Affiliations: [a] School of Mechanical Engineering, Xi’an Shiyou University, Xi’an, Shaanxi, China | [b] Chengdu North Petroleum Exploration and Development Technology Company Limited, China ZhenHua Oil Co., Ltd, Chengdu, Sichuan, China
Correspondence:
[*]
Corresponding author: Tao Ren, School of Mechanical Engineering, Xi’an Shiyou University, Xi’an, Shaanxi, China. E-mail: [email protected].
Abstract: Gear’s vibration signal contains the state information of the gear. Different types of gear faults have different vibration features in the time domain. Fault feature extracted from the vibration feature can be used to diagnose the gear fault. In this paper, the vibration feature of the gear signal in the time domain is analyzed, and the fault feature of the gear is extracted by using the kernel density estimation along with the probability statistical method. Then the vibration signal of the gear is processed and the probability density estimation of the amplitude is obtained through the function of kernel density estimation. Then the probability of the sample point falling to each vibration range is calculated. Finally, the fault diagnosis is achieved by identifying the fault feature based on the fault statistics in various amplitude ranges. According to the experimental results, it is shown that the fault feature extracted in this paper can be applied to the fault diagnosis of a gear.
Keywords: Kernel density estimation, gear, extract feature, fault diagnosis
