A method for measuring junction temperature of power semiconductor

Abstract

The present invention discloses a method for measuring the junction temperature of a power semiconductor, which belongs to the field of semiconductor measurement technology. The method specifically comprises: constructing a non-invasive circuit to collect temperature-sensitive parameters in real time, using deep learning to establish a nonlinear model of temperature-sensitive parameters and junction temperature from historical data, and then fusing and processing the temperature-sensitive parameters through a machine learning algorithm to calculate the real-time junction temperature; combining thermal resistance, thermal capacitance and preset thermal management indicators to comprehensively evaluate the thermal state of the power semiconductor, and feeding back the evaluation results to a control system; the control system automatically puts forward optimization suggestions based on this, realizes intelligent monitoring and adjustment of the working state of the power semiconductor, and improves the stability and efficiency of the system.

Description

Technical Field

[0001] The invention belongs to the technical field of semiconductor measurement, in particular to a method for measuring the junction temperature of a power semiconductor. Background Art

[0002] In the application of power semiconductor devices, junction temperature is a key parameter. Junction temperature refers to the actual operating temperature of semiconductors in electronic devices, that is, the temperature inside the semiconductor chip. In power semiconductor devices, junction temperature is an important indicator to measure the thermal state of the device. Since power semiconductors generate a lot of heat during operation, if it cannot be dissipated in a timely and effective manner, the junction temperature will increase. The increase in junction temperature will directly affect the electrical performance, reliability and life of the device.

[0003] The main methods for measuring power semiconductor junction temperature can be summarized into the following categories: First, electrical temperature measurement method, which is a method of estimating junction temperature by measuring parameter changes of power semiconductor devices under specific electrical conditions. It has the advantages of non-contact, good real-time performance and high accuracy, but is greatly affected by equipment aging; Second, physical contact temperature measurement method, which is to directly contact the temperature sensing element with the medium of the power semiconductor device being measured, and measure the temperature of the device by measuring the temperature of the temperature sensing element. Although it is intuitive, the measurement accuracy is low; Third, optical temperature measurement method, which uses infrared radiation radiated to the external environment by power semiconductor devices due to power consumption and heat generation during operation to measure temperature. It has the advantages of non-contact and non-destructive to device structure, but different sealing types and shell material factors will affect the transmission and measurement accuracy of infrared radiation.

[0004] For example, the patent with publication number CN112946450A discloses a power semiconductor device and a junction temperature measurement method thereof that facilitates junction temperature detection, including: the power semiconductor device is an IGBT device or a MOSFET device, the semiconductor device includes at least two unit cells, at least one of which is a junction temperature monitoring unit, and the other unit cells are completely the same in structure as the IGBT device or the MOSFET device; the difference in structure between the junction temperature monitoring unit and the IGBT device or the MOSFET device is that the surface of the P+ contact area is metallized to form an anode, and the surface of the N+ source area is metallized to form a cathode. The junction temperature monitoring unit is directly manufactured and integrated into the power semiconductor chip, so that the junction temperature value of the power semiconductor device can be directly measured in real time, and the real-time and accurate value can be obtained.

[0005] For example, the patent with the authorization announcement number CN115993516B discloses a method and system for measuring the internal junction temperature distribution of a press-fit power semiconductor device, including: measuring and obtaining the initial contact force value, contact force fluctuation value and initial junction temperature value of each sub-module in the press-fit power semiconductor device; obtaining the junction temperature fluctuation value of each sub-module according to the initial contact force value and contact force fluctuation value of each sub-module, and the functional relationship between the initial contact force value, junction temperature fluctuation value and contact force fluctuation value of a single sub-module inside the press-fit power semiconductor device; obtaining the internal junction temperature value of each sub-module based on the junction temperature fluctuation value and the initial junction temperature value of each sub-module; obtaining the internal junction temperature distribution of the press-fit power semiconductor device based on the internal junction temperature value of each sub-module. This technical solution is suitable for measuring the internal junction temperature distribution of press-fit power semiconductor devices, and there is no need to destroy the packaging structure of the press-fit power semiconductor device or open the device package.

