Inverter IGBT junction temperature prediction method and related equipment

Abstract

The invention relates to an inverter IGBT junction temperature prediction method and related equipment, and the method comprises the following steps: generating a loss characteristic spectrum through collecting IGBT electrical parameters and analyzing switching loss; calculating surface heat flux density, and constructing a heat flux distribution matrix. And performing thermal-electric coupling analysis by combining the two to obtain a junction temperature dynamic response sequence. Fourier transform is utilized to extract temperature fluctuation harmonic features, and a feature set is constructed; and extrapolating and predicting a junction temperature change trend based on the feature set, and outputting an IGBT junction temperature early warning signal through threshold judgment. The technical problem that the service life of a device is affected due to the fact that a large amount of loss is generated due to frequent switching actions in the working process of the inverter, the junction temperature of a chip rises and even temperature fluctuation occurs is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of inverter IGBTs, and particularly to a method for predicting the junction temperature of an inverter IGBT and related devices. Background Art

[0002] With the continuous improvement of industrial automation levels, inverters, as core components in motor drive systems, directly affect the stability and safety of the entire system in terms of their operating reliability. In an inverter, an insulated gate bipolar transistor (IGBT), as a key power device, undertakes the important task of electric energy conversion. During its operation, a large amount of loss is generated due to frequent switching actions, resulting in an increase in the chip junction temperature and even temperature fluctuations, thereby affecting the device life and system performance. Therefore, accurate prediction and real-time monitoring of the IGBT junction temperature have become one of the key technologies for improving the reliability of inverters and realizing condition-based maintenance.

[0003] Currently, traditional IGBT temperature detection methods mainly rely on temperature sensors on the package housing or radiator to estimate the chip temperature through indirect measurement. However, this off-line and single-point temperature measurement method is difficult to reflect the true and dynamically changing junction temperature distribution inside the chip. Especially under load mutations or high-frequency switching conditions, there are problems such as response lag and low accuracy. In addition, existing methods mostly ignore the loss characteristics during the IGBT switching process and their influence on the thermal distribution, lacking research on establishing a thermal-electrical coupling model based on electrical characteristics, which limits the accuracy and practicality of junction temperature prediction.

[0004] Although some studies have attempted to introduce loss calculation and thermal simulation means to improve the accuracy of IGBT temperature analysis, most of them stay at the static or steady-state analysis level and fail to fully consider the dynamic thermal effects brought by complex operating conditions in actual operation. At the same time, there is still a lack of effective signal processing means for extracting the characteristics of junction temperature fluctuations and extrapolating trends, making it difficult to achieve a forward-looking warning of the IGBT health state. Therefore, there is an urgent need for an IGBT junction temperature prediction method that can integrate electrical parameter acquisition, heat flow distribution modeling, and dynamic temperature prediction to solve the problems of poor dynamic response, low prediction accuracy, and weak warning ability existing in current research. Summary of the Invention

[0005] The main object of the present invention is to provide a method for predicting the junction temperature of an inverter IGBT and related devices, which solves the technical problem that in an inverter, a large amount of loss is generated due to frequent switching actions during operation, resulting in an increase in the chip junction temperature and even temperature fluctuations, thereby affecting the device life.

[0006] To achieve the above object, the present invention provides a method for predicting the junction temperature of an inverter IGBT, including the following steps: The electrical parameters of the inverter IGBT are collected through a voltage-current sensor and the switching losses are analyzed to obtain the IGBT loss characteristic map, where the IGBT loss characteristic map includes an on-state loss component and an off-state loss component; The heat flux density of the surface of the inverter IGBT is calculated through a thermocouple array to obtain the IGBT heat flux distribution matrix; Based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix, a thermo-electrical coupling analysis is performed on the inverter IGBT to obtain the IGBT junction temperature dynamic response sequence; The harmonic component analysis is performed on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation characteristic set; Based on the IGBT temperature fluctuation characteristic set, temperature trend extrapolation is performed to obtain the IGBT junction temperature prediction sequence, and threshold judgment analysis is performed on the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature warning signal.

[0007] Further, the heat flux density of the surface of the inverter IGBT is calculated through a thermocouple array to obtain the IGBT heat flux distribution matrix, including: Multiple micro-region grid cells are divided on the surface of the inverter IGBT, and the temperature gradients of each micro-region grid cell are collected through a thermocouple array to obtain the surface temperature distribution data set; Based on the surface temperature distribution data set, the thermal conductivity tensor is decomposed to obtain the thermal conductivity eigenvector, and the anisotropic heat conduction calculation is performed on the thermal conductivity eigenvector to obtain the heat flux density component matrix; The heat flux divergence analysis is performed on the heat flux density component matrix through the Laplace operator to obtain the heat flux field topological structure, and the heat flux integral operation is performed based on the heat flux field topological structure to obtain the IGBT heat flux distribution matrix.

[0008] Further, the thermo-electrical coupling analysis is performed on the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain the IGBT junction temperature dynamic response sequence, including: The power density distribution calculation is performed on the IGBT loss characteristic map to obtain the instantaneous power density distribution map on the surface of the inverter IGBT, and the heat source distribution characteristics are extracted based on the instantaneous power density distribution map to obtain the IGBT heat source spatial distribution eigenvector; Based on the IGBT heat source spatial distribution eigenvector and the IGBT heat flux distribution matrix, the heat diffusion equation is solved to obtain the temperature field distribution function of the inverter IGBT, and the boundary condition constraint analysis is performed on the temperature field distribution function to obtain the transient temperature response characteristic set of the inverter IGBT; By means of the non - linear thermal resistance network technology, calculate the dynamic thermal resistance of the transient temperature response feature set of the IGBT of the inverter, obtain the thermal impedance matrix of the IGBT multi - layer structure, and construct a temperature transfer function based on the thermal impedance matrix to obtain the IGBT junction temperature transfer characteristic equation; Based on the IGBT junction temperature transfer characteristic equation, conduct a transient thermal response analysis on the IGBT of the inverter to obtain the IGBT junction temperature dynamic response sequence, and extract the time - domain characteristics of the IGBT junction temperature dynamic response sequence to obtain the IGBT junction temperature fluctuation characteristic spectrum.

[0009] Furthermore, perform harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi - dimensional Fourier transform to obtain the IGBT temperature fluctuation feature set, including: Conduct a time - frequency domain mapping transformation on the IGBT junction temperature dynamic response sequence to obtain the IGBT temperature fluctuation time - frequency characteristic matrix, and perform wavelet packet decomposition operation on the temperature fluctuation time - frequency characteristic matrix to obtain the IGBT temperature harmonic energy distribution map; Conduct decoupling calculation of harmonic components on the IGBT temperature harmonic energy distribution map to obtain the IGBT temperature harmonic feature vector group, and perform non - linear phase modulation analysis on the temperature harmonic feature vector group to obtain the IGBT temperature harmonic modulation feature set; Extract the instantaneous frequency of the IGBT temperature harmonic modulation feature set through multi - dimensional Fourier transform to obtain the IGBT temperature harmonic frequency response curve, and analyze the harmonic interference characteristics of the temperature harmonic frequency response curve to obtain the IGBT temperature harmonic interference characteristic matrix; Based on the IGBT temperature harmonic interference characteristic matrix, conduct harmonic component reconstruction operation to obtain the IGBT temperature fluctuation feature set.

[0010] Furthermore, conduct temperature trend extrapolation based on the IGBT temperature fluctuation feature set to obtain the IGBT junction temperature prediction sequence, including: Conduct non - linear trend decomposition operation on the IGBT temperature fluctuation feature set to obtain the IGBT temperature evolution feature vector, and perform dynamic singular value decomposition on the IGBT temperature evolution feature vector to obtain the IGBT temperature trend principal component matrix; Based on the IGBT temperature trend principal component matrix, extract temperature topological features to obtain the IGBT temperature topological feature map, and conduct phase - space reconstruction analysis on the IGBT temperature topological feature map to obtain the IGBT temperature dynamics feature set; Conduct stability assessment on the IGBT temperature dynamics feature set through Lyapunov exponent analysis to obtain the IGBT temperature stability index sequence, and identify bifurcation points based on the temperature stability index sequence to obtain the IGBT temperature evolution bifurcation feature map; Perform non - linear extrapolation calculation on the IGBT temperature evolution bifurcation characteristic diagram to obtain the IGBT temperature trend prediction matrix, and reconstruct the temperature trajectory based on the temperature trend prediction matrix to obtain the initial value of the IGBT junction temperature prediction sequence; Through the adaptive Kalman filtering technology, dynamically correct the initial value of the IGBT junction temperature prediction sequence to obtain the corrected value of the IGBT junction temperature prediction sequence, and analyze the confidence interval of the corrected value of the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature prediction sequence.

