Application of Probabilistic Causal-effect Model based Artificial Fish-Swarm Algorithm for Fault Diagnosis in Mine Hoist

摘要

This paper presents an intelligent methodologyfor diagnosing incipient faults in mine hoist. As ProbabilisticCausal-effect Model-Based diagnosis is an active branch ofArtificial Intelligent, in this paper, the feasibility of usingprobabilistic causal-effect model is studied and it is appliedin artificial fish-swarm algorithm (AFSA) to classify thefaults of mine hoist. In probabilistic causal-effect model, weemployed probability function to nonlinearly map the datainto a feature space, and with it, fault diagnosis is simplifiedinto optimization problem from the original complex featureset. And an improved distance evaluation technique isproposed to identify different abnormal cases. The proposedapproach is applied to fault diagnosis of friction hoist withmany steel ropes, and testing results show that the proposedapproach can reliably recognise different fault categories.Moreover, the effectiveness of the method of mappinghitting sets problem to 0/1 integer programming problem isalso demonstrated by the testing results. It can get 95% to100% minimal diagnosis with cardinal number of faultsymptom sets greater than 20.