IGBT module temperature abnormity identification method and device
By performing time series prediction and cluster analysis on the time series of IGBT module temperature, abnormal temperature is identified, and the problem of lack of flexibility in the prior art methods based on preset thresholds is solved, and adaptive recognition of complex working conditions is achieved.
Patent Information
- Application Number
- CN202411602208.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-11
- Publication Date
- 2025-02-14
AI Technical Summary
The preset threshold method in the prior art is difficult to adapt to complex and variable actual working conditions, resulting in lack of flexibility in identifying temperature abnormalities of IGBT modules.
By obtaining the time series of temperature and normalizing it, the original data set is formed; the residual term is calculated based on the timing prediction to form the residual data set; the weighted average distance is determined based on the K-nearest neighbor set and the preset weight function; Class evaluation indicators determine the best and minimum number of points and perform cluster analysis to identify abnormal temperatures.
It realizes adaptation to temperature changes in different environments and operating conditions, has better flexibility, and solves the problem of difficulty in selecting clustering algorithm parameters through unsupervised autonomous clustering.
Smart Images
Figure CN119441916A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of temperature anomaly identification, and in particular to a method and device for identifying temperature anomaly of an IGBT module. Background Art
[0002] During the operation of a photovoltaic power station, the IGBT module, as the core component of a centralized inverter, has a direct impact on the stability and efficiency of the entire photovoltaic power station system due to its working condition. Excessive operating temperature will accelerate the aging of the IGBT module and even cause it to fail. Therefore, real-time and accurate identification of temperature anomalies of the IGBT module is essential for preventive maintenance and improving system reliability.
[0003] At present, most of the monitoring methods based on preset thresholds are used to identify temperature anomalies of IGBT modules. However, this method lacks flexibility and is difficult to adapt to complex and changeable actual working conditions. Summary of the invention
[0004] The embodiment of the present application provides a method and device for identifying abnormal temperature of an IGBT module, which solves the technical problem that the method based on preset threshold monitoring in the prior art is used to identify abnormal temperature of the IGBT module, which lacks flexibility and is difficult to adapt to complex and changeable actual working conditions. In the first aspect, the embodiment of the present application provides a method for identifying abnormal temperature of an IGBT module, the method comprising: obtaining a time series of temperature and normalizing it to form an original data set; performing time series prediction based on the original data set to obtain a predicted value at each moment, and calculating the residual term between the predicted value and the original value in the original data set to form a residual data set; determining the weighted average distance of each residual term according to the K-nearest neighbor point set of each residual term and a preset weight function; determining the optimal minimum number of points by calculating the minimum value of the clustering evaluation index according to the preset value range of the minimum number of points; performing cluster analysis on the residual data set based on the weighted average distance and the optimal minimum number of points, and identifying abnormal temperature according to the result of the cluster analysis.
[0005] In combination with the first aspect, in a possible implementation, performing time series forecasting based on the original data set to obtain the forecast value at each moment includes: inputting the temperature data in the original data set into a cubic exponential smoothing model according to a time series, and obtaining the forecast value at each moment output by the cubic exponential smoothing model.
[0006] In combination with the first aspect, in a possible implementation method, the weighted average distance of each residual item is determined based on the K-nearest neighbor point set of each residual item and the preset weight, including: calculating the neighbor distance between each residual item and each point in the corresponding K-nearest neighbor point set; calculating the local density of each residual item based on the neighbor distance; constructing the preset weight function based on the local density and the neighbor distance; calculating the weighted average distance of each residual item based on the neighbor distance and the preset weight function.
[0007] In combination with the first aspect, in a possible implementation manner, the preset weight function is as follows: in, represents the local density, represents the nearest neighbor distance, Represents the parameter for adjusting the weight of each factor.
[0008] In combination with the first aspect, in a possible implementation manner, the performing cluster analysis on the residual data set includes: performing cluster analysis on the residual data set using a density-based clustering algorithm.
[0009] In combination with the first aspect, in a possible implementation manner, the clustering evaluation index is the Davidson-Borgin Index.
[0010] In combination with the first aspect, in a possible implementation manner, the method further includes: when an abnormal temperature state is identified, issuing a warning signal; wherein the warning signal includes location information of the temperature anomaly.
