Application Of Emd And SVM Model In The Prediction Of Time Series Load

There are many high or large structures in water conservancy engineering, such as high dams and dam construction, bridge and long-span aqueduct, while the influence of wind and earthquake loads on the structure of the time series, which is affected by some engineering accidents, even cause disaster. Through deeply study the characteristics and laws of the wind and earthquake and other time series loads, and the predictions will be helpful to the control of structures and avoiding or mitigating engineering disasters.. In this paper, empirical mode decomposition and support vector machine model are used to study the prediction of wind speed and earthquake acceleration. The main contents are as follows:1)This paper analysises the instability of the wind and earthquake and the method for predicting of non-stationary data sequences. The characteristics of traditional forecasting methods and the current prediction methods widely used are compared.2)This paper introduces the method of SVM based on statistical learning theory and LS-SVM based on structural risk minimization. The basic principle of empirical mode decomposition method, the detailed decomposition steps and some problem solving methods are also introduced. In order to improve the prediction accuracy of the model that the PSO is introduced in the article. The kernel function and parameter selection problem of the model are studied. Finally, taking the prediction of wind speed and earthquake acceleration as an example to verify the validity of EMD.3)The method based on empirical mode decomposition and multi-step least squares support vector machine is put forward to establish the model and forecast wind speed. Empirical mode decomposition is applied to decompose dynamics wind speed signal into several intrinsic mode functions with different characteristic scale(frequency),. and then multi-step least squares support vector machine forecasting model corresponding to each stationary IMF component is built,and the finally prediction result is the sum of each prediction component. In the end, the results of example analysis shows that the error of the composite method is smaller than that of single LS-SVM prediction method, so the efficiency of the proposed method for wind speed forecasting is verified.4)Because of the acceleration of the earthquake has the characteristic of non-stationary time series, so the combination method is also applied to the acceleration sequence modeling and forecast.. The computational results verify the applicability of the combination method to the prediction of non-stationary time series..In the last, the writer summaries the paper and prospects to the further research.