Affiliation:
1. Laboratory of Numerical Control of Industrial Processes, National School of Engineers of Gabes, Gabes University, Gabes, Tunisia
Abstract
This paper introduces a new method for detecting switching instants and designing data-driven FTC for stochastic switched systems susceptible to sensor faults. The approach uses clustering algorithms and classification techniques to establish sub-models based on input-output databases. A data-driven approach is used to estimate the sensor fault, and controllers are generated for each mode to counteract the effects of the fault and minimise noise. The controller gains are determined using a novel LMIs derived from the general Lyapunov function. This approach’s effectiveness is proven through numerical analysis, which features two simulation examples of stochastic switched systems. The first example demonstrates a faulty stochastic switched system with two modes, while the second example depicts a practical application of vehicle rollover prevention systems.