摘要
针对稻瘟病气象预警中样本含有模糊信息,支持向量机对含有模糊信息样本无法处理的问题,建立适合稻瘟病气象预警特点的分类预警算法(强模糊支持向量机)。以模糊事件的可信性测度为基础,将模糊分类问题转化为求解模糊机会约束规划问题;将模糊机会约束规划化转化为与其等价的二次规划,据此给出强模糊支持向量机。并且研究了强模糊支持向量机在稻瘟病气象预警中的应用方法。对浙江省宁波市某水稻种植区2004—2007年稻瘟病气象预警试验,数据结果与实际情况吻合。由此可说明强模糊支持向量机能较好地解决样本中含有模糊信息的分类问题,基于强模糊支持向量机的稻瘟病气象预警方法对于稻瘟病气象预警有较大的优越性。
For a support vector machine can not deal with samples of fuzzy information contained in the weather early warning of rice blast,the classification method for early warning is constructed to meet the features of the weather early warning of rice blast in this work.Based on the creditability measure of fuzzy event,fuzzy classification problems can be transferred into solving the fuzzy chance constrained programming problem.Fuzzy chance constrained programming is transferred into its equivalent quadratic programming and strong fuzzy support vector machine is developed,Application of strong fuzzy support machine in the weather early warning of rice blast is studied.According to the weather early warning experiment in rice-growing areas in Ningbo City,Zhejiang province,the numerical results are fit closely to the actual results.The support vector machine can deal with the classification problems well for samples of fuzzy information and the method of weather early warning of rice blast based on strong fuzzy support vector machine has greater superiority over weather early warning of rice blast.
出处
《中国农业大学学报》
CAS
CSCD
北大核心
2010年第3期122-128,共7页
Journal of China Agricultural University
基金
国家自然科学基金资助项目(10926198)
国家"十一五"科技支撑计划项目(2006BAJ07B09)
浙江省自然科学基金资助项目(Y606082)
关键词
稻瘟病
预警
机器学习
强模糊支持向量分类机
可信性测度
rice blast
early warning
machine learning
strong fuzzy support vector classification
creditability measure
