Reliability analysis of hydraulic system in a NC machining center based on Bayes network

Abstract

Due to the complex state of hydraulic system of modern numerical control equipment, the traditional dynamic fault tree quantitative analysis is solved by Markov model and state transition equation. However, when the data is too much and complex, the problem of combinatorial explosion will occur in the solution of Markov model, which increases the difficulty of solving. Bayesian network technology inherits the state description and reasoning method of fault tree, and also has the ability to describe the polymorphism of events and the logical relationship of faults, which effectively reduces the problem of space explosion. It is very suitable for the reliability analysis of the system with many states and fast solving speed. In this paper, the discrete-time Bayesian network is used to analyze the reliability of the hydraulic system of numerical control equipment. Firstly, the logical relationship of the system fault is analyzed to establish a dynamic fault tree. Then, it is transformed into a Bayesian network model and a conditional probability table. The time is divided into three sections for reliability analysis, reasoning and calculation, and the results are compared with the Markov method. The efficiency of discretetime Bayesian network is verified, which provides help for fault diagnosis.