Robust control approach for virtually coupled train set with uncertain communication

Virtually coupled train sets have been proposed to improve transportation efficiency and the flexibility of train formations. More importantly, virtual coupling enables trains to exhibit greater robustness when faced with uncertainties such as disturbances caused by upstream disturbances. However, trains within a set may lose communication capabilities due to packet loss or poor wireless environments. Communication failures between trains may lead to information loss and alterations in the internal communication topology of the virtual formation, thereby impacting its stability. This paper presents a new controller designed for multi-predecessor-following formations to ensure stable train control during communication failures. The proposed controller includes a train status prediction module and preceding train status information weight regulators. For the prediction module, we propose a dynamic high-accuracy low-rank tensor completion (D-HaLRTC) method to predict the train status information, which is missing due to uncertain communication. In the information weight regulators, a weight factor is incorporated as a simple and effective way to achieve reasonable weight assignments to the information acquired from multiple trains. The string stability conditions of this new controller under various communication delay and communication failure scenarios are determined by performing stability analysis in the frequency domain. Simulation results show that the two proposed modules both have beneficial impacts on string stability. Finally, we compare the proposed controller with other baseline controllers under various communication failure scenarios, and the results show that the proposed controller can handle communication-failure-induced disturbances more smoothly.