关键词: MPdist, 长短时记忆网络, 相似性度量, 健康指标构建, 剩余寿命预测

Abstract: In view of the low efficiency and manual detection existing in the health management process of industrial robot axis, a health index construction method based on action cycle degradation similarity measurement under the background of mechanical axis operation monitoring big data was proposed, and the remaining useful life prediction was carried out by combining Long Short-Term Memory (LSTM) network. Firstly, MPdist was used to focus on the similarity of sub cycle sequences between different action cycles of mechanical axis, and the comparison distance between normal cycle data and degenerate cycle data was calculated, and then health index was constructed. Secondly, the long short-term memory network model was trained by health index set, and the mapping relationship between health index and remaining useful life was established. Finally, MPdist-LSTM model was used to automatically calculate the remaining useful life and give early warning in time. The six-axis industrial robot was used in the experimental platform, and about 15 million pieces of data were collected. The monotonicity, robustness and trend of Health Index (HI) and the average absolute error, root mean square error, coefficient of determination, error interval, early prediction and late prediction of Remaining Useful Life (RUL) were tested. The experimental results were compared with those of Dynamic Time Warping (DTW), Euclidean Distance (ED) and Time Domain Eigenvalue (TDE), MPdist combined with RNN and LSTM. The experimental results show that: compared with other methods, the monotonicity and trend of the proposed method are 0.07 and 0.13 higher than other methods, and the RUL prediction accuracy is higher, and the error interval is smaller, which proves the effectiveness of the proposed method.

Key words: MPdist, long short-term memory, similarity measure, health index construction, remaining useful life