[0011] Further, the performing non - linear extrapolation calculation on the IGBT temperature evolution bifurcation characteristic diagram to obtain the IGBT temperature trend prediction matrix includes: Perform fractal dimension decomposition operation on the IGBT temperature evolution bifurcation characteristic diagram to obtain the fractal feature vector of the temperature evolution trajectory, and calculate the Hausdorff dimension of the fractal feature vector of the temperature evolution trajectory to obtain the IGBT temperature fractal feature matrix; Based on the IGBT temperature fractal feature matrix, perform phase - change point identification operation on the inverter IGBT to obtain the IGBT temperature phase - change characteristic sequence, and analyze the temperature critical exponent of the IGBT temperature phase - change characteristic sequence to obtain the IGBT temperature critical characteristic spectrum; Extract the periodic orbit through the Poincaré map for the IGBT temperature critical characteristic spectrum to obtain the IGBT temperature periodic feature vector group, and reconstruct the chaotic attractor based on the temperature periodic feature vector group to obtain the IGBT temperature chaotic feature matrix; Construct a non - linear predictor for the IGBT temperature chaotic feature matrix to obtain the IGBT temperature trend prediction operator, and perform temperature trajectory extrapolation operation based on the temperature trend prediction operator to obtain the IGBT temperature trend prediction matrix.

[0012] Further, the performing phase - change point identification operation on the inverter IGBT based on the IGBT temperature fractal feature matrix to obtain the IGBT temperature phase - change characteristic sequence includes: Perform entropy value dynamic decomposition operation on the IGBT temperature fractal feature matrix to obtain the entropy value feature vector of the temperature sequence, and detect the phase - change critical point of the entropy value feature vector of the temperature sequence to obtain the IGBT temperature phase - change threshold matrix; Based on the IGBT temperature phase - change threshold matrix, perform temperature gradient tensor decomposition to obtain the IGBT temperature gradient feature field, and reconstruct the topological structure of the temperature gradient feature field to obtain the IGBT temperature topological feature sequence; Analyze the heat conduction characteristics of the IGBT temperature topological feature sequence through the hyperbolic diffusion equation to obtain the IGBT heat diffusion eigenvector group, and reconstruct the heat flux density field based on the heat diffusion eigenvector group to obtain the IGBT temperature field evolution matrix; Conduct non-equilibrium thermodynamics analysis on the IGBT temperature field evolution matrix to obtain the IGBT temperature phase change feature sequence.

[0013] The present invention also provides an inverter IGBT junction temperature prediction device, including: An acquisition module, configured to collect electrical parameters of the inverter IGBT through a voltage-current sensor and analyze the switching losses to obtain an IGBT loss feature map, where the IGBT loss feature map includes turn-on loss components and turn-off loss components; A calculation module, configured to calculate the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain an IGBT heat flux distribution matrix; A first analysis module, configured to perform thermoelectric coupling analysis on the inverter IGBT based on the IGBT loss feature map and the IGBT heat flux distribution matrix to obtain an IGBT junction temperature dynamic response sequence; A second analysis module, configured to perform harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain an IGBT temperature fluctuation feature set; A prediction module, configured to perform temperature trend extrapolation based on the IGBT temperature fluctuation feature set to obtain an IGBT junction temperature prediction sequence, and perform threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain an IGBT junction temperature warning signal.

[0014] The present invention also provides a computer device, including a memory and a processor, where a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method described in any one of the above are implemented.

[0015] The present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0016] A method for predicting the junction temperature of an inverter IGBT provided by the present invention includes the following steps: collecting electrical parameters of the inverter IGBT and analyzing the switching losses to obtain an IGBT loss characteristic map, calculating the heat flux density on the surface of the inverter IGBT to obtain an IGBT heat flux distribution matrix; performing a thermoelectric coupling analysis on the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain an IGBT junction temperature dynamic response sequence; performing a harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain an IGBT temperature fluctuation feature set; extrapolating the temperature trend based on the IGBT temperature fluctuation feature set to obtain an IGBT junction temperature prediction sequence, and performing a threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain an IGBT junction temperature warning signal, which solves the technical problem that in the inverter, a large amount of losses are generated due to frequent switching actions during operation, resulting in an increase in the chip junction temperature and even temperature fluctuations, thereby affecting the device life. The technical effect of realizing the harmonic component analysis of the junction temperature dynamic response sequence by using multi-dimensional Fourier transform can accurately identify the periodic and non-periodic components in the IGBT temperature fluctuation, which helps to reveal the thermal behavior law of the device under high-frequency switching or load mutation conditions and provides data support for the health state assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic diagram of the steps of a method for predicting the junction temperature of an inverter IGBT in an embodiment of the present invention; Figure 2 is a structural block diagram of an inverter IGBT junction temperature prediction device in an embodiment of the present invention; Figure 3 is a schematic structural diagram of a computer device in an embodiment of the present invention.

[0018] The implementation, functional features and advantages of the object of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0020] As Figure 1 shown, Figure 1 is a schematic diagram of the steps of a method for predicting the junction temperature of an inverter IGBT in an embodiment of the present invention; An embodiment of the present invention provides a method for predicting the junction temperature of an inverter IGBT, including the following steps: Step S1, collect electrical parameters of the inverter IGBT through a voltage-current sensor and analyze the switching losses to obtain an IGBT loss characteristic map, where the IGBT loss characteristic map includes turn-on loss components and turn-off loss components.

[0021] Specifically, the process of collecting electrical parameters of the inverter IGBT through a voltage-current sensor and analyzing the switching losses to obtain an IGBT loss characteristic map is a key step to ensure the efficient and safe operation of the IGBT module in the inverter. In actual operation, first, the voltage-current sensor needs to be accurately installed at the input and output ends of the inverter IGBT to monitor the voltage and current changes during its operation in real time. These sensors can capture the transient responses of the IGBT at the moment of turn-on and turn-off, including phenomena such as voltage dips and current surges, thus providing basic data for subsequent switching loss analysis. Then, by deeply analyzing the collected data, the turn-on loss components and turn-off loss components can be separated, and then an IGBT loss characteristic map can be constructed. This process not only requires accurate measurement of the energy loss in each switching cycle but also needs to consider the dynamic characteristics under different working conditions. For example, in an industrial automation system, a motor drive device controlled by an inverter, when it starts or the load changes, the IGBT needs to be frequently turned on and off to adjust the speed and torque of the motor. During this process, the IGBT will experience significant voltage and current fluctuations. Using the above method, the voltage-current sensor is used to collect these fluctuation data in real time and analyze the turn-on and turn-off losses contained in them, and we can obtain a detailed IGBT loss characteristic map. This map reflects the energy consumption pattern of the IGBT under different load conditions, helping engineers identify potential risk points, such as local overheating problems that may be caused by excessive switching losses. Based on this information, the design and control strategy of the inverter can be further optimized to improve the overall efficiency and reliability of the system. At the same time, this method also provides key input parameters for subsequent thermo-electric coupling analysis, making it possible to more accurately predict the IGBT junction temperature.

[0022] Step S2, calculate the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain an IGBT heat flux distribution matrix.

[0023] Specifically, the process of calculating the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain the IGBT heat flux distribution matrix is an important link in accurately perceiving the thermal behavior of the IGBT module in the inverter. In this step, multiple micro-thermocouples need to be integrated on the surface of the inverter IGBT according to a predetermined spatial layout to form a thermocouple array, so as to achieve multi-point synchronous measurement of the temperature distribution in different regions of the chip. By collecting the temperature data of these measurement points in real time and performing spatial interpolation processing, an image of the temperature field distribution on the chip surface can be obtained. On this basis, combined with the heat transfer theory model, the finite difference or finite element method is further used to reverse deduce the heat conduction process, so as to calculate the heat flux density distribution in each region. Finally, these heat flux density values are arranged in matrix form according to the spatial position, that is, the IGBT heat flux distribution matrix is obtained, providing the key spatial heat source input for the subsequent thermal-electrical coupling analysis. For example, in an industrial automation system, a motor device driven by an inverter may generate uneven heat distribution inside the inverter IGBT during operation due to load fluctuations or control strategy changes. Especially under high-frequency switching and high-current conditions, local hot spots may appear in some regions. At this time, through the thermocouple array arranged on the surface of the inverter IGBT, the temperature gradient information of these hot spot regions can be accurately captured, and the corresponding heat flux density distribution can be calculated accordingly. The obtained IGBT heat flux distribution matrix not only reflects the overall heat generation situation of the chip, but also can reveal the heat conduction path and heat accumulation trend, providing a physical basis and data support for the modeling of the junction temperature dynamic response sequence, and further improving the accuracy and reliability of the entire junction temperature prediction system.

[0024] Step S3: Based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix, perform thermal-electrical coupling analysis on the inverter IGBT to obtain the IGBT junction temperature dynamic response sequence.

