Accelerated life analysis and evaluation method for power module

Abstract

The invention relates to the technical field of power module testing, in particular to a power module accelerated life analysis and evaluation method. The method comprises the following steps: acquiring power module acceleration test data; performing temperature acceleration factor calculation according to the power module acceleration test data to obtain temperature acceleration factor data, and performing normal temperature life calculation according to the temperature acceleration factor data to obtain normal temperature life data; performing failure time extraction according to the power module acceleration test data to obtain failure time data; building a failure probability model according to the failure time data to obtain a failure probability model; performing failure life prediction according to the failure probability model to obtain failure life data; performing acceleration factor calculation according to the normal-temperature life data and the failure life data to obtain acceleration factor data; and obtaining accelerated life data of the power module according to the acceleration factor data and the failure life data. According to the invention, power module life prediction is closer to an actual use environment.

Description

Technical Field

[0001] The present invention relates to the technical field of power module testing, and in particular to a method for analyzing and evaluating the accelerated life of a power module. Background Art

[0002] With the popularization of modern electronic devices, especially in the fields of power electronics, automation control, transportation, and new energy vehicles, power modules, as key components, play a vital role in power conversion, regulation, and control. Power modules carry high voltage and high current loads and are in extreme working conditions such as high temperature and high load for a long time. The reliability of their performance directly affects the stability and life of the entire system. Therefore, life assessment and reliability analysis of power modules is one of the core tasks to ensure the long-term stable operation of the system. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a power module accelerated life analysis and evaluation method to solve at least one of the above technical problems.

[0004] The present application provides a method for analyzing and evaluating the accelerated life of a power module, comprising the following steps: Obtain power module accelerated test data; Calculate the temperature acceleration factor according to the power module accelerated test data to obtain the temperature acceleration factor data, and calculate the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data; Failure time is extracted according to the power module accelerated test data to obtain failure time data; failure probability model is constructed according to the failure time data to obtain failure probability model; failure life is predicted according to the failure probability model to obtain failure life data; Calculate the acceleration factor based on the normal temperature life data and the failure life data to obtain the acceleration factor data; The actual use condition life prediction is carried out based on the acceleration factor data and failure life data to obtain the accelerated life data of the power module.

[0005] In the present invention, by combining the temperature acceleration factor and the failure time data, the temperature and load conditions in the accelerated test can be converted into the life prediction results under the actual working environment. Through the accurate failure probability model, the real life of the power module in the actual use environment can be effectively evaluated. Through the integration of the life prediction of the actual use conditions, adaptive life prediction can be performed for different temperatures, loads and use scenarios. Through the dynamic calculation of the acceleration factor and the failure life, the method can adapt to different working environment conditions. For different working temperatures and load conditions, the model can automatically adjust the relevant parameters, and can predict the failure life at any time, ensuring that real-time and effective life prediction is provided in a variety of use scenarios. Traditional life tests usually require long-term actual tests and a large number of samples, while the present invention can significantly reduce the test cycle and the number of test samples by simulating the accelerated life calculation under different environments. At the same time, since a more accurate life prediction can be obtained in a shorter time, failures caused by insufficient testing in actual use can be avoided, thereby saving testing and maintenance costs.

[0006] Preferably, the obtaining of power module accelerated test data includes: Acquire test condition data, and perform accelerated simulation test operations under actual working conditions according to the test condition data, and record parameters during the process of the accelerated simulation test operations under actual working conditions to obtain accelerated test data of the power module.

[0007] In the present invention, by performing accelerated simulation tests under actual working conditions, the actual working state of the power module in the real application environment can be more closely matched. The parameters are recorded in detail during the accelerated simulation test, and the performance of the power module under different working conditions can be recorded in detail. By simulating actual working conditions for accelerated testing, the life data of the power module can be obtained in a shorter time, avoiding the time consumption problem in traditional long-term testing.

[0008] Preferably, the step of calculating the temperature acceleration factor according to the power module accelerated test data to obtain the temperature acceleration factor data, and calculating the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data includes: Extract the life of the power module according to the power module accelerated test data to obtain the life data of the power module; Select temperature points according to the power module accelerated test data to obtain temperature point data; Calculate the temperature acceleration factor according to the temperature point data to obtain the temperature acceleration factor data; The normal temperature life calculation is performed based on the temperature acceleration factor data and the power module life data to obtain the normal temperature life data.

[0009] The present invention uses various test data recorded in the experiment (such as life test data under high temperature, overload and other conditions) to extract the effective life of the power module under these specific test conditions through data analysis. Based on the accelerated test data of the power module, a temperature range is selected, which covers the actual working environment of the module. According to the distribution of data points and the temperature change law, several temperature points are selected for subsequent analysis. The Arrhenius model is used to calculate the acceleration effect on the life of the power module at different temperatures. There is a certain exponential relationship between temperature and life. The temperature acceleration factor is obtained by fitting the actual test data. Using the temperature acceleration factor, the life data obtained under the high temperature test is converted into the expected life at room temperature. By applying the reverse acceleration factor, the life under accelerated conditions is corrected to the life at room temperature, thereby realizing the function of accelerated testing.

