Research On Dynamic Risk Prediction For The Deep Excavations Based On ANN And GIS

With the increasing scale and complexity of the projects underground, there aremore and more challenges as well as opportunities for the excavations engineering.However, on account of conditions restricts and inappropriate management, it wouldlead to huge economic losses and negative social impacts without risks discoveringand controlling timely during the process of construction, especially for some deepexcavations engineering.In this paper, it provides some achievements for the risks dynamic predictionwith the combination of GIS and artificial neutral network as follows:Firstly, the ranges of prediction index ω, determined by the sum and rate ofdeformation and their responding warning value respectively in each monitoring items,are graded for the judgment of the risks based on the risk prediction theory as well asthe deformation of supporting structures and buildings around.Secondly, according to the actual situation, three BP neural networks aremodeled by data training and prediction through the code of Matlab and compared bythe application of the cap-beams displacement in the supporting structure when HongZecheng Project in Tian Jin started.Thirdly, in order to get the distribution of the risks markedly, the risks thematicmaps of monitoring points are distinguished between different colors combined by thefunction of Mapinfo in GIS for each monitoring project, not only helpful for plansdirectly and safety but also reducing waste under the relative safety state.Fourthly, the risk level in each monitoring items is obtained by risk weight fittingin different monitoring points. Further, dynamic risks analyses are carried out for theexcavations by fitting the influence degree of all dominant projects monitored.Finally, taking the advantage of ANN, self-configuration and self-healing as wellas the ability of nonlinear processing, and GIS, spatial distribution, models of ANNand its responding thematic maps are established respectively, combined to predict therisks of excavations with a project.