[0011] In a second aspect, an embodiment of the present application provides an IGBT module temperature anomaly identification device, which includes: an acquisition module, used to acquire a time series of temperature and normalize it to form an original data set; a residual module, used to perform time series prediction based on the original data set, obtain a predicted value at each moment, and calculate the residual term between the predicted value and the original value in the original data set to form a residual data set; a distance module, used to determine the weighted average distance of each residual term based on the K-nearest neighbor point set of each residual term and a preset weight function; a minimum point number module, used to determine the optimal minimum number of points by calculating the minimum value of a clustering evaluation index based on a preset value range of the minimum number of points; an identification module, used to perform cluster analysis on the residual data set based on the weighted average distance and the optimal minimum number of points, and identify abnormal temperature based on the result of the cluster analysis.
[0012] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method described in the first aspect or any possible implementation method of the first aspect.
[0013] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, they implement the method described in the first aspect or any possible implementation method of the first aspect.
[0014] The technical solution provided in the embodiments of the present application has at least the following technical effects: The embodiment of the present application provides an IGBT module temperature anomaly identification method, which obtains the time series of temperature and normalizes it to form an original data set, calculates the residual term between the predicted value and the original value at each moment to form a residual data set, performs cluster analysis on the residual data set based on the weighted average distance and the optimal minimum number of points, and identifies abnormal temperature according to the result of cluster analysis. When using the IGBT module temperature anomaly identification method for abnormal temperature identification, there is no need to manually set a fixed threshold in advance, cluster analysis is performed on the residual data set based on the weighted average distance and the optimal minimum number of points, and the abnormal temperature is identified according to the result of cluster analysis, thereby realizing the adaptation to temperature changes under different environments and working conditions, so that the IGBT module temperature anomaly identification method has better flexibility. In addition, the method determines the weighted average distance of each residual term according to the K-nearest neighbor point set of each residual term and the preset weight function, determines the optimal minimum number of points by calculating the minimum value of the clustering evaluation index according to the preset value range of the minimum number of points, realizes the adaptive determination of clustering algorithm parameters, does not need to manually adjust the parameters after manually observing the clustering results, and realizes unsupervised autonomous clustering. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0016] Figure 1 A flow chart of a method for identifying abnormal temperature of an IGBT module provided in an embodiment of the present application; Figure 2 A schematic diagram of a density-based clustering algorithm provided in an embodiment of the present application; Figure 3A schematic diagram of the structure of an IGBT module temperature anomaly identification device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0017] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0018] The technical solution of the embodiment of the present application can be applied to a variety of electronic or electrical equipment with IGBT modules, such as a centralized inverter of a photovoltaic power station.
[0019] The following is a part of the embodiments of the present application for explanation and explanation to facilitate understanding by those skilled in the art.
[0020] Insulated Gate Bipolar Transistor (IGBT) is a composite fully controlled voltage-driven power semiconductor device composed of a bipolar transistor and an insulated gate field effect transistor. It combines the advantages of bipolar transistors and insulated gate field effect transistors, and has the characteristics of high input impedance and low on-state voltage drop. It is widely used in servo motors, inverters, variable frequency home appliances and other fields.
[0021] The triple exponential smoothing model is a time series forecasting method that combines linear trends and seasonal factors and is suitable for data series with obvious seasonal changes. It can predict future values and can adapt to the changing trend of data.
[0022] Density-Based Spatial Clustering of Applications with Noise (DBSCAN) is able to divide areas with sufficiently high density into clusters and regard those low-density points as noise or outliers.
[0023] The Davies-Bouldin Index (DBI) is an indicator for evaluating the quality of clustering algorithms. This indicator takes into account the similarity of samples within a class and the difference between samples between classes. The smaller the value, the higher the clustering effectiveness.
[0024] K-Nearest Neighbors (KNN) is a basic classification and regression algorithm. Its core idea is that similar sample points should be close to each other in the feature space, so the neighboring sample points can be used to predict the category or attribute of unknown sample points.
[0025] The embodiment of the present application provides a method for identifying abnormal temperature of an IGBT module, such as Figure 1 As shown, the method includes steps 101 to 105.
[0026] Step 101: Obtain a time series of temperature and normalize it to form an original data set.
[0027] Exemplarily, a temperature sensor is used to continuously detect the surface temperature of the IGBT module, and when step 101 is executed, a time series of the temperature of the surface of the IGBT module is obtained from the temperature sensor.