[0025] Specifically, the process of performing thermo-electric coupling analysis on the IGBT of the inverter based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain the IGBT junction temperature dynamic response sequence is a key step in unifying the electrical loss and heat conduction behavior in the time and space dimensions. In this step, the IGBT loss characteristic map obtained in the previous link is first used as the heat source input, and the turn-on loss component and turn-off loss component contained therein respectively correspond to the instantaneous energy dissipation of the IGBT in different switching states. At the same time, the IGBT heat flux distribution matrix calculated by the thermocouple array is combined, and this matrix reflects the heat flux density distribution of each surface region and its change trend over time. By establishing a thermo-electric coupling model, the electrical loss is converted into the local heat generation rate, and the heat conduction and diffusion processes in the chip interior and the packaging structure are simulated using the heat transfer equation, thereby realizing the dynamic simulation of the internal junction temperature of the inverter IGBT. For example, in an industrial automation system, when a motor device driven by an inverter is in a running state of frequent start-stop or load mutation, the inverter IGBT will generate corresponding temperature fluctuations due to the periodic change of the switching loss. At this time, by synchronously inputting the real-time updated IGBT loss characteristic map and the IGBT heat flux distribution matrix into the thermo-electric coupling model, the temperature change sequence of the key nodes (i.e., junctions) inside the chip can be dynamically solved to form the IGBT junction temperature dynamic response sequence. This sequence not only contains the change trend of the junction temperature over time, but also reflects the balance relationship between heat accumulation and heat dissipation under different working conditions, providing a high-precision data basis for subsequent temperature fluctuation feature extraction and junction temperature prediction, thereby effectively supporting the health status monitoring and fault warning capabilities of the inverter system.

[0026] Step S4: Perform harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation feature set.

[0027] Specifically, the process of performing harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation feature set is to transform the complex junction temperature change signal in the time domain into multiple harmonic components in the frequency domain, thereby revealing the inherent periodic and aperiodic laws of temperature fluctuations of the IGBT under different working conditions. In this step, first, based on the IGBT junction temperature dynamic response sequence obtained in the previous step, which is used as the input signal, multi-dimensional Fourier transform technology is adopted to perform spectral decomposition on the time series data of the junction temperature, and the main frequency components contained therein and their corresponding amplitude and phase information are identified. These harmonic components can reflect the periodic temperature fluctuation characteristics of the IGBT during the switching action, load change, and heat accumulation process, and then form an IGBT temperature fluctuation feature set containing multiple frequency components. For example, in an industrial automation system, for a motor device controlled by an inverter, under high-frequency modulation or periodic load impact, obvious periodic temperature fluctuations will occur inside the inverter IGBT due to the repetitive change of switching losses. At this time, by performing multi-dimensional Fourier transform on the IGBT junction temperature dynamic response sequence, the dominant harmonic components related to the switching frequency, load cycle, etc. can be effectively extracted, and the IGBT temperature fluctuation feature set can be constructed accordingly. This feature set not only helps to identify abnormal modes of the chip's thermal behavior but also provides key feature parameters for subsequent temperature trend extrapolation and health status assessment, improving the accuracy and early warning ability of junction temperature prediction.

[0028] Step S5: Based on the IGBT temperature fluctuation feature set, perform temperature trend extrapolation to obtain the IGBT junction temperature prediction sequence, and perform threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature early warning signal.

[0029] Specifically, the process of extrapolating the temperature trend based on the IGBT temperature fluctuation feature set to obtain the IGBT junction temperature prediction sequence and performing threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature warning signal is an important part of realizing the state monitoring and fault prevention of the inverter system. In this step, the IGBT temperature fluctuation feature set extracted previously is first used as the input feature of the time series prediction model. By fitting the historical temperature fluctuation law, mathematical methods such as ARIMA, LSTM, or polynomial regression are used to extrapolate the future junction temperature change trend, thereby generating a forward-looking IGBT junction temperature prediction sequence. This sequence reflects the evolution trend of the internal junction temperature of the inverter IGBT in the future under the current operating conditions. Then, the prediction sequence is compared with the set safety threshold in real time. Once the predicted junction temperature exceeds the preset temperature upper limit threshold or its rising rate accelerates abnormally, it is determined that there is an overheating risk, and the IGBT junction temperature warning signal is triggered. For example, in an industrial automation system, when a motor device driven by an inverter gradually increases the temperature of the inverter IGBT due to continuous high-load operation, this warning mechanism can issue an alarm before actual thermal damage occurs, reminding the control system to take protection measures such as frequency reduction, current limiting, or shutdown, thereby effectively avoiding device failure and ensuring the stability and safety of the system.

[0030] In a specific embodiment, the calculation of the heat flux density on the surface of the inverter IGBT by the thermocouple array to obtain the IGBT heat flux distribution matrix includes: Dividing multiple micro-region grid units on the surface of the inverter IGBT, and collecting the temperature gradients of each micro-region grid unit through the thermocouple array to obtain the surface temperature distribution data set. Among them, the surface temperature distribution data set includes the grid unit temperature value and the grid unit boundary temperature gradient; Based on the surface temperature distribution data set, perform thermal conductivity tensor decomposition to obtain the thermal conductivity eigenvector, and perform anisotropic heat conduction calculation on the thermal conductivity eigenvector to obtain the heat flux density component matrix. Among them, the heat flux density component matrix includes the radial heat flux density component and the tangential heat flux density component; Perform heat flux divergence analysis on the heat flux density component matrix through the Laplace operator to obtain the heat flux field topology structure, and perform heat flux integral operation based on the heat flux field topology structure to obtain the IGBT heat flux distribution matrix. Among them, the IGBT heat flux distribution matrix includes the chip center heat flux density and the edge heat flux density.

[0031] Specifically, in the above step, "calculating the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain the IGBT heat flux distribution matrix" is a technical process that integrates spatial discretization modeling, multi-dimensional temperature sensing, and heat conduction physical analysis. The core of this method lies in dividing the surface of the inverter IGBT into multiple micro-region grid cells to achieve refined spatial modeling and analysis of the temperature field and heat flux field. Specifically, first, according to the geometric dimensions and structural characteristics of the inverter IGBT, a two-dimensional grid system with a fixed resolution is constructed on its surface. For example, the 10mm×10mm chip surface is evenly divided into 100 1mm×1mm micro-region grid cells, and each grid cell corresponds to a thermocouple measurement point, thus forming a high-density thermocouple array. Through the thermocouple sensors arranged in each micro-region grid cell, the real-time temperature values at each position can be synchronously collected to form a surface temperature distribution data set, which includes not only the temperature values at the center points of each grid cell (such as the temperature at the center region is 92°C and at the edge region is 85°C at a certain moment), but also the boundary temperature gradients between adjacent grid cells (for example, there is a temperature gradient of 7°C / mm between two adjacent cells). These data constitute the basic input for subsequent thermal conductivity tensor decomposition and heat flux density calculation. Based on the above surface temperature distribution data set, the thermal conductivity tensor decomposition technology is further used to perform directional modeling of the heat conduction characteristics of the inverter IGBT material. Since the inverter IGBT is usually composed of multiple layers of different materials, such as silicon substrate, metal layer, and ceramic insulation layer, its thermal conductivity has significant anisotropic characteristics. Therefore, by extracting the eigenvectors of the thermal conductivity tensor, the main heat conduction directions on the chip surface and their corresponding heat conduction capabilities can be identified. For example, the thermal conductivity along the radial direction in the central region of the chip is 180W / (m·K), while it is only 60W / (m·K) in the tangential direction. Subsequently, using these thermal conductivity eigenvectors for anisotropic heat conduction calculation, the heat flux density component matrix inside each micro-region grid cell is obtained, which includes the radial heat flux density component (such as 3.2W / cm² in the central region) and the tangential heat flux density component (such as 1.1W / cm² in the edge region), thus reflecting the difference in heat transfer intensity in different directions on the chip surface. On this basis, the Laplace operator is further introduced to analyze the heat flux divergence of the heat flux density component matrix, which is used to reveal the topological structure characteristics of the overall heat flux field on the chip surface. This process can identify the heat source concentration region (i.e., the positive divergence region) and the heat convergence region (i.e., the negative divergence region). For example, a positive divergence value as high as 0.5W / cm³ may appear in the central region of the chip, indicating that this is the main heat generation area; while a negative divergence value of -0.3W / cm³ may be detected near the package edge, indicating that this region is the main heat diffusion path. The obtained heat flux field topological structure not only helps to understand the heat flow mechanism inside the chip, but also provides a spatial distribution basis for subsequent heat flux integral operation.Finally, by performing a heat flux integration operation on the topological structure of the heat flow field and accumulating the heat flux density values in all micro-region grid cells according to their spatial positions, a complete IGBT heat flow distribution matrix can be obtained. This matrix includes the heat flux density at the center of the chip (e.g., 3.5 W / cm² in the central region) and the heat flux density at the edge (e.g., 1.2 W / cm² in the edge region), thus achieving a leap from local measurement to global heat behavior characterization. This heat flow distribution matrix is not only an important input parameter for thermo-electrical coupling analysis but also provides crucial heat source distribution information for the subsequent junction temperature prediction model. For example, in an industrial automation system, a motor drive device controlled by an inverter generates a large amount of loss due to frequent switching operations during continuous heavy-duty operation, resulting in a local temperature rise. At this time, through the above-mentioned heat flux density calculation method based on the thermocouple array, the high-temperature hot spot in the central region of the chip can be accurately captured, and its corresponding high heat flux density distribution can be identified. If the heat flux density in this region exceeds the design threshold (e.g., 4.0 W / cm²), the system can issue a warning signal accordingly, prompting the control system to take corresponding cooling and protection measures, thereby effectively preventing device overheating failure and ensuring the stable operation and safety of the system.

