Signed-data reinforced observer-based fault diagnosis for virtually-coupled electric multiple units trains

Virtual coupling has been recognized as a promising yet challenging technology for the next generation of train control systems. As a mean of guaranteeing the safety of closer-running of virtually-coupled electric multiple units (EMU) trains, fault detection and diagnosis play a critical role in perceiving abnormal running conditions. This paper develops a new signed-data reinforced observer-based technique for dealing with actuator faults’ detection and diagnosis for virtually-coupled EMU trains. The introduction of novel signed-data reinforcement technique ensures that unbiased fault diagnosis can be guaranteed with any regressor vector composed of control signals on EMU trains. Fault detection and diagnosis observers, together with fault alarming principle, information transmission coding strategy are elaborated using Lyapunov stability theorem. Simulation results are given to demonstrate the effectiveness of proposed algorithm.