卡尔曼滤波算法在空气质量数据校准中的应用

针对自建点实时监控存在的零点漂移和量程漂移导致空气质量数据测量精度受到制约的问题,提出一种基于卡尔曼滤波算法对空气质量数据进行校准的模型。以PM2.5数据为例,结合其他空气质量数据及气象参数经逐步回归得到调整的数据后,使用改进的卡尔曼滤波算法对自建点测量数据进行校准。经同一时间区间内的自建点、国控点测量及校准数据的对比分析表明:通过改进的卡尔曼滤波校准空气质量数据能够很好地跟踪实际数据的变化趋势,平均误差比较小;使用模型将2019年5月、2018年12月数据的均方根误差分别从11.367、27.188降低至修正后的7.555、10.759,精度得到了显著提高。

Application of Kalman Filter Algorithm in Air Quality Data Calibration

Aiming at the problem that the measurement accuracy of air quality data is restricted due to zero drift and range drift in the real-time monitoring of self-built points, a model of air quality data calibration based on Kalman filtering algorithm is proposed. Taking PM2.5 data as an example, the improved Kalman filtering algorithm was used to calibrate the measured data in the self-built points after combining other air quality data and meteorological parameters adjusted by stepwise regression. The comparative analysis of the self-built points, national control points and calibration data in the same time interval shows that the improved Kalman filter can well track the variation trend of the actual data and the average error is relatively small. The root-mean-square error of the data in May 2019 and December 2018 was reduced from 11.367 and 27.188 respectively to 7.555 and 10.759 after correction, indicating that the precision of the calibration results was significantly improved.