Study of a Real-Time Filtering Algorithm Suitable for LQR Control

Linear quadratic regulator (LQR) is crucial in modern optimal control theory. However, its application to nonlinear unstable systems suffers from poor robustness and instability. To address this, this paper proposes a self-optimizing filtering algorithm from a signal processing perspective. The algorithm samples signal values via recursive queue, calculates averages, and limits them. Iterations find the optimal limit interval, and the average of interval values is the filtering result. Experimental verification on a first-order linear inverted pendulum shows this method outperforms others in accuracy, filtering quality, and stability, while meeting real-time control requirements. This improves LQR's robustness and control accuracy.