[0010] Preferably, the extracting failure time according to the power module accelerated test data to obtain failure time data; constructing a failure probability model according to the failure time data to obtain a failure probability model; and predicting failure life according to the failure probability model to obtain failure life data include: Extract failure time according to power module accelerated test data to obtain failure time data; The Weibull distribution model is constructed based on the failure time data to obtain the failure probability model; Failure life prediction is performed based on the failure probability model to obtain preliminary failure life data; The failure life data is corrected at normal temperature according to the temperature acceleration factor data to obtain the failure life data.

[0011] The Weibull model in the present invention can accurately characterize the failure law of the power module and predict future failure events based on the actual failure time data. Failure life prediction based on the failure probability model can provide preliminary prediction data for the service life of the power module. By correcting the temperature acceleration factor, a life assessment closer to the actual working conditions can be provided, ensuring the accuracy of the life prediction results.

[0012] Preferably, the calculation of the acceleration factor according to the normal temperature life data and the failure life data to obtain the acceleration factor data includes: The acceleration factor data is obtained by dividing the normal temperature life data and the failure life data.

[0013] The present invention uses the Arrhenius model to calculate the acceleration factor and applies it to normal temperature life data and failure life data, which can provide a reliable life prediction under different environmental conditions. Based on the acceleration factor calculation of the Arrhenius model, the accelerated life data obtained by the power module in a high temperature environment can be converted into reliability data in a normal temperature environment.

[0014] Preferably, the actual use condition life prediction is performed according to the acceleration factor data and the failure life data to obtain the power module accelerated life data, including: Obtain actual usage condition data; According to the acceleration factor data and failure life data, actual life prediction is performed to obtain actual life data; The actual use condition life calculation is performed based on the actual use condition data and the actual life data to obtain the power module accelerated life data.

[0015] By combining the acceleration factor data with the failure life data, the present invention can more accurately simulate the life of the power module in the actual use environment. The laboratory test results are converted into reliability data in actual applications, which avoids the high cost and time consumption caused by long-term actual use tests, and provides more accurate prediction results. By predicting the life according to the actual use conditions, it can ensure that the reliability and quality of the product are fully evaluated before it leaves the factory.

[0016] Preferably, the step of selecting temperature points according to the power module accelerated test data to obtain temperature point data includes: Select temperature points at equal intervals according to the power module accelerated test data to obtain temperature point data; or, Non-equidistant temperature points are selected according to the power module accelerated test data to obtain temperature point data.

[0017] The medium-spaced temperature points of the present invention can provide uniform temperature change data within a certain range, while the non-equally-spaced temperature points can focus on testing specific temperature segments and capture more detailed temperature change characteristics. This improves the accuracy and pertinence of temperature point selection, ensures that all key temperature ranges can be covered, and obtains more representative and accurate test data, thereby optimizing test results.

[0018] Preferably, the step of selecting non-equidistant temperature points according to the power module accelerated test data to obtain temperature point data includes: Perform exponential regression calculation based on the power module accelerated test data to obtain exponential regression data; Select a temperature point in the high temperature range according to the exponential regression data to obtain the first temperature point data; Select a temperature point in the normal temperature range according to the exponential regression data to obtain the second temperature point data; Select the temperature point in the temperature change range according to the exponential regression data to obtain the third temperature point data; Performing temperature point density calculation on the third temperature point data to obtain temperature point density data; The first temperature point data and the second temperature point data are screened according to the temperature point density data to obtain the temperature point data.

[0019] The present invention uses exponential regression calculation to effectively extract the relationship between temperature and power module life, thereby accurately selecting key temperature intervals (such as high temperature intervals, normal temperature intervals, and temperature change intervals). By selecting non-uniformly spaced temperature points, the temperature point density in the high temperature interval, normal temperature interval, and temperature change interval will be dynamically adjusted according to actual test requirements and the operating characteristics of the power module, avoiding overly uniform or overly concentrated selection of temperature points in a specific interval. The targeted nature of the test is improved, so that the test data can be more focused on the key operating temperature intervals faced by the power module, thereby obtaining more accurate accelerated life analysis results.

[0020] Preferably, the step of selecting a temperature point in a high temperature interval according to exponential regression data to obtain first temperature point data includes: According to the temperature test condition data corresponding to the power module accelerated test data, high temperature intervals are divided by using preset high temperature division interval parameter data to obtain high temperature interval condition data; According to the high temperature interval condition data, the exponential regression data is subjected to high temperature interval extraction to obtain the high temperature interval data; The life temperature sensitivity and the life change rate between temperatures are calculated based on the high temperature interval data, and the life temperature sensitivity data and the life change rate data between temperatures are obtained respectively; Divide the high temperature interval data according to the life temperature sensitivity data and select equidistant points to obtain the first high temperature point data; Perform clustering processing on the life span change rate data between temperatures to obtain the life span change rate clustering data; The high temperature interval data corresponding to the life span change rate cluster number with the maximum cluster center value in the life span change rate clustering data is regarded as the second high temperature point data; The first high temperature point data and the second high temperature point data are subjected to duplicate point removal and are determined as the first temperature point data.