[0028] When normalizing the acquired temperature in step 101, the following formula may be used: Where NT represents the normalized original value in the original data set; T represents the surface temperature of the IGBT module; T min Indicates the minimum operating temperature of the IGBT module; T max Indicates the maximum operating temperature of the IGBT module.
[0029] After the acquired temperature time series is normalized in step 101, the dimension effect between the temperature data is eliminated, so that the stability and generalization ability of the IGBT module temperature anomaly identification method are improved.
[0030] Step 102: Perform time series prediction based on the original data set to obtain the predicted value at each moment, and calculate the residual term between the predicted value and the original value in the original data set to form a residual data set.
[0031] It should be understood that the predicted value is calculated by the prediction model based on the original values at other times in the original data set, representing the theoretical value of the temperature at each time; the original value in the original data set represents the actual value of the temperature at each time; the residual term represents the difference between the actual value of the temperature at each time and the theoretical value.
[0032] Specifically, the calculation formula of the residual term is as follows: ; In the formula, res i is the residual term at the i-th moment; y i 、x i are the predicted value and original value at the i-th moment respectively.
[0033] Step 103: Determine the weighted average distance of each residual item according to the K-nearest neighbor point set of each residual item and a preset weight function.
[0034] The weighted average distance in step 103 is used as the radius of subsequent clustering. The constructed preset weight function can reflect the factors covered by the weighted average distance, so the preset weight function can be adjusted according to the factors emphasized by the weighted average distance.
[0035] Step 104: According to the preset value range of the minimum number of points, the optimal minimum number of points is determined by calculating the minimum value of the clustering evaluation index.
[0036] The minimum number of points refers to the number of points that each cluster in the clustering result must contain at least when clustering. In density-based clustering algorithms, it is used to determine whether the density of a cluster is high enough. For example, in the DBSCAN algorithm, the minimum number of points defines how many points are required within a specified radius to form a dense area and thus be considered part of a cluster.
[0037] Exemplarily, the clustering evaluation index in the embodiment of the present application is the Davidson-Borgin Index.
[0038] The calculation formula of Davidson-Botting Index is as follows:
[0039] Where n is the number of clusters, r i 、r j Cluster c i and c j The average radius of (the average distance from the residual term within the cluster to the cluster center), is the cluster center c i 、c j The smaller the DBI, the better the clustering quality, because a smaller DBI means that the samples within the cluster are closer and the clusters are more separated.
[0040] Step 105: Based on the weighted average distance and the optimal minimum number of points, cluster analysis is performed on the residual data set, and abnormal temperature is identified according to the result of the cluster analysis.
[0041] Step 105 performs cluster analysis on the residual data set, and some residual items form multiple clusters, while other residual items are located outside the cluster. These residual items located outside the cluster are deviation values, and the original values in the original data corresponding to them are abnormal temperatures. In addition, since step 105 performs clustering based on weighted average distance and optimal minimum number of points, the clustering in step 105 can fully consider the factors that are important and the clustering effect.
[0042] The IGBT module temperature anomaly identification method provided in the embodiment of the present application does not require manual setting of fixed thresholds in advance, can adapt to temperature changes in different environments and working conditions, and has better flexibility. In addition, the IGBT module temperature anomaly identification method overcomes the defects of difficulty in selecting clustering algorithm parameters and the need for manual setting. It uses the weighted average distance as the radius of the cluster, and determines the optimal minimum number of points by calculating the minimum value of the clustering evaluation index, thereby realizing adaptive determination of clustering algorithm parameters, eliminating the need for manual observation of clustering results and manual adjustment of parameters, and realizing unsupervised autonomous clustering.
[0043] Step 102 of the embodiment of the present application specifically includes: inputting the temperature data in the original data set into a cubic exponential smoothing model according to a time series, and obtaining a predicted value at each moment output by the cubic exponential smoothing model.
[0044] Specifically, the formula of the triple exponential smoothing model is as follows: Where i=2,3,…,m. m is the length of the input time series, m≤T; x i is the input data at the i-th moment; s i ,t i 、p i are the first, second and third smoothing values at the i-th moment; k is the period; α, β, γ are the corresponding smoothing coefficients, which are used to reflect the proportion of historical data at different moments in the exponential smoothing value, and α, β, γ∈(0,1).
