A multi-chip IGBT uneven aging monitoring method, system and medium

Abstract

The present application relates to the technical field of power device monitoring, and particularly relates to a multi-chip IGBT uneven aging monitoring method, system and medium, the method comprising: applying a periodic power excitation to a multi-chip IGBT power module, and collecting the corresponding shell temperature response of each IGBT chip; constructing a shell temperature matrix; performing multivariate variational modal decomposition on the shell temperature matrix to obtain phase information of different characteristic frequency bands to construct a phase matrix; determining a phase lag matrix according to the phase matrix and a reference phase matrix; matching a preset aging state mapping relationship according to the distribution characteristics of the phase lag matrix at different characteristic frequency bands and the positions of each IGBT chip to determine the aging type, aging degree and uneven aging state. Through the present application, the problem that the prior art cannot accurately decouple and quantitatively monitor the uneven aging of different heat transfer paths of a multi-chip IGBT power module based on a non-intrusive shell temperature response is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of power device monitoring technology, and in particular to a method, system and medium for monitoring uneven aging of multi-chip IGBTs. Background Technology

[0002] Multi-chip IGBT power modules consist of multiple parallel chips and packaging and heat dissipation structures. Both localized chip packaging degradation and degradation of common heat dissipation paths alter the casing temperature response. However, inter-chip thermal coupling causes the temperature characteristics of different aging modes to overlap. Existing technologies, such as CN112327124A, assess uneven thermal fatigue aging by analyzing the casing temperature drop curves and time constants at corresponding chip locations, while CN120035765A identifies degradation of different thermal paths in the power module through frequency and phase response under periodic excitation. However, the former relies on fitting cooling curves and struggles to accurately distinguish between aging of localized packaging paths and common heat dissipation paths under thermal coupling conditions. The latter lacks spatial correspondence among multiple chips and multi-channel temperature collaborative analysis, making it difficult to locate and quantify uneven aging within the module. Therefore, existing technologies still struggle to accurately decouple and quantitatively monitor uneven aging of different heat transfer paths in multi-chip IGBT power modules based on non-invasive casing temperature response.

[0003] The information disclosed in this background section is intended only to enhance the understanding of the general background of this disclosure and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] This invention provides a method, system, and medium for monitoring uneven aging of multi-chip IGBTs, which can effectively solve the problems in the background art.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for monitoring uneven aging of multi-chip IGBTs, the method comprising: Periodic power excitation is applied to the multi-chip IGBT power module, and the case temperature response of each IGBT chip in the multi-chip IGBT power module is collected. Construct a shell temperature matrix based on the shell temperature response; The shell temperature matrix is ​​subjected to multivariate variational mode decomposition, and the phase information corresponding to different characteristic frequency bands is obtained to construct the phase matrix; The phase hysteresis matrix is ​​determined based on the phase matrix and the reference phase matrix corresponding to the health state. Based on the distribution characteristics of the phase lag matrix in different characteristic frequency bands and at corresponding positions of each IGBT chip, it is matched with a preset aging state mapping relationship to determine the aging type, aging degree, and uneven aging state of the multi-chip IGBT power module.

[0006] Furthermore, the periodic power excitation is a sinusoidal half-wave loss excitation with a frequency of 1Hz applied to each of the IGBT chips, and the housing temperature response is obtained by synchronously collecting data at multiple temperature monitoring points located at the bottom of the housing of the multi-chip IGBT power module and corresponding to each of the IGBT chips at a preset sampling frequency.

[0007] Furthermore, the shell temperature matrix is ​​constructed by arranging the shell temperature responses of each shell according to the sampling time of the monitoring channels corresponding to the multiple temperature monitoring points.

[0008] Furthermore, the shell temperature matrix is ​​subjected to multivariate variational mode decomposition, including: The shell temperature matrix is ​​decomposed into multiple modal components, wherein each modal component includes the channel mode corresponding to each monitoring channel, and the channel modes belonging to the same modal component have the same center frequency; Perform Hilbert transform on each of the channel modes to obtain the analytic signal of each of the channel modes; The goal is to minimize the sum of the bandwidths of each analytical signal after demodulation at the center frequency, and the constraint is to reconstruct the shell temperature matrix by the sum of each modal component, thereby solving for multiple modal components.

