IGBT module junction temperature monitoring method and device

Abstract

The invention provides an IGBT module junction temperature monitoring method and device, and belongs to the technical field of power electronic devices. The method comprises the following steps: acquiring a collector-emitter saturation voltage drop, a direct current power cycle period number and a shell temperature of an on-line IGBT (Insulated Gate Bipolar Translator) module; inputting the collector-emitter saturation voltage drop, the direct current power cycle period number and the shell temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the on-line IGBT module; wherein the junction temperature monitoring model is obtained by training a whale optimized gradient elevator model based on a pre-constructed data set, and the pre-constructed data set is obtained according to aging data; the aging data comprises the direct-current power cycle aging period number of the IGBT module, the aging shell temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module. According to the invention, the junction temperature of the IGBT module can be monitored on line.

Description

Technical Field

[0001] The present invention relates to the technical field of power electronic devices, and in particular to a method and device for monitoring junction temperature of an IGBT module. Background Art

[0002] Power electronic transformers (PETs), due to their high power density, high controllability, and high efficiency, are widely used in a variety of fields, including photovoltaic power generation, energy storage systems, DC transmission, aerospace, locomotive traction, and electric vehicles. The reliability of PETs is directly related to the safe and stable operation of the entire system. Failures can lead to power outages, equipment damage, economic losses, and even fires and explosions, threatening personal safety. Research shows that power semiconductors are the most vulnerable component of the commutation system, with a failure rate of approximately 34%. Therefore, as a representative power semiconductor device, the reliability of IGBTs is crucial to the safe operation of the system.

[0003] Currently, high-frequency switching, such as the frequent switching of electrical equipment between power supply (connecting to a circuit) and power supply (disconnecting from a circuit), leads to a high failure rate for IGBT modules. Promptly identifying and replacing failed IGBT modules can effectively reduce this failure rate, thereby improving the reliability of flexible DC transmission systems. Research has found that junction temperature fluctuation is a key factor affecting IGBT module reliability, and implementing junction temperature monitoring for IGBT modules is an important means of improving their reliability.

[0004] Conventional technology typically uses cameras or temperature-sensitive electrical parameter methods to monitor IGBT module junction temperature. However, accurate camera measurements require opening the module package, while temperature-sensitive electrical parameter methods require injecting a small current (typically 100mA) to estimate junction temperature while the converter is offline. Both methods are unsuitable for online monitoring, creating an urgent need for an accurate online IGBT module junction temperature monitoring method. Summary of the Invention

[0005] The embodiments of the present invention provide a method and device for monitoring the junction temperature of an IGBT module, so as to solve the problem that the junction temperature of the IGBT module cannot be monitored online.

[0006] In a first aspect, an embodiment of the present invention provides a method for monitoring junction temperature of an IGBT module, comprising:

[0007] Obtain collector-emitter saturation voltage drop, DC power cycle number and case temperature of online IGBT modules;

[0008] The collector-emitter saturation voltage drop, DC power cycle number, and case temperature are input into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module.

[0009] Among them, the junction temperature monitoring model is obtained by training the whale optimized gradient boosting machine model based on a pre-constructed data set. The pre-constructed data set is obtained according to the aging data. The aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging case temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module.

[0010] In one possible implementation, the method further includes:

[0011] Build a power cycle accelerated aging experimental platform, conduct accelerated power cycle experiments on the experimental platform, and obtain aging data of IGBT modules based on the accelerated power cycle experiments.

[0012] In one possible implementation, the process of obtaining an aging dataset of an IGBT module based on an accelerated power cycling experiment includes:

[0013] Set the experimental parameters, including the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current, and the number of DC power cycles of the IGBT module.

[0014] Obtain the collector-emitter saturation voltage drop, aging case temperature and precise junction temperature of the IGBT module during each DC power cycle aging period;

[0015] The aging data of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging period.

[0016] In one possible implementation, the process of obtaining the precise junction temperature of the IGBT module includes:

[0017] Inject a small current into the IGBT module during shutdown;

[0018] Measure the collector-emitter saturation voltage drop of the IGBT module under low current;

[0019] The precise junction temperature of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module under low current.

