Design And Realization Of The Geological Disaster Infomation System Based On Neural Network Evaluation Model

Sichuan Province is prone to geological disasters, especially after the 2008 Wenchuan earthquake. The geological disasters, which frequently happen in this region, make great impacts on the lives and economies. To reinforce the geological disaster prevention and management, the State government has investigated on the geological disasters of about 100 counties of Sichuan. Meanwhile, the Province government has also carried out several special investigations on the geological disasters. During these works, a large amount of data have been produced. It is meaningful to figure out how to dynamically manage these data and timely update the prone ares by using the modern information technology. The findings and solutions would be valuable for supporting not only the geological disaster prevention and management, but also the future investigations works.This study intended to develop a set of criteria and a model for evaluating the prone areas of geological disasters. During this study, the applications of neural network in the studies of geological disasters were reviewed. The needs and requirements of an information system of geological disasters were analysed. Based on it, this study designed the architecture, functions, and databases of such a system and then realised it. The functional and performance tests were also carried out. At last, the system’s function of prone areas evaluation was verified at two example counties by applying the actual investigation data.As conclusions, this study:1. established an evaluation model for the prone areas of geological disasters, which consists of six factors including altitude, depth of faults cut, degree of terrain slope, lithology, geological structure, and current status of geological disasters. The quantitative criteria of these six factors were developed as well.2. The input layers of the six factors were divided by rectangle grids. To meet the precision requirement, the numbers of grids for each factors were determined to be between 4,000 to 10,000.3. The radial basis function neural networks was used as the basis to develop the evaluation model. The details of a feasible evaluation method were specified.4. A stable information system of geological disasters was developed. The functions of this system include view control, graphic administration, attribute administration, data query, model analysis, etc.5. The developed system was used in several real cases to evaluate and to predict the prone areas of geological disasters. The predict results produced by the system were consistent with the facts. It showed that the model for evaluating prone areas of geology disasters developed in this study can be considered as reliable and could be used in other actual cases.Further studies could focus on,1. the robustness of the model in the situations including and excluding the predisposing factors of geological disasters, and2. the comparisons of the evaluation models developed by the method of radial basis function and other methods commonly used in developing neural networks models.