Hybrid propagation modeling based clutch fault diagnosis of multi-mode electromechanical transmission system using particle filter

In order to reveal the failure mechanism of the clutch at the system level and realize online fault diagnosis and localization of the hybrid electric vehicle equipped with a specific multi-mode electromechanical transmission system (MM-EMTS), with particular emphasis on the behavior of clutch slippage failure. The fault propagation mechanism is analyzed at both the component and system levels, revealing that fault propagation is jointly driven by changes in continuous state variables and discrete system-constrained states. Based on the hybrid propagation characteristics of clutch faults, a systematic fault diagnosis strategy is constructed. This strategy establishes a fault hybrid propagation model based on the stochastic hybrid automaton (SHA), built upon a unified health and fault mode of the system. Furthermore, a hybrid estimation algorithm based on the particle filter is employed to estimate the discrete modes of the system, enabling the diagnosis and localization of the faulty clutch. The accuracy of fault isolation is validated through the estimated continuous state variables provided by the algorithm. Research results indicate that with 80 particles and resampling, the algorithm can accurately locate the faulty clutch within 0.02 s after failure. The mean absolute errors for the continuous state variables are 0.2756, 0.7193, and 0.5182, while the root mean square errors are 0.6282, 2.2705, and 1.1389.