[0020] In one possible implementation, the method further includes:

[0021] Obtain aging data, including the precise junction temperature of the IGBT module, the collector-emitter saturation voltage drop of the IGBT module, the number of DC power cycle aging cycles of the IGBT module, and the aging case temperature of the IGBT module.

[0022] A data set is constructed based on the aging data, and a junction temperature monitoring model is obtained by training the data set.

[0023] In a possible implementation, constructing a data set based on aging data includes:

[0024] The data in the dataset is classified according to the different collector-emitter saturation voltage drops, different DC power cycle aging cycles and different aging case temperatures of the IGBT modules, and sub-datasets corresponding to different data are constructed;

[0025] The junction temperature monitoring model is trained based on the data set, including:

[0026] The whale optimized gradient boosting machine model is trained according to each sub-data set to obtain the junction temperature monitoring model.

[0027] In one possible implementation, a junction temperature monitoring model is obtained by training the data set, including:

[0028] Use a preset proportion of data in the dataset as the training set and the remaining data as the test set;

[0029] The whale optimized gradient boosting machine model is trained according to the training set, and the whale optimized gradient boosting machine model is tested according to the test set to obtain the junction temperature monitoring model.

[0030] In a second aspect, an embodiment of the present invention provides an IGBT module junction temperature monitoring device, comprising:

[0031] Data acquisition module, used to obtain the collector-emitter saturation voltage drop, DC power cycle number and case temperature of the online IGBT module;

[0032] The junction temperature monitoring module is used to input the collector-emitter saturation voltage drop, the number of DC power cycles, and the case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module;

[0033] Among them, the junction temperature monitoring model is obtained by training the whale optimized gradient boosting machine model based on a pre-constructed data set. The pre-constructed data set is obtained according to the aging data. The aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging case temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module.

[0034] In one possible implementation, the junction temperature monitoring module is used to:

[0035] Build a power cycle accelerated aging experimental platform, conduct accelerated power cycle experiments on the experimental platform, and obtain aging data of IGBT modules based on the accelerated power cycle experiments.

[0036] In one possible implementation, the junction temperature monitoring module is used to:

[0037] Set the experimental parameters, including the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current, and the number of DC power cycles of the IGBT module.

[0038] Obtain the collector-emitter saturation voltage drop, aging case temperature and precise junction temperature of the IGBT module during each DC power cycle aging period;

[0039] The aging data of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging period.

[0040] Embodiments of the present invention provide a method and device for monitoring the junction temperature of an IGBT module. These methods obtain the collector-emitter saturation voltage drop, DC power cycle number, and case temperature of an online IGBT module, and then input the collector-emitter saturation voltage drop, DC power cycle number, and case temperature into a pre-trained junction temperature monitoring model to obtain the online junction temperature of the IGBT module. The junction temperature monitoring model is trained using a whale-optimized gradient boosting machine model based on a pre-constructed dataset. The pre-constructed dataset is derived from aging data, which includes the number of DC power cycle aging cycles, aging case temperature, the precise junction temperature of the IGBT module, and the collector-emitter saturation voltage drop of the IGBT module. This method enables online monitoring of the junction temperature of the IGBT module, and by considering the effects of the DC power cycle number and case temperature on the junction temperature of the IGBT module, the junction temperature of the IGBT module monitored by the junction temperature monitoring model is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a flow chart of an implementation method for monitoring junction temperature of an IGBT module provided by an embodiment of the present invention;

[0043] Figure 2 1 is a schematic structural diagram of a power cycle test circuit provided by an embodiment of the present invention;

[0044] Figure 3aSchematic diagram of collector-emitter saturation voltage drop and junction temperature under low current provided by an embodiment of the present invention;

[0045] Figure 3b This is a schematic diagram of a small current injection provided by an embodiment of the present invention;

[0046] Figure 4a Schematic diagram of the internal structure of the IGBT module provided by an embodiment of the present invention;

[0047] Figure 4b This is a schematic diagram of the IGBT module output characteristic test provided by an embodiment of the present invention;

[0048] Figure 4c This is a schematic diagram of collecting the collector-emitter saturation voltage drop of an IGBT module provided by an embodiment of the present invention;

[0049] Figure 4d Schematic diagram of an IGBT module output characteristic curve cluster provided by an embodiment of the present invention;

