基于多源数据融合的空气质量二次预报模型与计算方法

Mathematical Modeling and Analysis of Air Quality Prediction Based on Multi-Source Data Fusion

以监测点长期空气质量预报基础数据为研究对象,建立了一次预报基础上的空气质量二次预报模型与计算方法。首先,利用格拉布斯准则和相关系数法对数据进行预处理,得到污染物浓度与气象条件之间的皮尔斯相关系数;然后,利用多元线性回归、ARIMA模型和随机森林模型建立了空气质量二次预报的模型与计算方法;最后,通过实例检验了二次预报模型的有效性。经检验,提出的融合多元数据的二次预报模型的精度有明显的提高,且具有很强的鲁棒性,在各个监测点的污染物预测中表现出良好的性能。

Based on the basic data of long-term air quality prediction at monitoring points,an optimized WRF-CMAQ weather prediction quadratic model is established.In order to deal with the shortcomings in the performance of WRF-CMAQ model and fully combine the correlation between meteorological conditions and pollutant concentrations,three types of models,ARIMA,multiple linear regression and random forest model,are constructed from the autocorrelation characteristics,linear and nonlinear correlation of time series between data,and the prediction results are automatically weighted to obtain the final secondary prediction model.The optimization model method is to construct the minimum error function by the difference between the measured data in the test set and the first prediction.The test shows that the accuracy of the secondary prediction model has been significantly improved.The three models have greatly improved the robustness of the model after fusion,and have good performance in all indicators of each monitoring point.