Adaptive Denoising Study Based on Improved Variable Step Size LMS Algorithm

• Published in: 2024 Second International Conference on Cyber-Energy Systems and Intelligent Energy (ICCSIE) pp. 1 - 6
• Main Authors: Zhang, Hong, Li, Janhang, Cheng, Yue, Liu, Fang, Qu, Chunhang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.05.2024
• Subjects: Convergence; Filtering; Filtering algorithms; Grounding; Heuristic algorithms; Noise cancellation; Noise reduction; Particle swarm algorithm; Particle swarm optimization; Signal to noise ratio; Small current grounding system; Stability analysis; Steady-state; Variable step size LMS algorithm
• DOI: 10.1109/ICCSIE61360.2024.10698025

To overcome the signal noise problem in small-current grounding systems, an improved LMS adaptive filter algorithm using a symmetric nonlinear function to adjust the step size is proposed in this paper. This algorithm makes innovative improvements to the Sigmoid function by using the absolute value and the nonlinear stretch to construct a new symmetric nonlinear function, which realizes the nonlinear mapping between the step size factor and the steady state error. The method not only solves the problem of noise accumulation and amplification through the normalized variable step factor, but also significantly improves the filtering effect and accelerates the convergence process of the algorithm. In addition, the particle swarm optimization method is combined to finely adjust the step size factor to the optimal one to further improve the denoising performance. After rigorous theoretical analysis and simulation experiments under various conditions, the results demonstrate the superior performance of the proposed algorithm in suppressing noise, accelerating convergence, and improving the output signal-to-noise ratio, especially in the low signal-to-noise ratio environment and dynamic change conditions, which show remarkable stability and high efficiency.