[0021] In the present invention, the temperature test conditions of the power module accelerated test data are analyzed, and the high temperature interval is divided in combination with the preset high temperature division interval parameter data, ensuring that the selected temperature points can fully cover the working performance of the module in a high temperature environment. By extracting the exponential regression data according to the high temperature interval condition data, the life data in the high temperature interval can be accurately associated with the temperature effect, ensuring that the extracted data can better represent the performance of the power module in the actual use environment. The present invention improves the understanding of the relationship between temperature and life, so that the selected temperature points can more accurately reflect the specific impact of temperature changes on the life of the power module, thereby improving the scientificity and accuracy of the temperature point selection. Cluster analysis makes the temperature point selection more targeted and accurate, and can identify and focus on the temperature intervals that have a greater impact on the life of the power module. After removing duplicate points, the distribution of temperature points is more reasonable and uniform, avoiding excessive concentration of temperature points in a certain specific temperature range.

[0022] Preferably, the filtering of the first temperature point data and the second temperature point data according to the temperature point density data to obtain the temperature point data includes: Perform temperature density calculations of different scales on the first temperature point data and the second temperature point data to obtain a first temperature density data set and a second temperature density data set; Determine whether the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data; When it is determined that the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data, the first temperature point data corresponding to the first temperature density data is retained to obtain the first temperature point retained data; Determine whether the second temperature point data of the second temperature point data set is greater than or equal to the temperature point density data; When it is determined that the second temperature density data of the second temperature density data set is greater than or equal to the temperature point density data, the second temperature point data corresponding to the second temperature density data is retained to obtain second temperature point retained data; The first temperature point retention data and the second temperature point retention data are determined as the temperature point data.

[0023] In the present invention, by calculating the temperature density of the first temperature point data and the second temperature point data at different scales, it can be ensured that the selected temperature points cover the characteristics of the temperature range at different scales. By setting a threshold for the temperature point density (for example, whether the temperature density is greater than or equal to the temperature point density data), it is possible to accurately screen out temperature points with higher density. By screening out temperature points with higher density, it can be ensured that the selected temperature points represent the true performance of the power module under different temperature conditions to a certain extent. By screening and retaining temperature points that meet the conditions, the analysis results affected by temperature points that do not meet the conditions are avoided.

[0024] The beneficial effect of the present invention is that the life estimation corrected by the temperature acceleration factor reduces the life prediction error caused by the temperature difference of different working environments. The failure probability model can accurately reflect the failure law of the power module in the actual working environment. When dealing with different loads and usage environments, the model can simulate different types of failure modes to provide accurate life assessment. The present invention calculates the acceleration factor for the normal temperature life data and the failure life data, which not only makes up for the shortcomings of the life prediction under normal temperature conditions, but also makes the life prediction closer to the actual use environment through the correction of the acceleration factor. Through detailed life assessment, the stability and reliability of the power module under actual use conditions are enhanced, which can help manufacturers and engineers identify failure modes in advance, thereby improving the product's ability to operate in extreme environments such as high pressure and high temperature. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings: Figure 1 A flowchart showing a method for analyzing and evaluating the accelerated life of a power module according to an embodiment of the present invention is provided; Figure 2 A flowchart showing a method for calculating a temperature acceleration factor and a normal temperature life in accordance with an embodiment of the present invention is provided; Figure 3 A flowchart of a failure life prediction method according to an embodiment is shown; Figure 4 A flowchart of the steps of a method for predicting life under actual usage conditions according to an embodiment is shown. DETAILED DESCRIPTION

[0026] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0027] In addition, the drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0028] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0029] See also Figures 1 to 4 The present application provides a method for analyzing and evaluating the accelerated life of a power module, comprising the following steps: S1: Obtain power module accelerated test data; S2: Calculate the temperature acceleration factor according to the power module acceleration test data to obtain the temperature acceleration factor data, and calculate the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data; S3: extract failure time according to the power module accelerated test data to obtain failure time data; construct a failure probability model according to the failure time data to obtain a failure probability model; predict failure life according to the failure probability model to obtain failure life data; S4: Calculate the acceleration factor according to the normal temperature life data and the failure life data to obtain the acceleration factor data; S5: Perform life prediction under actual use conditions based on the acceleration factor data and the failure life data to obtain accelerated life data of the power module.

[0030] In one embodiment, the power module accelerated test data is usually composed of experimental data running under high temperature, high current or high voltage conditions, including the operating temperature of the power module during the test, the failure time of the module under accelerated test conditions, the operating current and voltage. It is obtained through high temperature aging experiments, and different ambient temperatures are set (such as 125°C, 150°C, 175°C). Rated or excess voltage and current are applied to put the module in an overload condition. The operating time is recorded and the failure samples are counted.

[0031] Temperature acceleration factor (AF) calculation: ; in, is the temperature acceleration factor, is an exponential function, is the activation energy of the power module (0.7-1.2eV), is the Boltzmann constant (8.617× eV / K), is the actual operating temperature (K), is the accelerated test temperature (K).

[0032] Failure time data is extracted from the test samples, and the failure time of multiple samples at the test temperature is recorded. The failure time is fitted using the Weibull distribution, and the specific calculation is: ; is the probability density function of the failure time, that is, at a certain time The probability of sample failure is is a shape parameter, if represents the wear failure stage, is the characteristic lifespan, is the failure time, which refers to the time from the start of the experiment to the failure of the sample. is the natural index. Based on the failure time data, the maximum likelihood estimation (MLE) is used to fit the Weibull distribution, and we get , .

[0033] By constructing a failure probability model, the lifespan under different reliability levels is calculated. Set the target reliability , calculate lifespan . ,in is the reliability life, For life reliability.

