Quantile Regression Analysis Of Lifetime Data On Fractional Factorial Design

In the Fractional Factorial Design,the traditional maximum likelihood method used to analyze the lifetime quantile may have the following problems.First,the method assumes that the data follows some known distribution,but the actual data distribution type is unknown.If different data distribution assumptions are used for maximum likelihood estimation,it may lead to different analysis results and increase the difficulty for engineers to make correct decisions.Secondly,the accuracy of maximum likelihood estimation analysis may be affected when there are heteroscedasticity and censoring of data.Therefore,other methods are needed to analyze the quantiles of life data more accurately.Based on the above possible problems,this paper uses the quantile regression method to model and analyze the quantile of life data,providing a more accurate and reliable modeling method for the data analysis of life quantile.The specific research contents are as follows:Firstly,the quantile regression of the normal Fractional Factorial Design D=ABC is given,and the unbiased and effective estimation of the model coefficient is proved by numerical simulation,and the model is applied to the life analysis of the point-to-point welding test.Secondly,quantile regression model is carried out for Plackett-Burman design in informal Fractional Factorial Design analysis.The prediction performance comparison between maximum likelihood estimation and quantile regression method is given,and the influence factors of diamond life quantile are analyzed by using this method.Thirdly,construct the censored quantile regression model based on Plackett-Burman design,compare the prediction performance of maximum likelihood estimation and censored quantile regression methods under different censoring percentages,and finally analyze the thermostat case.