ارزیابی ریسک حوادث فشار برق باز مبتنی بر شبکه های بیزی

Since multiple state events could not be described and accident probability could not be calculated by fault tree quantitatively for safety assessment, applications of Bayesian network (BN) in safety and reliability fields for the past few years were introduced and basic principle and inference algorithm of Bayesian network were presented. A Bayesian network was developed to model open press electric shock accident as a result of charged press enclosure. The multi-state nodes of the network were illustrated and accident probability was computed finally. The results show that Residual Current Operated Circuit-Breaker with Integral Overcurrent Protection (RCBO) installation can reduce accidents probability of phase-voltage and under-phase-voltage electric shock sharply for open press working in T - T system of low voltage power supply system. In a word, Bayesian network can model system which contains multiple state events, furthermore, accidents probability of that system can be computed correctly, in addition, it offers a comparative tool for various safety design of machine system to guarantee machine inherent safety.