[0050] Figure 5a Schematic diagram of the relationship between the collector-emitter saturation voltage drop and the junction temperature of the IGBT module under 150A current injection provided by an embodiment of the present invention;

[0051] Figure 5b Schematic diagram of the relationship between the collector-emitter saturation voltage drop and the junction temperature of the IGBT module under 165A current injection provided by an embodiment of the present invention;

[0052] Figure 5c Schematic diagram of the relationship between the collector-emitter saturation voltage drop and the junction temperature of the IGBT module under 180A current injection provided by an embodiment of the present invention;

[0053] Figure 5d Schematic diagram of the relationship between the collector-emitter saturation voltage drop and the junction temperature of the IGBT module under 195A current injection provided by an embodiment of the present invention;

[0054] Figure 6 It is a structural schematic diagram of an IGBT module junction temperature monitoring device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0056] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below with reference to the accompanying drawings.

[0057] Figure 1 The implementation flow chart of the IGBT module junction temperature monitoring method provided by the embodiment of the present invention is detailed as follows:

[0058] Step 101 : Obtain the collector-emitter saturation voltage drop, DC power cycle number, and case temperature of an online IGBT module.

[0059] In this embodiment, the collector-emitter saturation voltage drop of the IGBT module is the voltage drop between the collector and emitter of the IGBT module under high current operating conditions. The number of DC power cycles is the total number of complete switching cycles experienced by the IGBT module. Each cycle includes a fixed on-time and a fixed off-time. The case temperature is the temperature of the IGBT module casing. In power electronic equipment, the IGBT module is the core component of power conversion, and its case temperature is an important parameter for evaluating the thermal state of the module. The case temperature can reflect the heat dissipation and thermal management efficiency of the IGBT module, and is also one of the key indicators for monitoring the health of the IGBT module. The number of DC power cycles is determined based on the operating time of the IGBT module.

[0060] Step 102 : Input the collector-emitter saturation voltage drop, the number of DC power cycles, and the case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module.

[0061] In some embodiments, the junction temperature monitoring model is obtained by training a whale optimized gradient boosting machine model based on a pre-constructed data set, and the pre-constructed data set is obtained according to aging data, and the aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging case temperature of the IGBT module, the precise junction temperature of the IGBT module, and the collector-emitter saturation voltage drop of the IGBT module.

[0062] In this embodiment, the collector-emitter saturation voltage drop, the number of DC power cycles and the case temperature are input into a pre-trained junction temperature monitoring model. The junction temperature of the IGBT module can be obtained through the junction temperature monitoring model, thereby monitoring the junction temperature of the online IGBT module.

[0063] Specifically, first, aging data is obtained and a data set is obtained based on the aging data. Then, the whale optimized gradient boosting machine model is trained based on the data set to obtain a junction temperature monitoring model. Finally, the collector-emitter saturation voltage drop, DC power cycle number and case temperature are obtained online. The collector-emitter saturation voltage drop, DC power cycle number and case temperature are input into the trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module.

[0064] This approach utilizes the Whale Optimized Gradient Boosting Machine model to monitor junction temperature. The Gradient Boosting Machine algorithm employed is a highly efficient gradient boosting framework that excels at handling complex nonlinear relationships and achieves high prediction accuracy. By simulating the predatory behavior of whales, the Whale Optimization Algorithm effectively optimizes the hyperparameters of the Gradient Boosting Machine, further enhancing the model's predictive performance. Combining the efficiency of the Gradient Boosting Machine with the optimization capabilities of the Whale Optimization Algorithm, the Whale Optimized Gradient Boosting Machine model achieves high-precision, fast-convergence, robustness, and wide adaptability for IGBT module junction temperature monitoring.

[0065] This method is mainly used in the fields of high-voltage, large-capacity modular power electronic equipment and power converter power device module junction temperature monitoring in renewable energy systems.

[0066] In an embodiment of the present invention, the collector-emitter saturation voltage drop, number of DC power cycles, and case temperature of an online IGBT module are obtained and then input into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module. The junction temperature monitoring model is obtained by training a whale-optimized gradient boosting machine model based on a pre-constructed dataset. The pre-constructed dataset is obtained based on aging data, which includes the number of DC power cycle aging cycles, the aging case temperature, the precise junction temperature of the IGBT module, and the collector-emitter saturation voltage drop of the IGBT module. This method can monitor the junction temperature of the IGBT module online, and by considering the effects of the number of DC power cycles and the case temperature on the junction temperature of the IGBT module, the junction temperature of the IGBT module monitored by the junction temperature monitoring model is more accurate.

