060] = E 0 [ 1 − m α ( T − T 0 ) ] ;

[0061] Where E is the modified Young's modulus, is the initial temperature of the copper material, is Young's modulus, is the linear expansion coefficient, is the temperature of each structural component, is the ratio of the temperature coefficient of Young's modulus to the linear expansion coefficient, .

[0062] After correction, the expression of copper material conductivity is:

[0063] ;

[0064] in, is the conductivity of the corrected copper material, is the real-time temperature of the copper material.

[0065] Specifically, in the vibration simulation of the dry-type iron-core reactor, the temperature rise change has a certain influence on the material parameters of the dry-type iron-core reactor structural parts. Among them, the linear change of Young's modulus and conductivity with temperature on the vibration simulation results cannot be ignored. In order to obtain the vibration distribution characteristics of the dry-type iron-core reactor, it is necessary to indirectly couple the temperature rise and vibration. Therefore, it is necessary to integrate the temperature calculation results to correct the Young's modulus and conductivity of each component of the dry-type iron-core reactor.

[0066] Furthermore, except for copper in the dry-type reactor material, the conductivity of other components changes with temperature and has little effect on the vibration simulation results. The present invention mainly considers the temperature response of the conductivity of the copper material, and therefore corrects the conductivity of the copper material of the dry-type iron core reactor.

[0067] In fact, during the operation of dry-type iron-core reactors, the time-varying magnetic field will produce hysteresis loss and eddy current loss, causing the temperature of the dry-type iron-core reactor to rise. The heat transfer process of the dry-type iron-core reactor mainly includes heat conduction and heat convection. The temperature field of the dry-type iron-core reactor is simulated using the magnetic field-fluid-temperature multi-physics field coupling interface.

[0068] In some embodiments, the process of outputting the defect type in step S3 is implemented based on a random forest algorithm, and specifically includes:

[0069] S3-1. Randomly sample from the defect sample data set to obtain N defect sample subsets, where the defect sample subsets include M feature quantities, and the random sampling is random sampling without replacement;

[0070] S3-2, establishing a corresponding decision tree for each defect sample subset, and randomly selecting m feature dimensions from the M feature quantities, and inputting the feature dimensions into each decision tree to obtain N prediction results, where m≤M;

[0071] S3-3. Vote for the N prediction results to obtain the prediction result with the highest vote as the defect type.

[0072] Specifically, the process of the random forest algorithm diagnosis model is as follows: first, the reactor detection sample data is input, and multiple defect sample subsets are obtained with replacement sampling, including: defect sample subset 1, defect sample subset 2, , defect sample subset , and establish a corresponding decision tree for each defect sample subset. The decision tree contains multiple root nodes and leaf nodes. The defect sample subset is input into each corresponding decision tree. Starting from the root node, the defect sample data is judged according to the splitting rule to determine whether it is greater than the corresponding standard data to determine which child node the sample should enter, and the corresponding child node is created. This process will continue until a leaf node is reached. Finally, the reactor detection sample data is input into each decision tree and output The candidate prediction results are selected, and the most frequently appearing candidate prediction results (i.e. The mode of the alternative prediction results is taken as the final diagnosis result.

[0073] Specifically, when executing step S3-1, the defect sample data set is derived from the temperature and vibration data of the dry-type iron core reactor when it is running at different ambient temperatures obtained by multi-physics field coupling simulation calculation. Taking the turn-to-turn short circuit in the middle of the outer coil of phase A of the dry-type iron core reactor as an example, the ambient temperature is set to 20°C.

[0074] See also Figure 7 , Figure 7 (a) in the figure shows the temperature change of the dry-type iron core reactor. Figure 7 (b) in the figure shows the temperature change of the core and the core clamp. Figure 7 (c) in the figure shows the temperature change of the inner coil. Figure 7 (d) in the figure shows the temperature change of the outer coil. Figure 7 (e) in the figure indicates the vibration release of the dry-type iron core reactor.

[0075] Depend on Figure 7 It can be obtained that the hot spot is located in the short-circuit defect area, the temperature rise is 89 ℃, which is close to the temperature rise limit, and the temperature rise of other areas of the dry-type iron-core reactor is greatly increased compared with the dry-type iron-core reactor under normal operating conditions. The maximum vibration displacement of the dry-type iron-core reactor under this defect is 90μm.

[0076] According to different defect conditions (mainly including local overheating or abnormal vibration caused by inter-turn short circuit or excessive magnetic leakage, local overheating caused by airway blockage, abnormal vibration caused by loose bolts, etc.), combined with the temperature and vibration simulation analysis of the dry-type iron core reactor, a total of 200 sets of defect sample data were obtained, as shown in Table 1.