[0034] Based on the normal temperature life and failure life, calculate the total acceleration factor: ,in is the total acceleration factor, is the normal temperature life data, In order to obtain the life under accelerated test conditions, the traditional acceleration factor calculation relies on experimental data (such as life under high temperature) for accelerated calculation, and then the normal temperature life is corrected according to the Arrhenius model or other empirical formulas. However, considering the influence of the failure probability model, such calculation can make a more scientific correction to the acceleration factor, so that it is not just based on a simple multiple of temperature change, but more in line with factors such as workload and vibration in actual working conditions.

[0035] Combining the acceleration factor and failure life, the actual service life can be estimated: , Accelerated life data for power modules, is the total acceleration factor, is the reliability life (also known as failure life data).

[0036] Preferably, the obtaining of power module accelerated test data includes: S11: Acquire test condition data, and perform an accelerated simulation test operation under actual working conditions according to the test condition data, and record parameters during the accelerated simulation test operation under actual working conditions to obtain accelerated test data of the power module.

[0037] In one embodiment, the test condition data is the operating parameters of the power module in the accelerated aging test. Determine the main factors affecting the life of the power module, such as temperature, current, voltage, ambient humidity, etc. Start the power module aging test according to the set accelerated stress conditions. The test equipment includes a high-temperature oven (providing a constant temperature environment, such as 150°C, 175°C). A DC regulated power supply (providing constant current and voltage). A power load device (applying actual working loads, such as constant current loads). A data acquisition system (recording current, voltage, temperature and failure information in real time).

[0038] Place the power module in a constant temperature oven and maintain a stable temperature for 30 minutes. Apply a set high temperature (e.g. 150°C). Apply a set high current (e.g. 1.5 times the rated current). Apply a set high voltage (e.g. 1.2 times the rated voltage). Run the module continuously and monitor temperature, current, and voltage. Record data points every 10 minutes. Failure criteria are current deviation exceeding 10%, or voltage drop exceeding 5%, or internal circuit open / short circuit. Once a failure is detected, stop the test and record the failure time.

[0039] Preferably, the step of calculating the temperature acceleration factor according to the power module accelerated test data to obtain the temperature acceleration factor data, and calculating the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data includes: S21: extracting the life of the power module according to the power module accelerated test data to obtain the life data of the power module; S22: selecting temperature points according to the power module accelerated test data to obtain temperature point data; S23: Calculate the temperature acceleration factor according to the temperature point data to obtain the temperature acceleration factor data; S24: Calculate the normal temperature life according to the temperature acceleration factor data and the power module life data to obtain the normal temperature life data.

[0040] In one embodiment, the failure time of each power module at different temperatures is recorded, and the average failure time of multiple modules at the same temperature is counted.

[0041] Set specific time intervals to select temperature points.

[0042] Temperature acceleration factor (AF) calculation: ; in, is the temperature acceleration factor, is an exponential function, is the activation energy of the power module (0.7-1.2eV), is the Boltzmann constant (8.617× eV / K), is the actual operating temperature (K), is the accelerated test temperature (K).

[0043] Calculation at normal temperature: ,in, is the normal temperature life data, is the life obtained under accelerated test conditions.

[0044] Preferably, the extracting failure time according to the power module accelerated test data to obtain failure time data; constructing a failure probability model according to the failure time data to obtain a failure probability model; and predicting failure life according to the failure probability model to obtain failure life data include: S31: extracting failure time according to the power module accelerated test data to obtain failure time data; S32: constructing a Weibull distribution model based on the failure time data to obtain a failure probability model; S33: Predict the failure life according to the failure probability model to obtain preliminary failure life data; S34: Perform normal temperature life correction on the failure life data according to the temperature acceleration factor data to obtain failure life data.

[0045] In one embodiment, the failure criteria of the power module are set, such as output current drop of more than 10%. Output voltage drop of more than 5%. Internal short circuit or open circuit. Accelerated life test is performed at different temperatures (such as 125°C, 150°C, 175°C), and the failure time of each sample is recorded. The failure samples at each temperature point are counted to obtain failure time data.

[0046] Failure time data is extracted from the test samples, and the failure time of multiple samples at the test temperature is recorded. The failure time is fitted using the Weibull distribution, and the specific calculation is: ; is the probability density function of the failure time, that is, at a certain time The probability of sample failure is is a shape parameter, if represents the wear failure stage, is the characteristic lifespan, is the failure time, which refers to the time from the start of the experiment to the failure of the sample. is the natural index. Based on the failure time data, the maximum likelihood estimation (MLE) is used to fit the Weibull distribution, and we get , . is the shape parameter, which determines the failure mode: : Early failure (such as manufacturing defects). : Random failure (exponential distribution). : Wear failure (aging loss). is the characteristic life (i.e., the life at which 63.2% of the samples fail). is the expiration time.

[0047] Using maximum likelihood estimation (MLE) to solve and : ,in is the expiration time, is the sample size, is the shape parameter, is the characteristic life, which means the life when 63.2% of the samples fail. It represents the scale of the distribution and directly affects the time range of failure. For the The failure time of a sample, that is, the time from the start of the experiment to the failure of the sample. Usually it is the time data recorded in the experiment.

