AEセンシングと機械学習による通電摩耗のパターン解析

抄録

In recent years, EVs have been used in a wide variety of machines, from transmitting power to sending and receiving information. Mechanical elements that used to consist of shafts and wires have been replaced by electronic controls, allowing for miniaturization. However, with the shift to EVs, the effect of wear under electric current has become an issue. Unlike normal wear, wear under electric current causes arc discharge and melting due to the application of electricity, resulting in extensive damage. This affects signal transmission and reception when it occurs at terminals. In addition, natural phenomena such as rain must be considered for railway pantographs, and it is necessary to anticipate natural phenomena such as rain, and it is necessary to understand the mechanisms under various conditions for safe and stable operation. Therefore, material selection is important and evaluation methods are necessary. In this study, it was investigated whether arc discharge and melting could be evaluated by acoustic emission (AE) sensing and whether external factors could be sorted out by machine learning were. The results confirmed that there was a remarkable difference in the frequency range where melting and arc discharge occurred, and that machine learning could classify wear patterns under the three conditions of no electric current, with electric current, and with electric current and water drop.