[0006] The above existing technologies all have the following problems: in CN112946450A, once the chip manufacturing is completed, its built-in junction temperature monitoring unit is fixed, and it is difficult to adjust or upgrade according to subsequent application requirements or technological advances, which lacks flexibility and increases the complexity and risk of the design; in CN115993516B, it is specially designed for crimped power semiconductor devices. Its measurement method and system are not suitable for other types of power semiconductor devices. The scope of application is limited, and it relies on a specific functional relationship to describe the relationship between the contact force of the sub-module, junction temperature fluctuation and contact force fluctuation. The accuracy and applicability of this functional relationship are affected by many factors, such as manufacturing differences of devices, aging of materials, and changes in the working environment. At the same time, it is necessary to measure and record the initial contact force values, contact force fluctuation values ​​and initial junction temperature values ​​of multiple sub-modules, and perform complex calculations based on these data, which increases the complexity of the measurement and the workload of data processing. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention proposes a method for measuring the junction temperature of a power semiconductor, constructs a non-invasive circuit to collect temperature-sensitive parameters in real time, uses deep learning to establish a nonlinear model of temperature-sensitive parameters and junction temperature from historical data, and then fuses and processes the temperature-sensitive parameters through a machine learning algorithm to calculate the real-time junction temperature; combines thermal resistance, thermal capacitance and preset thermal management indicators to comprehensively evaluate the thermal state of the power semiconductor, and feeds the evaluation results back to the control system; the control system automatically puts forward optimization suggestions based on this, realizes intelligent monitoring and adjustment of the working state of the power semiconductor, and improves the system stability and efficiency.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for measuring junction temperature of a power semiconductor, comprising:

[0010] Step S1: construct a non-invasive measurement circuit to collect temperature-sensitive parameter values ​​of the power semiconductor in real time when the power semiconductor is in a working state;

[0011] Step S2: collecting historical temperature-sensitive parameter data and corresponding known junction temperature data of power semiconductors under different working conditions, and establishing a nonlinear mapping model between temperature-sensitive parameters and junction temperature using a deep learning method based on the historical data;

[0012] Step S3: Based on the nonlinear mapping model, a multi-parameter comprehensive evaluation algorithm is used to fuse the collected temperature-sensitive parameters to calculate the real-time junction temperature of the power semiconductor;

[0013] Step S4: comprehensively evaluating the thermal state of the power semiconductor according to the calculated real-time junction temperature value, combined with thermal resistance, thermal capacitance parameters and preset comprehensive thermal management indicators, and feeding back the evaluation results to the control system;

[0014] Step S5: The control system automatically proposes optimization suggestions based on the thermal state evaluation results.

[0015] Specifically, the temperature-sensitive parameters in step S1 refer to the carrier concentration, mobility, threshold voltage and saturation current temperature-sensitive parameters of the power semiconductor; the non-invasive measurement circuit is configured with an adaptive sensor array, and can be simultaneously connected to multiple measurement points of the power semiconductor to collect temperature-sensitive parameters in a non-contact manner; the adaptive sensor array is configured with a temperature-sensitive parameter sensor, an infrared temperature sensor and a micro-thermistor.

[0016] Specifically, the specific steps of step S3 include:

[0017] S3.1: Acquire the temperature-sensitive parameter value in real time and perform feature extraction on the temperature-sensitive parameter value;

[0018] S3.2: Use a multi-parameter comprehensive evaluation algorithm to fuse multiple temperature-sensitive parameters after feature extraction to generate temperature-sensitive feature parameters X = {x1,…,x N}, where x N represents the Nth temperature-sensitive characteristic parameter, where N is the number of temperature-sensitive characteristic parameters;

[0019] S3.3: Load the pre-trained nonlinear mapping model and convert the real-time collected X={x1,…,x N} is input into the trained nonlinear mapping model, and the nonlinear mapping model calculates the real-time junction temperature y of the power semiconductor based on the learned weights and relationships;

[0020] S3.4: Verify the calculated real-time junction temperature y, and optimize the nonlinear mapping model based on the verification result.

