摘要
针对空气质量预测中复杂的时空问题,本文构造了多站点间的交互时空特征,搭建了结合CNN和LSTM的深度时空模型,并引入注意力机制学习多特征之间的权重分布,找出对空气质量指数(AQI)影响较大的特征重点关注,构造了融合CNN-LSTM和注意力机制的AQI预测模型。使用2019年1月至2020年12月间运城市各站点的小时粒度数据进行实验,结果表明,该模型对空气质量指数的预测较基模型具有更优的性能。
Aiming at the complicated spatiotemporal problems in air quality prediction,this paper constructs interactive spatiotemporal features between multiple sites,builds a deep spatiotemporal model combining CNN and LSTM,and introduces an attention mechanism to learn the weight distribution between multiple features to find out the features that have a greater impact on the quality index,so as to construct an AQI prediction model combining CNN-LSTM and attention mechanism.Experiments were conducted using hourly granularity data of various stations in Yuncheng City from January 2019 to December 2020.The results show that the model has better performance in predicting air quality index than the base model.
作者
刘媛媛
Liu Yuanyuan(Department of Mathematics and Information Technology,Yuncheng University,Yuncheng,Shanxi 044000,China)
出处
《计算机时代》
2022年第1期58-60,共3页
Computer Era
基金
运城学院应用研究项目(YQ-2020019)。
