Application Of Optimized GA-BP Model In Deformation Prediction Of Foundation Pit Engineering Of Metro Station

During the excavation of foundation pit of metro station,there will be a series of deformations: such as ground settlement,settlement of surrounding buildings(or structures),horizontal or vertical displacement of retaining piles,etc.If the deformation is too large,it may cause damage to the foundation pit structure,even collapse,also may cause building(structure)wall cracking,even collapse.At present,in order to ensure the safety of construction,construction monitoring will be carried out in foundation pit engineering.Based on the project overview,surrounding environment and geological conditions of Sanxiaokou station of Hefei metro line 5,this paper analyzed the risk factors of the project,determined the key points of the deformation monitoring,and formulated the construction monitoring scheme for deformation monitoring according to the relevant specifications.However,relying on construction monitoring alone is not enough,it is also necessary to take certain measures to simulate or predict the deformation development in advance.There are many uncertain and random factors affecting the deformation of foundation pits,it is difficult to fully understand the effect of various factors.Therefore,the essence of deformation prediction is to construct the relationship between monitoring data and time,and predict the deformation value through time.The thesis analyzed and studied the advantages and characteristics of BP algorithm and genetic algorithm,and based on the monitoring data of Sanxiaokou station foundation pit engineering,using MATLAB programming,respectively established BP neural network model,GA-BP combined model and optimized GA-BP combined model.By comparing the prediction results of the three models and analyzing the accuracy evaluation indexes of the models,the results show that the prediction results of the three models are not much different from the monitoring data,and the three models are all applicable to the deformation prediction of metro station foundation pit engineering.The prediction accuracy of the optimized GA-BP combined deformation prediction model is higher than the other two.