摘要
文章分析了地质灾害风险评估的基本内容,包括风险源危险性评价及受险对象易损性评价;以此为基础,结合地质灾害特征,对地质灾害风险评估指标体系、地质灾害风险评估BP神经网络的建立、训练与仿真方法进行探讨。以天水地区为例,运用MATLAB软件,调用newff命令,建立基于BP神经网络的地质灾害风险评估模型。该神经网络分3层,输入层有7个节点,输出层有3个节点,隐含层有6个节点。选择8个典型评价单元、以定性分析结果为预期输出,调用train命令对所建立的神经网络进行训练,对其余单元调用sim命令进行仿真,得出天水地区地质灾害风险评估分区图。理论分析与实例验证表明:BP神经网络具有自身调节连接权值的优点,是一种较为理想的地质灾害风险评估方法。
The paper analyzed the contents of geologic hazard risk valuation, including risk valuation of risk origin and lost degree valuation of the object threatened. Combining with the characteristic of the geologic hazard, the geologic hazard risk valuation index system, the building up, training, simulation of the geologic hazard risk valuation based on BP -neural network are discussed. Taking Tianshui area as a example, using newff order in Matlab soft, the geologic hazard risk valuation BP -neural network model is built up. The network included 3 layers, input layer has 7 nodes, output has 3 nodes, conceal layer has 6 nodes. Selecting 8 valuation units, taking qualitative analysis result as anticipative output, the nerve network is trained by train order and the other units are simulated by sim order, so the geologic hazard risk valuation zoning is gotten. The theoretical analysis and example validate result shows that BP-neural network is an ideal geologic hazard risk evaluation ; method due to its advantage of automatic adjustment connection weigh.
出处
《中国地质灾害与防治学报》
CSCD
2008年第2期100-104,共5页
The Chinese Journal of Geological Hazard and Control
基金
中国地质调查局项目(200013000154)
