Study On The Dam Deformation Prediction Method Of Multi Scale Based On Empirical Mode Decomposition

The safety monitoring of dam has been essential for dam construction and operation.Deformation monitoring,analysis and prediction are the most important parts of dam safety monitoring.Master the deformation regularity of the dam,achieve reliable and accurate prediction of dam deformation can resolve the potential crisis and provide scientific basis for the design of dam.Therefore the dam deformation prediction is very important.The main work as follows.(1)Introduce the research status of deformation prediction methods of monitoring dam and the research status at home and abroad,summarize the method of wavelet denoising,empirical mode decomposition,fruit fly optimization algorithm and BP neural network,and these algorithms are applied to dam deformation monitoring.(2)Introduce the monitoring method and technology of dam deformation monitoring,the principles and requirements of dam deformation monitoring,the prediction model of dam deformation and the technical scheme for external deformation of a Hydropower Station.(3)Introduce the wavelet denoising theory and Empirical Mode Decomposition theory,adopt the improved hierarchical wavelet denoising method,combine the EMD and wavelet method,propose the IMF1-Wavelet denoising method,and it is applied to dam deformation denoising successfully.(4)Introduce the theory of the fruit fly optimization algorithm and the back propagation neural network.The defects of the fruit fly optimization algorithm itself is improved,and the simulation experiment is carried out on its optimization ability,aiming at the defects of BP neural network system,this paper proposed a method of the BP neural network weights and threshold optimized by the improved fruit fly optimization algorithm.(5)Combine with the example of a dam deformation monitoring,denoise the deformation data by the IMF1-Wavelet denoising method,then decompose the data byEmpirical Mode Decomposition.Establish the FOA-BP prediction modle for the decomposed components,the prediction results of the components are added as the final prediction results,realize the the prediction of isolation.By comparing the predicted results,proved that using wavelet denoising and EMD algorithm of internal structure and fruit fly optimization BP neural network of external platform,the prediction accuracy relative to the traditional model has better effect.