关键词: RV减速器, 动力学模型, 数值模拟, 人工智能, 故障诊断

Abstract: The conditional monitoring of rotary vector (RV) reducers using machine learning has great significance to guarantee the reliability and security of RV reducer. However, in engineering applications, fault samples missing is rationale for restricting the development of artificial intelligence diagnosis. Because the problem of the insufficient fault sample of the RV reducer that have to be resolved, an intelligent diagnosis methods using dynamic model is proposed to detect faults in these reducers. In the first step, dynamic model of the healthy RV reducer is constructed, and pearson correlation coefficient (PCC)-based model updating technique is employed to obtain dynamic model with a certain precision. In the second step, the mathematical model of gear fault is established and inserted into the dynamic model to calculate the fault samples, which served as training samples of artificial intelligence diagnostic models to classify testing samples (fault to be diagnosed). Finally, the simulated fault samples are used as training samples, and the unknown fault samples obtained from test are utilized as test sample, fault classification using convolutional neural network (CNN) are obtained, which show that dynamic model driving fault diagnosis method for RV reducer can be guided to solve fault samples missing problem.

Key words: RV reducer, dynamic model, numerical simulation, artificial intelligent, fault diagnosis