Multisensor-Driven Cross-Domain Motor Fault Diagnosis Based on Multibasis Energy Pattern

Motor fault diagnosis methods based on visual image features have made certain achievements. However, there are still some limitations. First, image features have poor adaptability to changes in operating conditions. Second, multisensor information is seldom considered, resulting in some industrial data not being fully utilized. Thus, a new multisensor-driven cross-domain motor fault diagnosis method is proposed in this article to solve the above problems. To be specific, a multibasis energy pattern (MBEP) method is proposed to reduce the impact of changes in operating conditions on image features. Then, to fully utilize the information of multiple sensors, a fault diagnosis framework based on manifold-embedded distribution alignment-evidence theory (MEDA-ET) is proposed. The validity of the proposed method is evidenced under multiple and time-varying operating conditions, with the average accuracy in fault diagnosis reaching 99.21% and 96.27%, respectively.