CN-122020982-A - Loss junction temperature prediction method of power module

CN122020982ACN 122020982 ACN122020982 ACN 122020982ACN-122020982-A

Abstract

The invention provides a loss junction temperature prediction method of a power module, which relates to the technical field of power electronics and comprises the steps of reading working condition files containing voltage, current, temperature ratio and resistance, importing characteristic parameters of a power module data manual, constructing a loss and thermal resistance model which dynamically changes along with the temperature, the current and the resistance based on the characteristic parameters by utilizing an interpolation algorithm, calculating a pulse sequence in a current half period, calculating conduction loss and switching loss corresponding to each pulse by an accumulation method to obtain total loss, calculating preliminary junction temperature based on the loss and the thermal resistance model, feeding back the preliminary junction temperature to a loss calculation process for cyclic iteration until the junction temperature difference value calculated by two times is smaller than a preset threshold value, and taking the last calculation result as the life prediction of the highest junction temperature. The method has the advantages that the pulse loss is accurately calculated through the accumulation method, the multidimensional dynamic model is constructed, and the thermoelectric coupling feedback iteration logic is introduced, so that the high efficiency and the accuracy of junction temperature calculation under mass working conditions are realized.

Inventors

LI SAINAN
SHEN LI

Assignees

嘉兴斯达微电子有限公司

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (10)

1. The loss junction temperature prediction method of the power module is characterized by comprising the following steps of: Step S1, reading a working condition file containing voltage, current, temperature ratio and resistance, and importing characteristic parameters of a power module data manual; s2, constructing a loss and thermal resistance model which dynamically changes along with temperature, current and resistance by utilizing an interpolation algorithm based on the characteristic parameters and the working condition file; Step S3, calculating a pulse sequence in a current half period, and calculating the conduction loss and the switching loss corresponding to each pulse by an accumulation method based on the loss and the thermal resistance model so as to obtain the total loss; And S4, calculating a preliminary junction temperature based on the loss and thermal resistance model, feeding back the preliminary junction temperature to a loss calculation process for cyclic iteration until the junction temperature difference value calculated twice successively is smaller than a preset threshold value, and taking the last calculation result as the highest junction temperature for a worker to predict the service life of the power module.
2. The method according to claim 1, wherein the loss and thermal resistance model includes an output characteristic model, and the process of constructing the output characteristic model in step S2 includes: Firstly, extracting output characteristic curve data at different temperatures in a data manual, and then interpolating between a normal temperature curve and a high temperature curve by adopting a linear interpolation method according to the working condition temperatures read in real time to generate a function curve which accords with the saturated conduction voltage drop and the current of the working condition temperatures as an output characteristic model.
3. The method according to claim 1, wherein the loss and thermal resistance model includes a loss characteristic model, and the process of constructing the loss characteristic model in step S2 includes: A1, acquiring current loss curve data and resistance loss curve data under normal temperature and high temperature conditions in a data manual, performing temperature correction interpolation by adopting a linear interpolation algorithm, and fitting to obtain a current loss initial curve and a resistance loss initial curve under corresponding working condition temperatures; A2, if the working condition current is in the numerical range of the current loss initial curve, determining left and right sampling points closest to the working condition current on the curve to conduct linear interpolation, if the working condition current is smaller than the current minimum value in the current loss initial curve, selecting two sampling points at the forefront end of the current loss initial curve to conduct processing to obtain a loss value, and if the working condition current is larger than the current maximum value in the current loss initial curve, selecting two sampling points at the tail end of the current loss initial curve to conduct processing to obtain a loss value, and multiplying the loss value obtained by the ratio of the working condition voltage to the specification voltage to obtain an intermediate loss value; Step A3, interpolation of resistance dimensions is carried out on the resistance loss initial curve by adopting the interpolation method same as the step A2, and a first resistance loss component under the working condition resistance and a second resistance loss component under the specification resistance are respectively obtained; And step A4, dividing the intermediate loss value by the second resistance loss component, and multiplying the obtained result by the first resistance loss component, so as to finally synthesize a target loss model which simultaneously covers four dimensions of working condition voltage, working condition current, working condition resistance and working condition temperature.
4. The method according to claim 1, wherein the loss and thermal resistance model includes a thermal characteristic model, and the process of constructing the thermal characteristic model in step S2 includes: Step B1, obtaining transient thermal resistance curve data in a power module data manual, and calculating to obtain a time parameter corresponding to half of a positive half period of a current sine wave according to output frequency in a working condition file; And B2, searching sampling points corresponding to the time parameter in the transient thermal resistance curve data, and if the time parameter is not a direct sampling point, selecting two adjacent sampling points before and after the time parameter to conduct linear interpolation processing, and fitting to obtain an equivalent thermal resistance coefficient corresponding to the time parameter, so that the thermal characteristic model construction aiming at the current working condition is completed.
5. The method of predicting a junction temperature of a loss according to claim 4, wherein the step S3 includes: Step S31, dividing the switching frequency in the working condition file by twice the output frequency to determine the number of voltage pulses in the positive half period of the current sine wave, and calculating to obtain an instantaneous current point corresponding to each pulse according to the position information of each voltage pulse in the period; Step S32, the instant current point is brought into the output characteristic model to interpolate, corresponding instant conduction voltage drop is obtained, and the conduction loss component of a single pulse is calculated by combining the duty ratio corresponding to the pulse under the current working condition; step S33, performing multidimensional interpolation processing in the loss characteristic model by utilizing the instantaneous current point to obtain a switching energy consumption value corresponding to a single pulse as a switching loss component; step S34, summing and accumulating the conduction loss components and the switching loss components of all pulses in the positive half period of the current to obtain the total loss of the power module; And step S35, taking the total loss as a thermophysical input quantity, and combining the equivalent thermal resistance coefficient obtained by fitting in the thermal characteristic model to calculate the junction temperature rise of the power module.
6. The method of predicting a junction temperature of a loss according to claim 5, wherein the step S4 includes: step S41, multiplying the total loss by the equivalent thermal resistance coefficient, and superposing the obtained product result on the shell temperature read in real time in the working condition file to obtain a preliminary junction temperature; step S42, the preliminary junction temperature is used as a feedback temperature parameter and is fed back to the loss calculation process, and an updated current loss initial curve and an updated resistance loss initial curve are obtained through re-fitting of an interpolation algorithm; Step S43, calculating updated total loss according to the current loss initial curve and the resistance loss initial curve obtained by re-fitting, and calculating the absolute value of the difference between the current junction temperature and the junction temperature obtained by the last calculation by using the updated total loss to recalculate the current junction temperature; Step S44, judging whether the absolute value is smaller than a preset threshold value: if yes, stopping the iterative process, and determining the current junction temperature obtained by the last calculation as the highest junction temperature of the power module; if not, the current junction temperature is updated to the feedback temperature parameter, and the step S42 is returned to continue to be executed until the convergence condition is met.
7. The method according to claim 6, wherein the preset threshold in step S44 is 2 ℃.
8. The loss junction temperature prediction method according to claim 1, further comprising, after performing the step S4: And S5, performing rain flow counting processing on junction temperature curves formed by the calculated junction temperatures, and extracting temperature rise fluctuation characteristics for life prediction of the power module.
9. The method according to claim 8, wherein the rain counting process in step S5 includes extracting all peaks and valleys in the junction temperature curve, and calculating the difference between adjacent peaks and valleys, and when the difference is greater than a predetermined minimum peak-valley value range difference, reserving the junction temperature curve for life prediction.
10. The method for predicting the junction temperature of loss according to claim 5, wherein the calculating the duty ratio corresponding to the pulse in the step S32 under the current working condition includes selecting a corresponding calculation formula according to a modulation mode selected by a user at the main interface, where the modulation mode includes SPWM modulation or SVPWM modulation.

