Freight train speed active disturbance rejection tracking and wheel anti-slip based on fuzzy neural network with DBO optimization

In this paper, we present a proposed scheme for speed tracking control specifically designed for freight trains. This innovative speed tracking approach effectively prevents wheel slippage and ensures optimal speed control of the traction motor. The strategy integrates the use of Direct Torque Control (DTC), a technique employed by the HXD1 electric traction locomotive to regulate the asynchronous motor. To achieve velocity tracking control, we implement a Predictive Auto Disturbance Rejection Control (PADRC) system. Notably, the PADRC system includes an output prediction module estimator that enables accurate forecasting of time-delay system responses. Additionally, we develop an Unscented Kalman Filter (UKF) observer and seamlessly integrate an adaptive parameter adjustment mechanism, powered by Dung Beetle Optimizer-Fuzzy Neural Network (DBO-FNN), into the observer architecture. By utilizing the anti-slip parameters obtained through this observer, we determine the control scheme for anti-skid control. We validate the efficacy of this scheme through simulations using an actual speed curve of a freight train under both wet and dry pavement conditions. The results show a remarkable improvement in speed tracking accuracy, with respective increases of 66.45% and 56.29% over Adaptive Model Predictive Control (AMPC) and Non-Linear Auto Disturbance Rejection Control (NLADRC), in terms of speed tracking. Moreover, the tracking stationarity witnesses notable enhancements, increasing by 42.07% and 60.87% respectively. Additionally, the anti-slip performance of trains running on dry and wet tracks increased by 12.45% and 26.56%, respectively.