基于匹配追踪的伺服系统谐振抑制研究

Research on Resonance Suppression of Servo System Based on Matching Pursuit

针对于自适应陷波器(ANF)的谐振频率估计法收敛速度慢、初始值敏感和搜索精度不高的限制,提出一种基于匹配追踪(MP)算法的谐振频率提取方案。首先,建立双惯量伺服系统的动力学方程,采用三参数陷波器,设计陷波频率、陷波深度和陷波宽度的整定过程;然后,设置频率和相位为分解参数,构建以正弦波谐振模型为字典的过完备原子库,通过MP算法对速度误差信号和字典中的原子逐一匹配,内积最大的原子频率即是谐振频率;最后,通过谐振频率对陷波器3个参数进行整定,完成对谐振的在线抑制。仿真实验表明,与基于ANF的谐振频率估计法相比,所提方案无论在谐振频率的收敛速度还是辨识精度上都有很大的优势,并且陷波效果也验证了所提方案的优越性。

Aiming at the limitation of slow convergence speed,sensitive initial value and low search accuracy of resonant frequency estimation of adaptive notch filter(ANF),a resonant frequency extraction scheme based on matching pursuit(MP)algorithm was proposed.First,the dynamic equation of the two-inertia servo system is established,and the three-parameter notch filter is used to design the setting process of notch frequency,notch depth and notch width.Then,set frequency and phase as decomposition parameters,build an over-complete atom library with sine wave resonance model as dictionary,and match the velocity error signal with the atoms in the dictionary one by one through MP algorithm,and the atomic frequency with the largest inner product is the resonant frequency.Finally,the three parameters of the wave trap are adjusted by the resonance frequency to complete the online suppression of resonance.Simulation experiments show that compared with the resonant frequency estimation method based on ANF,the proposed scheme has great advantages in both the convergence speed and the identification accuracy of the resonant frequency,and the notch effect also verifies the superiority of the proposed scheme.