基于人工智能的空气质量预测系统的开发及应用

Development and Application of Artificial Intelligence-Based System for Air Quality Prediction

文章针对传统人工神经网络在预测环境空气质量时难以挖掘数据中的内在关系并加以学习,以及收敛速度慢等问题,提出了一种基于深度学习的人工智能空气质量预测系统。该系统由空气质量预测神经网络和空气质量网格化神经网络组成。通过训练大数据来挖掘影响空气污染物浓度各因子之间的内在关系,从而提高区域空气质量预测准确性。利用人工智能技术较强的扩展性,将传统数值预报方法集成到该系统中,得到整个区域空气质量的网格化结果。将人工智能空气质量预测系统应用到唐山地区,预测了未来7 d的空气质量。结果表明,该系统做出了较为准确的预测,PM2.5的预测数据与观测数据之间的相关系数R2为0.79,预测结果的均方根误差值低于25μg/m3,较好地反映了PM2.5每小时的变化趋势。在环境大数据背景下,该研究技术方法为大气污染防控决策提供了有效支撑。

An artificial intelligence-based system for air quality prediction is proposed to solve the problem that the traditional artificial neural networks have difficulty getting the internal relationships in the data,as well as the low speed of convergence.The system consists of neural networks for air quality prediction and grid-based air quality.The internal relationships between factors affecting air pollutant concentration were analyzed by training big data,and thereby improving the accuracy of regional air quality prediction.Traditional numerical forecasting method was integrated into the system by the strong scalability of artificial intelligence technology.And the gridded results for air quality across the region was obtained.The artificial intelligence-based air quality prediction system was applied to the Tangshan city.The results showed that the R2 value of PM2.5 between the observed and predicted data was 0.79,and the root mean square error was less than 25μg/m3,which reflects the hourly change trend of PM2.5.In the context of environmental big data,the technical approach of this study provides effective support for air pollution prevention and control decision making.