Variable structure T–S fuzzy model and its application in maneuvering target tracking

To realize the adaptive identification of T–S fuzzy model structure, we propose a variable structure T–S fuzzy model algorithm. Compare to traditional multi-input single-output in the T–S fuzzy model, we extend single-output fuzzy rules to multi-dimensional output fuzzy rules, which has the advantage that all multi-dimensional outputs share the same premise parameter; Then the joint block structure sparse ridge regression model is used to realize the identification of the consequent parameter, which provides a regression model. In this model, some regression coefficient blocks with small contribution will be reduced to zero accurately, while maintaining high prediction accuracy. Otherwise, the Fuzzy Expectation Maximization (FEM) is proposed to coarse fine the premise parameter. Finally, the variable structure T–S fuzzy model is applied to the maneuvering target tracking without filter. The simulation results show that the proposed algorithm is more accurate and stable than the Interacting Multiple Model (IMM), Interacting Multiple Model Unscented Kalman Filtering (IMMUKF), Interacting Multiple Model Rao-Blackwellized Particle Filtering (IMMRBPF) and T–S Fuzzy semantic Model (TS-FM) algorithms in dealing with uncertain problems in nonlinear maneuvering target tracking systems.