作　　者： (文琴）; (罗飞）;

机构地区： 成都信息工程大学软件工程学院,成都610225

出　　处： 《四川理工学院学报（自然科学版）》 2017年第4期96-100,共5页

摘　　要： 为了更准确地找出影响空气质量指数的气象因子与提高其预测精度,提出了基于熵、BP神经网络和时间序列模型的组合预测模型。该方法利用增加了特征变量的转移熵方法,得到影响AQI的气象因子及其影响度,将得到的气象因子与AQI实测值作为BP神经网络的输入因子和时间序列分析模型的特征因子,影响度作为BP神经网络输入因子的初始权重,构建BP神经网络预测模型和时间序列分析预测模型,最后用熵值法组合各个预测模型的预测结果。实验表明利用该方法对空气质量指数进行预测可提高其预测精度。 In order to accurately extract the meteorological factors that affect the air quality index and improve the prediction accuracy,a prediction model based on entropy,BP neural network and time series model is proposed. This method uses the information transfer entropy with the characteristic variables to obtain the characteristic factor and the specific influence degree. The obtained characteristic factor and measured values of AQI are used as the input factor of the BP neural network and the characteristic factors of the time series analysis model,the influence degree is the initial weight of the BP neural network,construct BP neural network and time series analysis model,finally,the results of each prediction model are composed by the entropy method. The experiment shows that the This method can improve the stability and the predict accuracy of the forecast of air quality index.

关 键 词： 空气质量指数预测 转移熵 熵值法