摘要
目前的大气污染程度变化及气象特征分析方法未标准化处理连续雾霾环境数据,导致其分析精度较低、相对误差较大。为此,提出连续雾霾环境下大气污染程度变化及气象特征分析方法。在预处理连续雾霾环境数据后,综合玻尔兹曼机、受限玻尔兹曼机和BP三种神经网络建立深度置信神经网络分析模型,从而分析大气污染程度变化及气象特征。实验中,对比不同方法的相对误差后发现,此次研究的连续雾霾环境下大气污染程度变化及气象特征分析方法具有较高的分析精度和较小的相对误差。
Current atmospheric pollution degree change and meteorological characteristics analysis methods are not standardized to deal with continuous haze environmental data,which leads to low analysis accuracy and large relative error.Therefore,the analysis method of atmospheric pollution degree change and meteorological characteristics under continuous haze environment is proposed.After pre-processing the data of continuous haze environment,the deep confidence neural network analysis model was established by integrating Boltzmann machine,limited Boltzmann machine and BP neural network,so as to analyze the change of air pollution degree and meteorological characteristics.After comparing the relative errors of different methods,it is found that the analysis method proposed in this study has higher analysis accuracy and smaller relative errors.
作者
邵清军
Shao Qingjun(Baiyin Meteorological Bureau, Baiyin 730900, China)
出处
《环境科学与管理》
CAS
2021年第8期82-86,共5页
Environmental Science and Management
关键词
连续雾霾
雾霾环境
大气污染
污染程度
气象特征
continuous haze
haze environment
air pollution
pollution degree
meteorological characteristics