Description

Loss junction temperature prediction method of power module Technical Field The invention relates to the technical field of power electronics, in particular to a loss junction temperature prediction method of a power module. Background Loss calculation and junction temperature prediction of power modules (such as IGBT, diode and the like) are core links of power electronic system design and reliability evaluation. In the prior art, loss junction temperature calculation generally faces the following technical difficulties: First, the calculation accuracy is limited. The traditional loss calculation mostly adopts an integrating method or a simple average value method, and when nonlinear waveforms under complex modulation modes (such as SPWM and SVPWM) are processed, loss characteristics in dynamic pulses cannot be accurately captured, so that a calculation result has larger deviation from an actual measurement value. Second, model adaptation is poor. Existing computational models are often based on a single temperature, voltage, or resistance reference in a data book. However, in actual working conditions, current, temperature and gate resistance fluctuate in real time, and a static model cannot dynamically adjust loss and thermal resistance parameters according to the working conditions, so that the static model is difficult to adapt to complex and changeable operating environments. Finally, the calculation process lacks feedback logic. Because of the strong thermo-electric coupling relation between the power consumption of the power module and the junction temperature (namely, the loss changes along with the temperature rise, and the temperature is also influenced by the loss), the traditional open loop calculation method ignores the real-time correction process between the loss and the temperature, and often leads to inaccurate prediction results of the highest junction temperature. In addition, when facing massive working condition files (such as hundreds of thousands of lines of data), the manual or semi-automatic processing mode is extremely low in efficiency, and the requirement of large-scale life prediction is difficult to meet. Disclosure of Invention Aiming at the problems in the prior art, the invention provides a loss junction temperature prediction method of a power module, which comprises the following steps: Step S1, reading a working condition file containing voltage, current, temperature ratio and resistance, and importing characteristic parameters of a power module data manual; s2, constructing a loss and thermal resistance model which dynamically changes along with temperature, current and resistance by utilizing an interpolation algorithm based on the characteristic parameters and the working condition file; Step S3, calculating a pulse sequence in a current half period, and calculating the conduction loss and the switching loss corresponding to each pulse by an accumulation method based on the loss and the thermal resistance model so as to obtain the total loss; And S4, calculating a preliminary junction temperature based on the loss and thermal resistance model, feeding back the preliminary junction temperature to a loss calculation process for cyclic iteration until the junction temperature difference value calculated twice successively is smaller than a preset threshold value, and taking the last calculation result as the highest junction temperature for a worker to predict the service life of the power module. Preferably, the loss and thermal resistance model includes an output characteristic model, and the process of constructing the output characteristic model in step S2 includes: Firstly, extracting output characteristic curve data at different temperatures in a data manual, and then interpolating between a normal temperature curve and a high temperature curve by adopting a linear interpolation method according to the working condition temperatures read in real time to generate a function curve which accords with the saturated conduction voltage drop and the current of the working condition temperatures as an output characteristic model. Preferably, the loss and thermal resistance model includes a loss characteristic model, and the process of constructing the loss characteristic model in step S2 includes: A1, acquiring current loss curve data and resistance loss curve data under normal temperature and high temperature conditions in a data manual, performing temperature correction interpolation by adopting a linear interpolation algorithm, and fitting to obtain a current loss initial curve and a resistance loss initial curve under corresponding working condition temperatures; A2, if the working condition current is in the numerical range of the current loss initial curve, determining left and right sampling points closest to the working condition current on the curve to conduct linear interpolation, if the working condition current is smaller than the current minimum value in the c