Research And Implementation On Key Technology Of Life Prognosis For Brake Pads In EMUs

With the rapid development of modern industrial technologies, the life prognosis of has increasingly prominent role in the production and operation of enterprises. The braking system is a key system of the Electric Multiple units (EMUs), its reliability, stability and economy is the focus of attention of braking system manufacturer, repair factory and other related departments. Brake pads is the key component of the braking system, when braking, the friction process with high temperature environment and large external force will cause some wear on it. When the wear quantity reaches a certain level, we should replace it to ensure the normal operation of the braking system. Therefore, based on the life prognosis of brake pads to make maintenance and replacement policy plays an important role in ensuring the safety of the EMUs and improving utilization rate of materials.This paper takes brake pads of braking system in EMUs as the research object, using machine learning algorithms, taking comprehensive consideration of the factors related to the friction process, based on the real-time status data of EMUs collected by the EMUs train-ground communication system, studied the wear estimation technique which is the key technology of life prognosis and established a wear quantity estimation model for brake pads in EMUs. The following parts can be focused.(1) Based on the analysis of the key technology of life prognosis, the research of wear quantity estimation model for brake pads is taken as the core content of this paper. Considering the specific Situation of friction process of brake pads, the scheme of establishing the wear quantity estimation model is determined.(2) Based on the existing problems in EMUs real-time running data collection, the data collection system, EMUs train-ground communication system is introduced. This part focus on the data collected process, the selection of wireless network for data transmission and the process of receiving and handling these data.(3) This paper studied the BP neural network model in machine learning and proposed some improving methods for existing defects in BP algorithm. Based on EMUs real-time status data collected by the EMUs train-ground communication system, this study trained the BP neural network using the improved BP algorithm and established a wear estimation model of brake pads in EMUs. And the implementation of taking the wear estimation model as the software component of life prognosis is introduced.The experiment results show that for the data collected in this study, the improved training algorithm achieves much better performance than the traditional training algorithm.