【作者】 彭岩； 王万森； 涂序彦；

【Author】 Peng Yan, Wang Wan-sen Tu Xu-yan (Information Engineering College. Capital Normal University, Beijing 100037, China); (Information Engineering College, USTB, Beijing 100083, China)

【机构】 首都师范大学信息工程学院； 北京科技大学信息工程学院；

【摘要】 在预警系统中建立有效地风险预测模型是十分重要的。我国进行风险预警起步较晚,预警时有些因素带有模糊性,而神经网络带有高度并行处理信息的机制且有自学习、自适应能力,因而使系统灵活性更强。本文提出并建立了一种基于SOM的风险预警模型,同时,解决了网络训练及样本选用中的问题; 并对预警结果进行了分析。更多还原

【Abstract】 In the Risk Early-Warning System, risk prediction model is very important that, it is regarding as the core of risk early warning. Compared with traditional models, neural network has self-study, self-organizing as well as self-adaptive advantages. This paper introduced SOM neural network into the Risk Early-Warning System; The methods are improved to select sample data and train the network. Experimental resuljs show that the SOM model is accurate and can give visual prediction results as well.更多还原