[0009] Furthermore, phase information corresponding to different characteristic frequency bands is obtained to construct a phase matrix, including: Perform a discrete Fourier transform on each of the channel modes in each of the modal components to obtain the frequency domain signal of each of the channel modes; Within the first characteristic frequency band of 0 to 0.1 Hz (excluding 0 Hz) and the second characteristic frequency band of 0.1 to 100 Hz, the frequency with the largest frequency domain amplitude in each monitoring channel is determined, and the determined frequencies are used to form the characteristic frequency set of the corresponding monitoring channel. For each characteristic frequency in the set of characteristic frequencies, determine the target modal component with the largest frequency domain amplitude at the corresponding characteristic frequency; The phase of the target modal component at the corresponding characteristic frequency is obtained, and the phase matrix is ​​constructed according to the phase corresponding to each monitoring channel.

[0010] Furthermore, the phase lag matrix is ​​constructed based on the phase difference at the corresponding characteristic frequencies of each monitoring channel in the current state and the healthy state. The phase lag corresponding to the first characteristic frequency band is used to characterize the heat transfer path state from the housing to the environment, and the phase lag corresponding to the second characteristic frequency band is used to characterize the heat transfer path state from the chip junction to the housing.

[0011] Furthermore, the phase lag matrix is ​​determined based on the phase matrix and the reference phase matrix corresponding to the health state, including: The phase lag of each monitoring channel in the current state is obtained by subtracting the reference phase of the corresponding monitoring channel in the healthy state at the same characteristic frequency. This is done by taking the phase lag of each monitoring channel in the first characteristic frequency band and the second characteristic frequency band. The phase lag matrix is ​​constructed based on the phase lag corresponding to each monitoring channel, wherein the phase lag in the first characteristic frequency band is used to characterize the change in thermal resistance from shell to environment, and the phase lag in the second characteristic frequency band is used to characterize the change in thermal resistance between the junction and shell.

[0012] Furthermore, matching is performed with a preset aging state mapping relationship, including: Calculate the Euclidean distance between the phase lag matrix in the current state and the phase lag matrix of each sample in the multidimensional lookup table, and determine the mapping sample with the smallest Euclidean distance as the target mapping sample; Based on the aging degree of the solder layer and the aging degree of the heat sink corresponding to the target mapping sample, the aging type and aging degree of the multi-chip IGBT power module are determined. In response to the fact that the phase lag of some monitoring channels increases in the second characteristic frequency band but the phase lag in the first characteristic frequency band does not increase accordingly, it is determined that the corresponding IGBT chip has uneven aging of the solder layer. In response to the fact that the phase hysteresis of each of the monitoring channels increases in the second characteristic frequency band and the increase is similar, it is determined that the multi-chip IGBT power module has uniform aging of the solder layer. In response to the fact that the phase lag of each of the monitoring channels increases in the first characteristic frequency band but does not increase accordingly in the second characteristic frequency band, it is determined that the multi-chip IGBT power module has heat sink aging.

[0013] A multi-chip IGBT non-uniform aging monitoring system, the system comprising: The power excitation acquisition module applies periodic power excitation to the multi-chip IGBT power module and acquires the case temperature response of each IGBT chip in the multi-chip IGBT power module. The shell temperature matrix construction module constructs a shell temperature matrix based on the shell temperature response. The phase feature extraction module performs multivariate variational mode decomposition on the shell temperature matrix and obtains phase information corresponding to different characteristic frequency bands to construct the phase matrix; The phase lag calculation module determines the phase lag matrix based on the phase matrix and the reference phase matrix corresponding to the health state. The aging state determination module matches the distribution characteristics of the phase lag matrix in different characteristic frequency bands and the corresponding positions of each IGBT chip with the preset aging state mapping relationship to determine the aging type, aging degree and uneven aging state of the multi-chip IGBT power module.

[0014] A computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the multi-chip IGBT non-uniform aging monitoring method.