[0067] In a possible implementation, the following processing may be performed: building a power cycle accelerated aging experimental platform, performing an accelerated power cycle experiment on the experimental platform, and obtaining aging data of the IGBT module based on the accelerated power cycle experiment.

[0068] In this example, accelerated power cycling experiments were conducted on an experimental platform. During the experiment, aging data of the IGBT module was collected. The Whale Optimized Gradient Boosting Machine model was trained based on the aging data to generate a junction temperature monitoring model. By analyzing the data obtained from the accelerated aging experiments, the junction temperature monitoring method can be optimized to improve its accuracy and reliability.

[0069] In one possible embodiment, the process of obtaining an aging data set of an IGBT module according to an accelerated power cycling experiment is specifically processed as follows: setting experimental parameters; wherein the parameters include the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current and the number of DC power cycle cycles of the IGBT module; obtaining the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module for each DC power cycle aging cycle; obtaining the aging data of the IGBT module based on the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging cycle.

[0070] In this embodiment, the fatigue process of IGBT modules is usually slow. To improve research efficiency, it is necessary to apply extraordinary cyclic stress impact and conduct fatigue accelerated power cycle test. This method significantly shortens the fatigue process, improves research efficiency, and more accurately reveals the performance of the module under extreme conditions. The IGBT module fatigue accelerated power cycle test platform, its main circuit is as follows Figure 2 The aging accelerated test uses a DC power cycle with a control strategy of constant switching time T on / T off , T on =T off =1s. Set the gate voltage VG to 15V, the heating current IHeat to 195A, the precise junction temperature test current IMeas to 100mA, and perform 8000 cycles of DC power cycling. During the experiment, collect the IGBT collector-emitter saturation voltage drop V CE (sat), and record the case temperature for each cycle.

[0071] In one possible implementation, the process of obtaining the precise junction temperature of the IGBT module is specifically handled as follows: a small current is injected into the IGBT module during shutdown; the collector-emitter saturation voltage drop of the IGBT module under the small current is measured; and the precise junction temperature of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module under the small current.

[0072] In this embodiment, due to the low current injection, the collector-emitter saturation voltage drop V CE (sat) is insensitive to module fatigue and has a good linear relationship with the junction temperature Tj, so this method can achieve accurate monitoring of the junction temperature Tj.

[0073] Specifically, accurate junction temperature measurement requires a K-curve test before the aging test. The slope M and intercept B of the K-curve of this module are -0.00227 and 0.606089 respectively. The K-curve test method is to measure the junction temperature during the module shutdown period (t off) Inject a small current into the module and measure V CE (sat), and use the collector-emitter saturation voltage drop V under small current CE The good linear relationship between (sat) and junction temperature Tj can be used to accurately estimate the IGBT junction temperature, such as Figure 3a , as shown. Figure 3b Schematic diagram of small current injection.

[0074] In one possible implementation, the following processing may also be performed: obtaining aging data; wherein the aging data includes the precise junction temperature of the IGBT module, the collector-emitter saturation voltage drop of the IGBT module, the number of DC power cycle aging cycles of the IGBT module, and the aging case temperature of the IGBT module; constructing a data set based on the aging data, and obtaining a junction temperature monitoring model through training based on the data set.

[0075] In this embodiment, a data set is constructed based on aging data, and the junction temperature monitoring model trained based on the data set can be applied in actual power electronic equipment to monitor and predict the junction temperature of the IGBT module in real time, thereby improving the reliability and safety of the equipment.

[0076] In one possible implementation, a data set is constructed based on the aging data, and the specific processing is as follows: the data in the data set is classified according to the different collector-emitter saturation voltage drops, different numbers of DC power cycle aging cycles and different aging shell temperatures of the IGBT modules, and sub-data sets corresponding to different data are constructed; a junction temperature monitoring model is obtained by training the data set, and the specific processing is as follows: the whale optimized gradient boosting machine model is trained according to each sub-data set to obtain the junction temperature monitoring model.

