Statistical Analysis Of Blocked Data Of Three Unit Hybrid Systems

In the process of life test, because of the lack of effective tools, data record error, insufficient funds, and other factors, We are often not able to obtain the exact cause of system failure, we called the data which we got the masked data. The masked data can help us save time and costs, but also get better estimate of the unknown parameters in the case of data records error. So research about the masked data and statistical analysis has been made good achievements. People usually research about units in series, parallel or mixed systems, whose life obey Exponential distribution, Weibull distribution or other distribution, they provide the Bayes estimation, Maximum likelihood estimation and approximate interval estimation, etc.This paper study about the mixed systems after masked data appeared. Respectively give out the maximum likelihood estimation of the parameters of mixed system life distribution under two different cases of the parameters of mixed system life distribution.The first chapter is introduction, we mainly describe the research background of this article, and then made some basic assumptions about the following model.In the second chapter, we study about the three-unit series-parallel system when masking occurs. In view of the two different cases of the parameters of mixed system life distribution, we respectively get the maximum likelihood function under the cases of complete sample and censored samples, and then get the proof the uniqueness of solution of the Maximum likelihood equation under the case of the four parameters are equal. Finally, some examples are given by using Matlab software to generate simulated data and get the estimation of the parameters of Maximum likelihood, and then get simulation results by1000or500times of simulation.In the third chapter, we study about the three-unit parallel-series system when masking occurs. In view of the two different cases of the parameters of mixed system life distribution, we respectively get the maximum likelihood function under the cases of complete sample and censored samples, and then get the proof the uniqueness of solution of the Maximum likelihood equation under the case of the four parameters are equal by drawing. Finally, some examples are given by using Matlab software to generate simulated data and get the estimation of the parameters of Maximum likelihood, and then get simulation results by1000or500times of simulation.