[0048] By constructing a failure probability model, the lifespan under different reliability levels is calculated. Set the target reliability , calculate lifespan . ,in is the reliability life, is the life reliability. Combining the acceleration factor and failure life, the actual service life is calculated: , Accelerated life data for power modules, is the total acceleration factor, is the reliability life (also known as failure life data).

[0049] Preferably, the calculation of the acceleration factor according to the normal temperature life data and the failure life data to obtain the acceleration factor data includes: S41: Perform phase division processing on the normal temperature life data and the failure life data to obtain acceleration factor data.

[0050] In one embodiment, the total acceleration factor is calculated based on the normal temperature life and the failure life: ,in is the total acceleration factor, is the normal temperature life data, is the life obtained under accelerated test conditions.

[0051] Preferably, the actual use condition life prediction is performed according to the acceleration factor data and the failure life data to obtain the power module accelerated life data, including: S51: Acquiring actual usage condition data; S52: performing actual life prediction according to the acceleration factor data and the failure life data to obtain actual life data; S53: Perform actual use condition life calculation according to the actual use condition data and the actual life data to obtain accelerated life data of the power module.

[0052] In one embodiment, the operating temperature and load conditions of the power module are obtained from the actual application environment. The working cycle of the power module is recorded, including the on / off frequency, working time, etc., to obtain the working cycle and frequency. Life to failure Make corrections, ,in Actual life span data.

[0053] Under high load conditions, the life of the power module will be shortened accordingly. The life is adjusted by the load factor LF (the load factor is obtained based on experimental data or standard literature). The load factor LF is determined by experiment. When the load is 80%, the load factor is 0.8. Divide the actual life data by the load to obtain the accelerated life data of the power module.

[0054] Preferably, the step of selecting temperature points according to the power module accelerated test data to obtain temperature point data includes: Select temperature points at equal intervals according to the power module accelerated test data to obtain temperature point data; or, Non-equidistant temperature points are selected according to the power module accelerated test data to obtain temperature point data.

[0055] In one embodiment, the equally spaced temperature point selection refers to selecting a number of temperature points at the same interval within the desired temperature range (e.g., 25°C to 175°C). Select N temperature points. Set N=5 as an example. Calculate the interval between each temperature point in equal proportion. Select the temperature points by interval.

[0056] The selection of non-equidistant temperature points is based on the working characteristics of the module within this temperature range. According to the failure characteristics of the power module, the module failure is more significant in certain temperature ranges (such as high temperature areas), so you can choose to increase the test frequency near these temperature points (the temperature range can be obtained by curve fitting and inflection point calculation to obtain the demarcation point, and the data interval is divided by the demarcation point, the low temperature is the low temperature area, and the high temperature is the high temperature area). For example, according to the experimental requirements or module characteristics, a more intensive temperature point selection is performed in the range of 85℃ to 175℃. Select multiple temperature points based on the module performance characteristics and experimental requirements. In this case, select the following temperature points: low temperature area (25℃ to 85℃): interval is 20℃, high temperature area (85℃ to 175℃): interval is 10℃.

[0057] Preferably, the step of selecting non-equidistant temperature points according to the power module accelerated test data to obtain temperature point data includes: Perform exponential regression calculation based on the power module accelerated test data to obtain exponential regression data; Select a temperature point in the high temperature range according to the exponential regression data to obtain the first temperature point data; Select a temperature point in the normal temperature range according to the exponential regression data to obtain the second temperature point data; In one embodiment, more importantly, a second-order derivative is calculated based on the exponential regression data to obtain curvature data; a curvature change is calculated based on the curvature data to obtain curvature change data; a threshold is screened based on the curvature change data to obtain inflection point data; a temperature point in the normal temperature interval is selected based on the inflection point data to obtain temperature point data in the normal temperature interval; a temperature point device parameter association is constructed based on the temperature point data in the normal temperature interval to obtain primary second temperature point data; a time equivalent calculation is performed based on the primary second temperature point data to obtain secondary second temperature point data; an information entropy calculation is performed based on the secondary second temperature point data to obtain temperature point entropy value data; the normal temperature interval temperature point data is iteratively pruned based on the temperature point entropy value data to obtain second temperature point data; The exponential regression model is expressed as: , is exponential regression data, generally accelerated test life data, is the regression coefficient, is the natural exponential term, is another regression coefficient, It is the temperature test data in the power module accelerated test data; second-order derivative calculation: , the meaning of the parameters is the same as above. , is the curvature. Curvature change calculation (calculate the rate of change of adjacent curvature values): , is the curvature change, For the +1 temperature point curvature value, For the The curvature value of each temperature point is used to calculate the curvature change rate. : ,in The time parameter data corresponding to the curvature change. Set the curvature change rate threshold , select the average value 3 times the standard deviation : ,in is the curvature change rate threshold, is the mean value of the curvature change, is the standard deviation of the curvature change, and the point that meets the conditions is selected as the inflection point: . Select the temperature point in the normal temperature range according to the inflection point data: set the physical meaning of the normal temperature range (such as 25℃-85℃). Determine the boundary point of the normal temperature range based on the inflection point data and : , , select a temperature point within this interval: , obtain the temperature point data in the normal temperature range for device parameter correlation analysis.

