Studies On Deep Foundation Pit Monitoring And Deformation Prediction Based On Neural Networks Method

The establishment of deep foundation pit monitoring was studied systemically. By practical process of a project, the matters needing attention during pit monitoring are put forward. And applications of the back propagation neural network, the grey system and their union model employed in the horizontal deformation prediction of deep foundation pit are studied and discussed. After that, the characteristics of the three methods when used to predict the deformation of foundation pit are contrasted. Some conclusions reached from the above include:1. The deformation monitoring is a key link in the informational construction and an important way to make sure the foundation pit in safety. Besides, it is the precondition in foundation pit deformation prediction by system methods as well. In order to lessen the influence of retaining structure, the monitoring scheme must be established perfectly, and the survey points should be laid rationally. For detecting the danger in time, the advanced apparatus and the reasonable alarm value must be used. The monitoring value can afford basic data for deformation prediction.2. It is strong in fault-tolerant and generalization capability of back propagation neural network. Studying enough though the original monitoring value, it can be put into predicting the deformation of foundation pit. The results of the predicting are exact enough and satisfied in engineering use.3. The predicting curves by grey system method are better in simulating the measured ones, especially when they are continual ones. But, as compared with the measured curves, the peak and the valley of the predicting ones are lag. These refers that the predicting values are bad if this method is used in predicting the immediate affairs.4. The union model of the back propagation neural network and the grey system can make use of their advantages and get a good result when used to the simples which don't have enough data. 5. Though the relative analysis among the three methods, it can be concluded that the predicting results by the methods of back propagation neural network and the union model of them are better than those by the method of grey system in most cases. Compared with the method of back propagation neural network, the union model can get a good result when used to the simples which don't have enough data. And it is more timesaving.