基于RBF神经网络滑坡灾害发生概率预报方法

摘要

为了解决滑坡传统预报中参数单一问题,基于RBF神经网络,采用MIV算法进行成灾因子的筛选;通过因子的历史数据训练、泛化建立模型;最后,模型经过自学习功能,输出滑坡成灾概率结果,并与预警等级相结合进行预报功能.以汉阴县滑坡灾害为例验证模型的预测结果.结果表明,预测与实际结果吻合度达到91.12%,表明该方法具有一定的可行性.%In order to solve the problem that the parameter of landslide prediction is onefold in the traditional forecasting methods,MIV algorithm is used to select catastrophic factors based on RBF neural network algorithm.Then the prediction model of landslide is established through training,and generalizing the historical data of selecting factors.Finally,the probabil-ity of landslide disaster is output through the model's self-learning function,and the forecas-ting function is realized according to the warning grade.The model is used in Hanyin county, and the result shows that the agreement between model predictions and actual result reaches 91 .12%,and the feasibility of the method is thus verified.