A novel fault prevention model for metro overhead contact system

Abstract

As an indispensable part of metro traction power supply system, the metro overhead contact system (OCS) is complex in composition and has various causes of faults, so its fault prevention is too complicated to be handled by using traditional methods. However, over the last few years, the fault data of metro OCS has been through explosive growth. Consequently, the method of high utility mining, which is capable of pointedly mining fault association rules from historical fault data, is adapted to address the issue. According to the fault association rules, once the precursor of the specified fault occurs, we can immediately conduct preventive maintenance to achieve the purpose of fault prevention. In this paper, we proposed a novel fault prevention model for metro OCS. First, we proposed a concept hierarchical database structure with a detailed encoding method, which can efficiently store the massive fault data and facilitate the subsequent mining. Then an improved algorithm for mining multiple-level top-K high utility fault-sets (MTHU) for metro OCS is proposed. At last, a case study is conducted and the mining result is analyzed to reveal the inner connections between faults. Based on the analysis result, the detailed preventive maintenance suggestions are given to effectively achieve fault prevention for OCS.