[0032] In a specific embodiment, the thermo-electrical coupling analysis of the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flow distribution matrix to obtain the IGBT junction temperature dynamic response sequence includes: Performing a power density distribution calculation on the IGBT loss characteristic map to obtain an instantaneous power density distribution map on the surface of the inverter IGBT, and extracting the heat source distribution characteristics based on the instantaneous power density distribution map to obtain an IGBT heat source spatial distribution feature vector, where the IGBT heat source spatial distribution feature vector includes the hot spot distribution coordinates on the chip surface and the heat source intensity distribution coefficient; Solving the heat diffusion equation based on the IGBT heat source spatial distribution feature vector and the IGBT heat flow distribution matrix to obtain the temperature field distribution function of the inverter IGBT, and performing boundary condition constraint analysis on the temperature field distribution function to obtain the transient temperature response feature set of the inverter IGBT, where the transient temperature response feature set of the inverter IGBT includes the temperature gradient change rate and the heat diffusion time constant; Calculating the dynamic thermal resistance of the transient temperature response feature set of the inverter IGBT through the nonlinear thermal resistance network technology to obtain the IGBT multi-layer structure thermal impedance matrix, and constructing a temperature transfer function based on the thermal impedance matrix to obtain the IGBT junction temperature transfer characteristic equation, where the IGBT junction temperature transfer characteristic equation includes the thermal resistance time constant and the heat capacity coupling coefficient; Based on the IGBT junction temperature transfer characteristic equation, transient thermal response analysis is carried out on the IGBT of the inverter to obtain the IGBT junction temperature dynamic response sequence, and the time-domain characteristics of the IGBT junction temperature dynamic response sequence are extracted to obtain the IGBT junction temperature fluctuation characteristic spectrum, where the IGBT junction temperature fluctuation characteristic spectrum includes the temperature fluctuation amplitude and the phase delay characteristic.

[0033] Specifically, the process of performing thermo - electrical coupling analysis on the IGBT of the inverter based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain the IGBT junction temperature dynamic response sequence is a comprehensive technical process that integrates power loss modeling, heat conduction physical simulation, and multi - layer structure thermal resistance network modeling. This method first starts from the electrical characteristics, calculates the power density distribution by analyzing the IGBT loss characteristic map obtained previously, converts the turn - on loss component and turn - off loss component during the switching process into energy input per unit area, and forms the instantaneous power density distribution map on the surface of the inverter IGBT. For example, during a certain high - frequency switching process, the total loss of the IGBT within one cycle is measured as 2.5W by voltage and current sensors, and combined with the chip area (such as 10mm×10mm), the average power density can be calculated to be approximately 25W / cm². Further, by combining the distribution characteristics of the loss over time, high - loss regions can be identified, and thus the IGBT heat source spatial distribution characteristic vector can be extracted, which includes the hot spot distribution coordinates (such as the temperature is the highest at the center point (5,5)) and the heat source intensity distribution coefficient (such as the intensity in the central region is 1.8 times that of the edge region), used to describe the local heating differences in each region of the chip surface. On this basis, by combining the IGBT heat source spatial distribution characteristic vector extracted above with the IGBT heat flux distribution matrix obtained previously (including the heat flux density of 3.5W / cm² at the chip center and 1.2W / cm² at the edge), a heat diffusion equation is constructed and numerically solved to simulate the heat propagation process within the chip. In this step, the partial differential equation of heat conduction is discretized by the finite - difference method or the finite - element method, considering parameters such as material thermal conductivity, density, and specific heat capacity, and finally the temperature field distribution function of the inverter IGBT is obtained. For example, at a certain moment, the temperature in the central region of the chip reaches 96℃, while the temperature in the edge region is 87℃, and there is a temperature gradient of approximately 9℃ between them. Subsequently, boundary conditions are imposed on the temperature field distribution function, such as the heat dissipation condition of the package bottom surface, ambient temperature setting, etc., and the IGBT transient temperature response characteristic set is further extracted, which includes the temperature gradient change rate (such as rising 0.4℃ / mm per millisecond) and the heat diffusion time constant (such as the heat diffusion time in the central region is 2.3ms), and these parameters reflect the thermal response ability of the chip at different positions and different times. Then, the non - linear thermal resistance network technology is used to model the above - mentioned transient temperature response characteristic set, and the multi - layer structure of the chip (such as silicon layer, solder layer, copper substrate, etc.) is abstracted into multiple series - connected and parallel - connected thermal resistance nodes, so as to calculate its dynamic thermal resistance characteristics. For example, the thermal resistance of the silicon layer is 0.5K / W, the solder layer is 0.3K / W, and the copper substrate is 0.1K / W, and the IGBT multi - layer structure thermal impedance matrix is constructed through superposition and coupling relationships.On this basis, a temperature transfer function model is further established, namely the IGBT junction temperature transfer characteristic equation, which can describe the dynamic mapping relationship between the input power fluctuation and the output junction temperature. It includes the thermal resistance time constant (such as 1.8 ms) and the heat capacity coupling coefficient (such as 0.7 J / (K·cm³)), and these parameters reflect the thermal inertia and response delay of the chip under dynamic load changes. Finally, based on the IGBT junction temperature transfer characteristic equation, the transient thermal response analysis of the inverter IGBT is carried out to simulate the junction temperature evolution process under actual working conditions and generate the IGBT junction temperature dynamic response sequence. For example, under continuous PWM modulation, the junction temperature shows periodic fluctuations, with a peak temperature of up to 115 °C, a valley value of 90 °C, and a fluctuation amplitude of 25 °C. Further extracting the time-domain characteristics of this sequence can obtain the IGBT junction temperature fluctuation characteristic spectrum, which includes the temperature fluctuation amplitude (such as ±12.5 °C) and the phase delay characteristic (such as lagging 0.6 ms relative to the power pulse). These characteristics not only reveal the thermal behavior law of the chip under complex operating conditions but also provide key data support for subsequent harmonic analysis and trend extrapolation. For example, in an industrial automation system, when a motor device driven by an inverter undergoes frequent start-stop or load mutation, the inverter IGBT will experience severe thermal shock. At this time, through the above thermal-electrical coupling analysis process, the change trend of the junction temperature can be predicted in real time, and potential overheating risks can be identified. If it is detected that the junction temperature fluctuation amplitude exceeds the design threshold (such as ±15 °C) after a certain operation, the system can issue an early warning signal in advance to prompt the controller to take protection measures such as frequency reduction and current limiting, thereby effectively avoiding device failure due to overheating and ensuring the long-term stable operation and safety reliability of the system.

[0034] In a specific embodiment, the harmonic component analysis of the IGBT junction temperature dynamic response sequence is performed through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation characteristic set, including: Perform a time-frequency domain mapping transformation on the IGBT junction temperature dynamic response sequence to obtain the IGBT temperature fluctuation time-frequency characteristic matrix, and perform a wavelet packet decomposition operation on the temperature fluctuation time-frequency characteristic matrix to obtain the IGBT temperature harmonic energy distribution map, where the IGBT temperature harmonic energy distribution map includes the fundamental frequency energy density coefficient and the harmonic energy attenuation rate; Perform a decoupling calculation of the harmonic components on the IGBT temperature harmonic energy distribution map to obtain the IGBT temperature harmonic characteristic vector group, and perform a non-linear phase modulation analysis on the temperature harmonic characteristic vector group to obtain the IGBT temperature harmonic modulation characteristic set, where the IGBT temperature harmonic modulation characteristic set includes the harmonic amplitude modulation coefficient and the phase modulation depth parameter; Instantaneously extract the frequency of the IGBT temperature harmonic modulation feature set through multi-dimensional Fourier transform to obtain the IGBT temperature harmonic frequency response curve, and analyze the harmonic interference characteristics of the temperature harmonic frequency response curve to obtain the IGBT temperature harmonic interference feature matrix, where the IGBT temperature harmonic interference feature matrix includes harmonic intermodulation components and harmonic crosstalk coefficients; Perform harmonic component reconstruction operations based on the IGBT temperature harmonic interference feature matrix to obtain the IGBT temperature fluctuation feature set.

