Rolling bearing fault diagnosis based on wireless sensor network data fusion

Abstract

With the continuous development of intelligent manufacturing, mechanical equipment is developing in the direction of large-scale, integration, precision and intelligence. The coupling between different equipment in the operation of the system also makes the mechanical equipment increasingly become a whole, which puts forward higher requirements for the fault diagnosis of mechanical equipment, especially rolling bearings. With the rapid development of wireless network technology, rolling bearing fault diagnosis has become a reality. In this paper, the wireless sensor network will be innovatively used to collect the data of key parts of mechanical equipment, so as to improve the problem of insufficient and accurate collected data in the traditional convolution neural network fault detection algorithm, convert the corresponding time-domain signal into image signal, and combine the global average pool layer on the basis of the combination of nonlinear convolution layers, So as to optimize the corresponding wireless network structure, and finally realize the adaptive extraction and fault diagnosis of rolling bearing fault features. The experimental results show that the fault detection accuracy of the optimized convolution neural network algorithm is about 99.8%, and has good robustness and practical value.