首页|长短时记忆神经网络模型在空气污染预测中的研究

长短时记忆神经网络模型在空气污染预测中的研究

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
基于K均值(K-Means)聚类算法进行聚类分析，将气象条件分为三类，并且分析和阐述各类气象条件的特征。针对气象监测数据和空气污染物的时间序列特点，设计基于长短时记忆(LSTM)神经网络的空气污染预测模型。将时空相关性与长短时记忆神经网络算法进行有效的融合，提出基于时空相关性的长短时记忆(SK-LSTM)神经网络的空气污染预测模型。通过空间划分，空间聚集以及空间插值，获得目标区域和周围区域的历史空气质量检测数据和历史气象监测数据，然后通过等权融合方法将时间数据和空间数据进行融合，并将其作为SK-LSTM神经网络算法的输入，最终输出的结果为带有区域协调的污染物浓度预测值。该算法能有效对空气中污染物的浓度进行更准确、高效的预测。最后通过数值仿真验证所提算法的有效性。
Research on Long Short-Term Memory Neural Network Model in Air Pollution Prediction
Based on the clustering analysis utilizing K-means clustering algorithm,we di-vide the meteorological conditions into three categories,and then analyze and describe the characteristics of the three categories.According to the meteorological monitoring data and time series characteristics of air pollutants,we design an air pollution prediction model based on long short-term memory(LSTM)neural network.Then a SK-LSTM algorithm based on spatio-temporal correlation is proposed by combining spatio-temporal correlation with LSTM algorithm.After spatial partitioning,spatial aggregation and spatial interpolation,we obtain the historical air quality detection data and historical meteorological monitoring data of the target area and the surrounding area.Then using the equal-weight fusion method,we fuse the temporal data and the spatial data,and use it as the input of the SK-LSTM neural network algorithm.The final output is the predicted value of pollutant concentration with regional coordination.The algorithm can effectively predict the concentration of pollutants in the air,accurately and efficiently.Finally,the effectiveness of the proposed algorithms is verified by numerical simulation.

K-Means algorithmlong and short term memory neural networkair pollution forecasttemporal correlation

周宁、李铁军、郝崇清、路继勇、刘朝阳

展开 >

河北科技大学 电气工程学院,河北 石家庄 050018

K-Means聚类算法 长短时记忆神经网络 空气污染预测 时空相关性

河北省科技厅自然科学研究项目面上项目河北省教育厅基础研究重点培育专项河北省科技厅科技重大专项河北省科技厅重点研发项目河北科技大学基础研究质量提升专项(重点)河北科技大学博士科研基金

F2022208007JZX202300522280802Z23311803D2021YWF121181439

2024

数学的实践与认识
中国科学院数学与系统科学研究院

数学的实践与认识

CSTPCD北大核心
影响因子：0.349
ISSN：1000-0984
年,卷(期)：2024.54(4)
  • 30