[0058] Temperature point-device parameter association construction: calculate the device parameters corresponding to each temperature point (such as life, stress, deformation, etc.), calculate the sensitivity of temperature to device parameters: ,in, Temperature point Sensitivity to device parameters, are device parameters (such as life, stress, deformation, etc.), is the temperature point. Construct the temperature-device parameter correlation matrix: ,in Represent temperature Parameters The sensitivity of , construct the temperature-device parameter correlation matrix : ; is the temperature-device parameter correlation matrix, Temperature point Device parameters sensitivity, is the number of selected temperature points, is the number of device parameters studied; choose The most sensitive temperature point in: ,in The sensitivity screening threshold is determined by the global sensitivity mean and standard deviation. , is the mean sensitivity of all temperature points, The temperature points that meet the condition are selected to form the primary second temperature point data for time equivalent calculation.

[0059] The time equivalent calculation uses the acceleration factor (AF) model to calculate the time equivalent coefficient of temperature: , is the acceleration factor, is the base of natural logarithms, is the activation energy, is the Boltzmann constant, is the reference temperature, is the actual test temperature. Calculate the time weighting coefficient: , is the time weighting coefficient, is the power adjustment value of the acceleration factor, It is a material property index, determined based on the material aging characteristics. It is used to adjust the weights of high temperature points and low temperature points so that the contribution of different temperature points to reliability analysis is more balanced. For example, a high temperature point has a larger acceleration factor, which will greatly shorten the experimental time, while a low temperature point has a smaller acceleration factor, which is closer to the actual working environment. The influence weight of the high temperature point is reduced through the time weighting coefficient. Indicates temperature point Corresponding device parameters (such as life, stress, deformation, etc.). Calculate the corrected parameter values: in, is the adjusted parameter value, As a time equivalent correction factor. Set the time equivalent impact threshold (preset or calculated based on the previous threshold calculation method), and select the temperature point that meets the conditions (greater than the threshold). After time equivalent calculation and screening, the system obtains the secondary second temperature point data. These temperature points are adjusted by time weighting so that the test data can more accurately represent the actual usage. Compared with the primary second temperature point data, the secondary second temperature point data considers the impact of different temperature points on device aging in a more balanced manner, improving the reliability of the experimental data and engineering applicability.

[0060] Calculate the probability distribution for each temperature point: , At the temperature point Next, the characteristic variable The conditional probability distribution of Temperature point The characteristic variable The observation frequency, is the characteristic variable, indicating the device under temperature A characteristic under the condition, such as current change, is the temperature point, which refers to the different temperature conditions measured during the experiment. Calculate the information entropy: , Temperature point Information entropy.

[0061] Calculate the global entropy mean and standard deviation: , is the global entropy mean, is the total number of temperature points, Temperature point The corresponding information entropy. Calculate the standard deviation: , is the standard deviation of the global entropy. , is the pruning threshold, which is used to remove redundant temperature points with low information entropy. , is the pruning threshold, if a temperature point The information entropy is lower than the pruning threshold , then the temperature point will be deleted. Calculate the information entropy of all temperature points . Calculate the global entropy mean and standard deviation . Set the pruning threshold . Delete all the Recalculate the information entropy of the remaining temperature points. If there are still low entropy points, repeat the above steps until the information entropy of all temperature points is higher than the pruning threshold.

[0062] Select the temperature point in the temperature change range according to the exponential regression data to obtain the third temperature point data; Performing temperature point density calculation on the third temperature point data to obtain temperature point density data; The first temperature point data and the second temperature point data are screened according to the temperature point density data to obtain the temperature point data.

[0063] In one embodiment, an exponential model is used for regression fitting: , is exponential regression data, generally accelerated test life data, is the regression coefficient, is the natural exponential term, is another regression coefficient, This is the temperature test data in the power module acceleration test data.

[0064] Divide the exponential regression data by a threshold value (the threshold value can be obtained by binary division) to obtain exponential regression normal temperature interval data and exponential regression high temperature interval data. Perform proportional temperature point selection based on the exponential regression normal temperature interval data and the exponential regression high temperature interval data to obtain first temperature point data and second temperature point data respectively. Divide the exponential regression normal temperature interval data and the exponential regression high temperature interval data into a smaller scale (such as further dividing into three equal parts) to obtain normal temperature interval refined data and high temperature interval refined data; perform slope calculation based on the normal temperature interval refined data and the high temperature interval refined data to obtain normal temperature interval slope data and high temperature interval slope data; perform neighboring interval ratio calculation based on the normal temperature interval slope data and the high temperature interval slope data to obtain neighboring interval ratio data; perform normalization based on the neighboring interval ratio data to obtain ratio normalized data; perform weighted calculation and rounding on the preset temperature point quantity data based on the ratio normalized data to obtain the temperature point quantity data corresponding to each interval (i.e., interval temperature point quantity data); perform temperature point selection (proportional selection) based on the interval temperature point quantity data to obtain the third temperature point data; Perform temperature point density calculation on the third temperature point data to obtain temperature point density data; perform smaller scale division (such as further dividing into three equal parts) according to the first temperature point data and the second temperature point data to obtain first temperature point refined data and second temperature point refined data; perform temperature point density calculation on the scale data corresponding to the scale division on the first temperature point refined data and the second temperature point refined data to obtain first temperature point density data and second temperature point density data respectively; compare the first temperature point density data and the second temperature point density data with the temperature point density data respectively, and only retain the first temperature point data / second temperature point data corresponding to the first temperature point density data / second temperature point density data greater than the temperature point density data, thereby obtaining temperature point data.

