Taking the foundation pit project of a municipal people's hospital as an example, a prediction model was established based on measured data combined with BP neural network. Wavelet denoising combined with BP neural network model and wavelet denoising combined with improved genetic algorithm optimized BP neural network model were used to evaluate the prediction effect of foundation pit deformation data using error analysis theory. The results show that comparing the actual processing and prediction data capabilities of the three models, the prediction accuracy of measured data combined with BP neural network model is between 1％ and 4％, the prediction accuracy of wavelet denoising combined with BP neural network model is between 1％ and 2％, the prediction accuracy of BP neural network model optimized by wavelet denoising combined with improved genetic algorithm is within 1％, and the prediction accuracy of BP neural network model optimized by wavelet denoising combined with improved genetic algorithm is the highest. For the deformation monitoring of foundation pit, the BP neural network model optimized by wavelet denoising combined with improved genetic algorithm has higher prediction accuracy, which can provide practical reference for similar projects.
关键词
基坑监测/组合模型/BP神经网络/小波分析/改进遗传算法
Key words
foundation pit monitoring/combination model/BP neural network/wavelet analysis/improved genetic algorithm
维普
万方数据