[0015] The technical solution of this invention can achieve the following technical effects: By applying periodic power excitation and collecting the shell temperature response of each chip in a multi-chip IGBT power module, a shell temperature matrix is ​​constructed. Phase information of different characteristic frequency bands is extracted through multivariate variational mode decomposition to form a phase hysteresis matrix. Then, by combining the spatial distribution characteristics with the preset aging state mapping relationship, the aging type, aging degree and non-uniform aging state can be identified. This effectively solves the problem that existing technologies are difficult to accurately decouple and quantitatively monitor the non-uniform aging of different heat transfer paths of multi-chip IGBT power modules based on non-invasive shell temperature response.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A flowchart illustrating the method for monitoring uneven aging of multi-chip IGBTs; Figure 2 This is a schematic diagram of the vertical heat transfer process of an IGBT module. Figure 3A schematic diagram of a multi-chip IGBT uneven aging monitoring logic; Figure 4 A schematic diagram of the framework for a multi-chip IGBT non-uniform aging monitoring model. Figure 5 This is a schematic diagram showing the specific spatial distribution of temperature monitoring points. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1; like Figure 1 As shown, this application provides a method for monitoring uneven aging of multi-chip IGBTs, the method comprising: S10: Apply periodic power excitation to the multi-chip IGBT power module and collect the case temperature response of each IGBT chip in the multi-chip IGBT power module. S20: Construct the shell temperature matrix based on the shell temperature response; S30: Perform multivariate variational mode decomposition on the shell temperature matrix and obtain phase information corresponding to different characteristic frequency bands to construct the phase matrix; S40: Determine the phase lag matrix based on the phase matrix and the reference phase matrix corresponding to the health state; S50: Based on the distribution characteristics of the phase lag matrix in different characteristic frequency bands and the corresponding positions of each IGBT chip, match it with the preset aging state mapping relationship to determine the aging type, aging degree and uneven aging state of the multi-chip IGBT power module.

[0022] Specifically, this embodiment uses a multi-chip IGBT power module as the monitoring object. Temperature sensors are arranged at the bottom of the module casing, corresponding to the positions of each IGBT chip, to acquire the temperature response of each chip after passing through the solder layer, substrate, and casing. The number of temperature monitoring points is determined according to the number of chips and their spatial distribution. Each temperature sensor uses the same installation method and sampling conditions. The heat transfer process of the IGBT module is a vertical heat transfer process, such as... Figure 2As shown, heat is transferred through multiple material layers from the chip junction to the module casing, involving components such as the chip, solder, substrate, and heat sink. When the solder or heat sink ages, it leads to an increase in the phase lag and amplitude of the system's thermal response. The increased phase lag manifests as an increase in the delay time of the temperature response relative to the heat source excitation (power loss), while the increased amplitude reflects a rise in steady-state temperature. These changes in frequency domain characteristics directly affect the temperature distribution and thermal stress distribution of the device, ultimately impacting its reliability and lifespan. Based on the above analysis, this invention provides a method for monitoring the non-uniform aging of multi-chip IGBT power modules based on casing temperature response to identify module aging types and quantitatively assess the degree of non-uniform aging. Figure 3 , Figure 4 As shown; A sinusoidal half-wave loss excitation with a frequency of 1Hz is applied to the power module to cause periodic power loss in each IGBT chip; the shell temperature of each temperature monitoring point is collected synchronously at continuous sampling time, and the temperature data is arranged according to the temperature monitoring point and sampling time to construct a shell temperature matrix containing spatial location features and temporal variation features; the excitation duration meets the observation requirements of low-frequency temperature response, and the sampling frequency meets the frequency resolution requirements of the target analysis frequency band. A multivariate variational mode decomposition method is used to jointly decompose each temperature monitoring channel in the shell temperature matrix to obtain multiple multi-channel modal components. Each multi-channel modal component contains the channel modes corresponding to each temperature monitoring channel, and the channel modes in the same multi-channel modal component have corresponding center frequencies. The decomposition process aims to minimize the sum of the bandwidths of each mode and constrains all modal components to be able to reconstruct the shell temperature matrix, thereby maintaining the modal correspondence between different temperature monitoring channels. Frequency domain analysis was performed on each channel mode to determine the characteristic frequencies with the largest amplitudes in the first characteristic frequency band (greater than 0 Hz and no higher than 0.1 Hz) and the second characteristic frequency band (greater than 0.1 Hz and no higher than 100 Hz). For each characteristic frequency, the modal component with the largest amplitude at that frequency was selected and the corresponding phase was extracted. The phases were arranged according to the temperature monitoring location and characteristic frequency band to form the phase matrix corresponding to the current state. When the power module is in a healthy state, a reference phase matrix is ​​obtained using the same excitation, sampling, and data processing conditions. The current phase matrix is ​​compared with the phase of the corresponding monitoring position and corresponding characteristic frequency in the reference phase matrix to obtain the phase hysteresis matrix. The phase hysteresis in the first characteristic frequency band is used to characterize the common heat dissipation path state from the casing to the environment, and the phase hysteresis in the second characteristic frequency band is used to characterize the local heat transfer path state from the chip junction to the casing. During offline calibration, the shell temperature response corresponding to different solder layer aging degrees and different heat sink aging degrees is collected respectively, and the sample phase hysteresis matrix is ​​obtained in the same way. The sample phase hysteresis matrix is ​​associated with aging type, aging location and aging degree and stored to form an aging state mapping library, and a multi-dimensional search relationship is established through gridding. During online monitoring, the characteristic distance between the current phase hysteresis matrix and the phase hysteresis matrices of each sample in the aging state mapping library is calculated. The sample with the smallest distance is taken as the matching result, and the aging type and aging degree are determined accordingly. When only some temperature monitoring channels show obvious phase hysteresis in the second characteristic frequency band, it is determined that the corresponding chip has local solder layer aging. When the phase hysteresis of each temperature monitoring channel in the second characteristic frequency band increases and the degree of change is similar, it is determined that there is uniform aging of the solder layer. When the phase hysteresis of each temperature monitoring channel in the first characteristic frequency band increases and the change in the second characteristic frequency band is small, it is determined that there is heat sink aging. Thus, the aging type, aging degree and non-uniform aging state of the power module are obtained.

