Support vector machine-based optimisation of traction gear modifications for multiple-condition electric multiple unit

Abstract

The traction gear train is subjected to various internal excitations under different traction conditions, such as stiffness excitation, error excitation, meshing shock excitation, and tooth side clearance. The optimal modification scheme is different for each state, and the optimal modification plan based on a single working condition may not be applicable to every working condition. In this paper, we investigate the vibration response characteristics of electric multiple unit gearing under multiple conditions and propose a multi-condition modification scheme. Under different traction conditions, the mapping between gear modification parameters and vibration acceleration in gear transmissions is investigated using support vector machines. Genetic algorithms are used to solve the gear-modifying parameters to minimise the maximum vibration acceleration. A weight assignment principle is proposed to calculate the electric multiple unit traction gear transmission under different conditions, with the operating time and the amount of vibration in each condition as the measurement index. The results of the simulation show that the vibration acceleration under continuous conditions is reduced by 3.55 m/s2, a decrease of 69.88%; the vibration acceleration under high-speed conditions is reduced by 3.301 m/s2, a reduction of 58.74%, and the results show that the overall index of the electric multiple unit traction gear transmission system under different conditions has been improved after the optimisation of the multi-conditions modification, the gear transmission system's vibration is significantly reduced.