[0035] Specifically, the process of performing harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation feature set is a high-order signal analysis method that integrates time-frequency signal processing, nonlinear modulation analysis and harmonic reconstruction technology. The core of this step is to convert the IGBT junction temperature dynamic response sequence that originally only exists in the time domain into a multi-dimensional representation in the frequency domain or even the time-frequency domain, thereby revealing the periodicity, non-periodicity and nonlinear coupling components hidden inside it, and finally constructing the IGBT temperature fluctuation feature set for subsequent prediction and early warning. Specifically, firstly, the IGBT junction temperature dynamic response sequence obtained in the previous sequence is transformed into a time-frequency domain mapping, that is, the one-dimensional time series signal is converted into a two-dimensional time-frequency feature matrix to form an IGBT temperature fluctuation time-frequency feature matrix. For example, during a certain inverter operation process, a 2-second IGBT junction temperature dynamic response sequence was collected, the sampling frequency was 1kHz, and a total of 2000 data points were included, and its temperature value fluctuated periodically between 90°C and 115°C. Through short-time Fourier transform (STFT) or continuous wavelet transform, it can be converted into a temperature fluctuation time-frequency feature matrix with dual dimensions of time and frequency, in which each row corresponds to the frequency distribution at a certain moment, and each column reflects the energy change of a certain frequency component over the entire time range. Subsequently, the temperature fluctuation time-frequency feature matrix is ​​subjected to wavelet packet decomposition to further refine the energy distribution characteristics of each frequency component. Wavelet packet decomposition can divide the original signal into different frequency bands, thereby extracting more detailed frequency sub-band information. For example, after completing the wavelet packet decomposition, it can be identified that the fundamental frequency energy density coefficient is 0.72 (that is, the fundamental frequency component accounts for 72% of the total energy), and the high-order harmonic energy shows an exponential decay trend with increasing frequency, of which the second harmonic accounts for about 18%, the third harmonic accounts for about 6%, and the fourth and above harmonics account for about 4% in total, thereby obtaining the IGBT temperature harmonic energy distribution diagram. This distribution diagram not only reflects the main frequency component corresponding to the main heating mechanism, but also reveals the degree of influence of high-frequency fluctuations on the thermal behavior of the chip. Next, the harmonic component decoupling calculation is performed on the above IGBT temperature harmonic energy distribution diagram, aiming to separate each independent harmonic component from the overall signal, and then construct the IGBT temperature harmonic feature vector group. The vector group consists of multiple feature vectors representing different frequency components, and each feature vector contains the corresponding amplitude, frequency and phase information. On this basis, nonlinear phase modulation analysis is further carried out to capture the interaction between harmonics caused by load mutation or control strategy adjustment.For example, during a certain motor startup process, a significant phase jitter phenomenon was detected in the IGBT temperature fluctuation signal. Among them, the second harmonic had a phase shift of approximately 12° relative to the fundamental frequency, and at the same time, its amplitude was modulated and amplified by 1.3 times. From this, the IGBT temperature harmonic modulation feature set can be extracted, including the harmonic amplitude modulation coefficient (such as 1.3) and the phase modulation depth parameter (such as ±12°). These parameters have important reference value for judging whether the chip is in an abnormal working condition. To further explore the transient features in the temperature fluctuation signal, multi-dimensional Fourier transform is used to extract the instantaneous frequency of the IGBT temperature harmonic modulation feature set, thereby generating the IGBT temperature harmonic frequency response curve. This curve can visually display the frequency drift of each harmonic component within different time periods. For example, within the first 50 ms after a load mutation occurs, the fundamental frequency briefly rises from 50 Hz to 53 Hz, while the second harmonic drops from 100 Hz to 97 Hz, indicating that the system is in a transitional state. By analyzing the harmonic interference characteristics of this frequency response curve, new frequency components generated due to harmonic intermodulation and their intensities can be identified. For example, new intermodulation components such as 75 Hz and 125 Hz are generated between 50 Hz and 100 Hz, and their amplitudes are 8% and 5% of the original fundamental frequency respectively. From this, the IGBT temperature harmonic interference feature matrix can be constructed, including the harmonic intermodulation components and the corresponding harmonic crosstalk coefficients (such as 0.08 and 0.05). Finally, after obtaining the complete IGBT temperature harmonic interference feature matrix, through harmonic component reconstruction operation, all the identified harmonic components are recombined into a new temperature fluctuation signal model, thereby obtaining the final IGBT temperature fluctuation feature set. This feature set not only contains the main frequency components of the original signal, but also integrates the nonlinear modulation effect and the influence of harmonic intermodulation, and can more comprehensively describe the thermal behavior characteristics of the IGBT under complex operating conditions. For example, the error of the reconstructed temperature fluctuation signal at the peak is less than ±1.5 °C, and the overall correlation coefficient reaches above 0.96, indicating that it has high reduction accuracy and characterization ability. For example, in an industrial automation system, a motor device driven by an inverter will generate significant temperature fluctuations inside the inverter IGBT due to the periodic change of switching losses during frequent start-stop or load mutation operating states. At this time, through the above-mentioned harmonic component analysis process based on multi-dimensional Fourier transform, the dominant harmonic components related to the switching frequency, load cycle, and control strategy can be accurately identified, and a high-precision IGBT temperature fluctuation feature set can be constructed based on this. This feature set can not only be used to monitor the health status of the chip in real time, but also serve as the input basis for the subsequent junction temperature prediction and warning system. For example, when it is detected that after a certain operation, the harmonic crosstalk coefficient exceeds the set threshold (such as 0.1), the system can issue a warning signal in advance, prompting the controller to take protection measures such as frequency reduction, current limiting, or shutdown, thereby effectively avoiding device failure due to overheating and ensuring the long-term stable operation and safety reliability of the system.

[0036] In a specific embodiment, the temperature trend extrapolation based on the IGBT temperature fluctuation feature set to obtain the IGBT junction temperature prediction sequence includes: Performing a non-linear trend decomposition operation on the IGBT temperature fluctuation feature set to obtain an IGBT temperature evolution feature vector, and performing a dynamic singular value decomposition on the IGBT temperature evolution feature vector to obtain an IGBT temperature trend principal component matrix, where the IGBT temperature trend principal component matrix includes a temperature change eigenvalue sequence and a temperature evolution feature vector group; Extracting IGBT temperature topology features based on the IGBT temperature trend principal component matrix to obtain an IGBT temperature topology feature map, and performing a phase space reconstruction analysis on the IGBT temperature topology feature map to obtain an IGBT temperature dynamics feature set, where the IGBT temperature dynamics feature set includes a temperature trajectory embedding dimension and a phase space reconstruction delay time; Evaluating the stability of the IGBT temperature dynamics feature set through Lyapunov exponent analysis to obtain an IGBT temperature stability index sequence, and identifying bifurcation points based on the temperature stability index sequence to obtain an IGBT temperature evolution bifurcation feature map, where the IGBT temperature evolution bifurcation feature map includes a temperature stability threshold and a bifurcation point time series; Performing a non-linear extrapolation calculation on the IGBT temperature evolution bifurcation feature map to obtain an IGBT temperature trend prediction matrix, and reconstructing a temperature trajectory based on the temperature trend prediction matrix to obtain an initial value of the IGBT junction temperature prediction sequence, where the initial value of the IGBT junction temperature prediction sequence includes a temperature prediction reference value and a temperature prediction error range; Dynamically correcting the initial value of the IGBT junction temperature prediction sequence through an adaptive Kalman filter to obtain a corrected value of the IGBT junction temperature prediction sequence, and performing a confidence interval analysis on the corrected value of the IGBT junction temperature prediction sequence to obtain an IGBT junction temperature prediction sequence, where the IGBT junction temperature prediction sequence includes a temperature prediction mean value and a temperature prediction variance.

