支持向量机在凝冻日数预测中的应用

Applications of Support Vector Machine to Freezing

分析影响凝冻天气的相关因素:温度、降水量、湿度、气压和风速,对这些数据进行归一化处理,并以这些数据为输入样本,以是否有凝冻为输出样本,选取1969-2000年的数据作为训练样本,以2001-2008年的数据作为测试样本,然后分别用支持向量机和BP神经网络对凝冻天气进行识别预测。通过数据分析和实验比较,采用支持向量机对凝冻的识别率比BP神经网络更精确,这对凝冻天气的预测有很大帮助。

This paper analyzes some factors about the impact of Freezing weather： temperature, precipitation, humidity, air pressure and wind speed, then makes this data normalized, uses from 1969 to 2000 data as input data and uses from 2001 to 2008 data as the test data, finally uses support vector machine and BP neural network method to identify and forecast Freezing weather. Compared with BP neural net work predictor, experimental results show that the support vector machine predictor has higher precision.