Mearsured Record Analysis Method Applies In Predicting Of Monitoring Deep Foundation Excavation

With the rapid development of Chinese economy, the construction of infrastructure steps into its peak. Being part of architectural engineering, deep foundation excavation is one task people will frequently be confronted. The excavation of deep foundation will unavoidably caused deformation, the monitoring and predicting of which is the most important section of information construction, is one of the important questions to be solved in the construction of deep foundation engineering. However, owing to the deficiencies of traditional theoretical calculation method, and the wide-spread finite-element method, such a complicated non-linear problem can not be solved easily. How to explore a scientific, accurate way to forecast the deformation of deep-foundation possesses significant engineering and practical meaning.Based on the researches of predecessors, this paper gives a detailed analysis the fundamental theories of grey theory model (GM) and grey Markov chain model (SCGM) in measured record analysis methods, and on the basis of this it presents relative predicting models. After the brief introduction of the development of artificial neural network and its application in engineering , the paper penetrates the structure, characteristics and learning algorithms of artificial neural network, and sets up the BP neural network model to predicting the deformation of deep foundation.In the end, with the help of one engineering example, on the basis of MATLAB 7.0 tool box, the paper programs the predicting procedure of the theoretical methods. With the comparison and analysis of different controlling parameter segments and training function selected, the papers manifests that it is feasible to apply BP artificial neutral network model to the predicting of deformation of deep foundation, which can effectively forecasts the future development tendency. With littler monitoring statistics and under strict safety control, GM and SCGM models can be used as a supplementary deformation-predicting device.