Data Analysis on Incomplete Failure of Ship Electromechanical System Based on Bayesian Method

摘要

As a result of the influence of man-made or other factors, the recorded fault data in the process of keeping maintenance of ship electromechanical system are often inaccurate with incomplete record information, low-quality data and small size sample. For the incomplete failure data, this paper will analyze the failure data of the certain type of ship electromechanical system thoroughly and comprehensively through Bayesian method. Sorting the data firstly and then analyzing graphical statutory representation to determine the failure trend, I will adopt the homogeneous Poisson process and the non-homogeneous Poisson process (power law process (PLP) and Duane) on quantitative analysis to establish failure data analysis model with Bayesian method and obtain the analysis result through PyMC3 probability programming. Finally, the Bayesian estimation of the average failure frequency of ship electromechanical system will be concluded in this paper.