Research On Calculation Of The Effective Reinforce Depth Of Dynamic Compaction By Use Of BP Network Model

For a long time, dynamic compaction is widely used because of its superiority. However, due to strengthening mechanism is very complex, it makes calculation of the effective reinforcement depth which is based on the theoretical results be difficultly to achieve. Most of the existing methods with strong empirical. Simplifying assumptions and the evils affecting the popularization and application of them. At present, how to use the data accumulated in engineering practice, to achieve the most optimized designment of construction by using effective reinforcement depth as the evaluation,it has become into an important content of study for dynamic compaction.This paper analyzes the nonlinear relationship between the effective reinforcement depth and impact factors, introduces the basical features of artificial neural network. It poposals the technical feasibility of the program which building a BP neural network model to forecast the effective reinforcement depth. The BP network model established ultimately provides some help for selecting construction parameters of dynamic compaction.The articles include the following:1. It describs the mechanism of dynamic compaction in foundation treatment, clears the background, significance and current development status for the prediction of effective depth of dynamic compaction.2. With the numerous collections for the data of examples of dynamic compaction as a foundation treatment method,it analysises the influence on reinforcement depth which the construction parameters and the foundation soil parameters affect.3. By analyzing test data of the new logistics center project in Dalian Railway Container field by detection of dynamic consolidation, show that the dynamic compaction has an obvious effect on this kind of soil in foundation treantment project. It can get higher bearing capacity of foundation, and improves the deformation of soil.4. Analysises the characteristics of existing calculation methods,and preparates a computing software; compares calculation accuracy of common method by using the measured data from the Dalian project.; proposes the superiority of calculation for artificial neural network.5. Eigenvectors of the network sample selection and network setting options, implements network training and the testing process for testing samples. Final results shows that that the BP network model constructed for the prediction of effective depth of dynamic compaction have a good reliability.