[0077] In this embodiment, constructing sub-datasets corresponding to each classification condition can more accurately capture the behavioral characteristics of the IGBT module under different conditions. By training the model on different sub-datasets, it can ensure that the model has good prediction performance under various conditions and enhance the model's generalization ability.

[0078] In one possible implementation, a junction temperature monitoring model is obtained by training the data set, and the specific processing is as follows: a preset proportion of data in the data set is used as a training set, and the remaining data is used as a test set; the whale optimized gradient boosting machine model is trained according to the training set, and the whale optimized gradient boosting machine model is tested according to the test set to obtain a junction temperature monitoring model.

[0079] In this embodiment, the test data is divided into 80% training set and 20% test set. Through a strict training and testing process, the practicality and effectiveness of the model in actual application are ensured, providing a scientific basis for thermal management of power electronic equipment.

[0080] In one possible implementation, the reliability of monitoring the junction temperature based on the collector-emitter saturation voltage drop under high current is verified by the following process:

[0081] Schematic diagram of the internal structure of the IGBT module, the principle diagram of the IGBT module output characteristic test, the collector-emitter saturation voltage drop acquisition and the measured output characteristic curve cluster are as follows: Figures 4a-4d As shown. Measure the collector-emitter saturation voltage drop V CE At (sat), a gate voltage of 15V is applied to the IGBT gate and a current of a specific magnitude I is applied to the collector. C I C The current was gradually increased from 0 A to 300 A in increments of 5 A. To ensure measurement accuracy, data acquisition was performed after the current and voltage stabilized, with each acquisition lasting 300 μs.

[0082] In order to analyze the junction temperature T j V CE To understand the influence of (sat), it is necessary to obtain the module output characteristic curves at different temperatures. In the experiment, the power module was placed on a heating table, and the chip temperature was sensed in real time by the module's built-in NTC resistor sensor. The NTC resistance decreases as the temperature rises, and the temperature it senses is between the chip junction temperature T j and case temperature T C The present invention adopts the Infineon FF150R12ME3G IGBT module, and the NTC resistor is located between the module's No. 5 and No. 6 terminals. Figure 4a 1-11 are the terminals of the IGBT module.

[0083] The heating platform was used to adjust 12 groups of temperatures (27.70°C, 41.65°C, 53.05°C, 60.15°C, 72.35°C, 78.50°C, 88.80°C, 98.80°C, 108.90°C, 117.70°C, 127.80°C, and 137.60°C). The output characteristic curve of the module was measured at each temperature. The intersection of the 12 output characteristic curves corresponds to a collector current of 50A. When I C When the current is lower than 50A, the temperature characteristic is mainly dominated by the diode, showing a negative temperature characteristic, V CE (sat) decreases with increasing temperature; and when I C When the current is higher than 50A, the temperature characteristic is mainly dominated by MOSFET, showing a positive temperature characteristic, V CE (sat) increases with increasing temperature.

[0084] With collector current I cTaking 150A (INom), 165A (1.1INom), 180A (1.2INom), and 195A (1.3INom) as examples, the relationship between the collector-emitter saturation voltage drop VCE (sat) and the junction temperature Tj of the IGBT module is fitted under large current injection conditions. Figures 5a-5d As shown in the figure, under the four groups of large current injection conditions, the VCE(sat) and Tj of the IGBT module show a good linear relationship, and the linear fit R 2 All of them are higher than 0.999, which shows high fitting accuracy. Therefore, it is feasible to estimate the junction temperature Tj by measuring the collector-emitter saturation voltage drop VCE(sat) under large current injection.

[0085] In one possible implementation, the construction of the gradient boosting machine model is specifically processed as follows:

[0086] Suppose the kth tree is for sample x i The predicted value is f k (x i ), then when the kth tree is trained, the sample x i The predicted value of is the linear sum of the predicted values of k trees:

[0087]

[0088] Where F is a set of k decision trees, f k For an independent tree model in the forest, For sample x i The model predicted value.

[0089] The objective function of the gradient boosting machine model is:

[0090]

[0091] Where obj is the objective function, is the loss function, which is used to measure the error between the model prediction value and the actual value. is a regularization term used to control the complexity of the model.

