Adaptive parameterized control for line spectrum vibrations using a genetic algorithm-optimized variable forgetting...

Line-spectrum vibrations originating from periodically operating equipment such as power transmission systems concentrate energy at specific frequencies, posing significant challenges for conventional control algorithms. A Youla(Q)-parameterized adaptive controller for unknown time-varying line-spectrum disturbance rejection is proposed, with its forgetting factors optimized by a genetic algorithm (GA). The base controller is augmented with tunable Q-parameters through the Youla parameterization. The Q-parameters are tuned using the recursive least squares (RLS) algorithm and converge to satisfy the internal model principle (IMP) interpolation condition, enabling adaptive suppression of line-spectrum disturbances without explicit disturbance model identification. In this study, a performance-driven optimization method for the forgetting factor variation law using a genetic algorithm is proposed. This approach surpasses previous heuristic methods and resolves the conflict between fast convergence and small steady-state error in adaptive algorithms. The experimental results on an active vibration isolator demonstrate that the GA-enhanced adaptive controller not only accelerates the convergence speed but also enhances the steady-state control performance. The system successfully adapts to and suppresses five simultaneous time-varying line-spectrum vibrations, with an average peak attenuation of 39 dB, confirming the practical efficacy and robustness of the method.