A Neurofuzzy Function Approximator Approach to Nonlinear Systems State-Estimation

A novel fuzzy inference system is introduced with desirable approximation properties for highly nonlinear systems that can be expressed in linear parameter varying form. This fuzzy inference system uses a hashing function to eliminate unnecessary computations and is compared with existing fuzzy inference systems. Furthermore, an application to nonlinear systems state estimation is provided, and the result is compared with that of the Extended Kalman filter. The main benefit of this novel fuzzy inference system is its suitability to resource-constrained embedded control and estimation applications.