Research On Identification Methods Of Nonlinear Systems With Clearance Based On Output Feedback

With the development of engineering applications,more and more complex structures with nonlinear characteristic are applied,and the nonlinear dynamics problem becomes more and more prominent.The research of nonlinear system identification method has become a hot spot,and has made great progress in theory,calculation and test.However,there are still some limitations.In order to meet the increasing demand of technology and environmental performance on structure and equipment,nonlinear system identification methods need to be further developed and perfected.In this paper,the identification method of the nonlinear system based on output feedback is mainly studied for the nonlinear system with clearance.Based on the existing relevant methods,the method is improved or coupled with other methods to improve the identification accuracy and enhance the applicability.The main content of the paper include the following:(1)The first chapter introduces the research significance of nonlinear system identification methods,and summarizes the research status of nonlinear system identification methods based on output feedback in frequency domain and time domain respectively.The identification problems of the nonlinear system with clearance are briefly summarized,and the necessity of developing and improving the identification theory and method is pointed out.(2)The second chapter focuses on the coupling errors caused by the large differences in the numerical magnitude between the excitation force and the nonlinear description functions when both of them are simultaneously considered as an input vector,and proposes a nonlinear separation identification through feedback of the outputs method.The proposed method needs two excitation tests including the low-level excitation test and the high-level excitation test.The underlying linear frequency response function matrix is firstly identified under low-level excitation,and only nonlinear description functions are considered as an input to identify nonlinear parameters under high-level excitation by using separation strategy.The validity and accuracy of the proposed method are verified by the numerical examples of the three degrees-of-freedom nonlinear structure with clearance and the cantilever beam nonlinear structure with clearance.(3)The third chapter focuses on the actual conditions in which the external excitation of the system is difficult to measure,and proposes a output-only nonlinear subspace identification method.By solving the linear part and the nonlinear part separately in the subspace identification process,the nonlinear parameters can be identified only by measuring the output data on the premise that the characteristic matrix or state space matrix of the underlying linear system is known.The validity and feasibility of the proposed method are verified by the numerical examples of the three degrees-of-freedom nonlinear structure with clearance and the cantilever beam nonlinear structure with clearance.(4)In the fourth chapter,due to the difficulty in selecting the nonlinear description functions,the identification process based on time-domain nonlinear subspace was cumbersome under different excitation level.The neural network models trained based on nonlinear restoring force data reconstructed by time-domain nonlinear subspace method and response data at the nonlinear position under different excitation level are used to equivalently represent the nonlinear restoring force-response mapping relationship.As a result,the identification process is no longer dependent on the system models.In other words,the nonlinear restoring force can be obtained as long as the response at the nonlinear position is known and the computational efficiency is improved.Furthermore,regarding the problem that it is difficult to measure the external force of nonlinear system,a load identification method based on neural network and subspace method is proposed.The validity and feasibility of the proposed load identification method are verified by the numerical examples of the three degrees-of-freedom nonlinear structure with clearance.(5)The fifth chapter designs and sets up an three degrees-of-freedom experimental structure with clearance.The proposed method of the improved frequency domain nonlinear system based on separation strategy and the time-domain output-only subspace identification method of the nonlinear system,as well as the load identification method based on the neural network and subspace method are respectively applied to the experimental system.The feasibility of the proposed method is validate via the comparison and analysis the result of the experiment.