Reliability Analysis on the Injection System by Mapping T-S Fault Trees into Bayesian Networks

摘要

A novel reliability analysis technique based on Bayesian network and T-S FTA is proposed in this paper. In the proposed technique, the nodes in Bayesian network can be expressed in terms of fuzzy possibilities and the magnitudes of the failure in the system are expressed in term of fuzzy variables. The paper shows that any T-S fault tree can be directly mapped into a BN and that forward inference techniques on the latter may be used to obtain classic parameters computed from the former (i.e. reliability of Top Event or any sub-system, the importance of Basic Events, etc). Furthermore, by using BN, a general diagnostic analysis (backward inference) can be performed, in which posterior probability is computed. The proposed technique is applied to analyze the reliability of the injection system of chemotherapy pharmacy robot.