Study On Lifespan Evaluation Method Of Polyimide Films Modified By Nano-TiO₂ Under Pulsed Voltage

Flexible DC transmission engineering is an important technology to solve the spatial reverse distribution of power load center and new energy resources and large-scale grid absorption of new energy power generation.However,as the key equipment of flexible DC converter valve,high voltage IGBT devices,the outermost Polyimide（PI）films of chip passivation layer is the weak link of insulation system,which directly affects the safety and reliability of high voltage IGBT devices and even flexible DC transmission system.With the development of nano control technology,the electrical and physicochemical properties of PI films modified by nano have been significantly improved.However,with the improvement of voltage level in flexible HVDC engineering,it is urgent to improve the accuracy of lifespan assessment of PI films modified by nano under pulse voltage,so as to guide the maintenance and replacement of electrical components in flexible HVDC engineering.In this paper,PI films modified by nano-TiO₂ are prepared by ultrasonic oscillation method and in situ polymerization method.The electric aging test platform under pulse voltage is built.Eight lifespan evaluation characteristics are established from two dimensions.Taking the fitting correlation coefficient of the lifespan model as the evaluation standard,the synergistic influence law of probability distribution model,empirical failure probability formula,failure data sample size and statistical inspection steps on the lifespan evaluation results of PI films modified by nano-TiO₂is revealed by fitting the inverse power function electric lifespan model,the method of lifespan evaluation is gradually optimized.The main work and conclusions of this paper are as follows:（1）PI films modified by nano-TiO₂ are prepared,failure data are obtained and lifespan evaluation is carried out.It is found that the lifespan evaluation results of Weibull distribution are generally larger than those of lognormal distribution,in which the lifespan evaluation results of unbiased estimation formula are the largest and that of Hessian formula is the least.It is shown that when the same set of failure data is used to evaluate the lifespan of PI films modified by nano-TiO₂using the combined lifespan evaluation scheme of Weibull distribution with unbiased estimation formula,this lifespan evaluation combination yields the smallest probability of failure and the longest lifespan for the same pressurization time.Conversely,the combined scheme of lifespan evaluation with lognormal distribution and Hessian formula is more likely to yield the conclusion of insulation failure.（2）Descriptive statistics of PI films modified by nano-TiO₂ are established from four dimensions,and the overall law of failure data of this batch of samples is obtained.Probability distribution test,consistency test of failure mechanism,significance and correlation test of regression equation are introduced.It is found that when 1200V is applied,the failure data of PI films modified by nano TiO₂ could not pass the consistency test of failure mechanism,but it still passe the probability distribution test and the significance and correlation test of regression equation of inverse power function lifespan model.By comparing whether 1200V failure data is applied for lifespan evaluation results,it is found that using the data set that failed to pass the consistency test of failure mechanism would change the correlation coefficient value of lifespan evaluation equation,and then affect the combination of optimal probability distribution model and empirical failure formula,and finally bring a large error to the lifespan evaluation result value.Therefore,the validity and necessity of statistical test are verified,and the lifespan evaluation model of PI films modified by nano TiO₂ is revise.（3）The grey neural network model is built by mapping the grey time response function into the neural network.According to the characteristics of lifespan evaluation of PI films modified by nano TiO₂,the process and algorithm of lifespan evaluation base on the expanded sample capacity of gray neural network are established for the purpose,and the influence of different probability distribution models,empirical failure probability formula and sample size of failure data on the lifespan evaluation of PI films modified by nano TiO₂ is revealed.The results show that when the failure data is of small capacity.The inverse power function lifetime evaluation model,Weibull distribution and mathematical expectation formula should be selected to evaluate the electric lifespan of PI films modified by nano TiO₂.The error of this scheme is minimum in small samples.When the failure data is large,the inverse power function lifespan evaluation model,lognormal distribution and median formula are the best combination of lifespan evaluation schemes.