基于 HY-FMV 模型的雷暴天气预测研究

本文以极端天气中的雷暴天气为研究对象,基于历史气象数据预测未来三小时是否发生雷暴。为预测雷暴是否发生,本文分别对极端天气气象数据的采样、数据预处理、特征选择,以及建模分析进行了研究,最终提出一种基于机器学习方法的HY-FMV模型框架对雷暴天气进行预测。该模型采用混合模型进行数据预处理,基于概率分布与模型评价进行特征的选择和构建,并使用梯度提升树算法对极端天气进行预测分类。最后,本文以2010年到2015年福建和广东两省数据为例,分别使用本文所提出的HY-FMV模型,和随机森林算法等进行雷暴天气预测,结果表明,本文所提出的HY-FMV模型在F1指标上精度达到78%,相比其他算法,在雷暴天气预测精度上提高了0.5%-0.6%。

Research on An HY-FMV Model for Thunderstorm Weather Forecasting

This paper takes the thunderstorm weather in extreme weather as the research object. Based on historical meteorological data, whether thunderstorms occur in the next three hours is forecasted. A HY-FMV ((Hybrid Fill Missing Value) model is proposed to forecast thunderstorm weather and the effectiveness of the model is verified by experiments. A hybrid model is used to preprocess the data. Features are selected and constructed based on probability distribution and model evaluation. And gradient boosting decision tree algorithm is used to predict and classify extreme weather. In the end, taking extreme weather data of Fujian and Guangdong provinces from 2010 to 2015 as experimental data, the HY-FMV model proposed in this paper and random forest model were used respectively to forwecast thunderstorms. Results show that the accuracy of the HY-FMV model proposed in this paper reaches 78% on the F1 index. Compared with other models, the F1 index of thunderstorm weather forecast has increased by 0.5%-0.6%.