[0021] Specifically, the calculation formula for the real-time junction temperature of the power semiconductor in S3.3 is:

[0022]

[0023] Where y represents the real-time junction temperature of the power semiconductor, g(·) represents the nonlinear mapping function implemented by the comprehensive evaluation model, σ(·) represents the nonlinear activation function, and w i represents the weight of the i-th temperature-sensitive characteristic parameter, x i represents the i-th temperature-sensitive characteristic parameter, b i represents the bias term of the i-th temperature-sensitive characteristic parameter, α j represents the weight of the time-related temperature-sensitive characteristic parameter, h j (t) represents the jth temperature-sensitive characteristic parameter associated with time t, m represents the number of temperature-sensitive characteristic parameters associated with time, and t represents the current time.

[0024] Specifically, the specific steps of step S4 include:

[0025] S4.1: Obtain the thermal resistance and thermal capacitance parameters of the power semiconductor from the power semiconductor manufacturer's manual, and establish a thermal model of the power semiconductor based on the thermal resistance and thermal capacitance parameters;

[0026] S4.2: According to the application scenarios and performance requirements of power semiconductors, set the thermal management indicators of the maximum allowable junction temperature, junction temperature fluctuation range, and thermal management efficiency of power semiconductors. The formula is:

[0027]

[0028] Among them, y max Indicates the maximum allowable junction temperature value set. represents the maximum allowable junction temperature given in the power semiconductor manufacturer's manual, y ± Indicates the junction temperature fluctuation range, represents the minimum allowable junction temperature given in the power semiconductor manufacturer's manual, η represents the thermal management efficiency target, and R min Denotes the minimized thermal resistance, ΔT min represents minimizing the temperature difference between the node and the environment, Q max represents the maximum air flow, ∪ represents the union, {R min ∪ΔT min ∪Q max} means that R min , ΔT min , Q max The thermal management efficiency of the situation.

[0029] Specifically, the specific steps of step S4 also include:

[0030] S4.3: Substitute the real-time junction temperature value y into the thermal management index condition. If the real-time junction temperature value meets the thermal management index condition, the real-time junction temperature value is input into the thermal model, and the thermal state equation of the power semiconductor is calculated by combining the thermal resistance and thermal capacitance parameters. The thermal state equation formula is:

[0031]

[0032] Where C(y(t)) represents the heat capacity function that changes with junction temperature, y(t) represents the junction temperature that changes with time t, that is, the thermal state, T d (t) represents the dynamic ambient temperature, R(y(t), P(t)) represents the nonlinear thermal resistance, and P(t) represents the heat dissipation power that varies with time. Represents the thermal coupling term between the power semiconductor and the heat sink, W k (t) represents the temperature of the kth coupling element, R k (·) represents the thermal coupling impedance of the kth coupling element, P r (y(t)) represents the power lost due to radiation, P con (y(t),T d (t)) represents the power dissipated due to convection and M represents the number of coupling elements.

[0033] Specifically, the specific steps of step S4 also include:

[0034] S4.4: Use numerical methods to solve the thermal state equation and obtain the thermal state of the power semiconductor And judge whether y(t) meets the thermal management index conditions, where Δt s represents the time step, y(Δt s ) represents the time step Δt s Junction temperature at

[0035] S4.5: If the thermal management index conditions are met, the thermal state results of the power semiconductor are fed back to the control system;

[0036] S4.6: The control system is based on the thermal status results.

[0037] like It is judged that the power semiconductor is in an overheated state;

[0038] like Or if η(y(t))<η, it is judged that there is a thermal risk in the power semiconductor, where η(y(t)) represents the junction temperature efficiency at y(t).

[0039] Specifically, the specific steps of step S4 also include:

[0040] S4.7: Based on the judgment result, if it is in an overheating state or there is a thermal risk, a thermal management strategy is formulated based on the thermal state calculation result;

[0041] S4.8: converting the formulated thermal management strategy into a control instruction and sending the command to a cooling fan or a power regulator;

[0042] S4.9: Monitor the execution effect, monitor the thermal state changes of power semiconductors in real time, and adjust the thermal management strategy according to the execution effect.

