Strong tracking pseudo-measurement adaptive Kalman filtering based on reinforcement learning

Kalman filtering is a fundamental technique in state estimation and data denoising and strong tracking filter is a classic improvement method of it for solving systematic errors or model mismatches. This study is based on the basic idea of strong tracking filters. Building upon the concept of pseudo-measurement and leveraging the optimization capabilities of reinforcement learning, an adaptive strong tracking filtering algorithm named Reinforcement Learning Pseudo Kalman Filter (RLPKF) is developed based on the Deep Deterministic Policy Gradient (DDPG) framework. Numerical simulation results confirm that the proposed method improves both the accuracy and convergence speed of state estimation in the presence of biased initial conditions or systematic errors, demonstrating superior performance compared to conventional approaches.