The Study Of Vibration Active Control Based On Variable Noise Online Modeling

This thesis proposes a new method which combines improved gradient LMS algorithmwith online secondary path modeling, simulations and experiments are conducted on thisbasis.Firstly, this thesis derivates the improved gradient for LMS algorithm by using theweighted time average instead of the instantaneous value to estimate the gradient, and thisthesis compares this method with the filter-x LMS algorithm on this basis. Simulationinvestigations are conducted to compare the improved gradient LMS algorithm with thefilter-x LMS algorithm. The results of simulation demonstrate that the convergence speed ofthe improved gradient LMS algorithm is faster than the filter-x LMS algorithm and theimproved gradient LMS algorithm can be used in the case of impulsive disturbance. Secondly,this thesis derivates the variable injection noise on-line identification method on the basis ofthe existed method and proposes a new active vibration control strategy by combining thismethod with the gradient LMS algorithm. Simulation investigations are conducted. Theresults of simulations demonstrated that the convergence speed of the method proposed bythis paper is faster and residual error is less than the existed method. Thirdly, experiments areconducted on the real vibration isolation mounting to demonstrate the effect of the activevibration control, the results of the experiments demonstrated that good control results areachieved for different exciting frequencies conditions. Finally, this thesis derivatesmultichannel on-line identification active vibration control method. Simulations andexperiments are conducted to verify the effectiveness of the method proposed by this paper.The results of experiments demonstrated that the method proposed by this paper can also beused in the case of multichannel on-line identification active vibration control.