System and method for parameter calibration of igbt model based on neural network

Abstract

The invention discloses a neural network-based IGBT model parameter calibration system and method. The system is provided with a data input module, a circuit actual measurement module, a parameter simulation module and a parameter calibration module; the base drive voltage Vgs and the set voltage are set through the data input module. The electrode input current IT; the circuit actual measurement module measures the actual output value VCE of the voltage between the IGBT collector and the emitter according to Vgs and IT; the parameter simulation module determines the voltage between the IGBT collector and the emitter according to Vgs, IT and the IGBT model parameters. The voltage simulation output value V'CE; the parameter calibration module uses the neural network model to calibrate the IGBT model parameters according to the deviation between VCE and V'CE; the effect is: it is enough to calibrate the model parameters of the IGBT device, and the opening The IGBT device is modeled in the two states of switching off and off, fully utilizing the parameter optimization effect of the neural network model, and more accurately describing the internal changes of the IGBT.

Description

technical field [0001] The invention relates to a parameter calibration technology of a semiconductor device, in particular to a neural network-based IGBT model parameter calibration system and method. Background technique [0002] The insulated gate bipolar transistor (Insulated Gate Bipolar Transistor, IGBT) has both the high input impedance of the Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET) and the power transistor (Giant Transistor, GTR ) has the advantages of low conduction voltage drop and has been widely used in various electronic applications. Currently available device information mainly comes from public device data sheets and software simulation models. For example, the ideal model of IGBT can be obtained in Simulink, and the specific model and parameters of a certain type of IGBT can be obtained in Pspice. In practical applications, in order to better analyze the dynamic characteristics of the device and optimize its driving and protection circuit...

AI Technical Summary

Problems solved by technology

At the same time, Hefner only provided a set of reference model parameters, and did not give the method of parameter extraction; existing scholars obtain parameters through manuals and empirical formulas, with many estimated values, and the data acquisition is not systematic; some people obtain them through empirical formulas The initial value of the parameters, and then use a set of experimental data to optimize the parameters, the method is simple and easy to implement, but the optimization algorithm is not mentioned, the optimization accuracy is not high, and the direct relationship between the optimization objective function and the model parameters cannot be obtained from the model formula

Images