Non-Affine Fault-Tolerant Control for Multi Euler-Lagrange Systems Based on Adaptive Neural Network
Shitong Zhang,Shuai Cheng,Bin Xin,Qing Wang,Junzhe Cheng
DOI: https://doi.org/10.23919/ccc63176.2024.10662017
2024-01-01
Abstract:This paper proposes an advanced control strategy that combines adaptive backstepping control with Radial Basis Function Neural Network (BFNN) to effectively handle nonlinear dynamics and uncertainties in Euler-Lagrange (EL) systems, particularly during actuator failure. The adaptive backstepping control provides flexibility for complex control problems, and RBFNN enhances adaptability to unknown faults. Compared to traditional linear fault models, the non-affine fault modeling method used here accurately captures the actual fault complexity. Considering the nonlinear relationship between faults and system states provides a realistic representation, crucial for precise controller adaptation to dynamic system characteristics and fault responses, improving overall control effectiveness and system robustness. To address the algebraic ring problem in the control law, a Butterworth low-pass filter (BLF) is employed, effectively reducing high-frequency oscillations and ensuring smooth and stable control signals. BLF prove effective in avoiding instability and performance degradation, particularly with non-affine fault models, significantly enhancing the control system’s adaptability to complex fault scenarios.