Correction method for air quality forecast

The invention discloses a correction method for air quality forecasting, which comprises the following steps of: inputting air quality historical data and meteorological data of all monitoring stations in a target area for training on the basis of four existing air quality forecasting modes; a random forest algorithm, an extreme random tree algorithm and a gradient lifting regression tree algorithm are used for carrying out first optimization on four single-mode prediction results, then a BP neural network is used for carrying out second optimization, and finally a set mode correction prediction result is obtained through weighted average. According to the method and the device, through incorporating forecasting data of different air quality modes, different nested regions and different time efficiencies, the number of set members is increased as much as possible, and the forecasting result is optimized to the maximum extent on the existing system. 本发明公开了一种用于空气质量预报的订正方法，基于已有的4种空气质量预报模式，输入目标区域内所有监测站点空气质量历史数据与气