[0043] Specifically, the thermal management strategy formulated in S4.7 includes: reducing power output, increasing heat dissipation, and turning on the heat dissipation equipment of the water cooling system; the increasing heat dissipation includes: increasing the fan speed and adjusting the heat sink layout.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] 1. The present invention proposes a method for measuring the junction temperature of a power semiconductor, and proposes to simultaneously incorporate multiple temperature-sensitive parameters such as carrier concentration, mobility, threshold voltage, and saturation current into the evaluation system, and realize accurate measurement of the junction temperature of the power semiconductor through a multi-parameter comprehensive evaluation algorithm. Compared with single parameter measurement, this multi-parameter fusion method can more comprehensively and accurately reflect the thermal state of the power semiconductor; an accurate mathematical model between temperature-sensitive parameters and junction temperature is established, and the multi-parameter comprehensive evaluation algorithm is optimized to ensure that high measurement accuracy and stability can be maintained under complex working conditions. This algorithm optimization not only improves the measurement accuracy, but also enhances the adaptability and robustness of the algorithm.

[0046] 2. The present invention proposes a method for measuring the junction temperature of a power semiconductor, which adopts a non-invasive measurement circuit, avoids the damage that may be caused to the power semiconductor by the traditional invasive measurement method, and improves the safety and reliability of the measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A framework diagram of a method for measuring junction temperature of a power semiconductor according to the present invention;

[0048] Figure 2 This is a flowchart of an algorithm for measuring junction temperature of a power semiconductor according to the present invention;

[0049] Figure 3 A junction temperature process flow chart of a method for measuring junction temperature of a power semiconductor according to the present invention;

[0050] Figure 4 The present invention is a thermal state assessment flow chart of a method for measuring junction temperature of a power semiconductor. DETAILED DESCRIPTION

[0051] Example 1

[0052] See also Figure 1-Figure 2 , an embodiment of the present invention provides: a method for measuring the junction temperature of a power semiconductor, comprising the following steps:

[0053] Step S1: construct a non-invasive measurement circuit to collect temperature-sensitive parameter values ​​of the power semiconductor in real time when the power semiconductor is in a working state;

[0054] Among them, the temperature-sensitive parameter values ​​include: carrier concentration, mobility, threshold voltage and saturation current.

[0055] Step S2: collecting historical temperature-sensitive parameter data and corresponding known junction temperature data of power semiconductors under different working conditions, and establishing a nonlinear mapping model between temperature-sensitive parameters and junction temperature using a deep learning method based on the historical data;

[0056] Among them, different working conditions include: different currents, voltages, frequencies, duty cycles, and ambient temperatures.

[0057] Furthermore, the specific steps of constructing the nonlinear mapping model include:

[0058] S2.1: Obtain temperature-sensitive parameter data of power semiconductors and corresponding known junction temperature data, and perform preprocessing;

[0059] S2.2: Construct a neural network model, where the temperature-sensitive parameter data is used as input features in the input layer, and the predicted junction temperature value is output in the output layer. The number of hidden layers and the number of neurons in each layer need to be adjusted and optimized according to the actual situation;

[0060] S2.3: The mean square error loss function evaluates the difference between the predicted value and the true value of the neural network model, and the stochastic gradient descent optimization algorithm is selected to update the neural network model parameters;

[0061] S2.4: Use the collected historical data to train the neural network model. During the training process, the neural network model parameters are updated through continuous iterations so that the loss function value gradually decreases until the maximum number of iterations is reached;

[0062] S2.5: Use the reserved test set to evaluate the trained neural network model and calculate the prediction accuracy, recall rate, and F1 score indicators to measure the performance of the neural network model. At the same time, observe the difference between the prediction results of the neural network model and the actual value through visualization methods;

[0063] S2.6: Optimize the model based on the evaluation results, including adjusting the neural network structure, changing the loss function or optimization algorithm, and increasing the amount of data.

[0064] Step S3: Based on the nonlinear mapping model, a multi-parameter comprehensive evaluation algorithm is used to fuse the collected temperature-sensitive parameters to calculate the real-time junction temperature of the power semiconductor;

[0065] Step S4: comprehensively evaluating the thermal state of the power semiconductor according to the calculated real-time junction temperature value, combined with thermal resistance, thermal capacitance parameters and preset comprehensive thermal management indicators, and feeding back the evaluation results to the control system;

[0066] Step S5: The control system automatically proposes optimization suggestions based on the thermal state evaluation results.

