Feasibility study on wheelset fatigue damage with NOFRFs-KL divergence detection method in SIMO

Abstract Train wheelset is an important part of the train. After long-time and high load working, fatigue damage will appear. Then the micro-cracks will form and expand into macro cracks, which may cause a serious accident. Therefore, it is necessary to study the fatigue damage detection methods for train wheelset. Fatigue damage will cause nonlinear effects. The Nonlinear Output Frequency Response Functions (NOFRFs) is an effective tool to characterize the nonlinearity of the system. In this paper, based on the concept of Kull Back-Leibler (KL) divergence and NOFRFs, a detection method denoted as NOFRFs-KL (NKL) is proposed to study the feasibility of comprehensively evaluate the fatigue damage of wheelset. The train wheelset is considered as a single input multiple output (SIMO) nonlinear system while hammering on one point in the hammering test. And the estimated NOFRFs of multiple positions in the SIMO system are used to evaluate the fatigue damage at each output position. To verify the effectiveness of the NKL, some hammering tests are carried out. The NOFRFs of the wheelsets with micro-cracks, repaired wheelsets are calculated. The NOFRFs of the wheelsets with different served time are calculated. And then the corresponding NKL values are obtained. The results show that the fatigue damage in the wheelsets can be identified by the NKL index, and the NKL index can be proportional to the service time of the wheelset.