[0077] Table 1

[0078] Phase Measuring point location Ambient temperature / ℃ Hot spot temperature / ℃ Measuring point temperature / ℃ Measuring point vibration displacement / μm A upper part 10 107 106.58 508 A Central 20 118 116.41 1175 B lower part 10 106 74.17 0.48 C lower part 12 80.6 74.5 2.1 B Central 0 90 66.17 3.1 C upper part 40 107.11 136 7

[0079] Furthermore, when executing step S3-2, the number of decision trees is equal to the number of defect sample subsets, both of which are N.

[0080] Furthermore, when executing step S3-3, the expression of the highest vote prediction result is:

[0081] ;

[0082] in, is the input sample, is the indicative function, is a single decision tree, The predicted result for the highest vote is is the number of defect sample sets, Represents all vote prediction results.

[0083] In some embodiments, see Figure 2 First, a simulation model is constructed, and boundary conditions are set for temperature simulation. The material parameters of the temperature simulation are used to correct the material parameters of the vibration simulation to obtain the temperature and vibration distribution characteristics. Then, data sampling is performed according to the temperature and vibration distribution characteristics to obtain a defect sample data set. The defect sample data set is divided into a training set and a validation set. The training set is input into the established initial random forest model for training, and the parameters of the random forest model are optimized and adjusted based on the validation set to obtain the prediction model to be detected. Finally, the positions of the temperature monitoring points and the vibration monitoring points are determined according to the temperature and vibration distribution characteristics, and sensors (including temperature and vibration sensors) are installed at the positions to obtain the test set. The test set is used to perform accuracy detection on the prediction model to be detected to obtain the optimal diagnosis model, and the diagnosis results are output according to the optimal diagnosis model.

[0084] In some embodiments, when executing step S3, the basis for determining the defect type is composed of a first determination condition and a second determination condition, the first determination condition being: when the hot spot temperature of the dry-type iron core inductor is lower than 155°C and the maximum vibration amplitude of its clamp is lower than the vibration peak value, the operating state of the dry-type iron core inductor is normal, otherwise there is overheating or mechanical defects, the second determination condition being: when the hot spot temperature rise is lower than 95K and the maximum vibration amplitude of the core or winding of the dry-type iron core inductor is lower than the vibration peak value, the operating state of the dry-type iron core inductor is normal, otherwise there is overheating or mechanical defects, wherein the first determination condition is equivalent to the second determination condition.

[0085] Specifically, the vibration peak is 100 μm.

[0086] In some embodiments, when executing step S3 to adjust the parameters of the initial defect diagnosis model, it includes: obtaining the objective function based on the difference between the actual defect type value and the predicted value of the dry-type iron core reactor, and iteratively adjusting the parameters of the initial defect diagnosis model based on the verification set until the objective function remains unchanged and the iterative adjustment is stopped.

[0087] Specifically, the expression of the objective function is:

[0088] ;

[0089] in, is the objective function, is the actual defect type value of the dry-type iron core reactor, is the predicted value of the random forest model, is the number of samples in the defect sample set, is an indicative function.

[0090] In some embodiments, when executing step S3, the defect type diagnosis effect diagram is as follows: Figure 8 As shown, including real faults and diagnosed faults, Figure 8 It can be seen that the diagnostic accuracy of turn-to-turn short circuit and airway blockage reached 100% in all diagnostic times, among which the loose clamps and normal defect rates were as high as 87.5%. The defect diagnosis accuracy of dry-type iron core reactors using the multi-factor fusion algorithm reached 93.75%, and the defect diagnosis effect was good.

[0091] Furthermore, the present invention compares the influence of different indicators on the defect diagnosis results of dry-type iron core reactors, which mainly include three categories: indicator A: only using temperature indicator; indicator B: only using vibration indicator; indicator C: using both temperature and vibration indicators.

[0092] Common evaluation criteria used are: Accuracy (Acc), Precision (P), Recall (R) and F1 score (F1 Score).

[0093] The impact of the number of monitoring points on model performance is shown in Table 2.

[0094] Table 2

[0095] index Acc P R F1 A 0.896 0.902 0.915 0.908 B 0.815 0.823 0.806 0.814 C 0.932 0.918 0.925 0.921

[0096] It can be seen from Table 2 that when the present invention uses both temperature and vibration as defect diagnosis input conditions, its accuracy is higher than the defect accuracy when only temperature or only vibration is used as the feature quantity, and its accuracy rate reaches 0.932. Through this comparison, it is further verified that the defect diagnosis method of dry-type iron core reactor proposed by the present invention with a multi-factor fusion greatly improves the accuracy rate of dry-type iron core reactor defect diagnosis.

