A planetary gear reducer backlash identification based on servo motor current signal and optimized fisher discriminant analysis. (June 2021)

• Record Type: Journal Article
• Authors: Yang, Qichao; Liu, Tao; Wu, Xing; Deng, Yunnan; Chen, Qing
• Is Part Of: ISA transactions. Volume 112(2021)
• Page Start: 350
• Page End: 362
• Publication Date: 2021-06
• Subjects: Motor current signal analysis -- Gear backlash detection -- Fisher discriminant analysis -- Sensitivity weight ratio -- Feature extraction; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2020.12.016
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: Planetary gear reducer is widely used in industrial automation, and its performance highly affects the equipment reliability. The backlash and stiffness may cause the performance decline of planetary, hence the vibration, temperature, current and other signals are applied in planetary condition monitoring. The purpose of this paper is to develop a practical and effective method based on motor current signal analysis (MCSA) to identify backlash faults of planetary gear reducers. The sensitivity weight ratio (SWR) is proposed to optimize the introduced fisher discriminant analysis (FDA) algorithm, which is used to extract and screen the current signal characteristics of the servo motor. The motor is connected to the reducer, so the changes in the operating conditions of the planetary gears can be observed in the motor current. Compared with the traditional detection method of equipment health status, the Hall current sensor is a non-invasive method with lower cost and easy installation. Besides, the support vector machine (SVM) classifier and some published methods are utilized to classify the backlash of the planetary gear. Finally, experimental tests were carried out under different backlashes and loads to verify the effectiveness of the method. Highlights: A planetary gear backlash recognition method based on SWR optimized FDA and SVM. Based on MCSA, extract multiple features to construct a feature library. Use the SWR optimized FDA algorithm to select the features. Test to verify the effect of the proposed method on fault identification.