[0067] Among them, optimization suggestions include but are not limited to adjusting the design of the heat sink, increasing the fan speed, optimizing the air duct, changing the coolant flow rate or changing the working mode of the power semiconductor to reduce the junction temperature of the power semiconductor.

[0068] Example 2

[0069] See also Figure 3-Figure 4 In this embodiment, the temperature-sensitive parameters in step S1 refer to the carrier concentration, mobility, threshold voltage and saturation current temperature-sensitive parameters of the power semiconductor; the non-invasive measurement circuit is configured with an adaptive sensor array, and can be connected to multiple measurement points of the power semiconductor at the same time to collect temperature-sensitive parameters in a non-contact manner; the adaptive sensor array is configured with a temperature-sensitive parameter sensor, an infrared temperature sensor and a micro-thermistor.

[0070] The specific steps of step S3 include:

[0071] S3.1: obtaining the value of the temperature-sensitive parameter in real time, and performing feature extraction on the value of the temperature-sensitive parameter, wherein the feature extraction method in the present invention adopts the principal component analysis method, which is the prior art content in this field and is not the inventive solution of the present application, and will not be described in detail here;

[0072] S3.2: Use a multi-parameter comprehensive evaluation algorithm to fuse multiple temperature-sensitive parameters after feature extraction to generate temperature-sensitive feature parameters X = {x1,…,x N}, where x N represents the Nth temperature-sensitive characteristic parameter, N is the number of temperature-sensitive characteristic parameters, wherein the multi-parameter comprehensive evaluation algorithm in the present invention adopts a weighted average method, and the extracted features are fused by setting relevant parameters of the weighted average method, such as a weight coefficient, a learning rate, and the number of iterations. However, the specific calculation process of the weighted average method is the prior art content in this field, and is not the inventive solution of the present application, and is not described in detail here;

[0073] S3.3: Load the pre-trained nonlinear mapping model and convert the real-time collected X={x1,…,x N} is input into the trained nonlinear mapping model. The nonlinear mapping model calculates the real-time junction temperature y of the power semiconductor based on the learned weights and relationships. The calculation formula for the real-time junction temperature of the power semiconductor is:

[0074]

[0075] Where y represents the real-time junction temperature of the power semiconductor, g(·) represents the nonlinear mapping function implemented by the comprehensive evaluation model, which is used to further nonlinearly transform the internal summation result, σ(·) represents the nonlinear activation function, which is used to introduce nonlinearity, and w i represents the weight of the i-th temperature-sensitive characteristic parameter, obtained through machine learning algorithm training, x i represents the i-th temperature-sensitive characteristic parameter, b i represents the bias term of the i-th temperature-sensitive characteristic parameter, α j represents the weight of the time-related temperature-sensitive characteristic parameter, h j (t) represents the jth temperature-sensitive characteristic parameter associated with time t, m represents the number of temperature-sensitive characteristic parameters associated with time, including historical data, time trends or periodic factors, and t represents the current time;

[0076] In the present invention, the calculation formula of the real-time junction temperature of the power semiconductor is based on the original weighted sum, and the consideration of nonlinear activation, bias term and time-related characteristics is added, so that the nonlinear mapping model can handle complex data and relationships more flexibly. At the same time, by introducing additional functions g(·) and h j (t), enhancing the expressive power of nonlinear mapping models.

[0077] It should be noted that the calculation of the nonlinear mapping model is carried out through forward propagation, mainly the input layer receives the temperature-sensitive feature parameters, and calculates through the hidden layer layer by layer, including weighted summation and activation function, and finally reaches the output layer to obtain the predicted value of the real-time junction temperature.

[0078] S3.4: Verify the calculated real-time junction temperature y, and optimize the nonlinear mapping model based on the verification result.

[0079] Furthermore, the specific steps of S3.4 include:

[0080] S3.41: Real-time junction temperature verification:

[0081] 1) Use a high-precision temperature sensor as a temperature measurement device to measure the actual junction temperature y of the power semiconductor and obtain the true junction temperature value

[0082] 2) Compare the real-time junction temperature y calculated by the nonlinear mapping model with the actual junction temperature value Make a comparison and assess the differences between the two;

[0083] S3.42: Analyze the real-time junction temperature y and the actual junction temperature value The differences between the two models can be identified to identify the sources of error, such as sensor accuracy, model complexity, and accuracy of input data, and to assess the significance of the differences to determine whether further model optimization is needed.