[0097] Although the embodiments of the present invention are described in detail above, it is obvious to those skilled in the art that various modifications and variations can be made to these embodiments. However, it should be understood that such modifications and variations are within the scope and spirit of the present invention as described in the claims. Moreover, the present invention described herein may have other embodiments and may be implemented or realized in a variety of ways.

Claims

1. A multi-factor integrated dry-type iron core reactor defect diagnosis method, characterized in that: include: Acquire a temperature field and a structural field simulation model of a dry-type iron core reactor based on physical characteristics, acquire a temperature distribution curve and a vibration distribution curve of the dry-type iron core reactor based on the simulation model, and acquire a temperature monitoring point and a vibration monitoring point of the dry-type iron core reactor based on the temperature distribution curve and the vibration distribution curve; The temperature distribution characteristics are obtained after the magnetic field, fluid and temperature multi-physics field coupling simulation is performed based on the temperature monitoring point and the vibration monitoring point, and the vibration distribution characteristics are obtained after the material parameters in the vibration simulation are corrected according to the temperature distribution characteristics. A defect sample data set is established by combining the temperature distribution characteristics and the vibration distribution characteristics, and the defect sample data set is divided into a training set, a verification set and a test set, wherein the process of correcting the material parameters in the vibration simulation includes correcting Young's modulus and the electrical conductivity of the copper material, wherein after correction, the expression of Young's modulus is: ； in, E is the corrected Young's modulus, is Young's modulus, is the initial temperature of the copper material, is the linear expansion coefficient, is the temperature of each structural component, is the ratio of the temperature coefficient of Young's modulus to the coefficient of linear expansion; After correction, the expression of copper material conductivity is: ； in, is the conductivity of the corrected copper material, is the real-time temperature of the copper material; Construct an initial defect diagnosis model, train, adjust parameters and test the initial defect diagnosis model based on the training set, validation set and test set to obtain a trained prediction model, input the defect sample data set into the trained prediction model to output the defect type.

2. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 1 is characterized in that: The process of obtaining the temperature monitoring points and vibration monitoring points of the dry-type iron core reactor includes: A temperature detection path is set in the axial direction and radial direction on the three-phase coil of the dry-type iron core reactor, a temperature distribution curve of the temperature change with displacement on the temperature detection path is obtained based on the temperature sensor, and a position corresponding to the highest temperature is obtained as a temperature monitoring point based on the temperature distribution curve; The vibration displacement distribution curve of the core and the coil of the dry-type iron core reactor in the X, Y and Z axis directions is obtained based on the vibration sensor, and the position with the maximum vibration displacement is obtained as the vibration monitoring point based on the vibration displacement distribution curve.

3. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 1 is characterized in that: The temperature monitoring points are deployed at the lower, middle and upper parts of the three-phase coil, which are 13.3%, 49.7% and 84.4% of the length of the three-phase coil from bottom to top respectively. The number of temperature monitoring points and vibration monitoring points are both 9.

4. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 1 is characterized in that: The process of outputting defect types includes: Random sampling from defect sample data set N A defect sample subset, the defect sample subset includes M feature quantity; Build a corresponding decision tree for each defect sample subset, and M Randomly select from m feature dimensions, and the feature dimensions are input into each decision tree to obtain N prediction results, among which, m ≤ M ; Regarding the N The prediction results are voted for and the prediction result with the highest vote is taken as the defect type.

5. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 1 is characterized in that: The basis for determining the defect type consists of a first determination condition and a second determination condition. The first determination condition is: when the hot spot temperature of the dry-type iron core reactor is lower than 155°C and the maximum vibration amplitude of its clamp is lower than the vibration peak value, the operating state of the dry-type iron core reactor is normal, otherwise there is overheating or mechanical defects. The second determination condition is: when the hot spot temperature rise is lower than 95K and the maximum vibration amplitude of the reactor core or winding is lower than the vibration peak value, the operating state of the dry-type iron core reactor is normal, otherwise there is overheating or mechanical defects. The first determination condition is equivalent to the second determination condition.

6. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 1 is characterized in that: The process of adjusting the parameters of the initial defect diagnosis model includes: obtaining an objective function based on the difference between the actual defect type value and the predicted value of the dry-type iron core reactor, and iteratively adjusting the parameters of the initial defect diagnosis model based on the verification set until the iterative adjustment is stopped when the objective function remains unchanged.

7. The defect diagnosis method of a dry-type iron core reactor with multi-factor integration according to claim 6 is characterized in that: The expression of the objective function is: ； in, is the objective function, is the actual defect type value of the reactor, is the predicted value of the random forest model, is the number of samples in the defect sample set, is an indicative function.

Images