Research Of Detection And Characteristic Analysis Of Nonlinear Structure Under Random Excitation

Nowadays, with the rapidly progress of the astronautics technic, the compond material is widely used in spaceflight field. Owing to the complex environment, lots of nonlinear problems appeare. The linear method can not meet the need of identification, and the study of nonlinear is necessary. Meanwhile, the traditional mechanical environment testing can not meet the need of ensure the reliability of aerocraft. The theories and methods of detecting and characteristic analysis of nonlinear of dynimic system under random excitation are studied deeply in this paper, and using the random vibration test and take aluminum honeycomb sandwich, which has been widely used in aerocraft, as study object.In random test, if the input is white noise single, when the system is nonlinear system, the output is not white noise single anymore. Thus, the nonlinear detection of system become the nonlinear detection of output data. Using the higher 2order statistic spectrum and largest Lyapunov exponent to detect the nonlniear existence. To observe whether the bispectrum plot of output data is smooth to detect the nonlinear. If the possibility of the presence of chaotic behavior in the time-series data is explored, and the existence of nonlinear is proved.To deal with prolblems of detecting and characteristic analysis of nonlinear of dynimic system under random excitation. A method combining Fokker-PlancL-Ko1mogorov(FPK) transformation and equivalent principle method is used to caculate equivalent linear frequency response function (FRF) model of nonlinear systems containning nonlinear damping and nonlinear siffness, respectively. The shapes of equivalent linear FRF and Nyquist plots will change when the excitation level changed.and the rule is explored. The result can be the method of nonlinear detection and characteristic.Taking aluminum honeycomb sandwich board as subject, the project of random test is designed. The existence of nonlinear is detected by using FRF, the higher 2order statistic spectrum and the Lyapunov exponent methods. Owing to the chaonic phenomena of the data, and the state space reconstruction of the test data is done and Volterra adaptive filter is used to predict the test data. it is shown the test data could be accurately predicted.The FRF is gained, and thus the nonlinear characterise can be done by using acceleration data directly, thus the process of data tansformation is avoided. The nonlinear is characterised by using the classical FRF and the FRF expressed by acceleration of both input and output. The phenomena of natural frequency decreasing and other nonlinear phenomenon are explored. The applicability of the methods metined in this paper is proved.