Adaptive filtered x-least mean square algorithm with improved convergence for resonance suppression. (October 2014)

• Record Type: Journal Article
• Authors: Wang, Yong-Qing; Huang, Fu-Chang; Liu, Hai-Bo
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 228:Number 9(2014)
• Page Start: 668
• Page End: 676
• Publication Date: 2014-10
• Subjects: Resonance suppression -- adaptive filtered x-least mean square algorithm -- convergence speed -- elastic drive systems; Mechanical engineering -- Periodicals; Automatic control -- Periodicals; Systems engineering -- Periodicals; 621.3
• DOI: 10.1177/0959651814541883
• Languages: English
• ISSNs: 0959-6518

Abstract

The existence of the resonance is usually a trouble causing instability for most elastic drive systems. Generally, the measurement of original resonance of load side in a drive system is a direct solution for resonance suppression, but exact data are difficult to come by, such as torsional torque, load speed and disturbance torque. Therefore, a developed method for resonance suppression based on adaptive filtered x-least mean square algorithm with improved convergence is presented in this research. The proposed method obtains the resonance iteratively and reduces the significant resonance oscillations through a finite impulse response filter adapted by the least mean square error principle. In order to tackle the convergence speed problem caused by the high dynamics of the forward path model, another finite impulse response filter is inserted into the control structure to smooth the current control reference signal and the speed error signal, so that the dynamics features of the forward path model are improved. Furthermore, a filtered x-least mean square control structure for elastic drive systems is also developed. From the simulation and experimental results, the resonance is more effectively suppressed with proposed modified filtered x-least mean square structure compared with notch filters, and the inserted finite impulse response improves the convergence speed.