[0065] The first temperature point data and the second temperature point data are screened according to the temperature point density data to obtain the temperature point data.

[0066] Preferably, the step of selecting a temperature point in a high temperature interval according to exponential regression data to obtain first temperature point data includes: According to the temperature test condition data corresponding to the power module accelerated test data, high temperature intervals are divided by using preset high temperature division interval parameter data to obtain high temperature interval condition data; According to the high temperature interval condition data, the exponential regression data is subjected to high temperature interval extraction to obtain the high temperature interval data; The life temperature sensitivity and the life change rate between temperatures are calculated based on the high temperature interval data, and the life temperature sensitivity data and the life change rate data between temperatures are obtained respectively; Divide the high temperature interval data according to the life temperature sensitivity data and select equidistant points to obtain the first high temperature point data; Perform clustering processing on the life span change rate data between temperatures to obtain the life span change rate clustering data; The high temperature interval data corresponding to the life span change rate cluster number with the maximum cluster center value in the life span change rate clustering data is regarded as the second high temperature point data; The first high temperature point data and the second high temperature point data are subjected to duplicate point removal and are determined as the first temperature point data.

[0067] In one embodiment, the high temperature interval division parameters are preset according to the condition data of the power module accelerated test. The high temperature interval is defined as temperature , and further divided into different small intervals according to different temperature sections (such as the principle of three equal divisions). The preset high temperature division parameters are: high temperature and low interval: , high temperature medium range: , high temperature and high range: , according to the temperature information in the test data, the data is grouped according to the high temperature intervals divided above. According to the high temperature interval condition data, the exponential regression data is subjected to high temperature interval extraction to obtain high temperature interval data; Lifetime temperature sensitivity calculation: ; is the lifetime temperature sensitivity, and are the life data of adjacent temperature points, and are adjacent temperature points respectively.

[0068] Life span change rate between temperatures calculate: ; According to the life temperature sensitivity data, the high temperature interval data is divided according to the trend of sensitivity change. For example, if the sensitivity value changes greatly, more temperature points are selected in this interval. Set a threshold. When the rate of change of sensitivity is greater than a predetermined threshold, it indicates that the temperature change has a greater impact on the life. At this time, more dense temperature points should be selected to divide the interval into multiple smaller temperature sub-intervals. The temperature interval of each sub-interval is small (for example: 5°C or less per interval) to capture rapid changes in sensitivity. For areas where the sensitivity change rate is less than the threshold, fewer temperature points can be selected, such as larger temperature intervals (for example, 20°C or more per interval), because in these intervals, the impact of temperature changes on life is small, so too many test points are not needed.

[0069] Use a clustering algorithm to process the data of the life change rate between temperatures and divide the data into different clusters. For example, a distance-based clustering method is used to cluster temperature points with similar change rates into one cluster. The data is clustered into three categories based on the similarity of the change rate: Class 1: small change rate, representing a lower sensitivity interval, Class 2: medium change rate, representing a medium sensitivity interval, Class 3: large change rate, representing a high sensitivity interval. Based on the clustering results, select the cluster with the largest cluster center value. The change rate cluster of class 3 is the largest cluster, and the data points in the corresponding high temperature interval data are regarded as the second high temperature temperature point data. The first high temperature temperature point data and the second high temperature temperature point data are merged to remove duplicate temperature points.

[0070] Preferably, the filtering of the first temperature point data and the second temperature point data according to the temperature point density data to obtain the temperature point data includes: Perform temperature density calculations of different scales on the first temperature point data and the second temperature point data to obtain a first temperature density data set and a second temperature density data set; Determine whether the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data; When it is determined that the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data, the first temperature point data corresponding to the first temperature density data is retained to obtain the first temperature point retained data; Determine whether the second temperature point data of the second temperature point data set is greater than or equal to the temperature point density data; When it is determined that the second temperature density data of the second temperature density data set is greater than or equal to the temperature point density data, the second temperature point data corresponding to the second temperature density data is retained to obtain second temperature point retained data; The first temperature point retention data and the second temperature point retention data are determined as the temperature point data.

[0071] In one embodiment, when calculating the temperature density, a suitable scale is selected according to different temperature intervals. For example, in areas where the temperature changes slowly (such as low temperature areas), a larger interval is used to calculate the density. In areas where the temperature changes more drastically (such as high temperature areas), a smaller interval is used to calculate the density in order to capture the changes more accurately. Temperature density is to evaluate the temperature changes in a certain interval by calculating the intervals between adjacent temperature points. The selected low temperature zone (25 degrees to 85 degrees) and high temperature zone (85 degrees to 175 degrees) need to adopt different calculation scales respectively. In the low temperature zone, select a temperature interval with an interval of 30 degrees or two equal parts to calculate the temperature density. In the high temperature zone, select a temperature interval with an interval of 15 degrees or five equal parts to calculate the temperature density. Determine whether the temperature density corresponding to each interval in the low temperature zone or the high temperature zone is greater than the threshold. If it is greater, it is retained, and if it is not greater, it is removed.