[0023] The technical solution of this invention applies periodic power excitation and collects the shell temperature response of each chip in a multi-chip IGBT power module, constructs a shell temperature matrix, extracts phase information of different characteristic frequency bands through multivariate variational mode decomposition and forms a phase hysteresis matrix, and then combines spatial distribution characteristics with a preset aging state mapping relationship to achieve the identification of aging type, aging degree and non-uniform aging state. This effectively solves the problem that existing technologies are difficult to accurately decouple and quantitatively monitor the non-uniform aging of different heat transfer paths of multi-chip IGBT power modules based on non-invasive shell temperature response.

[0024] Furthermore, the periodic power excitation is a sinusoidal half-wave loss excitation with a frequency of 1Hz applied to each IGBT chip, and the casing temperature response is obtained synchronously by multiple temperature monitoring points set at the bottom of the multi-chip IGBT power module casing and corresponding to each IGBT chip according to a preset sampling frequency.

[0025] As a preferred embodiment of the above, a sinusoidal half-wave loss excitation with a frequency of 1Hz is applied to each IGBT chip in the multi-chip IGBT power module, so that each chip generates periodic power loss and the heat transfer path from the chip junction to the casing and from the casing to the heat sink forms a periodic temperature response; the sinusoidal half-wave loss excitation can be achieved by controlling the drive signal and load current of the IGBT chip, and each chip uses the same excitation frequency and excitation period to ensure that the temperature response of different chips is comparable; Six temperature monitoring points are arranged at the bottom of the power module housing. Each of the six temperature monitoring points corresponds to the position of each IGBT chip inside the module and is used to collect the housing temperature under each chip. The number of temperature monitoring points can be adjusted according to the number of chips, and each chip to be monitored must correspond to at least one temperature monitoring point. After the power excitation begins, the shell temperature of each temperature monitoring point is collected synchronously at the preset sampling frequency during continuous sampling time to obtain the temperature time series data of each temperature monitoring channel; the sampling duration meets the observation requirements of periodic temperature response and low-frequency thermal response, and the sampling frequency meets the frequency resolution requirements of subsequent characteristic frequency band analysis; the same excitation parameters, sensor position, sampling frequency and sampling duration are used for health status calibration and aging status monitoring.

[0026] Furthermore, the shell temperature matrix is ​​constructed by arranging the monitoring channels corresponding to multiple temperature monitoring points according to the sampling time to construct the shell temperature response.

