基于大数据分析的气象观测数据质量控制算法研究  

Research on meteorological observation data quality control algorithm based on big data analysis

作　　者：尤嘉铖 YOU Jiacheng(CAAC East China Regional Administration,Shanghai 200335,China)

机构地区：[1]中国民用航空华东地区空中交通管理局,上海200335

出　　处：《电子设计工程》2022年第11期103-107,共5页Electronic Design Engineering

基　　金：国家自然科学基金委员会与中国民用航空局联合基金(U1511328)。

摘　　要：目前,常用的气象数据质量控制算法均是依据统计学原理对数据的变化趋势进行分析,针对其算力无法对大数据进行处理以及实现质量控制的问题。文中将BP神经网络模型与粒子滤波算法结合,有效增强了神经网络的综合计算性能,将该算法应用于海量气象数据处理,可以显著减少外在因素对数据精确度的影响。实验结果表明,所提模型的预测数据误差平均值和异常数据监测误差平均值在对比算法中最小,证明该模型可以准确地监测气象数据,同时能够对异常气象数据进行修补拟合,进而实现对气象数据的质量控制。At present,the commonly used meteorological data quality control algorithms are based on the statistical principle to analyze the change trend of the data,aiming at the problem that its computational power can not process the big data and realize the quality control.In this paper,the BP neural network model is combined with particle filter algorithm,which effectively enhances the comprehensive computing performance of neural network.The application of this algorithm to massive meteorological data processing can significantly reduce the impact of external factors on data accuracy.The experimental results show that the average error of prediction data and the average error of abnormal data monitoring of the proposed model are the smallest in the comparison algorithm,which proves that the model can accurately monitor the meteorological data,and can repair and fit the abnormal meteorological data,so as to realize the quality control of the meteorological data.

关 键 词：气象数据 数据质量控制 BP神经网络 粒子滤波 数据关联度 自动气象站 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]