Research On Intelligent Prediction Of Structural Health State Based On FESN

In real life,large-scale building structures and equipment will be more or less damaged in the course of the service.If not promptly discovered and dealt with,it often result in serious consequences of both people and property.So health monitoring,diagnosis,evaluation and prediction of the structure is particularly important.In this paper,under the premise of structural health monitoring,structural damage feature is analyzed and extracted and the structural damage trend is studied for the purpose of predicting the trend of structural health status.The specific research contents are as follows:The signal preprocessing method based on Empirical Wavelet Transform(EWT)is studied.The EWT is used to segment the spectrum of the acceleration vibration signal of the structural damage.The appropriate and orthogonal wavelet bandpass filter bank is constructed,and the Amplitude Modulated-Frequency Modulated(AM-FM)component of the Fourier spectrum with compact support is obtained,and then extract the components containing rich damage information.Its transformed by Hilbert,and calculate the instantaneous frequency and instantaneous amplitude.The experimental results show that the instantaneous frequency can reflect the change of the stiffness before and after the damage of the structure,and there are obvious differences of the instantaneous frequency in the different nodes or different damage conditions.Therefore,it can be used as a predictor of structural health status which can reflects the trend of structural health status and lays the foundation for the next trend of damage forecast.The nonlinear time series forecasting method according to fuzzy theory and echo state network is studied.The reasoning algorithm,the training process and the key parameters of the network are studied in detail,and the stability of the network algorithm is strictly defined.The experimental results show that the selection of the appropriate parameters has some influence on the prediction accuracy.The prediction accuracy of the tantaloid neuron activation function is higher than that of the leaky state.Compared with the traditional echo state network,the fuzzy echo State network(FESN)has stronger nonlinear approximationability,higher prediction accuracy and can handle larger sample data,and the training efficiency is also improved,which provides the theoretical basis for the prediction of the health state of the actual engineering structure.The trend prediction method of structural health status based on FESN is studied,the EWT method is used to extract the AM-FM component with internal structure damage information,and the Hlibert transform is carried out to obtain the instantaneous frequency and then as the input of prediction model.FESN is used to predict the engineering simulation of single-degree-of-freedom structure and multi-degree-of-freedom structural model,and it is applied to the prediction of actual engineering data.The experimental results show that the FESN prediction model is closer to the real value and the prediction accuracy is higher.