[0027] Furthermore, the shell temperature matrix is ​​subjected to multivariate variational mode decomposition, including: The shell temperature matrix is ​​decomposed into multiple modal components, where each modal component includes the channel mode corresponding to each monitoring channel, and the channel modes belonging to the same modal component have the same center frequency; Perform Hilbert transform on each channel mode to obtain the analytic signal of each channel mode; The goal is to minimize the sum of the bandwidths of all analytical signals after demodulation at the center frequency, and the constraint is to reconstruct the shell temperature matrix by the sum of all modal components. Multiple modal components are then solved.

[0028] As a preferred embodiment of the above embodiments, in one embodiment, the casing temperature data is collected by arranging six temperature monitoring points (labeled T) at the bottom of the power module casing. C1 -T C6 The specific spatial distribution of monitoring points is as follows: Figure 5 As shown, temperature monitoring points were sampled at intervals, with sampling times ranging from t1 to t2. n The sampling frequency is f; Casing temperature matrix T C The temperature of the casing is determined by the temperature monitoring points obtained at the corresponding sampling time. ; Furthermore, the multivariate variational mode decomposition (MVMD) method is adopted, based on: ; and constraints: ; Shell temperature matrix Decomposed into Each mode is represented and a matrix is ​​constructed. ,satisfy: ; ; Among them, U k Let {U} represent the k-th mode. k} represents the set of modes to be optimized. ω represents the k-th mode of the p-th channel; k Let {ω} be the center frequency of the k-th mode. k} represents the set of center frequencies; ∂ t The time derivative characterizes the local changes in the mode. This is the analytic signal representation of the Hilbert transform. The Dirac impulse function; This is the center frequency demodulation term, used to shift the k-th mode to the baseband; The norm is used to quantize the bandwidth of the mode; This indicates that the sum of all modal components can reconstruct the original multi-channel shell temperature signal.

[0029] Furthermore, obtaining phase information corresponding to different characteristic frequency bands to construct a phase matrix includes: Perform Discrete Fourier Transform on each channel mode in each modal component to obtain the frequency domain signal of each channel mode; Within the first characteristic frequency band of 0 to 0.1 Hz (excluding 0 Hz) and the second characteristic frequency band of 0.1 to 100 Hz, the frequency with the largest frequency domain amplitude in each monitoring channel is determined, and the determined frequencies are used to form the characteristic frequency set of the corresponding monitoring channel. For each characteristic frequency in the characteristic frequency set, determine the target modal component with the largest frequency domain amplitude at the corresponding characteristic frequency; The phase of the target modal component at the corresponding characteristic frequency is obtained, and a phase matrix is ​​constructed based on the phase of each monitoring channel.

[0030] Furthermore, the phase lag matrix is ​​constructed based on the phase difference at the corresponding characteristic frequencies of each monitoring channel in the current state and the healthy state. The phase lag corresponding to the first characteristic frequency band is used to characterize the heat transfer path state from the shell to the environment, and the phase lag corresponding to the second characteristic frequency band is used to characterize the heat transfer path state from the chip junction to the shell.