[0037] Specifically, the process of extrapolating the temperature trend based on the IGBT temperature fluctuation feature set to obtain the IGBT junction temperature prediction sequence is a high-order prediction process that integrates non-linear time series analysis, dynamic system modeling, and adaptive filtering techniques. The core objective of this step is to construct a forward-looking and robust junction temperature prediction model by deeply mining the temperature evolution law of the inverter IGBT under complex operating conditions, so as to provide accurate data support for the subsequent warning mechanism. In the specific implementation process, first, a non-linear trend decomposition operation is performed on the previously extracted IGBT temperature fluctuation feature set to remove the periodic components and noise interference in the signal, and extract the IGBT temperature evolution feature vector that reflects the true temperature evolution trend. For example, during a certain process of the inverter driving the motor, the collected IGBT junction temperature data shows obvious periodic fluctuations (fundamental frequency 50Hz) and non-linear mutation characteristics. By methods such as empirical mode decomposition (EMD) or variational mode decomposition (VMD), the original temperature fluctuation signal can be decomposed into multiple intrinsic mode functions (IMFs), where the low-frequency part represents the long-term trend and the high-frequency part reflects the transient disturbance. On this basis, further perform dynamic singular value decomposition on the IGBT temperature evolution feature vector to extract the main characteristic components that dominate the temperature change, and form the IGBT temperature trend principal component matrix. For example, the cumulative contribution rate of the first three principal components reaches 92%, corresponding to the temperature rising trend, the change in the fluctuation amplitude, and the heat accumulation effect respectively. These eigenvalues constitute the temperature change eigenvalue sequence, and the corresponding eigenvector group reveals the spatial distribution characteristics of each component. Next, based on the above IGBT temperature trend principal component matrix, temperature topological feature extraction is carried out, that is, the temperature evolution process is mapped into a high-dimensional space to form the IGBT temperature topological feature map. By performing phase space reconstruction analysis on this map, the internal dynamic structure of the IGBT temperature evolution can be identified. For example, when using the delay embedding method to reconstruct the temperature time series, the phase space reconstruction delay time is set to 3ms and the embedding dimension is 7, which means that the temperature state at the current moment is jointly determined by the states of the past 6 time points. Thus, the IGBT temperature dynamics feature set includes the temperature trajectory embedding dimension (such as 7) and the phase space reconstruction delay time (such as 3ms). These parameters reflect the degrees of freedom and memory length of the temperature evolution system, and are of great significance for judging whether the system is in a stable state. Subsequently, Lyapunov exponent analysis is introduced to evaluate the stability of the IGBT temperature dynamics feature set to quantify the sensitivity of the temperature trajectory to changes in the initial conditions. If the Lyapunov exponent is positive, it indicates that there is chaotic behavior in the system and the temperature change is unpredictable; if it is negative, it indicates that the system tends to be stable. For example, in a certain experiment, the calculated Lyapunov exponent is -0.12s⁻¹, indicating that the current IGBT temperature evolution is in a convergent state and the system is overall stable.On this basis, the bifurcation point in the temperature evolution path is further identified, that is, the critical moment when the temperature trajectory undergoes a structural transformation due to changes in system parameters. For example, during the continuous load increase process, it was found that the IGBT junction temperature had an obvious inflection point at the 12th second, when the temperature rise rate suddenly increased from 2°C per second to 5°C per second, indicating that the system entered a new working mode. From this, the IGBT temperature evolution bifurcation characteristic diagram can be constructed, which includes the temperature stability threshold (such as 110°C) and the bifurcation point time series (such as 12s, 18s, 24s, etc.), providing key turning point information for subsequent predictions. Next, the IGBT temperature evolution bifurcation characteristic diagram is nonlinearly extrapolated to predict the temperature change trend in the future. This process usually uses methods such as neural networks, support vector regression (SVR) or nonlinear autoregressive models (NAR), combining historical bifurcation point information with current evolution trends to generate an IGBT temperature trend prediction matrix. For example, when it is known that the temperature bifurcation points in the first 20 seconds are the 6th second, the 12th second, and the 18th second, the model predicts that the next bifurcation may occur at the 24th second, and estimates that the temperature reference value at this time will reach 112°C, with an error range of ±3°C. Thus, the initial value of the IGBT junction temperature prediction sequence can be obtained, including the temperature prediction reference value (such as 112°C) and the temperature prediction error range (such as ±3°C), which provides an initial reference for subsequent correction. Finally, in order to improve the prediction accuracy and suppress the influence of measurement noise, the adaptive Kalman filter is used to dynamically correct the initial value of the IGBT junction temperature prediction sequence. This filtering method can establish a feedback adjustment mechanism between the predicted value and the actual measured value, and continuously correct the prediction result. For example, at a certain moment, the predicted temperature is 112°C, while the actual value of the sensor is 110°C, and the Kalman gain is 0.8, then the updated correction value is 110.4°C. Through continuous iteration, the corrected value of the IGBT junction temperature prediction sequence is finally obtained, and a confidence interval analysis is performed on it to determine the credibility range of the prediction result. For example, after multiple rounds of filtering, the final predicted temperature mean is 111.5℃, the standard deviation is 1.2℃, and the confidence interval is [109.1℃, 113.9℃]. The resulting IGBT junction temperature prediction sequence not only contains the temperature prediction mean, but also provides temperature prediction variance information, which can effectively support risk assessment and early warning decisions. For example, in an industrial automation system, during the continuous heavy-load operation of an inverter-driven motor device, the inverter IGBT produces significant temperature fluctuations due to frequent switching actions. At this time, through the above-mentioned temperature trend extrapolation process based on the IGBT temperature fluctuation feature set, the junction temperature change trend in the next few seconds can be accurately predicted. For example, the system detects that the current temperature is close to the design upper limit (such as 115℃), and the prediction shows that the temperature may reach 118℃ in the next minute, exceeding the safety threshold.At this time, the control system can issue a warning signal in advance, prompting protective measures such as frequency reduction, current limiting or shutdown, so as to avoid overheating failure of the device and ensure the long-term stable operation and safety reliability of the system.

[0038] In a specific embodiment, the non-linear extrapolation calculation of the IGBT temperature evolution bifurcation characteristic diagram to obtain the IGBT temperature trend prediction matrix includes: Performing a fractal dimension decomposition operation on the IGBT temperature evolution bifurcation characteristic diagram to obtain a fractal feature vector of the temperature evolution trajectory, and performing a Hausdorff dimension calculation on the fractal feature vector of the temperature evolution trajectory to obtain an IGBT temperature fractal feature matrix, where the IGBT temperature fractal feature matrix includes a temperature trajectory fractal dimension and a fractal scale exponent; Based on the IGBT temperature fractal feature matrix, performing a phase change point identification operation on the inverter IGBT to obtain an IGBT temperature phase change characteristic sequence, and analyzing the temperature critical exponent of the IGBT temperature phase change characteristic sequence to obtain an IGBT temperature critical characteristic diagram, where the IGBT temperature critical characteristic diagram includes a temperature phase change critical point and a phase change diffusion coefficient; Extracting a periodic orbit from the IGBT temperature critical characteristic diagram through a Poincaré map to obtain a group of IGBT temperature periodic feature vectors, and performing a chaotic attractor reconstruction based on the temperature periodic feature vector group to obtain an IGBT temperature chaotic feature matrix, where the IGBT temperature chaotic feature matrix includes a temperature orbit period coefficient and a chaotic attractor parameter; Constructing a non-linear predictor for the IGBT temperature chaotic feature matrix to obtain an IGBT temperature trend prediction operator, and performing a temperature trajectory extrapolation operation based on the temperature trend prediction operator to obtain an IGBT temperature trend prediction matrix, where the IGBT temperature trend prediction matrix includes a temperature prediction trajectory coefficient and a prediction confidence parameter.

[0039] Specifically, in the process of performing non - linear extrapolation calculations on the IGBT temperature evolution bifurcation feature map to obtain the IGBT temperature trend prediction matrix, it is first necessary to conduct a detailed mathematical analysis of the IGBT temperature evolution bifurcation feature map. This process begins with the fractal dimension decomposition operation on the IGBT temperature evolution bifurcation feature map, aiming to capture the subtle changes and complexities in the temperature evolution trajectory, so as to extract the fractal feature vector of the temperature evolution trajectory. For example, in the application scenario of an inverter driving a motor, the IGBT module shows a complex dynamic change pattern in its junction temperature due to being in a high - load state for a long time. Through the fractal dimension decomposition operation, we can identify the self - similarity and scale - invariance features in these patterns, and then obtain the fractal feature vector describing the temperature change characteristics. Next, the Hausdorff dimension calculation is performed on the obtained fractal feature vector of the temperature evolution trajectory, which is one of the important indicators for measuring the complexity of the fractal structure. Through this calculation, we can obtain the IGBT temperature fractal feature matrix, which contains key information such as the fractal dimension of the temperature trajectory and the fractal scaling exponent. Suppose that under a specific experimental condition, the calculated fractal dimension is 1.65, which indicates the degree of irregularity of the IGBT temperature evolution over time. The fractal scaling exponent further quantifies how this irregularity changes with the change of spatial or temporal scales, providing basic data for subsequent analysis. Based on the above - mentioned IGBT temperature fractal feature matrix, a phase - change point identification operation can be carried out on the inverter IGBT, aiming to find out the key turning points that may occur during the temperature change process. This step is crucial for understanding the thermal behavior of the IGBT under different working conditions. For example, in a continuously operating power - electronic device, as the load current increases, the junction temperature of the IGBT gradually rises, but there may be a sharp change at some critical points. By analyzing the IGBT temperature phase - change characteristic sequence, especially by calculating the temperature critical exponent, an IGBT temperature critical feature map can be obtained, which contains important parameters such as the temperature phase - change critical point and the phase - change diffusion coefficient. If in a certain instance, a critical temperature value of 120 °C is determined, and it is found that when the temperature exceeds this value, the temperature rise rate significantly accelerates, it indicates that a phase - change phenomenon occurs at this time. To further reveal the periodicity and potential chaotic characteristics of the IGBT temperature change, the Poincaré mapping method is used to extract the periodic orbits from the IGBT temperature critical feature map. The Poincaré mapping is an effective tool for studying the periodic behavior of non - linear dynamic systems. At this stage, a set of IGBT temperature periodic feature vectors will be obtained, which represent the stable cyclic patterns in the temperature change. Then, based on these groups of periodic feature vectors, a chaotic attractor reconstruction is carried out to construct the IGBT temperature chaotic feature matrix. This matrix includes the temperature orbit period coefficient and the chaotic attractor parameters, which jointly describe the internal law of the IGBT temperature fluctuation.For example, through analysis, it is found that the parameters of a typical chaotic attractor are approximately 0.7, which means that the IGBT temperature fluctuations have a certain degree of unpredictability but still follow some underlying rules. Finally, using the information in the IGBT temperature chaos feature matrix, a non-linear predictor is constructed, and based on this, temperature trajectory extrapolation operations are carried out to finally obtain the IGBT temperature trend prediction matrix. This process involves combining known historical temperature data with a prediction model to infer future temperature trends. For example, assuming that based on historical data analysis, the average temperature increase rate within the next 30 minutes is 0.5 °C per minute and the prediction confidence parameter is 90%, then we can estimate the junction temperature change of the IGBT in the next time period. In this way, not only the temperature prediction trajectory coefficient is obtained, but also the reliability range of the prediction result is clarified, which is crucial for real-time monitoring and preventing IGBT overheating. To sum up, through a series of complex processes on the IGBT temperature evolution bifurcation feature map, including fractal dimension decomposition, Hausdorff dimension calculation, phase transition point identification, Poincaré mapping application, and non-linear predictor construction, etc., we can effectively predict the junction temperature change trend of the IGBT. This method not only improves the prediction accuracy but also enhances the understanding of IGBT thermal management, helps to take timely measures to prevent failures caused by overheating, and ensures the safe and reliable operation of power electronic devices. For example, in an industrial automation system, when it is detected that the IGBT is about to reach its safe operating limit, the system can give an early warning and automatically adjust the working parameters to avoid equipment damage.

