基于深度神经网络的空气质量预测系统

Study on air quality predictions system based on deep neural network

提出了使用深度栈式自编码模型进行空气质量预测.选择了PM2.5、PM10等污染物数据作为样本.本模型基于Java平台构建,进行了训练和参数调整,建立了最优的空气预测模型.根据北京市的实验结果表明,该模型具有良好的精度.与支持向量回归(SVR)模型和线性回归模型相比,本文提出的模型具有优越的性能.

This paper proposed a deep stacker auto-encoder learning model to predict the air quality.Selected PM 2.5,PM 10 and other pollutant data as a sample.The model was constructed on the JAVA platform,and the training and parameter adjustment were carried out to establish the optimal air prediction model.The experimental results based on the Beijing region showed that the proposed method had a good accuracy.Moreover,a comparison with the support vector regression(SVR)models and linear regression(LR)models demonstrated that the proposed method of performing air quality predictions had a superior performance.