摘要
用机器学习中有监督学习模型支持向量机SVM来进行强对流天气的识别和预报。强对流天气的发生可以看作是小概率事件,因此强对流天气的预警问题可以作为不平衡数据分类问题来处理。在SVM的应用上结合判别准则来对不平衡数据进行处理,更好的对强对流天气进行预警。本文从数据的获取、训练算法的选择、算法的应用、实验结果的评估几个方面进行了详细的描述。通过采用丹佛地区的数据进行大量试验,排除了不平衡数据对分类的干扰,提高了强对流天气识别的准确度。
The present study was designed to use a supervised learning method-support vector machines SVM of machine learning to recognize and forecast the strong convective weather. The occurrence of strong convective weather can be seen as a small probability event, so this problems can be handled as imbalanced data classification. To make better forecast,on the application of SVM we proposed a new criterion for processing data on imbalances. This paper described the algorithm in several aspects：the data obtained, the training algorithm, the application of the algorithm, the assessment results. This paper used Denver area data, eliminated the interference of imbalanced data classification, and improved the accuracy of recognition of severe convective weather.
出处
《电子设计工程》
2016年第9期4-7,11,共5页
Electronic Design Engineering
基金
国家自然科学基金(41005024)
关键词
强对流天气预警
SVM
不平衡数据分类
机器学习
SVM
strong convective weather warning
SVM
unbalanced data classification
machine learning