[0040] In a specific embodiment, performing a phase transition point identification operation on the inverter IGBT based on the IGBT temperature fractal feature matrix to obtain an IGBT temperature phase transition feature sequence, including: Performing entropy value dynamic decomposition operation on the IGBT temperature fractal feature matrix to obtain a temperature sequence entropy value feature vector, and detecting the phase transition critical point of the temperature sequence entropy value feature vector to obtain an IGBT temperature phase transition threshold matrix, where the IGBT temperature phase transition threshold matrix includes a heat diffusion critical value and a temperature jump coefficient; Performing temperature gradient tensor decomposition based on the IGBT temperature phase transition threshold matrix to obtain an IGBT temperature gradient feature field, and performing topological structure reconstruction on the temperature gradient feature field to obtain an IGBT temperature topological feature sequence, where the IGBT temperature topological feature sequence includes temperature field singular point coordinates and a temperature gradient curl coefficient; Performing heat conduction characteristic analysis on the IGBT temperature topological feature sequence through a hyperbolic diffusion equation to obtain an IGBT heat diffusion feature vector group, and performing heat flux density field reconstruction based on the heat diffusion feature vector group to obtain an IGBT temperature field evolution matrix, where the IGBT temperature field evolution matrix includes a heat flux density distribution coefficient and a temperature field divergence parameter; Perform non-equilibrium thermodynamics analysis on the IGBT temperature field evolution matrix to obtain the IGBT temperature phase change characteristic sequence.

[0041] Specifically, in the process of identifying the phase change points of the inverter IGBT based on the IGBT temperature fractal feature matrix and obtaining the IGBT temperature phase change feature sequence, it is first necessary to perform entropy value dynamic decomposition operation on the IGBT temperature fractal feature matrix. This process aims to extract the entropy value feature vectors that can reflect the system state changes from the complex temperature data. For example, in an industrial automation system, when a motor device controlled by an inverter operates under high load, the junction temperature of its IGBT module shows a non-linear fluctuation pattern. By performing entropy value dynamic decomposition on these temperature data, the entropy value feature vectors related to the temperature changes can be identified. Suppose the analysis results show that when the temperature rising rate exceeds 2°C per second, the entropy value of the system increases significantly, which may indicate an impending phase change. By detecting the phase change critical points in these entropy value feature vectors, we can construct the IGBT temperature phase change threshold matrix, which includes the heat diffusion critical value (such as 105°C) and the temperature jump coefficient (such as 2.3), to identify the key turning points of temperature changes. Next, based on the obtained IGBT temperature phase change threshold matrix, further carry out the temperature gradient tensor decomposition operation to reveal the fine structure of the internal temperature distribution of the IGBT. The temperature gradient tensor decomposition can transform the temperature field information into multi-dimensional gradient features, thus forming the IGBT temperature gradient feature field. For example, under a certain experimental condition, it is calculated that the temperature gradient in the center area of the chip is 5°C / mm, while that in the edge area is 2°C / mm. On this basis, the topological structure of the temperature gradient feature field is reconstructed to generate the IGBT temperature topological feature sequence, which not only includes the coordinates of the singular points of the temperature field (such as there is a singular point at the center point (5,5)), but also reflects the temperature gradient curl coefficient (such as 0.4). These parameters help to understand the thermal behavior characteristics of the IGBT under different working conditions and provide key inputs for the subsequent analysis of heat conduction characteristics. Subsequently, the hyperbolic diffusion equation is used to deeply analyze the IGBT temperature topological feature sequence to evaluate its heat conduction characteristics. The hyperbolic diffusion equation can effectively simulate the heat propagation process in materials, especially for a rapidly changing temperature field, this method is particularly important. For example, in the above scenario, when the IGBT is in a high-load state, by applying the hyperbolic diffusion equation analysis, the IGBT heat diffusion feature vector group can be obtained, including the heat flux density distribution coefficient (such as 3.5 W / cm² in the center area) and the temperature field divergence parameter (such as a positive value indicates heat diffusion outward, and a negative value indicates heat aggregation). Based on these feature vector groups, the heat flux density field is further reconstructed to form the IGBT temperature field evolution matrix. This matrix not only depicts the heat flow path inside the IGBT, but also provides important information about the evolution trend of the temperature field. Finally, perform non-equilibrium thermodynamics analysis on the IGBT temperature field evolution matrix to comprehensively understand the energy conversion mechanism during the temperature change process and its impact on the system stability.Non-equilibrium thermodynamics analysis can reveal the behavioral laws of a system when it is far from equilibrium, especially the unique properties exhibited during phase change processes. For example, in the aforementioned case, through non-equilibrium thermodynamics analysis, we found that as the load continued to increase, the temperature of the IGBT gradually approached its design upper limit (such as 115 °C). At this time, some parameters in the temperature field evolution matrix showed significant changes, indicating that the system was approaching a critical state. By comprehensively analyzing these parameters, the IGBT temperature phase change characteristic sequence was finally obtained, which included multiple key indicators, such as the temperature phase change critical point, phase change diffusion coefficient, etc. These indicators not only provided a basis for predicting the future working state of the IGBT but also laid a foundation for the design of real-time monitoring and early warning mechanisms. For example, in an actual industrial automation environment, in a motor device driven by an inverter, when it starts and stops frequently or the load changes suddenly, the IGBT module will generate a large amount of heat due to switching losses, resulting in a rapid increase in the junction temperature. At this time, through the above-mentioned phase change point identification operation process based on the IGBT temperature fractal characteristic matrix, the key turning points in the temperature change process can be accurately captured. For example, when it is detected that the temperature reaches 110 °C and the temperature rising rate exceeds the set threshold (such as 2 °C per second), the system can determine that the IGBT is in a state close to phase change and issue an early warning signal accordingly. This precise temperature monitoring and prediction mechanism can not only effectively prevent the device from failing due to overheating but also optimize the overall performance of power electronic equipment, ensuring the stability and safety of the system. By continuously adjusting control strategies, such as reducing the frequency or limiting the current, the temperature can be prevented from continuing to rise to a dangerous level, thereby extending the service life of the equipment and improving the operating efficiency.

[0042] The inverter IGBT junction temperature prediction method in the embodiments of the present invention has been described above. Next, the inverter IGBT junction temperature prediction device in the embodiments of the present invention will be described. Please refer to Figure 2 , an embodiment of the inverter IGBT junction temperature prediction device in the embodiments of the present invention includes: An acquisition module 21, configured to collect electrical parameters of the inverter IGBT through voltage and current sensors and perform switching loss analysis to obtain an IGBT loss characteristic map, where the IGBT loss characteristic map includes an on-state loss component and an off-state loss component; A calculation module 22, configured to calculate the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain an IGBT heat flux distribution matrix; A first analysis module 23, configured to perform thermoelectric coupling analysis on the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain an IGBT junction temperature dynamic response sequence; A second analysis module 24, configured to perform harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain an IGBT temperature fluctuation feature set; A prediction module 25, configured to perform temperature trend extrapolation based on the IGBT temperature fluctuation feature set to obtain an IGBT junction temperature prediction sequence, and perform threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain an IGBT junction temperature warning signal.

[0043] In this embodiment, for the specific implementation of each unit in the above device embodiment, please refer to the description in the above method embodiment, and details are not described herein again.

[0044] Refer to Figure 3 , in an embodiment of the present invention, a computer device is further provided. The internal structure of the computer device may be as Figure 3 shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected through a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0045] Those skilled in the art can understand that Figure 3 the structure shown in

[0046] is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0047] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0048] It should be noted that in this article, the terms "including", "comprising", or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, device, article, or method including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, device, article, or method. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, device, article, or method including that element.

[0049] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be similarly included in the patent protection scope of the present invention.

