Adaptive Minimum-Entropy Hybrid Compensation for Compound Faults of Non-Gaussian Stochastic Systems

This study investigated minimum-entropy hybrid fault-tolerant control (FTC) theory for non-Gaussian stochastic systems with compound faults. After fuzzy linearization for the singular systems, the output probability density function (PDF) is generated by rational square root B-splines. To deal with the compound faults consisting of single sensor fault and intermittent multiple actuator faults, an active-passive hybrid adaptive FTC scheme is proposed: A passive compensation function can directly reconstruct the algorithm to mask the sensor fault; then, actuator fault estimation accurately tracks the multiple actuator faults. Hence, the hybrid FTC combines estimated information and passive compensation simultaneously implements active actuator fault repair and passive sensor fault shielding. A novel variable parameter algorithm that mimics animal predation behavior is designed and incorporated into learning rates, making the controller more sensitive to the incipient deviations in actuator faults. Finally, with the optimal indicators containing entropy and mean of non-Gaussian PDF, the minimum-entropy FTC is achieved. Lyapunov and indicator functions prove the stability, simulation verifies the effectiveness of the methods.