[0084] S3.43: Based on the results of the difference analysis, formulate strategies for optimizing the nonlinear mapping model, including adjusting the model structure, adding input features, and optimizing model parameters;

[0085] S3.44: Modify and adjust the nonlinear mapping model according to the developed optimization strategy, including retraining the model and adjusting model parameters;

[0086] S3.45: Recalculate the real-time junction temperature using the optimized nonlinear mapping model and compare it with the actual junction temperature value to verify the optimization effect.

[0087] S3.46: Repeat steps S3.41 to S3.45 until the performance of the nonlinear mapping model meets the requirements.

[0088] The specific steps of step S4 include:

[0089] S4.1: Obtain the thermal resistance and thermal capacitance parameters of the power semiconductor from the power semiconductor manufacturer's manual, and establish a thermal model of the power semiconductor based on the thermal resistance and thermal capacitance parameters;

[0090] Furthermore, the specific implementation steps of establishing the thermal model of the power semiconductor include:

[0091] (1) Select the resistor-capacitor thermal model based on the application scenario and performance requirements of the power semiconductor;

[0092] (2) constructing a framework of the resistor-capacitor thermal model according to the selected resistor-capacitor thermal model, including determining nodes, thermal resistance and thermal capacitance elements in the resistor-capacitor thermal model, and connection relationships between them;

[0093] (3) applying the acquired thermal resistance and thermal capacitance parameters to the resistor-capacitor thermal model and setting specific values ​​of each thermal resistance and thermal capacitance element, wherein these parameters will directly affect the accuracy and reliability of the thermal model;

[0094] (4) Verify the resistor-capacitor thermal model using experimental data or simulation data, and evaluate the accuracy and reliability of the resistor-capacitor thermal model by comparing the differences between the predicted results of the resistor-capacitor thermal model and the actual measurement results;

[0095] (5) The resistor-capacitor thermal model is optimized based on the verification results. If there is a large difference between the predicted results and the actual measurement results, the parameters or structure of the resistor-capacitor thermal model can be adjusted to improve the accuracy and reliability of the resistor-capacitor thermal model.

[0096] S4.2: According to the application scenarios and performance requirements of power semiconductors, set the thermal management indicators of the maximum allowable junction temperature, junction temperature fluctuation range, and thermal management efficiency of power semiconductors. The formula is:

[0097]

[0098] Among them, y max Indicates the maximum allowable junction temperature value set. represents the maximum allowable junction temperature given in the power semiconductor manufacturer's manual, y ± Indicates the junction temperature fluctuation range, represents the minimum allowable junction temperature given in the power semiconductor manufacturer's manual, η represents the thermal management efficiency target, and R min Denotes the minimized thermal resistance, ΔT min represents minimizing the temperature difference between the node and the environment, Q max represents the maximum air flow, ∪ represents the union, {R min ∪ΔT min ∪Q max} means that R min , ΔT min , Q max Thermal management efficiency of the situation;

[0099] S4.3: Substitute the real-time junction temperature value y into the thermal management index condition. If the real-time junction temperature value meets the thermal management index condition, the real-time junction temperature value is input into the thermal model, and the thermal state equation of the power semiconductor is calculated by combining the thermal resistance and thermal capacitance parameters. The thermal state equation formula is:

[0100]

[0101] Where C(y(t)) represents the heat capacity function that changes with junction temperature, y(t) represents the junction temperature that changes with time t, that is, the thermal state, T d (t) represents the dynamic ambient temperature, R(y(t), P(t)) represents the nonlinear thermal resistance, and P(t) represents the heat dissipation power that varies with time. Represents the thermal coupling term between the power semiconductor and the heat sink, W k (t) represents the temperature of the kth coupling element, R k (·) represents the thermal coupling impedance of the kth coupling element, P r (y(t)) represents the power lost due to radiation, P con (y(t),Td (t)) represents the power dissipated due to convection, M represents the number of coupling elements;

[0102] S4.4: Use numerical methods to solve the thermal state equation and obtain the thermal state of the power semiconductor And judge whether y(t) meets the thermal management index conditions, where Δt s represents the time step, y(Δt s ) represents the time step Δt s The junction temperature at the time of the invention is 0.0407 W, wherein the specific steps of solving the equation using the numerical method are the prior art content in the art, which is not the inventive solution of the present application and will not be described in detail here;

[0103] S4.5: If the thermal management index conditions are met, the thermal state results of the power semiconductor are fed back to the control system;

[0104] S4.6: The control system is based on the thermal status results.

