Feedforward Adaptive Control Algorithm And Hybrid Adaptive Control Algorithm Research For Active Vibration Control System

Active vibration control is the most important issue in the area of vibration engineering, for its remarkable advantages. It involves multiple disciplines, such as electronics, structural mechanics, dynamics, acoustics, fluid mechanics, thermodynamics, materials science, aesthetics. As the core of active vibration control studies, active vibration control algorithm researches are in full wing involving almost all control theories and control techniques. But there are still many unsolved applicable and practical problems. There is no paradigm methodology for active vibration control system design, and it needs further research.Based on the background of a major project of the National Natural Science Foundation of China, this dissertation focuses on stable active adaptive vibration control methods and techniques to improve the robustness and control performance for mechanical vibration systems with inherent feedback. Both feedforward and hybrid adaptive control paradigms are provided. Both theoretical and technical researches are done:(1) A typical mechanical active vibration control system with inherent feedback is constructed using inertial actuators as exciter and suppresser, and acceleration sensors as residual sensor and reference sensor, for real-time control algorithm verification.(2) A class of improved recursive least squares algorithm is proposed for adaptive parameter updating. And U-D factorization technique is employed for long term real time parameter updating.(3) Recursive least squares algorithm, extended least squares algorithm, and output error with extended prediction model algorithm are proposed for vibration control system identification, and whiteness test and model verification method is provided.(4) A new adaptive feedforward active vibration control algorithm is proposed, while detailed algorithm derivation and analysis process is given based on hyperstability theory. Strictly positive real condition is provided and relaxing requirements are given. Simulation examples are shown to demonstrate the effectiveness of the proposed algorithms for the deterministic perfect matching case, stochastic perfect matching case and non-perfect matching case. The simulation results show employing specific pre-filter could improve the control performance of the proposed feedforward algorithm significantly, and the proposed algorithm could give a good control performance and rapid convergence rate even there exists large measuring noise and the adaptive controller order could not fulfill the perfect matching condition.(5) To improve the convergence and control performance of the proposed feedforward control algorithm, a new adaptive hybrid active vibration control algorithm is proposed. Detailed algorithm derivation and analysis process is provided. Also strictly positive real condition for the proposed algorithm is provided, as well as relaxing strictly positive real condition requirements. Simulation examples are shown to demonstrate the effectiveness of the proposed algorithms for the deterministic perfect matching case, stochastic perfect matching case and non-perfect matching case. Also the simulation results show the hybrid methods could provide a much better convergence and control peroformance than the feedforward methods, with lower order controller.(6) Experimental verification and analysis were done to verify all the proposed algorithms, comparing with the experimental results using feedforward filtered-U least mean square algorithm (FULMS algorithm), hybrid FULMS algorithm and feedforward adaptive controller combined with pole placement feedback controller. Experimental results confirm the proposed algorithms perform excellent control performance rather than other control methods.