Optimal Speed Tracking of Freight Trains Combined with Segmented Soft-Switching Control

Smoothness of travel speed and stopping accuracy are important for freight trains. However, due to the large mass of freight trains, their driving speed is easy to jitter at the operating condition switching point. For these purposes, this paper designs a Dual Mode Optimal Control (DMOC) for tracking the target travel speed of freight trains. This controller contains two sub-controllers, Adaptive Model Predictive Control (AMPC) and Preview control (PC). An Elman Neural Network (ENN) is incorporated in AMPC to adjust the control weights of MPC in real time to output the optimal driving speed. The Affinity propagation-Fast-minimum covariance determinant algorithm, combined in ENN identifies the noisy samples in the training samples and improves the fitting effect of the network. PC and AMPC are fused together by a soft-switching control method. The soft switching control method based on Tanh function can achieve smooth switching of controllers and obtain a good control effect. By comparing with active disturbance rejection control and fuzzy proportional-integral-derivative under two speed profiles, DMOC can effectively reduce the speed jitter of speed tracking, improve the stopping accuracy and timeliness of freight trains, and reduce energy consumption.