[0045] The predicted value of the temperature at the i+h moment is: Among them, y i+h is the predicted value of the temperature at the i+hth moment; s i ,t i 、p i is the first, second, and third smoothing value at the i-th moment; k is the period; h is the prediction step, h=1, 2, 3, .... Since the embodiment of the present application requires the predicted value of the temperature at each moment, the prediction step h is set to 1, that is, prediction is performed step by step.
[0046] The quasi-Newton optimization method (BFGS) is used to optimize the mean squared error (MSE), and α, β, and γ with the minimum MSE are taken as smoothing coefficients. The MSE calculation formula is as follows: Where n is the number of data points in the selected time series; y i 、x i are the predicted value and original value of the temperature at the i-th moment in the time series, respectively.
[0047] Step 103 in the embodiment of the present application specifically includes steps 201 to 203 .
[0048] Step 201: Calculate the neighbor distance between each residual term and each point in the corresponding K-nearest neighbor point set.
[0049] Specifically, when executing step 201, the neighbor distance is calculated according to the following formula: Where P i Represents the residual term in the residual data set; is the residual term P i The corresponding K-nearest neighbor point set; Indicates P i With P j The distance between neighbors.
[0050] Step 202: Calculate the local density of each residual term based on the nearest neighbor distance.
[0051] Specifically, when executing step 202, the local density is calculated according to the following formula: In the formula, Residual term P i The local density, K represents The number of midpoints.
[0052] Step 203: construct a preset weight function based on local density and neighbor distance.
[0053] Exemplarily, the preset weight function may be constructed as follows: Parameters for factor weights.
[0054] β is used to avoid zero division errors when the local density is 0; α is used to control the contribution of local density to the weight; γ is used to control the contribution of neighbor distance to the weight; δ is used to control the speed of neighbor distance decay. The larger δ is, the faster the weight of neighbor distance decays.
[0055] Step 204: Calculate the weighted average distance of each residual term according to the nearest neighbor distance and a preset weight function.
[0056] When executing step 204, the weighted distance of each residual term is first calculated using the following formula: Then, the weighted mean distance of each residual term is calculated by the following formula: In step 105 of the embodiment of the present application, cluster analysis is performed on the residual data set, which specifically includes: using a density-based clustering algorithm to perform cluster analysis on the residual data set.
[0057] Density-based clustering algorithms have the advantages of not requiring a preset number of clusters, being able to handle clusters of any shape, and being robust to noise. Figure 2 shown.
[0058] Of course, when performing cluster analysis on the residual data set in step 105 of the embodiment of the present application, other clustering algorithms, such as hierarchical clustering, may also be used.
[0059] The IGBT module temperature anomaly identification method provided in the embodiment of the present application also includes: issuing a warning signal when a temperature anomaly state is identified; wherein the warning signal includes location information of the temperature anomaly.
[0060] After receiving the early warning signal, the operation and maintenance personnel can clearly know the location information of the temperature abnormality, and can then quickly reach the location of the temperature abnormality and carry out repairs.
[0061] The embodiment of the present application also provides an IGBT module temperature anomaly identification device 300, such as Figure 3 As shown, the IGBT module temperature anomaly identification device 300 includes an acquisition module 301 , a residual module 302 , a distance module 303 , a minimum point module 304 and an identification module 305 .
[0062] The acquisition module 301 is used to acquire the time series of temperature and normalize it to form an original data set. The residual module 302 is used to perform time series prediction based on the original data set, obtain the predicted value at each moment, and calculate the residual term between the predicted value and the original value in the original data set to form a residual data set. The distance module 303 is used to determine the weighted average distance of each residual term based on the K-nearest neighbor point set of each residual term and a preset weight function. The minimum point module 304 is used to determine the optimal minimum number of points by calculating the minimum value of the clustering evaluation index based on the preset value range of the minimum number of points. The identification module 305 is used to perform cluster analysis on the residual data set based on the weighted average distance and the optimal minimum number of points, and identify abnormal temperatures based on the results of the cluster analysis.
[0063] The residual film module 302 performs time series prediction based on the original data set to obtain the predicted value at each moment, including: inputting the temperature data in the original data set into the cubic exponential smoothing model according to the time series to obtain the predicted value at each moment output by the cubic exponential smoothing model.
[0064] The distance module 303 is specifically used to: calculate the neighbor distance between each residual item and each point in the corresponding K-nearest neighbor point set; calculate the local density of each residual item based on the neighbor distance; construct a preset weight function based on the local density and the neighbor distance; calculate the weighted average distance of each residual item based on the neighbor distance and the preset weight function.
