A Type-3 Fuzzy Actuator-Fault-Tolerant Control Strategy with Lyapunov Predictive Learning

This paper introduces an active fault-tolerant control (AFTC) strategy for detecting and eliminating the actuator faults in nonlinear control systems. The developed scheme consists of two subsystems: a main control system and a prallel dynamic virtual system. The main subsystem uses a Type-3 (T3) fuzzy logic system (FLS) for the online identification of system uncertainties, a T3-FLS-based predictive controller, and a supplementary compensator. Simultaneously, the virtual system uses a similar parallel configuration, including its own T3-FLS, predictive control structure, supplementary compensator, and a fault detecting mechanism. The actuator fault detection block consists of two T3FLSs. In fact, in the virtual structure, assuming a virtual sensor and considering a virtual fault for this sensor, the value of the actuator control signal will be estimated. The method relies just on measurable input–output data, and eliminates the need for explicit pre-defined dynamics. The identification, fault detection and compensation are applied in a real-time scheme by adaptively updating the T3-FLSs based on feedbacks obtained from the parallel virtual structure. The continuous and accurate fault detection is ensured by the suggested mechanism. The stability is analytically proven using Lyapunov-based methods. The suggested approach is applied on two different non-identical applications: a blood-sugar regulation in diabetes, and the inflation control in a dynamic economic model. The simulations show the effectiveness of the developed AFTC.