[0031] As a preferred embodiment of the above, the phase matrix of each monitoring point at the characteristic frequency... The calculation steps are as follows: ; in, The frequency domain component of the k-th mode of the p-th monitoring channel at frequency f; Let be the time-domain sampled value of the k-th mode of the p-th monitoring channel at the i-th sampling time; is the frequency to be analyzed; i is the time-domain sampling point number; n is the total number of time-domain sampling points; k is the modal number; p is the monitoring channel number; j is the imaginary unit; The complex exponential basis functions are given by the discrete Fourier transform; in the time domain, each channel of each mode is computed. The discrete Fourier transform; ; In the formula, This represents the amplitude of the corresponding frequency domain component; This indicates the frequency at which the amplitude in the frequency domain reaches its maximum value; The characteristic frequency of the p-th monitoring channel is determined within the first characteristic frequency band, which is greater than 0 Hz and not higher than 0.1 Hz. The characteristic frequency of the p-th monitoring channel is determined within the second characteristic frequency band greater than 0.1Hz and not higher than 100Hz; the 0~0.1Hz frequency band (excluding 0Hz) and the 0.1~100Hz frequency band are respectively taken as the frequencies with the largest amplitudes among the two frequency bands of all channels. A set of characteristic frequencies is obtained for each channel. Let p be the set of characteristic frequencies of the p-th monitoring channel; ; In the formula, For the p-th monitoring channel at the characteristic frequency The target mode number with the largest amplitude in the frequency domain; This indicates the mode number that determines the maximum amplitude in the frequency domain among all modes; i is the characteristic frequency band number, i=1 indicates the first characteristic frequency band, i=2 indicates the second characteristic frequency band; The characteristic frequency determined for the p-th monitoring channel within the i-th characteristic frequency band; The amplitude of the corresponding frequency domain component is given by the above formula; the frequency f is determined by the above formula. p,i The corresponding mode with the largest amplitude, k*; Calculate the mode at frequency f p,i Phase at: ; This leads to the phase set of all channels. The phase of each channel is... ; The formula for calculating the phase is: ; Among them, R jc For junction-shell thermal resistance; R caFor the thermal resistance from the shell to the environment; C j and C c These are the heat capacities of the junction and the shell, respectively; in the low-frequency range (0~0.1Hz), the phase is mainly determined by the steady-state thermal resistance R. ca Decision, and R ca The larger the value of R, the greater the phase lag. jc The change in R has a relatively small impact on the phase; in the mid-frequency range (0.1~100Hz), when R jc When R increases, the numerator term in the phase expression increases and the denominator term decreases, resulting in an increase in phase lag; when R... ca When the value increases, both the numerator and denominator terms in the phase expression decrease simultaneously, and the phase does not change significantly.

[0032] Furthermore, the phase lag matrix is ​​determined based on the phase matrix and the reference phase matrix corresponding to the health state, including: The phase lag of each monitoring channel in the first and second characteristic frequency bands is obtained by subtracting the phase of each monitoring channel in the current state from the reference phase of the corresponding monitoring channel in the healthy state at the same characteristic frequency. A phase lag matrix is ​​constructed based on the phase lag corresponding to each monitoring channel. The phase lag in the first characteristic frequency band is used to characterize the change in thermal resistance from shell to environment, and the phase lag in the second characteristic frequency band is used to characterize the change in thermal resistance between the junction and shell.

[0033] Furthermore, matching with the preset aging state mapping relationship includes: Calculate the Euclidean distance between the phase lag matrix in the current state and the phase lag matrix of each sample in the multidimensional lookup table, and determine the mapping sample with the smallest Euclidean distance as the target mapping sample; Based on the aging degree of the solder layer and the aging degree of the heat sink corresponding to the target mapping sample, determine the aging type and aging degree of the multi-chip IGBT power module; In response to the fact that the phase lag of some monitoring channels increases in the second characteristic frequency band but does not increase accordingly in the first characteristic frequency band, it is determined that the corresponding IGBT chip has uneven aging of the solder layer. The phase hysteresis of each monitoring channel in the second characteristic frequency band increases and the increase is similar, indicating that the solder layer of the multi-chip IGBT power module is uniformly aged. The phase lag of each monitoring channel increased in the first characteristic frequency band but did not increase accordingly in the second characteristic frequency band, indicating that the multi-chip IGBT power module has heat sink aging.

[0034] As a preferred embodiment of the above, shell temperature matrix measurement and characteristic frequency phase extraction are performed on samples with different aging degrees, and the phase hysteresis matrix is... With aging degree matrix Dp The mapping relationship satisfies: ; ; ; in The characteristic frequency phase represents the solder layer under aging conditions; The characteristic frequency phase representing the healthy state of the solder layer; The characteristic frequency phase represents the aging state of the heat sink; The characteristic frequency phase representing the radiator in a healthy state; D p_solder Used to indicate the degree of aging of the solder layer in channel p; D p_sink Used to indicate the degree of aging of the heat sink in channel p; This is the error term; Offline mapping library The storage format is: ; in, Let z represent the aging degree of the i-th sample corresponding to the p-th temperature monitoring channel; z is the total number of samples; the mapping library is gridded to establish a multidimensional lookup table; When identifying the degree of aging, the phase offset matrix in the current state is extracted. In the mapping library, the nearest neighbor search algorithm is used. Find the most similar samples to obtain different types of aging. Through the formula: ; Where i* is the index of the target sample most similar to the current state; This indicates the sample number corresponding to the smallest distance selected from all samples; P is the temperature monitoring channel or the corresponding IGBT chip location number; This represents the phase offset difference between the current state and the i-th sample. is the 2-norm, used to characterize the Euclidean distance between two phase offset matrices.