[0065] Specifically, the preset weight function is as follows: Represents the parameter for adjusting the weight of each factor.
[0066] The identification module 305 performs cluster analysis on the residual data set, including: performing cluster analysis on the residual data set using a density-based clustering algorithm.
[0067] Among them, the clustering evaluation index is the Davidson-Botting index.
[0068] The IGBT module temperature anomaly identification device 300 further includes an early warning module, which is used to send out an early warning signal when an abnormal temperature state is identified; wherein the early warning signal includes location information of the temperature anomaly.
[0069] The devices or modules described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. For the convenience of description, the above system is described by dividing it into various modules according to its functions. When implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware. Of course, the module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.
[0070] Some modules in the apparatus described in the present application may also be computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc. that perform specific tasks or implement specific abstract data types.
[0071] An embodiment of the present application also provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned IGBT module temperature anomaly identification method is implemented.
[0072] The embodiment of the present application also provides a computer-readable storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, the above-mentioned IGBT module temperature anomaly identification method is implemented.
[0073] The above-mentioned storage media include but are not limited to random access memory (RAM), read-only memory (ROM), cache, hard disk drive (HDD) or memory card (Memory Card).
[0074] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other. Each embodiment focuses on the differences from other embodiments.
[0075] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit the present application. Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present application.
Claims
1. A method for identifying abnormal temperature of an IGBT module, characterized in that: include: Obtain the temperature time series and normalize it to form the original data set; Performing time series prediction based on the original data set to obtain a predicted value at each moment, and calculating a residual term between the predicted value and the original value in the original data set to form a residual data set; Determining a weighted average distance of each residual item according to a K-nearest neighbor point set of each residual item and a preset weight function; According to the preset value range of the minimum number of points, the optimal minimum number of points is determined by calculating the minimum value of the clustering evaluation index; Based on the weighted average distance and the optimal minimum number of points, cluster analysis is performed on the residual data set, and abnormal temperature is identified according to the result of the cluster analysis.
2. The method according to claim 1, characterized in that The performing time series prediction based on the original data set to obtain a prediction value at each moment includes: The temperature data in the original data set is input into a cubic exponential smoothing model according to a time series, and the predicted value at each moment output by the cubic exponential smoothing model is obtained.
3. The method according to claim 1, characterized in that Determining the weighted average distance of each residual item according to the K-nearest neighbor point set of each residual item and a preset weight includes: Calculate the nearest neighbor distance between each residual term and each point in the corresponding K-nearest neighbor point set; Calculating the local density of each residual term according to the nearest neighbor distance; Based on the local density and the neighbor distance, construct the preset weight function; The weighted average distance of each of the residual items is calculated according to the nearest neighbor distance and the preset weight function.
4. The method according to claim 3, characterized in that The preset weight function is as follows: ; in, represents the local density, represents the nearest neighbor distance, Represents the parameter for adjusting the weight of each factor.
5. The method according to claim 1, characterized in that The performing cluster analysis on the residual data set comprises: A density-based clustering algorithm is used to perform cluster analysis on the residual data set.
6. The method according to claim 1, characterized in that The clustering evaluation index is the Davidson-Borgin index.
7. The method according to claim 1, characterized in that The method further comprises: When an abnormal temperature state is identified, an early warning signal is issued; wherein the early warning signal includes location information of the temperature anomaly.
8. An IGBT module temperature anomaly identification device, characterized in that: include: The acquisition module is used to obtain the time series of temperature and normalize it to form the original data set; A residual module, used to perform time series prediction based on the original data set, obtain a predicted value at each moment, and calculate a residual term between the predicted value and the original value in the original data set to form a residual data set; A distance module, used to determine the weighted average distance of each residual item according to the K-nearest neighbor point set of each residual item and a preset weight function; The minimum point module is used to determine the optimal minimum point by calculating the minimum value of the clustering evaluation index according to the preset value range of the minimum point; The identification module is used to perform cluster analysis on the residual data set based on the weighted average distance and the optimal minimum number of points, and identify abnormal temperature according to the result of the cluster analysis.
9. An electronic device, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Test data processing method and system for motor stator wire lapping
CN121008165A
A test data processing method and system for motor stator winding
CN121008165B