关键词: 微弱故障检测, 旋转机械, 辛主成分分析, 超圆盘模型, 单分类超圆盘

Abstract: Aiming at the problems that early weak faults of rotating machinery were difficult to detect in time and accurately, an intelligent detection method was proposed based on SPCA and OCHD. Firstly, SPCA was used to map vibration signals to a symplectic space, and the symplectic eigenvalues which might best characterize the main energy and effective information of the signals were extracted as the fault features of rotating machinery. Then, the hyperdisk model was introduced into the one-class classification domain to propose the OCHD model. OCHD used the hyperdisk model to evaluate the class distribution of known samples, and the optimal one-class hyperplane was constructed by finding the closest points on the geometric model to the origin, so as to realize the intelligent detection of early weak faults. Finally, the effectiveness of the proposed method was verified by the bearing life cycle data from the university of Cincinnati. The experimental results show that SPCA may effectively extract the sensitive fault information of bearings, and the fault detection performance of OCHD is significantly better than that of other one-class models.

Key words: weak fault detection, rotating machinery, symplectic principal component analysis(SPCA), hyperdisk model, one-class hyperdisk(OCHD)