Foundation pit deformation prediction method

Abstract

The invention provides a foundation pit deformation prediction method. The foundation pit deformation prediction method comprises the steps of 1, setting an initial weight and a threshold value of a neural network; 2, determining parameters of a network structure model, and compiling a neural network operation program; 3, training the neural network, and inputting training samples into a training network in the neural network; 4, performing training sample output on foundation pit deformation data of the training samples by running the neural network; 5, processing a prediction sample by adopting an error grading iteration method, inputting the prediction sample into the trained neural network for sample prediction, and carrying out grading iteration until an error iteration value is smaller than a set error value; 6, outputting a final prediction sample result. According to the method, the error result between a predicted value and an actual monitoring value is optimized through processing of the error grading iteration method, the predicted value result better conforms to the actual monitoring value, the prediction accuracy is improved, and therefore the prediction performance of foundation pit deformation is improved.

Description

technical field [0001] The invention belongs to the technical field of foundation pit engineering, and in particular relates to a method for predicting deformation of foundation pits. Background technique [0002] The scale of urban construction is expanding, and the scale of urban underground space construction is increasing. With the increase of underground construction scale, deep foundation pit projects are constantly appearing, and the safety accidents caused by deep foundation pits are also increasing year by year. The monitoring of deep foundation pits is particularly important, and the on-site monitoring of foundation pits cannot show the deformation of foundation pits very well. Therefore, it is necessary to predict the deformation of deep foundation pits, manage key parts, prevent accidents, and improve scientific research. ability to mitigate disasters. [0003] Many scholars have conducted research in the field of foundation pit deformation prediction, and have ...

AI Technical Summary

Problems solved by technology

[0003] Many scholars have conducted research in the field of foundation pit deformation prediction, and have made great progress. Many scholars use neural network for prediction combined with algorithms to optimize traditional neural network, wavelet, gray model, quantum particle swarm, least square method, etc. for prediction , the results show that although it can meet the needs of engineering, there are shortcomings at the same time. The classic BP neural network has the problems of slow convergence speed, easy to fall into local minima and the selection of the number of hidden layers. It is necessary to use algorithms to optimize the neural network. Improving the predictive accuracy and predictive performance of neural networks

Images