摘要
将粗集理论与支持向量机结合起来,研究了针对大坝安全监测数据序列中出现小样本、短序列、不确定等情况时的监控模型,利用粗集理论中的知识约简,引入属性重要度概念,对输入数据预处理,简化大坝工作性态影响因素和效应量之间的映射关系,实现了支持向量机模型输入的优化设计,使模型更能体现大坝的工作机制。以某混凝土大坝为例,分别采用统计模型、BP神经网络模型、标准SVM模型以及RS-SVM模型进行建模分析,对比验证了RS-SVM模型方法的可行性。
In this paper,the safety monitoring model of small sample,short sequence and uncertainty in dam monitoring sequence wasresearched by combining rough sets with support vector machines. Using the knowledge reduction in rough sets and introducing the concept ofattribute importance to preprocess the input data,it simplified the relationship between influence factors and effect quantity of dam workingstate and realized the optimization design of input vector of SVM model,which could reflect working mechanism of the dam more explicitly.Taking a concrete dam as an example,the feasibility and characteristics of the RS-SVM method was verified respectively by the statisticalmodel,BP neural network model and standard SVM model in the paper.
出处
《人民黄河》
CAS
北大核心
2016年第7期130-133,共4页
Yellow River
关键词
大坝安全
支持向量机
粗集理论
监控模型
dam safety
support vector machines
rough sets
monitoring model
