Extracting relevant features for diagnosing machine tool faults in cloud architecture

This paper presents a cloud diagnosis architecture to support diagnosis of different machine tool faults with similar abnormal events. Lacking the corresponding features of failure historical data, similar abnormal events are insufficient to be used for identifying the root causes of faults. On the basis of a novel event-oriented process monitoring and backtracking (EOPMB) method and the clustering non-dominated sorting genetic algorithm (CNSGA), this paper proposes a cloud diagnosis architecture for identifying failure causes by extracting relevant features of various faults from different machine tools. Results show that the proposed architecture can assist users in improving diagnosis performance.