[0035] Example 2; Based on the same inventive concept as the multi-chip IGBT non-uniform aging monitoring method in the foregoing embodiments, the present invention also provides a multi-chip IGBT non-uniform aging monitoring system, the system comprising: The power excitation acquisition module applies periodic power excitation to the multi-chip IGBT power module and acquires the case temperature response of each IGBT chip in the multi-chip IGBT power module. The shell temperature matrix construction module constructs a shell temperature matrix based on the shell temperature response. The phase feature extraction module performs multivariate variational mode decomposition on the shell temperature matrix and obtains phase information corresponding to different characteristic frequency bands to construct the phase matrix; The phase lag calculation module determines the phase lag matrix based on the phase matrix and the reference phase matrix corresponding to the health state. The aging state determination module matches the distribution characteristics of the phase lag matrix in different characteristic frequency bands and the corresponding positions of each IGBT chip with the preset aging state mapping relationship to determine the aging type, aging degree and uneven aging state of the multi-chip IGBT power module.

[0036] The system described above in this invention can effectively implement a method for monitoring uneven aging of multi-chip IGBTs, and the technical effects it can achieve are as described in the above embodiments, and will not be repeated here.

[0037] Example 3; Based on the same inventive concept as the multi-chip IGBT non-uniform aging monitoring method in the foregoing embodiments, the present invention also provides a computer-readable storage medium storing a computer program, the computer program including program instructions, which, when executed by a processor, can realize the multi-chip IGBT non-uniform aging monitoring method.

[0038] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for monitoring uneven aging of multi-chip IGBTs, characterized in that, The method includes: Periodic power excitation is applied to the multi-chip IGBT power module, and the case temperature response of each IGBT chip in the multi-chip IGBT power module is collected. Construct a shell temperature matrix based on the shell temperature response; The shell temperature matrix is ​​subjected to multivariate variational mode decomposition, and the phase information corresponding to different characteristic frequency bands is obtained to construct the phase matrix; The phase hysteresis matrix is ​​determined based on the phase matrix and the reference phase matrix corresponding to the health state. Based on the distribution characteristics of the phase lag matrix in different characteristic frequency bands and at corresponding positions of each IGBT chip, it is matched with a preset aging state mapping relationship to determine the aging type, aging degree, and uneven aging state of the multi-chip IGBT power module.

2. The method for monitoring uneven aging of multi-chip IGBTs according to claim 1, characterized in that, The periodic power excitation is a sinusoidal half-wave loss excitation with a frequency of 1Hz applied to each of the IGBT chips. The housing temperature response is obtained by synchronously collecting data at multiple temperature monitoring points located at the bottom of the housing of the multi-chip IGBT power module and corresponding to each of the IGBT chips at a preset sampling frequency.

3. The method for monitoring uneven aging of multi-chip IGBTs according to claim 2, characterized in that, The shell temperature matrix is ​​constructed by arranging the shell temperature responses of each shell according to the sampling time of the monitoring channels corresponding to multiple temperature monitoring points.

4. The method for monitoring uneven aging of multi-chip IGBTs according to claim 1, characterized in that, The shell temperature matrix is ​​subjected to multivariate variational mode decomposition, including: The shell temperature matrix is ​​decomposed into multiple modal components, wherein each modal component includes the channel mode corresponding to each monitoring channel, and the channel modes belonging to the same modal component have the same center frequency; Perform Hilbert transform on each of the channel modes to obtain the analytic signal of each of the channel modes; The goal is to minimize the sum of the bandwidths of each analytical signal after demodulation at the center frequency, and the constraint is to reconstruct the shell temperature matrix by the sum of each modal component, thereby solving for multiple modal components.

