| 基于自编码网络的空气污染物浓度预测 |
| |
| 引用本文： | 秦东明,丁志军,金玉鹏,赵勤. 基于自编码网络的空气污染物浓度预测[J]. 同济大学学报(自然科学版), 2019, 47(5): 0681-0687 |
| |
| 作者姓名： | 秦东明 丁志军 金玉鹏 赵勤 |
| |
| 作者单位： | 同济大学 嵌入式系统与服务计算教育部重点实验室，上海 200092,同济大学 嵌入式系统与服务计算教育部重点实验室，上海 200092,上海师范大学 信息与机电工程学院， 上海 200234,上海师范大学 信息与机电工程学院， 上海 200234 |
| |
| 基金项目： | 国家自然科学(61572326, 61702333, 61772366),上海市自然科学基金(18ZR1428300),上海市科委创新项目(17070502800, 16JC1403000),上海市教委项目(C160049), 嵌入式系统与服务计算国家教育部重点实验室开放课题(项目编号ESSCKF 2016-01) |
| |
| 摘 要： | 深度学习为城市空气污染物浓度预测提供了更为强大的数据拟合能力,为空气污染预测提供全新的智能计算方法.为此,提出了一个基于自编码神经网络的污染物浓度预测模型AEPP(auto-encoder-based pollutant prediction).该模型包括编码器和解码器两个部分.其中,编码器用于提取出时间序列污染物浓度数据分布特征,即语境向量;解码器利用提取的特征预测未知时间内污染物浓度数据.模型中编码器和解码器采用多层LSTM(long short-term memory)模型结构,实现长时间依赖预测目标.实验表明,提出的模型可以提高对污染物浓度的预测水平.
|
| 关 键 词： | 空气污染预测 自编码模型 深度学习 数值分析 |
| 收稿时间： | 2018-07-04 |
| 修稿时间： | 2019-03-18 |
An Air Pollutant Prediction Model Based on Auto-Encoder Network |
| |
| QIN Dongming,DING Zhijun,JIN Yupeng and ZHAO Qin. An Air Pollutant Prediction Model Based on Auto-Encoder Network[J]. Journal of Tongji University(Natural Science), 2019, 47(5): 0681-0687 |
| |
| Authors: | QIN Dongming DING Zhijun JIN Yupeng ZHAO Qin |
| |
| Affiliation: | Key Laboratory of Embedded System and Service Computing of the Ministry of Education, Tongji University, Shanghai 200092, China,Key Laboratory of Embedded System and Service Computing of the Ministry of Education, Tongji University, Shanghai 200092, China,College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China and College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 200234, China |
| |
| Abstract: | In this paper, an autoencoder-based pollutant prediction (AEPP) model is proposed based on the auto-encoder neural network, which is composed of an encoder and a decoder. First, the encoder extracts the distribution characteristics of the time series of pollutant concentration data, namely the context vector. Secondly, the decoder uses the extracted characteristics to predict the pollutant concentration data in the next unknown time. Both the encoder and the decoder in the model can adopt several LSTM structures for long time prediction. Experiments show that the AEPP model proposed in this paper can improve the effect of pollutant prediction. |
| |
| Keywords: | air pollutant prediction auto-encoder model deep learning numerical analysis |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《同济大学学报(自然科学版)》浏览原始摘要信息 |
|
点击此处可从《同济大学学报(自然科学版)》下载全文 |
|
