Research On The Method Of Eigenvalue Extraction Of Nonlinear System Based On State Signal

In the development of today's society,the research of robotic arms has attracted more and more attention from all walks of life.In the field of robotic arm research,since the robotic arm system is a complex non-linear system and is interfered by many factors,it has always been a difficult problem to accurately model the robotic arm.The research on the extraction of nonlinear features of the manipulator system has important guiding significance for the follow-up precise modeling of the manipulator.In recent years,the research on non-linear feature extraction has reached a certain depth.In the field of feature extraction studied in this article,it is not the process of feature extraction on images.The manipulator system studied in this thesis is a complex nonlinear system,and its state signal is one or more sets of data.The new data obtained may not have a clearer physical meaning like the effect in image processing,and it may only be expressed as a transformation relationship or result.This thesis studies the nonlinear characteristics of the flexible joint manipulator,and uses different nonlinear feature extraction algorithms to analyze the flexible joint manipulator model.This thesis derives the dynamic equations of the flexible joint manipulator based on the Lagrangian dynamic equation theory of the manipulator,and analyzes its state signals.It is found that the flexible joint manipulator has some non-linear disturbances compared to the rigid manipulator.The simulation analyzes the difference and connection between the state signals under different joint stiffnesses,and establishes a database for further research on the robotic arm system in the future.In terms of feature extraction,the basic principles and algorithms of the manifold algorithm are studied.Among them,the LLE algorithm is deeply studied,combined with the multi-class algorithm to study the principle of support vector machine,the LLE-SVM algorithm is used to extract the different stiffness characteristics of the flexible joint manipulator,and the characteristic interval of the joint stiffness is analyzed.Through the simulation of this algorithm,the advantages and disadvantages of this algorithm are analyzed.Through the in-depth study of the PCA algorithm,combined with the ELM algorithm,the PCA-ELM algorithm,the LASSO-PCA-ELM algorithm,the LLE-KELM algorithm and other algorithms are used to comprehensively analyze the characteristics of the state signal,and an improved LASSO-PCA-KELM is proposed.algorithm.Comparing this algorithm with other algorithms,the analysis shows that this algorithm is superior to other algorithms,and this algorithm can improve the accuracy of classification and recognition after feature extraction.And through the simulation experiment results analysis of the algorithm,a comprehensive comparison of other algorithms used in this article,analysis of the advantages and disadvantages of the improved LASSO-PCA-KELM algorithm,the results show that this method compared with other methods,the algorithm is in the recognition rate,Stability and other aspects are better than other algorithms.