Research On Dam Deformation Prediction Based On Multi-factor Neural Network Model

China has many rivers,abundant runoff,and a huge drop,and it contains very rich hydropower resources.There are 3886 rivers with theoretical hydropower resources of 10 MW and above,and more than 98,000 dams have been built respectively.While providing clean energy,it also plays an important role in flood control,water supply and irrigation.The old dams built before the 1980 s are basically earth-rock dams,and there is a high probability of danger.China 's high dams above 200 meters are mainly concentrated in the western region.The conditions are extremely complex and it is necessary to study the safety monitoring of the dam.Because the deformation of the dam is a long-term,non-linear and complex change process that is affected by many factors such as water pressure factor,time-efficiency factor and temperature factor,the traditional method is to use a single theoretical method and algorithm model to deform the dam Prediction,so this traditional model based on ideal strict assumptions is difficult to achieve good prediction results.Under this circumstance,an artificial neural network,a nonlinear information processing model that can carry out complex logic operations and nonlinear relationships,is introduced to construct a deformation prediction model for the dam.To this end,this paper uses the extreme learning machine,which is a research hotspot in recent years,as the basic research method,combined with genetic algorithm and particle swarm optimization algorithm,proposes a combined model,and innovatively incorporates empirical mode decomposition,Complementary Ensemble Empirical Mode Decomposition(CEEMD)filtering,and influence factor selection Based on theoretical knowledge such as Autoregressive Integrated Moving Average model(ARIMA)error correction model and so on,a dam deformation prediction model based on multi-factor neural network and its error correction and accuracy evaluation are constructed.The specific research content includes the following aspects:(1)The study has studied the relevant theories and main characteristics of empirical mode decomposition,and has focused on the problems of empirical mode decomposition(EMD),in-depth research on set empirical mode decomposition and supplementary set empirical mode decomposition.Since the deformation of the dam is the result of a combination of factors,it is possible to better analyze the influence of its physical factors through decomposition.On this basis,scientific and reasonable filtering of dam deformation monitoring data improves the accuracy of the dam deformation prediction model.(2)The Spearman rank correlation coefficient is used to analyze and evaluate the main impact factors in the input layer of the intelligent algorithm,and then select the impact factors accordingly,and finally confirm the impact factors of the dam deformation prediction model.At the same time,in order to eliminate the influence of dimensional and numerical gaps,the influence factors and deformation monitoring data of the dam deformation prediction model are normalized.(3)Combining genetic algorithm,particle swarm optimization and extreme learning machine,using Genetic Algorithm(GA)algorithm and Particle Swarm optimization(PSO)algorithm to optimize the input weight and hidden layer threshold of Extreme Learning Machine(ELM)neural network,construct GA-ELM and PSOELM combination models respectively,and combine A variety of dam deformation influencing factors selected before are used to construct a dam deformation prediction model based on a multi-factor neural network,combined with an example to verify.(4)The ARIMA error correction model is used to correct the residuals of the prediction results,weakening the prediction residuals of the dam deformation prediction model,and further improving the accuracy of the model.Combining the three evaluation coefficients of root mean square error RMSE,average absolute error MAE and average relative error MRE,all the predictio n results are evaluated and analyzed to verify the rationality of the dam deformation prediction model based on the multi-factor neural network.