摘要
空气质量预测能够预知区域空间内的大气污染物浓度,对污染防治、环境保护和人身健康等具有非常重要的意义。针对现有空气污染物预测模型未能充分挖掘和利用上下文因素的影响和作用,提出了一种上下文特征注入的空气污染物预测模型。首先,通过循环神经网络和深度置信网络分别学习和提取空气污染物浓度数据的时间序列特征和上下文特征。然后,使用向量融合机制将提取到的上下文特征注入到时间序列特征中,生成新的融合特征。最后,将新的高阶融合特征送入预测器,对空气污染物浓度做出准确可靠的预测。实验选用2017年1月至2021年7月共55个月的PM2.5污染物浓度数据,并与LSTM、GRU、BiLSTM预测模型相比较,结果表明提出的特征注入模型在多种场景下都能够准确地拟合空气污染物浓度的真实值,预测精度优于传统循环神经网络模型,各项评价指标均较好,表现出较强的适应性和准确性。
Air quality prediction can predict the concentration of air pollutants in regional space,which is of great significance to pollution prevention,environmental protection and human health.Aiming at the failure of existing air pollutant prediction models to fully exploit the influence and function of context factors,an air pollutant concentration time series prediction model based on contextual features injection was proposed.Firstly,the time series features and context features of air pollutant concentration data were learned and extracted through recurrent neural networks and deep belief networks,respectively.Then,the vector fusion mechanism was used to inject the extracted context features into the time series features to generate new fusion features.Finally,the new high-level fusion features were sent to the predictor to make an accurate and reliable prediction of the future air pollutant concentration.55 months of PM2.5 pollutant concentration data from January 2017 to July 2021 were selected for the experiment,and compared with LSTM,GRU and BiLSTM prediction models.It is showed that the proposed feature injected model can accurately approximate the true value of the air pollutant concentration in a variety of scenarios,the prediction accuracy is better than that of traditional cyclic neural network models,and all evaluation indicators are the best,showing strong adaptability and accuracy.
作者
魏思
李欣泽
郤丽媛
刘紫君
董哲为
WEI Si;LI Xin-ze;XI Li-yuan;LIU Zi-jun;DONG Zhe-wei(School of Information Engineering,Chang’an University,Xi’an 710064,China;School of Economic and Management,Shaanxi Xueqian Normal University,Xi’an 710100,China)
出处
《计算机技术与发展》
2023年第9期196-201,共6页
Computer Technology and Development
基金
国家重点研发计划(2019YFE0108300)
国家自然科学基金项目(62001058)
陕西省自然科学基础研究计划资助项目(2020JM-258)。
