Reinforced Lattice Kalman Filters: A Robust Nonlinear Estimation Strategy

This paper introduces the Sliding Innovation Lattice Filter (SILF), a robust extension of Kalman (LKF) that leverages sliding mode theory. SILF incorporates boundary layer in measurement update formulation, enabling filter innovation to slide within predefined upper and lower bounds. enhances robustness SILF, making it resilient model uncertainties noise. Additionally, derivative-free formulation is developed using statistical linear regression, eliminating need for Jacobian calculations. To further improve accuracy, robustness, convergence behavior presence abrupt changes system model/parameters, reinforced with Iterated Sigma Point Filtering Strong Tracking strategies, resulting Reinforced (RLKF). The experimental findings estimation distorted power waveforms illustrate superior performance RLKF over competing methods, especially when operating scenarios characterized by noisy environments.