[0072] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for analyzing and evaluating the accelerated life of a power module, characterized in that: The following steps are involved: Obtain power module accelerated test data; Calculate the temperature acceleration factor according to the power module accelerated test data to obtain the temperature acceleration factor data, and calculate the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data; Failure time extraction is performed based on the power module accelerated test data to obtain failure time data; a failure probability model is constructed based on the failure time data to obtain a failure probability model; failure life prediction is performed based on the failure probability model to obtain failure life data; Calculate the acceleration factor based on the normal temperature life data and the failure life data to obtain the acceleration factor data; The actual use condition life prediction is carried out based on the acceleration factor data and failure life data to obtain the accelerated life data of the power module.

2. The method according to claim 1, characterized in that The obtaining of power module accelerated test data includes: Acquire test condition data, and perform accelerated simulation test operations under actual working conditions according to the test condition data, and record parameters during the process of the accelerated simulation test operations under actual working conditions to obtain accelerated test data of the power module.

3. The method according to claim 1, characterized in that The method of calculating the temperature acceleration factor according to the power module accelerated test data to obtain the temperature acceleration factor data, and calculating the normal temperature life according to the temperature acceleration factor data to obtain the normal temperature life data includes: Extract the life of the power module according to the power module accelerated test data to obtain the life data of the power module; Select temperature points according to the power module accelerated test data to obtain temperature point data; Calculate the temperature acceleration factor according to the temperature point data to obtain the temperature acceleration factor data; The normal temperature life calculation is performed based on the temperature acceleration factor data and the power module life data to obtain the normal temperature life data.

4. The method according to claim 1, characterized in that: The failure time is extracted according to the power module accelerated test data to obtain the failure time data; A failure probability model is constructed according to the failure time data to obtain a failure probability model; Failure life prediction is performed based on the failure probability model to obtain failure life data, including: Extract failure time according to power module accelerated test data to obtain failure time data; The Weibull distribution model is constructed based on the failure time data to obtain the failure probability model; Failure life prediction is performed based on the failure probability model to obtain preliminary failure life data; The failure life data is corrected at normal temperature according to the temperature acceleration factor data to obtain the failure life data.

5. The method according to claim 1, characterized in that: The acceleration factor is calculated based on the normal temperature life data and the failure life data to obtain the acceleration factor data, including: The acceleration factor data is obtained by dividing the normal temperature life data and the failure life data.

6. The method according to claim 1, characterized in that The actual use condition life prediction is performed according to the acceleration factor data and the failure life data to obtain the power module accelerated life data, including: Obtain actual usage condition data; According to the acceleration factor data and failure life data, actual life prediction is performed to obtain actual life data; The actual use condition life calculation is performed based on the actual use condition data and the actual life data to obtain the power module accelerated life data.

7. The method according to claim 3, characterized in that The step of selecting temperature points according to the power module accelerated test data to obtain temperature point data includes: Select temperature points at equal intervals according to the power module accelerated test data to obtain temperature point data; or, Non-equidistant temperature points are selected according to the power module accelerated test data to obtain temperature point data.

8. The method according to claim 7, characterized in that The step of selecting non-equidistant temperature points according to the power module accelerated test data to obtain temperature point data includes: Perform exponential regression calculation based on the power module accelerated test data to obtain exponential regression data; Select a temperature point in the high temperature range according to the exponential regression data to obtain the first temperature point data; Select a temperature point in the normal temperature range according to the exponential regression data to obtain the second temperature point data; Select the temperature point in the temperature change range according to the exponential regression data to obtain the third temperature point data; Performing temperature point density calculation on the third temperature point data to obtain temperature point density data; The first temperature point data and the second temperature point data are screened according to the temperature point density data to obtain the temperature point data.

9. The method according to claim 8, characterized in that The step of selecting a temperature point in a high temperature range according to the exponential regression data to obtain first temperature point data includes: According to the temperature test condition data corresponding to the power module accelerated test data, high temperature intervals are divided by using preset high temperature division interval parameter data to obtain high temperature interval condition data; According to the high temperature interval condition data, the exponential regression data is subjected to high temperature interval extraction to obtain the high temperature interval data; The life temperature sensitivity and the life change rate between temperatures are calculated based on the high temperature interval data, and the life temperature sensitivity data and the life change rate data between temperatures are obtained respectively; Divide the high temperature interval data according to the life temperature sensitivity data and select equidistant points to obtain the first high temperature point data; Perform clustering processing on the life span change rate data between temperatures to obtain the life span change rate clustering data; The high temperature interval data corresponding to the life span change rate cluster number with the maximum cluster center value in the life span change rate clustering data is regarded as the second high temperature point data; The first high temperature point data and the second high temperature point data are subjected to duplicate point removal and are determined as the first temperature point data.

10. The method according to claim 9, characterized in that The step of screening the first temperature point data and the second temperature point data according to the temperature point density data to obtain the temperature point data includes: Perform temperature density calculations of different scales on the first temperature point data and the second temperature point data to obtain a first temperature density data set and a second temperature density data set; Determine whether the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data; When it is determined that the first temperature density data of the first temperature density data set is greater than or equal to the temperature point density data, the first temperature point data corresponding to the first temperature density data is retained to obtain the first temperature point retained data; Determine whether the second temperature point data of the second temperature point data set is greater than or equal to the temperature point density data; When it is determined that the second temperature density data of the second temperature density data set is greater than or equal to the temperature point density data, the second temperature point data corresponding to the second temperature density data is retained to obtain second temperature point retained data; The first temperature point retention data and the second temperature point retention data are determined as the temperature point data.

Images