Distributed optimization filtering for fuzzy quantization system with membership functions online learning

Abstract This paper investigates the distributed optimization filtering problem for nonlinear systems under resource constraints in wireless sensor networks based on T‐S fuzzy model. First, Sigma‐Delta dynamic quantizers are introduced to dual‐quantize the state estimation and measurement signals, decreasing bandwidth consumption. Second, a novel distributed fuzzy estimator with heterogeneous double‐subscript gains is constructed using a non‐PDC method to solve the problems of immeasurable premise variables and fuzzy asynchronous signals in fuzzy models. Distributed robust estimation method is applied to manage immeasurable premise variables, analyzing the mean‐square exponential stability and performance of the estimation error system. Third, the heterogeneous membership functions are optimally selected online using the gradient descent method within the feasible domain to reduce estimation error and improve interference attenuation performance. Finally, a tunnel diode circuit system and a numerical example are utilized to verify the effectiveness and superiority of the proposed method.