Q-learning-based adaptive resilient event-triggered fault detection for fuzzy stochastic systems: A novel asynchronous switching filter strategy

This paper develops a Q-learning-based adaptive resilient event-triggered fault detection framework for interval type-2 Takagi–Sugeno fuzzy semi-Markov jump systems by introducing a novel asynchronous switching filter under D -stability constraints. In order to deal with denial-of-service attacks, a novel multi-weighted parameter adaptive Q -learning strategy is proposed, which enables the selection of distinct modes in response to different patterns of data variations. This strategy alleviates channel congestion and enhances the transmission rate of critical data, thereby improving the fault detection performance. Furthermore, a co-design framework is established, in which the event-triggering scheme, plant, and filtering process are jointly modeled to ensure efficient and timely fault detection. Nevertheless, owing to external disturbances and related factors, asynchronous phenomena frequently arise. To capture this behavior, two hidden semi-Markov models are employed. The proposed approach employs multiple performance indices, including L 2 - L ∞ , passivity, H ∞ , and dissipativity. The sufficient conditions are proposed to guarantee exponential mean-square stability and determine the filter gains. Finally, the effectiveness and advantages of the proposed approach are validated through a tunnel diode circuit example.