Bayesian network approach based on fault isolation for power system fault diagnosis

This paper constitutes a component-oriented Bayesian model for power system fault diagnosis. The model identifies the fault by comparing whether the action of protection and breakers are accordant with the normal fault handling mode. Making use of time information to match the protections and breakers, the breakers can be classified into three layers according to the protection type. The serial characteristics of protection can revise the aberrant information and the type of breaker layers can rule out the components which are weak associated with the fault in power cut area. This paper identifies the short circuit fault from two respects - breaker isolation and protection isolation. The new Bayesian network structure reflects the breaker and protection operation mode more directly. The new Bayesian model keeps the error tolerance of the traditional one and augments the universality.