[0092] Expand the objective function into the following form:

[0093]

[0094] When training the kth tree, the prediction results of the first (k-1) trees are known, and the objective function can be simplified as:

[0095]

[0096] Perform a second-order Taylor expansion on the loss function:

[0097]

[0098] Where, is the loss function completed after the (k-1)th training, are the first-order and second-order partial derivatives of the loss function with respect to the predicted value, respectively. All three are constant terms. Let:

[0099]

[0100] Then the objective function can be simplified as:

[0101]

[0102] Therefore, when training the k-th tree model, g i 、h i are all known constants, {g i 、h i} Pass the information obtained from the training of the first (k-1) trees to the kth training.

[0103] To parameterize the tree model, define:

[0104] f k (x i )=W q (x i )

[0105] I j ={i|q(x i )=j}

[0106] In the formula, q(x i ) is the sample x i The position of the leaf node it falls into; W is the weight of the leaf node; I j Represents a set of tree models with j leaf nodes.

[0107] Through the definition of tree model parameterization, the function f k (x i )、Ω(f k ) parameterization. It is convenient to use the number of leaf nodes and leaf node values to control the complexity of a tree:

[0108]

[0109] Where T represents the number of leaf nodes, W j Represents the value of the j-th leaf node, and the parameters γ and λ are hyperparameters used to control the complexity of the tree model.

[0110] Convert the samples into a set of leaf node combinations to further simplify the objective function:

[0111]

[0112] Where, are all constants, and the parameters γ and λ are hyperparameters used to control the complexity of the tree model.

[0113] Perform quadratic optimization on the objective function:

[0114]

[0115] Where, are all constants, and the parameter λ is a hyperparameter used to control the complexity of the tree model.

[0116] Then the minimum value of the objective function can be expressed as:

[0117]

[0118] Where, are all constants, and the parameters γ and λ are hyperparameters used to control the complexity of the tree model.

[0119] For any tree model of known shape, the above formula can be used to derive the optimal solution of the objective function. To determine the shape of the tree model, a greedy algorithm is used to select features by maximizing the difference between leaf nodes before and after splitting, and then construct the tree model.

[0120]

[0121] Where, obj old * 、obj new * Respectively represent the minimum value of the objective function before and after introducing new features to the leaf node; {G R 、H L} and {G L 、H R} are the G of the left and right leaf nodes after introducing new features j and H j , parameters γ and λ are hyperparameters used to control the complexity of the tree model.

[0122] The greedy algorithm iteratively calculates the loss of the model nodes and selects the leaf node with the largest gain loss.

[0123] The Whale Optimization Algorithm (WOA) is a heuristic optimization algorithm based on the behavior of whale groups in nature. It simulates the hunting behavior of humpback whales and adjusts the position of whales to find the optimal solution. WOA has the advantages of simplicity, easy implementation and fast convergence, and is applicable to a variety of optimization problems. The present invention uses WOA to optimize the number of decision trees and learning rate of the gradient boosting model XGBoost. The optimization of the number of decision trees directly affects the complexity and prediction accuracy of the model. WOA avoids local optimality by searching for the optimal solution globally, thereby finding the optimal number of decision trees, reducing computational costs and improving prediction accuracy. The optimization of the learning rate controls the contribution of each decision tree. WOA accelerates the convergence of the model by adjusting the learning rate, maintaining or improving prediction accuracy. WOA's group behavior simulation effectively searches the parameter space and improves the overall performance of the model.

[0124] The gradient boosting machine model maximizes computational speed and accuracy based on the efficient implementation of the gradient boosting decision tree algorithm, but the step-by-step growth strategy results in unnecessary memory consumption. To obtain accurate prediction results more quickly, the present invention uses the whale optimization algorithm (WOA) to optimize two important parameters of the prediction model: the number of decision trees (n_estimators) and the learning rate (learning_rate). The optimal number of decision trees (n_estimators) and learning rate (learning_rate) obtained are 502 and 0.1, respectively.

[0125] In a possible implementation, the model can also be evaluated. The specific process is as follows: the comparison between the predicted value of the model and the true value shows that the fluctuations of the two are basically consistent, and the model fitting effect is good. Table 1 lists four model evaluation indicators: explained variance value (EV), mean absolute error (MAE), mean square error (MSE) and R square value (R 2 ), where R 2 It is 0.852734, indicating that the WOA-XGBoost model is better at predicting the IGBT junction temperature.

