Data-Driven Sliding Mode Optimal Control for Multi-Power Unit High-Speed Train: A Composite Approach

In this study, a multiple-input multiple-output (MIMO) data-driven sliding mode optimal control scheme is investigated for a multi-power unit high-speed train (HST) automatic driving system under disturbances. First, an integral terminal sliding mode control (ITSMC) law based on the dynamic linearization method is introduced to achieve error convergence in finite time. Subsequently, a parameter update law and an adaptive extended state observer (AESO) are designed to estimate the control gain and total uncertainty, respectively, which solves the problem of traditional sliding mode control requiring a large switching gain to handle the disturbance. Third, an optimal control signal is obtained by predictive control to achieve a higher level of tracking error accuracy and quickly reach the quasisliding mode state, and the compound optimal control scheme is derived under the combined action of the ITSMC and predictive control. The scheme considers and compensates for the total uncertainty caused by the error feedback, parameter estimation error, and unknown nonlinearity. The main advantages of this scheme include the following: controller design process uses only system data, provides excellent model adaptability, and disturbance rejection ability. Upon providing a stability-proof analysis of the proposed method, the proposed composite control method was compared and tested on a CRH380A train simulation test bench equipped in the laboratory. The simulation results show that the speed tracking errors of each power unit of the HST under the proposed control scheme are within [−0.123 km·h
⁻¹
, 0.144 km·h
⁻¹
]. The control forces and accelerations are within [−55 kN, 46 kN] and [−0.592 m·s −2, 0.521 m·s
⁻²
] respectively, which meets the requirements of safe, stable, and efficient operation of the train.