Intelligent Optimization BP Neural Network Model And Its Application In Deformation Monitoring Data Analysis

During the construction and operation of subway tunnel,the pressure of the external soil layer and the internal pressure generated by the subway operation act on the tunnel wall together,often leading to the lateral or longitudinal displacement and deformation of the tunnel inner wall.When the displacement deformation value exceeds the allowable deformation value,it may lead to train derailment and other safety accidents,so it is very necessary to conduct Deformation monitoring and predict the deformation development trend of subway tunnel.Because the displacement and deformation of subway tunnel are nonlinear and complex,support vector machine or BP neural network model is difficult to accurately predict its deformation trend.Aiming at the nonlinear characteristics of tunnel displacement,an improved adaptive noise complete set empirical mode decomposition algorithm(ICEEMDAN)is proposed to preprocess the deformation data,and the BP neural network model optimized by improved particle swarm optimization(IPSO-GA),and beetle antennae search-particle swarm optimization(BAS-PSO)algorithm is used to predict the deformation of subway tunnel.The main research content and results of this article are as follows:1.The basic principles and algorithm implementation process of empirical mode decomposition(EMD),ensemble empirical mode decomposition(EEMD),complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN),and improved complete ensemble empirical mode decomposition with adaptive noise(ICEEMDAN)are analyzed and studied.For nonlinear and non-stationary time series data,simulated signal analysis was used to compare the performance characteristics of the decomposition algorithms mentioned above.Simulation results showed that the ICEEMDAN algorithm outperformed the other three methods.2.Support vector machine(SVM),random forest,Adaboost algorithm and BP neural network model are analyzed and studied;The prediction of deformation monitoring data of subway tunnel inner wall segments using SVM,RF and Adaboost algorithm models is realized by Python;Based on Matlab,the BP neural network model is also used to predict the deformation monitoring data of subway tunnel inner wall segments;The prediction results and accuracy evaluation indicators indicate that the BP neural network model has better prediction performance than the other three commonly used machine learning methods.3.The basic principles and implementation processes of four intelligent optimization algorithms are analyzed and studied,including genetics,particle swarm optimization,cuckoo,and beetle antennae search.In response to the shortcomings of particle swarm optimization algorithm,a new particle swarm model is constructed by combining it with genetic algorithm and beetle antennae search algorithm based on the introduction of dynamic inertia weights.The test results of benchmark function show that the cosine function as a dynamic inertia factor can achieve better results,and the optimization effect of BAS-PSO algorithm is better than that of standard particle swarm optimization algorithm.4.Four neural network prediction models were constructed based on genetic algorithms,particle swarm optimization,cuckoo birds,and longicorn whiskers,including GA-BP,PSO-BP,CS-BP,and BAS-BP.The experimental analysis results indicate that the PSO-BP neural network model can achieve the optimal prediction performance.IPSO-GA-BP and BAS-PSO-BP neural network prediction models were constructed,and experimental analysis results showed that the two improved models had higher prediction accuracy compared to the standard PSO-BP model.Combining the ICEEMDAN algorithm,an ICEEMDAN-IPSO-GA-BP neural network prediction model was constructed.The experimental analysis results showed that the model had high prediction accuracy.Compared with the IPSO-GA-BP model,its prediction accuracy decreased by 69.05%and 70.58%,respectively,while R and R~2 increased by 27.01%and 72.32%,respectively.5.The C#and MATLAB language are used,the deformation monitoring prediction software is developed,which realizes the visualization of deformation monitoring data,intelligent algorithm parameter setting,deformation prediction and other functions.