[0126] Table 1 Model evaluation indicators

[0127]

[0128] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0129] The following are device embodiments of the present invention. For details not fully described therein, reference may be made to the corresponding method embodiments described above.

[0130] Figure 6 The following is a schematic diagram showing the structure of an IGBT module junction temperature monitoring device provided by an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, which are described in detail as follows:

[0131] like Figure 6 As shown, the IGBT module junction temperature monitoring device 6 includes:

[0132] The data acquisition module 61 is used to obtain the collector-emitter saturation voltage drop, DC power cycle number and case temperature of the online IGBT module;

[0133] The junction temperature monitoring module 62 is used to input the collector-emitter saturation voltage drop, the number of DC power cycles and the case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module;

[0134] Among them, the junction temperature monitoring model is obtained by training the whale optimized gradient boosting machine model based on a pre-constructed data set. The pre-constructed data set is obtained according to the aging data. The aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging case temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module.

[0135] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0136] Build a power cycle accelerated aging experimental platform, conduct accelerated power cycle experiments on the experimental platform, and obtain aging data of IGBT modules based on the accelerated power cycle experiments.

[0137] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0138] Set the experimental parameters, including the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current, and the number of DC power cycles of the IGBT module.

[0139] Obtain the collector-emitter saturation voltage drop, aging case temperature and precise junction temperature of the IGBT module during each DC power cycle aging period;

[0140] The aging data of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging period.

[0141] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0142] Inject a small current into the IGBT module during shutdown;

[0143] Measure the collector-emitter saturation voltage drop of the IGBT module under low current;

[0144] The precise junction temperature of the IGBT module is obtained based on the collector-emitter saturation voltage drop of the IGBT module under low current.

[0145] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0146] Obtain aging data, including the precise junction temperature of the IGBT module, the collector-emitter saturation voltage drop of the IGBT module, the number of DC power cycle aging cycles of the IGBT module, and the aging case temperature of the IGBT module.

[0147] A data set is constructed based on the aging data, and a junction temperature monitoring model is obtained by training the data set.

[0148] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0149] The data in the dataset is classified according to the different collector-emitter saturation voltage drops, different DC power cycle aging cycles and different aging case temperatures of the IGBT modules, and sub-datasets corresponding to different data are constructed;

[0150] The junction temperature monitoring model is trained based on the data set, including:

[0151] The whale optimized gradient boosting machine model is trained according to each sub-data set to obtain the junction temperature monitoring model.

[0152] In one possible implementation, the junction temperature monitoring module 62 is configured to:

[0153] Use a preset proportion of data in the dataset as the training set and the remaining data as the test set;

[0154] The whale optimized gradient boosting machine model is trained according to the training set, and the whale optimized gradient boosting machine model is tested according to the test set to obtain the junction temperature monitoring model.

[0155] An embodiment of the present invention obtains the collector-emitter saturation voltage drop, DC power cycle number, and case temperature of an online IGBT module and inputs the collector-emitter saturation voltage drop, DC power cycle number, and case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module. The junction temperature monitoring model is obtained by training a whale-optimized gradient boosting machine model based on a pre-constructed dataset. The pre-constructed dataset is obtained based on aging data, which includes the number of DC power cycle aging cycles, aging case temperature, the precise junction temperature of the IGBT module, and the collector-emitter saturation voltage drop of the IGBT module. This method can monitor the junction temperature of the IGBT module online, and by considering the impact of the DC power cycle number and case temperature on the junction temperature of the IGBT module, the junction temperature of the IGBT module monitored by the junction temperature monitoring model is more accurate.

[0156] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0157] Those skilled in the art will appreciate that the templates, units, and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0158] If the module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned embodiments of the IGBT module junction temperature monitoring method. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0159] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for monitoring junction temperature of an IGBT module, characterized in that: include: Obtain collector-emitter saturation voltage drop, DC power cycle number and case temperature of online IGBT modules; Inputting the collector-emitter saturation voltage drop, the number of DC power cycles, and the case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module; Among them, the junction temperature monitoring model is obtained by training the whale optimization gradient boosting machine model based on a pre-constructed data set, and the pre-constructed data set is obtained according to the aging data, and the aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging shell temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module.

