摘要
当今社会各种城市自然灾害频繁出现,灾害强度日趋加重,时刻威胁着地球上生存的人类,尤其威胁着人口密集的城市,正是由于灾害带来的破坏性,城市自然灾害损失评估具有其重大的意义。论文的核心是将数据挖掘技术中的粗糙集理论、遗传算法和神经网络相结合,尝试解决震害损失评估问题。该方法主要利用粗糙集软件Rosetta,及Matlab软件中的神经网络和遗传算法工具箱,建立了粗糙集对数据前置预处理,遗传算法优化神经网络初始权值、阈值从而利用网络进行预测评估的模型,并通过实例验证其可行。
In today＇s society, all kinds of urban natural disasters appear frequently. The intensity of disasters is increasing which threatens the survival of human beings, especially for the densely populated urban areas. Because of devastating disasters, urban natural disaster damage assessment is even more significant. The purpose of the paper is to integrate rough sets, Genetic Algorithm （GA） and neural networks of data mining technology to solve damage assessment problems. Through the utilization of rough sets software Rosetta, neural network and GA Toolbox in software Matlab, the method helps to obtain the threshold, to clarify the data preliminary process using rough sets, and to obtain the weights initialized by GA so that the network can be used to predict the model. The example was used for the verification.
出处
《工程管理学报》
2010年第1期29-32,共4页
Journal of Engineering Management
关键词
震害损失评估
粗糙集
遗传算法
BP神经网络
disaster damage assessment
rough sets
genetic algorithm
neural network