Claims

1. A method for predicting the junction temperature of an inverter IGBT, characterized in that Including the following steps: Collect electrical parameters of the inverter IGBT through a voltage-current sensor and analyze the switching losses to obtain the IGBT loss characteristic map, where the IGBT loss characteristic map includes turn-on loss components and turn-off loss components; Calculate the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain the IGBT heat flux distribution matrix; Perform thermo-electric coupling analysis on the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain the IGBT junction temperature dynamic response sequence; Perform harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation characteristic set; Perform temperature trend extrapolation based on the IGBT temperature fluctuation characteristic set to obtain the IGBT junction temperature prediction sequence, and perform threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature warning signal.

2. The inverter IGBT junction temperature prediction method according to claim 1, wherein The step of calculating the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain the IGBT heat flux distribution matrix includes: Divide multiple micro-region grid units on the surface of the inverter IGBT, and collect the temperature gradients of each micro-region grid unit through a thermocouple array to obtain the surface temperature distribution data set; Perform thermal conductivity tensor decomposition based on the surface temperature distribution data set to obtain the thermal conductivity eigenvector, and perform anisotropic heat conduction calculation on the thermal conductivity eigenvector to obtain the heat flux density component matrix; Perform heat flux divergence analysis on the heat flux density component matrix through the Laplace operator to obtain the heat flux field topology structure, and perform heat flux integral operation based on the heat flux field topology structure to obtain the IGBT heat flux distribution matrix.

3. The inverter IGBT junction temperature prediction method according to claim 1, characterized in that The step of performing thermo-electric coupling analysis on the inverter IGBT based on the IGBT loss characteristic map and the IGBT heat flux distribution matrix to obtain the IGBT junction temperature dynamic response sequence includes: Perform power density distribution calculation on the IGBT loss characteristic map to obtain the instantaneous power density distribution map on the surface of the inverter IGBT, and extract the heat source distribution characteristic based on the instantaneous power density distribution map to obtain the IGBT heat source spatial distribution characteristic vector; Solve the heat diffusion equation based on the IGBT heat source spatial distribution characteristic vector and the IGBT heat flux distribution matrix to obtain the temperature field distribution function of the inverter IGBT, and perform boundary condition constraint analysis on the temperature field distribution function to obtain the transient temperature response characteristic set of the inverter IGBT; Calculate the dynamic thermal resistance of the transient temperature response characteristic set of the inverter IGBT through the non-linear thermal resistance network technology to obtain the IGBT multi-layer structure thermal impedance matrix, and construct the temperature transfer function based on the thermal impedance matrix to obtain the IGBT junction temperature transfer characteristic equation; Perform transient thermal response analysis on the inverter IGBT based on the IGBT junction temperature transfer characteristic equation to obtain the IGBT junction temperature dynamic response sequence, and extract the time-domain characteristics of the IGBT junction temperature dynamic response sequence to obtain the IGBT junction temperature fluctuation characteristic spectrum.

4. The inverter IGBT junction temperature prediction method according to claim 1, characterized in that Performing harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain the IGBT temperature fluctuation feature set, including: Performing time-frequency domain mapping transformation on the IGBT junction temperature dynamic response sequence to obtain the IGBT temperature fluctuation time-frequency feature matrix, and performing wavelet packet decomposition operation on the temperature fluctuation time-frequency feature matrix to obtain the IGBT temperature harmonic energy distribution map; Performing harmonic component decoupling calculation on the IGBT temperature harmonic energy distribution map to obtain the IGBT temperature harmonic feature vector group, and performing non-linear phase modulation analysis on the temperature harmonic feature vector group to obtain the IGBT temperature harmonic modulation feature set; Performing instantaneous frequency extraction on the IGBT temperature harmonic modulation feature set through multi-dimensional Fourier transform to obtain the IGBT temperature harmonic frequency response curve, and analyzing the harmonic interference characteristics of the temperature harmonic frequency response curve to obtain the IGBT temperature harmonic interference feature matrix; Performing harmonic component reconstruction operation based on the IGBT temperature harmonic interference feature matrix to obtain the IGBT temperature fluctuation feature set.

5. The inverter IGBT junction temperature prediction method according to claim 1, characterized in that Performing temperature trend extrapolation based on the IGBT temperature fluctuation feature set to obtain the IGBT junction temperature prediction sequence, including: Performing non-linear trend decomposition operation on the IGBT temperature fluctuation feature set to obtain the IGBT temperature evolution feature vector, and performing dynamic singular value decomposition on the IGBT temperature evolution feature vector to obtain the IGBT temperature trend principal component matrix; Performing temperature topology feature extraction based on the IGBT temperature trend principal component matrix to obtain the IGBT temperature topology feature map, and performing phase space reconstruction analysis on the IGBT temperature topology feature map to obtain the IGBT temperature dynamics feature set; Performing stability evaluation on the IGBT temperature dynamics feature set through Lyapunov exponent analysis to obtain the IGBT temperature stability index sequence, and performing bifurcation point identification based on the temperature stability index sequence to obtain the IGBT temperature evolution bifurcation feature map; Performing non-linear extrapolation calculation on the IGBT temperature evolution bifurcation feature map to obtain the IGBT temperature trend prediction matrix, and performing temperature trajectory reconstruction based on the temperature trend prediction matrix to obtain the initial value of the IGBT junction temperature prediction sequence; Dynamically correcting the initial value of the IGBT junction temperature prediction sequence through adaptive Kalman filtering technology to obtain the corrected value of the IGBT junction temperature prediction sequence, and analyzing the confidence interval of the corrected value of the IGBT junction temperature prediction sequence to obtain the IGBT junction temperature prediction sequence.

6. The inverter IGBT junction temperature prediction method according to claim 5, characterized in that, Performing non-linear extrapolation calculation on the IGBT temperature evolution bifurcation feature map to obtain the IGBT temperature trend prediction matrix, including: Performing fractal dimension decomposition operation on the IGBT temperature evolution bifurcation feature map to obtain the fractal feature vector of the temperature evolution trajectory, and performing Hausdorff dimension calculation on the fractal feature vector of the temperature evolution trajectory to obtain the IGBT temperature fractal feature matrix; Based on the IGBT temperature fractal feature matrix, perform a phase transition point identification operation on the inverter IGBT to obtain an IGBT temperature phase transition feature sequence, and analyze the temperature critical exponent of the IGBT temperature phase transition feature sequence to obtain an IGBT temperature critical feature map; Extract the periodic orbit of the IGBT temperature critical feature map through Poincaré mapping to obtain an IGBT temperature periodic feature vector group, and reconstruct the chaotic attractor based on the temperature periodic feature vector group to obtain an IGBT temperature chaotic feature matrix; Construct a non-linear predictor for the IGBT temperature chaotic feature matrix to obtain an IGBT temperature trend prediction operator, and perform a temperature trajectory extrapolation operation based on the temperature trend prediction operator to obtain an IGBT temperature trend prediction matrix.

7. The inverter IGBT junction temperature prediction method according to claim 6, wherein The performing a phase transition point identification operation on the inverter IGBT based on the IGBT temperature fractal feature matrix to obtain an IGBT temperature phase transition feature sequence includes: Perform an entropy value dynamic decomposition operation on the IGBT temperature fractal feature matrix to obtain a temperature sequence entropy value feature vector, and detect the phase transition critical point of the temperature sequence entropy value feature vector to obtain an IGBT temperature phase transition threshold matrix; Perform a temperature gradient tensor decomposition based on the IGBT temperature phase transition threshold matrix to obtain an IGBT temperature gradient feature field, and reconstruct the topological structure of the temperature gradient feature field to obtain an IGBT temperature topological feature sequence; Analyze the heat conduction characteristics of the IGBT temperature topological feature sequence through a hyperbolic diffusion equation to obtain an IGBT heat diffusion feature vector group, and reconstruct the heat flux density field based on the heat diffusion feature vector group to obtain an IGBT temperature field evolution matrix; Perform a non-equilibrium thermodynamics analysis on the IGBT temperature field evolution matrix to obtain an IGBT temperature phase transition feature sequence.

8. An IGBT junction temperature prediction device for an inverter, characterized in that, Including: An acquisition module for collecting electrical parameters of the inverter IGBT through a voltage and current sensor and analyzing the switching loss to obtain an IGBT loss feature map, where the IGBT loss feature map includes an on-state loss component and an off-state loss component; A calculation module for calculating the heat flux density on the surface of the inverter IGBT through a thermocouple array to obtain an IGBT heat flux distribution matrix; A first analysis module for performing a thermo-electric coupling analysis on the inverter IGBT based on the IGBT loss feature map and the IGBT heat flux distribution matrix to obtain an IGBT junction temperature dynamic response sequence; A second analysis module for performing a harmonic component analysis on the IGBT junction temperature dynamic response sequence through multi-dimensional Fourier transform to obtain an IGBT temperature fluctuation feature set; A prediction module for extrapolating the temperature trend based on the IGBT temperature fluctuation feature set to obtain an IGBT junction temperature prediction sequence, and performing a threshold judgment analysis on the IGBT junction temperature prediction sequence to obtain an IGBT junction temperature warning signal.

9. A computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images