Prediction of subsidence deformation of subway project by improved markov chain and optimized wavelet neural network model

Abstract

In view of the fact that the factors affecting the subsidence deformation of subway engineering are complex and the subsidence monitoring data are random, this paper constructs an optimized wavelet neural network model with markov chain improvement to improve the accuracy of the predicted data. Firstly, the wavelet neural network is optimized by adding momentum and adjusting learning rate adaptively. Finally, based on markov chain theory, the residual between the predicted value and the actual monitoring value of the optimized wavelet neural network is improved. Taking the settlement data of a settlement monitoring point of xi'an metro line 14 as the research object, the prediction accuracy of the model was compared and analyzed by C++ program. The results of the program show that the additional momentum and the adaptive learning rate are effective for the optimization of the wavelet neural network model, and the prediction accuracy of the optimized wavelet neural network model improved by the markov chain is higher than that of the single model, which is suitable for the prediction of the subway engineering settlement.