Optimization Analysis Of Electrical Stress Simulation Method For Power Supply System

Circuit electrical stress analysis is a technology that test whether the components' strength in the circuit is standard in the circuit performance under the premise of stability.In a power system with high reliability requirements,electrical stress analysis is a step that must be performed.Due to the factors of manufacturing process,working environment and aging and so on,the component parameters will occur the phenomenon of drift,so the stress state of the circuit also has the characteristics of fluctuation.The traditional stress analysis is based on the stress of the components,the relationship between the parameters of each component in the circuit and the target is established.Through the search algorithm,the "worst" stress state of the circuit is obtained.This method of analysis ignores the relationship between circuit performance and circuit stress,and its analysis of the search for extreme values is too conservative.And another stress analysis method based on statistical tolerance also has the problem that the input distribution is difficult to measure and the analysis result of the complex circuit is low.At the same time,neither of these methods considers the relationship between circuits' performance and circuits' stress.In this paper,the traditional stress analysis method and the parameter distribution of the input components are studied.The research work can be composed of three parts: stress algorithms analysis and optimization,improvement and modeling of component parameter distribution and simulation verification.1)Based on the theory of circuit tolerance analysis,the method of Monte Carlo and the method of the worst case analysis used in traditional stress analysis is systematically studied.By comparing the advantages and disadvantages between the two algorithms,Monte Carlo method is used as the core algorithm to improve the method of the traditional stress analysis.The improved analytical method emphasizes the connection between the circuit stress and the circuit performance,and improves the simulation accuracy by comparing the results obtained through the method of extrapolating theoretical extreme value and extreme value analysis under the fixed sampling times.And finally to the circuit performance requirements as a constraint to correct the results of electrical stress analysis.2)When the model contained in the software device library does not meet the requirementsof simulation analysis,the method of component modeling under Saber platform is studied.Through the modeling of the diode and MOSFET,tolerance analysis as a constraint,established a set of component modeling method for circuit tolerance analysis.At the same time,the influence of the working environment on the drift mechanism of the parameters of different components is briefly introduced.The drift phenomenon of the parameters caused by various mechanisms is summarized and the three-level distribution model of component parameter drift is put forward for the first time.3)Based on Saber simulation software,a 20 W flyback switching power supply circuit was designed and verified.The feasibility and validity of the proposed theory are illustrated by two sets of simulations' contrast.First,the circuit components parameters drift are set to normal distribution and the working environment as a constraint to use three-level distribution to establish the parameter drift model for Monte Carlo stress analysis.The simulation results show that the stress extremes generated by the two input distributions are basically the same,but the standard deviation of the key stress parameters of the former is much smaller than that of the latter.Compared to the former,the ratio of the extreme value of the latter between the sampling extremes and extreme values method is big 6.4%?7.5%?4.8%?2.8%?3.6% and 3.7% respectively,which indicates that the stress analysis result with three distributions is more accurate.Secondly,by comparing the results of initial stress analysis,extreme stress analysis and improved Monte Carlo stress analysis,it can be seen that the improved stress analysis method reflects the relationship between performance and stress.Compared with the results of the "worst" extreme values method,stress ratio is small 27.9%,36.5%,18.9%,22.9%,20.9% and 17.6% respectively.