An Adaptive Particle Filter Technique for Nonlinear System State Estimation

This paper proposes an adaptive particle filter (APF) method to enhance the accuracy of indoor positioning and state estimation in nonlinear systems. Traditional particle filter methods are often hindered by issues such as particle degradation and impoverishment, which compromise the accuracy of state estimation, particularly in complex and nonlinear environments. To address these challenges, the proposed method dynamically adjusts the particle weights based on their distribution, ensuring a more balanced representation of the system’s state distribution and preventing the collapse of particle diversity. Experimental results validate the robustness and enhanced accuracy of the adaptive approach, with the mean absolute error (MAE) reduced to approximately $7-8 \mathrm{~cm}$, demonstrating the method’s effectiveness in achieving high-precision indoor positioning. This research provides a promising solution for accurate state estimation in nonlinear systems and contributes to the advancement of particle filtering techniques for complex practical applications.