Research On Intelligent Prediction Approach Of Dam Deformation Based On Deep Learning

Dam deformation prediction is one of the important scientific means of dam safety management.However,due to the coupling effect of many factors,the dam deformation is highly complex,which brings some difficulties to its deformation prediction.Therefore,it is necessary to study the dam deformation prediction method with strong adaptability and high prediction accuracy.Considering the excellent performance of deep learning method in the field of prediction in recent years,the research on dam deformation prediction method based on deep learning method is carried out in this paper.The main work is as follows:(1)Based on the basic calculation theory of deep learning model and dam deformation prediction theory,this paper studies the improvement method of deep learning model and how to apply deep learning hybrid model to dam deformation prediction.According to the idea of constructing integrated prediction method,the search and calculation methods of butterfly optimization algorithm(BOA)are studied.Aiming at the shortcomings of the original algorithm in optimizing the hyperparameters of deep learning model,quantum computing is introduced to improve the standard algorithm,and a quantum butterfly optimization algorithm(QBOA)for hyperparameter optimization is proposed.Based on the idea of depth combination,an attention mechanism(AM)is used to control the features contribution of multi factor deep learning autoregressive prediction model.Based on the idea of pre-processing layer,an empirical mode decomposition(EMD)is adopted to process the input sequence of the accuracy of deep learning autoregressive prediction model.(2)The nonlinear mapping relationship between various influencing factors and dam deformation is established by using the combined model of CNN,GRU and AM,and the CGA dam deformation regression prediction model is proposed.Integrating the QBOA algorithm and CGA prediction model,an integrated QBOA-CGA dam deformation prediction approach is proposed.The feasibility of the integrated prediction approach is verified by example analysis,and after the verification,the dynamic prediction is carried out considering the actual situation.(3)EMD is used to decompose the dam deformation time series.Considering the IMF are independent of each other,three ECG dam deformation autoregressive prediction models are proposed based on different model structures.Integrating QBOA algorithm and ECG autoregressive prediction model,an integrated QBOA-ECG dam deformation prediction approach is proposed.The optimal ECG model structure and the feasibility of the integrated prediction approach are verified through example analysis,and after the verification,the dynamic prediction is carried out considering the actual situation.