基于模型实时辨识自适应控制算法的时变机械系统振动主动控制

Active vibration control of time-varying mechanical systems based on an adaptive algorithm with model real-time identification

参数时变的现象广泛存在于机械系统。如果系统参数随着时间而发生较大变化,振动主动控制方案就需要考虑时变参数对控制算法的影响。针对动力学特性变化较大的时变机械系统振动,提出一种模型实时辨识自适应控制算法,该算法将传统的滤波自适应算法与递归预测误差方法相结合,利用改变梯度的递归预测误差方法实时估计控制通道模型。建立弹簧质量支承的非均匀截面杆纵向振动时域模型,模型中随时间而变化的弹簧刚度导致模型动力学特性发生较大变化。用模型实时辨识自适应控制算法对建立的杆模型进行振动控制数值仿真,仿真结果表明,所提出的控制算法能有效抑制时变系统的窄带和宽带振动。相对于现有的方法,该控制算法能实现更好的控制性能。最后,将所提出的控制算法应用到时变的摇摆系统振动控制,实验结果验证了所提出控制算法的可行性和有效性。

Phenomena of time-varying parameters widely exist in mechanical systems. If a system’s parameters change greatly with time, effects of time-varying parameters on control algorithm should be considered in active vibration control schemes. Here, an adaptive algorithm with model real-time identification was proposed for suppressing vibration of time-varying mechanical systems with larger variation of dynamic characteristics. The proposed algorithm combined the traditional filtered adaptive algorithm and the recursive prediction error method, the latter with gradient change was used to estimate the control channel model in real time. A time-domain model for longitudinal vibration of a non-uniform cross-section beam with spring-mass supports was established. The spring stiffness in the model changes with time to cause larger variation of the model’s dynamic characteristics. Numerical simulation for vibration control was performed for the established beam model using the proposed adaptive algorithm with model real-time identification. Simulation results showed that the proposed control algorithm can effectively suppress the time-varying system’s narrowband and broadband vibration;compared to the existing adaptive algorithms, the proposed algorithm can have a better control performance. Finally, the proposed control algorithm was applied in vibration control tests of a time-varying swing system. It was shown that test results verify the feasibility and effectiveness of the proposed control algorithm.