Dynamic Load Identification Based On Improved LMS Algorithm Applied Research

In 1960,Widrow and Hoff designed the Least Mean Square(Least Mean Square,LMS)algorithm.The LMS algorithm is an important research direction in the field of signal processing.Due to its wide range of applications,the LMS algorithm is still one of the most popular research problems.In many practical engineering problems,such as Strength analysis of the structure,system fault diagnosis,dynamic optimization design,etc.,dynamic load identification on the structure is very important.The traditional frequency domain dynamic load identification method needs to invert the system's frequency response function matrix.It is prone to ill-conditioned problems in the operation process.This method needs to understand the model related knowledge and the accuracy required by practical problems can't be well satisfied.In this paper,the dynamic load identification model established by the improved LMS algorithm don't need prior model parameters,which can avoid the problem that the frequency response function matrix is easy to be ill-conditioned.According to the idea of variable step size LMS algorithm,the original fixed step size factor of LMS algorithm is replaced by variable step size factor,and an improved LMS algorithm(New Variable Step Size LMS,NVSSLMS)is designed.In this paper,the optimal weight vector and the minimum mean square error of the improved algorithm are proved.The variable step size weight vector update formula is given.The theoretical analysis of the anti-noise performance of the NVSSLMS algorithm and the convergence of the algorithm are proved.The NVSSLMS algorithm is compared with the original LMS algorithm by numerical experiments.The comparison results can prove the effectiveness of the NVSSLMS algorithm.Based on the NVSSLMS algorithm,the NVSSLMS adaptive dynamic load model of SISO and MIMO systems is established and the time domain dynamic load is identified.The experimentally validated model can well identify the dynamic load of single point and double point,and the NVSSLMS adaptive dynamic load model of SISO system can better identify single frequency and double frequency dynamic load under noise interference.