GO-Bayes Method for System Modeling and Safety Analysis

Abstract

Safety analysis ensuring the normal operation of an engineering system is important. The existing safety analysis methods are limited to relatively simple fact description and statistical induction level. Besides, many of them enjoy poor generality, and fail to achieve comprehensive safety evaluation given a system structure and collected information. This work describes a new safety analysis method, called a GO-Bayes algorithm. It combines structural modeling of the GO method and probabilistic reasoning of the Bayes method. It can be widely used in system analysis. The work takes a metro vehicle braking system as an example to verify its usefulness and accuracy. Visual implementation by Extendsim software shows its feasibility and advantages in comparison with the Fault Tree Analysis (FTA) method.