[0105] like It is judged that the power semiconductor is in an overheated state;

[0106] like Or η(y(t))<η, it is judged that there is a thermal risk in the power semiconductor, where η(y(t)) represents the junction temperature efficiency at y(t);

[0107] S4.7: Based on the judgment result, if it is in an overheating state or there is a thermal risk, a thermal management strategy is formulated based on the thermal state calculation result;

[0108] S4.8: converting the formulated thermal management strategy into a control instruction and sending the command to a cooling fan or a power regulator;

[0109] S4.9: Monitor the execution effect, monitor the thermal state changes of power semiconductors in real time, and adjust the thermal management strategy according to the execution effect.

[0110] The thermal management strategy formulated in S4.7 includes: reducing power output, increasing heat dissipation, and turning on the heat dissipation equipment of the water cooling system; the increasing heat dissipation includes: increasing the fan speed and adjusting the heat sink layout.

[0111] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0112] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0113] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation modes, which are merely illustrative rather than restrictive. Under the guidance of the present invention, ordinary technicians in the field may also change, modify, replace and modify the above-mentioned embodiments without departing from the purpose and scope of protection of the present invention, and all of these are within the protection of the present invention.

Claims

1. A method for measuring the junction temperature of a power semiconductor, characterized in that: include: Step S1: construct a non-invasive measurement circuit to collect temperature-sensitive parameter values ​​of the power semiconductor in real time when the power semiconductor is in a working state; Step S2: collecting historical temperature-sensitive parameter data and corresponding known junction temperature data of power semiconductors under different working conditions, and establishing a nonlinear mapping model between temperature-sensitive parameters and junction temperature using a deep learning method based on the historical data; Step S3: Based on the nonlinear mapping model, a multi-parameter comprehensive evaluation algorithm is used to fuse the collected temperature-sensitive parameters to calculate the real-time junction temperature of the power semiconductor; Step S4: comprehensively evaluating the thermal state of the power semiconductor according to the calculated real-time junction temperature value, combined with thermal resistance, thermal capacitance parameters and preset comprehensive thermal management indicators, and feeding back the evaluation results to the control system; Step S5: the control system automatically proposes optimization suggestions based on the thermal state evaluation results; The specific steps of step S4 include: S4.1: Obtain the thermal resistance and thermal capacitance parameters of the power semiconductor from the power semiconductor manufacturer's manual, and establish a thermal model of the power semiconductor based on the thermal resistance and thermal capacitance parameters; S4.2: According to the application scenarios and performance requirements of power semiconductors, set the thermal management indicators of the maximum allowable junction temperature, junction temperature fluctuation range, and thermal management efficiency of power semiconductors. The formula is: Among them, y max Indicates the maximum allowable junction temperature value set. represents the maximum allowable junction temperature given in the power semiconductor manufacturer's manual, y ± Indicates the junction temperature fluctuation range, represents the minimum allowable junction temperature given in the power semiconductor manufacturer's manual, η represents the thermal management efficiency target, and R min Denotes the minimized thermal resistance, ΔT min represents minimizing the temperature difference between the node and the environment, Q max represents the maximum air flow, ∪ represents the union, {R min ∪ΔT min ∪Q max } means that R min , ΔT min , Q max Thermal management efficiency of the situation; The specific steps of step S4 also include: S4.3: Substitute the real-time junction temperature value y into the thermal management index condition. If the real-time junction temperature value meets the thermal management index condition, the real-time junction temperature value is input into the thermal model, and the thermal state equation of the power semiconductor is calculated by combining the thermal resistance and thermal capacitance parameters. The thermal state equation formula is: Where C(y(t)) represents the heat capacity function that changes with junction temperature, y(t) represents the junction temperature that changes with time t, that is, the thermal state, T d (t) represents the dynamic ambient temperature, R(y(t), P(t)) represents the nonlinear thermal resistance, and P(t) represents the heat dissipation power that varies with time. Represents the thermal coupling term between the power semiconductor and the heat sink, W k (t) represents the temperature of the kth coupling element, R k (·) represents the thermal coupling impedance of the kth coupling element, P r (y(t)) represents the power lost due to radiation, P con (y(t),T d (t)) represents the power dissipated due to convection, M represents the number of coupling elements; The specific steps of step S4 also include: S4.4: Use numerical methods to solve the thermal state equation and obtain the thermal state of the power semiconductor And judge whether y(t) meets the thermal management index conditions, where Δt s represents the time step, y(Δt s ) represents the time step Δt s Junction temperature at S4.5: If the thermal management index conditions are met, the thermal state results of the power semiconductor are fed back to the control system; S4.6: The control system is based on the thermal status results. like It is judged that the power semiconductor is in an overheated state; like Or if η(y(t))<η, it is judged that there is a thermal risk in the power semiconductor, where η(y(t)) represents the junction temperature efficiency at y(t).