2. The IGBT module junction temperature monitoring method according to claim 1, characterized in that: The method further comprises: A power cycle accelerated aging experimental platform is built, an accelerated power cycle experiment is performed on the experimental platform, and aging data of the IGBT module is obtained based on the accelerated power cycle experiment.

3. The IGBT module junction temperature monitoring method according to claim 2, characterized in that: The process of obtaining aging data of the IGBT module according to the accelerated power cycling experiment includes: Setting experimental parameters; wherein the parameters include the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current, and the number of DC power cycles of the IGBT module; Obtain the collector-emitter saturation voltage drop, aging case temperature and precise junction temperature of the IGBT module during each DC power cycle aging period; The aging data of the IGBT module is obtained according to the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging period.

4. The IGBT module junction temperature monitoring method according to claim 3, characterized in that: The process of obtaining the precise junction temperature of the IGBT module includes: Inject a small current into the IGBT module during shutdown; Measure the collector-emitter saturation voltage drop of the IGBT module under low current; The precise junction temperature of the IGBT module is obtained according to the collector-emitter saturation voltage drop of the IGBT module under the low current.

5. The IGBT module junction temperature monitoring method according to claim 1, characterized in that: The method further comprises: Obtaining aging data; wherein the aging data includes the precise junction temperature of the IGBT module, the collector-emitter saturation voltage drop of the IGBT module, the number of DC power cycle aging cycles of the IGBT module, and the aging case temperature of the IGBT module; A data set is constructed according to the aging data, and the junction temperature monitoring model is obtained by training according to the data set.

6. The IGBT module junction temperature monitoring method according to claim 5, characterized in that: The step of constructing a data set according to the aging data includes: Classifying the data in the dataset according to different collector-emitter saturation voltage drops, different numbers of DC power cycle aging cycles, and different aging case temperatures of the IGBT modules, and constructing sub-datasets corresponding to different data; The step of obtaining the junction temperature monitoring model through training according to the data set includes: The whale optimized gradient boosting machine model is trained according to each of the sub-data sets to obtain the junction temperature monitoring model.

7. The IGBT module junction temperature monitoring method according to claim 5, characterized in that: The step of obtaining the junction temperature monitoring model through training according to the data set includes: Using a preset proportion of data in the data set as a training set and the remaining data as a test set; The whale optimized gradient boosting machine model is trained according to the training set, and the whale optimized gradient boosting machine model is tested according to the test set to obtain the junction temperature monitoring model.

8. An IGBT module junction temperature monitoring device, characterized in that: include: Data acquisition module, used to obtain the collector-emitter saturation voltage drop, DC power cycle number and case temperature of the online IGBT module; a junction temperature monitoring module, configured to input the collector-emitter saturation voltage drop, the number of DC power cycles, and the case temperature into a pre-trained junction temperature monitoring model to obtain the junction temperature of the online IGBT module; Among them, the junction temperature monitoring model is obtained by training the whale optimization gradient boosting machine model based on a pre-constructed data set, and the pre-constructed data set is obtained according to the aging data, and the aging data includes the number of DC power cycle aging cycles of the IGBT module, the aging shell temperature of the IGBT module, the precise junction temperature of the IGBT module and the collector-emitter saturation voltage drop of the IGBT module.

9. The IGBT module junction temperature monitoring device according to claim 8, characterized in that: The junction temperature monitoring module is used for: A power cycle accelerated aging experimental platform is built, an accelerated power cycle experiment is performed on the experimental platform, and aging data of the IGBT module is obtained based on the accelerated power cycle experiment.

10. The IGBT module junction temperature monitoring device according to claim 8, characterized in that: The junction temperature monitoring module is used for: Setting experimental parameters; wherein the parameters include the switching time of the IGBT module, the gate voltage of the IGBT module, the heating current, the precise junction temperature test current, and the number of DC power cycles of the IGBT module; Obtain the collector-emitter saturation voltage drop, aging case temperature and precise junction temperature of the IGBT module during each DC power cycle aging period; The aging data of the IGBT module is obtained according to the collector-emitter saturation voltage drop of the IGBT module, the aging case temperature of the IGBT module and the precise junction temperature of the IGBT module corresponding to each DC power cycle aging period.

Images