5. The method for monitoring uneven aging of multi-chip IGBTs according to claim 4, characterized in that, To obtain phase information corresponding to different characteristic frequency bands and construct a phase matrix, including: Perform a discrete Fourier transform on each of the channel modes in each of the modal components to obtain the frequency domain signal of each of the channel modes; Within the first characteristic frequency band of 0 to 0.1 Hz (excluding 0 Hz) and the second characteristic frequency band of 0.1 to 100 Hz, the frequency with the largest frequency domain amplitude in each monitoring channel is determined, and the determined frequencies are used to form the characteristic frequency set of the corresponding monitoring channel. For each characteristic frequency in the set of characteristic frequencies, determine the target modal component with the largest frequency domain amplitude at the corresponding characteristic frequency; The phase of the target modal component at the corresponding characteristic frequency is obtained, and the phase matrix is ​​constructed according to the phase corresponding to each monitoring channel.

6. The method for monitoring uneven aging of multi-chip IGBTs according to claim 5, characterized in that, The phase lag matrix is ​​constructed based on the phase difference at the corresponding characteristic frequencies of each monitoring channel in the current state and the healthy state. The phase lag corresponding to the first characteristic frequency band is used to characterize the heat transfer path state from the housing to the environment, and the phase lag corresponding to the second characteristic frequency band is used to characterize the heat transfer path state from the chip junction to the housing.

7. The method for monitoring uneven aging of multi-chip IGBTs according to claim 6, characterized in that, The phase lag matrix is ​​determined based on the phase matrix and the reference phase matrix corresponding to the health state, including: The phase lag of each monitoring channel in the current state is obtained by subtracting the reference phase of the corresponding monitoring channel in the healthy state at the same characteristic frequency. This is done by taking the phase lag of each monitoring channel in the first characteristic frequency band and the second characteristic frequency band. The phase lag matrix is ​​constructed based on the phase lag corresponding to each monitoring channel, wherein the phase lag in the first characteristic frequency band is used to characterize the change in thermal resistance from shell to environment, and the phase lag in the second characteristic frequency band is used to characterize the change in thermal resistance between the junction and shell.

8. The method for monitoring non-uniform aging of multi-chip IGBTs according to claim 7, characterized in that, Matching with the preset aging state mapping relationship, including: Calculate the Euclidean distance between the phase lag matrix in the current state and the phase lag matrix of each sample in the multidimensional lookup table, and determine the mapping sample with the smallest Euclidean distance as the target mapping sample; Based on the aging degree of the solder layer and the aging degree of the heat sink corresponding to the target mapping sample, the aging type and aging degree of the multi-chip IGBT power module are determined. In response to the fact that the phase lag of some monitoring channels increases in the second characteristic frequency band but the phase lag in the first characteristic frequency band does not increase accordingly, it is determined that the corresponding IGBT chip has uneven aging of the solder layer. In response to the fact that the phase hysteresis of each of the monitoring channels increases in the second characteristic frequency band and the increase is similar, it is determined that the multi-chip IGBT power module has uniform aging of the solder layer. In response to the fact that the phase lag of each of the monitoring channels increases in the first characteristic frequency band but does not increase accordingly in the second characteristic frequency band, it is determined that the multi-chip IGBT power module has heat sink aging.

9. A multi-chip IGBT non-uniform aging monitoring system, characterized in that, The system includes: The power excitation acquisition module applies periodic power excitation to the multi-chip IGBT power module and acquires the case temperature response of each IGBT chip in the multi-chip IGBT power module. The shell temperature matrix construction module constructs a shell temperature matrix based on the shell temperature response. The phase feature extraction module performs multivariate variational mode decomposition on the shell temperature matrix and obtains phase information corresponding to different characteristic frequency bands to construct the phase matrix; The phase lag calculation module determines the phase lag matrix based on the phase matrix and the reference phase matrix corresponding to the health state. The aging state determination module matches the distribution characteristics of the phase lag matrix in different characteristic frequency bands and the corresponding positions of each IGBT chip with the preset aging state mapping relationship to determine the aging type, aging degree and uneven aging state of the multi-chip IGBT power module.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which includes program instructions that, when executed by a processor, can implement the multi-chip IGBT non-uniform aging monitoring method as described in any one of claims 1-8.

Images