2. A method for measuring junction temperature of a power semiconductor as claimed in claim 1, characterized in that: The temperature-sensitive parameters in step S1 refer to the carrier concentration, mobility, threshold voltage and saturation current temperature-sensitive parameters of the power semiconductor; the non-invasive measurement circuit is configured with an adaptive sensor array and can be simultaneously connected to multiple measurement points of the power semiconductor to collect temperature-sensitive parameters in a non-contact manner; the adaptive sensor array is configured with a temperature-sensitive parameter sensor, an infrared temperature sensor and a micro-thermistor.

3. A method for measuring junction temperature of a power semiconductor as claimed in claim 2, characterized in that: The specific steps of step S3 include: S3.1: Acquire the temperature-sensitive parameter value in real time and perform feature extraction on the temperature-sensitive parameter value; S3.2: Use a multi-parameter comprehensive evaluation algorithm to fuse multiple temperature-sensitive parameters after feature extraction to generate temperature-sensitive feature parameters X = {x1,…,x N }, where x N represents the Nth temperature-sensitive characteristic parameter, where N is the number of temperature-sensitive characteristic parameters; S3.3: Load the pre-trained nonlinear mapping model and convert the real-time collected X={x1,…,x N } is input into the trained nonlinear mapping model, and the nonlinear mapping model calculates the real-time junction temperature y of the power semiconductor based on the learned weights and relationships; S3.4: Verify the calculated real-time junction temperature y, and optimize the nonlinear mapping model based on the verification result.

4. A method for measuring junction temperature of a power semiconductor as claimed in claim 3, characterized in that: The calculation formula for the real-time junction temperature of the power semiconductor in S3.3 is: Where y represents the real-time junction temperature of the power semiconductor, g(·) represents the nonlinear mapping function implemented by the comprehensive evaluation model, σ(·) represents the nonlinear activation function, and w i represents the weight of the i-th temperature-sensitive characteristic parameter, x i represents the i-th temperature-sensitive characteristic parameter, b i represents the bias term of the i-th temperature-sensitive characteristic parameter, α j represents the weight of the time-related temperature-sensitive characteristic parameter, h j (t) represents the jth temperature-sensitive characteristic parameter associated with time t, m represents the number of temperature-sensitive characteristic parameters associated with time, and t represents the current time.

5. A method for measuring junction temperature of a power semiconductor as claimed in claim 4, characterized in that: The specific steps of step S4 also include: S4.7: Based on the judgment result, if it is in an overheating state or there is a thermal risk, a thermal management strategy is formulated based on the thermal state calculation result; S4.8: converting the formulated thermal management strategy into a control instruction and sending the command to a cooling fan or a power regulator; S4.9: Monitor the execution effect, monitor the thermal state changes of power semiconductors in real time, and adjust the thermal management strategy according to the execution effect.

6. A method for measuring junction temperature of a power semiconductor as claimed in claim 5, characterized in that: The thermal management strategy formulated in S4.7 includes: reducing power output, increasing heat dissipation, and turning on the heat dissipation equipment of the water cooling system; the increasing heat dissipation includes: increasing the fan speed and adjusting the heat sink layout.

Images