Neural network-enhanced Kalman filter for real-time active vibration control in nonlinear dynamic systems

Abstract

This study presents a novel Neural Network-Enhanced Kalman Filter (NN-KF) framework for real-time active vibration control in nonlinear dynamic systems. The proposed method integrates the state estimation capabilities of the Kalman filter with the adaptability and learning potential of neural networks, offering superior performance in complex and uncertain environments. Numerical simulations were conducted on a benchmark nonlinear vibration system to evaluate the effectiveness of the proposed approach. The results demonstrate that the NN-KF reduces vibration amplitude by 40.12% compared to traditional Kalman filtering methods. Additionally, the proposed method achieves a 14.07% improvement in response time and enhances system stability under varying operating conditions. When compared to conventional adaptive control techniques, the NN-KF framework improves energy efficiency by 8.35%, making it a robust and sustainable solution for vibration suppression. These findings highlight the potential of the NN-KF for applications in smart structures, aerospace systems, and high-precision machinery, where real-time control and adaptability are critical.