关于气象数据信息对降雨等级准确预测仿真  

Accurately Forecast Simulation of Rainfall Level with Meteorological Data

作　　者：王军[1,2] 彭明 程勇[2] WANG Jun;PENG Ming;CHENG Yong(College of Computer ＆ Software,Nanjing University of Information Science ＆ Technology,Nanjing Jiangsu 210044,China;Information Construction ＆ Management Department,Nanjing University of Information Science ＆ Technology,Nanjing Jiangsu 21004,China)

机构地区：[1]南京信息工程大学计算机与软件学院,江苏南京210044 [2]南京信息工程大学信息化建设与管理处,江苏南京210044

出　　处：《计算机仿真》2018年第9期445-449,479,共6页Computer Simulation

基　　金：国家自然科学基金资助项目(61373064;61402236);赛尔网络下一代互联网技术创新项目(NGII20160318);江苏省"六大人才高峰"项目(2013-DZXX-019;2015-DZXX-015);江苏省"六大人才高峰"创新人才团队项目(TD-XYDXX-004)

摘　　要：降雨是众多气象要素相互作用的结果,气象数据具有类型复杂、不定性、产生速度快、数据量大等特点,增大了降雨预测模型处理气象数据的难度。针对降雨预测方法要求气象要素之间相互独立,但很多气象要素相互关联,降低了预测的准确性和效率的问题,为了提高降雨等级的预测准确性与效率,提出一种基于Map Reduce的改进加权朴素贝叶斯算法(WMNBC)。首先根据气象数据相应条件属性、类属性及其它所有条件属性三者之间的相关程度,提出一种信息密度的概念,并结合对应条件属性取值数目设计一个新的属性赋权方法,然后建立基于MapReduc的WMNBC降雨等级预测模型,最后对实际降雨等级进行仿真测试。仿真结果表明,相比NBC和HWNN算法,WMNBC算法可以更加准确的预测降雨等级并提高预测效率,为降雨等级预测提供了新的途径。Rainfall is the result of the interaction of many meteorological elements, the meteorological data are of the character of complex type , uncertainty, fast producing, and large amount of data, which increases e the difficulty of using weather forecasting model to deal with meteorological data. The method of rainfall forecasting requires the meteorological data to be independent with each other, but the meteorological data are interrelated, which reduces the accuracy and efficiency of the forecast. In order to improve the accuracy and efficiency of rainfall classification forecasting, an improved weighted naive Bayesian classification algorithm was proposed based on MapReduce. Firstly, a concept of information density was proposed based on the meteorological factors corresponding condition attributes, class attributes and all other conditions of the relationship between the three attributes, and a new attribute empowerment method was designed according to the number of corresponding attribute values. Then the WMNBC rainfall level forecasting model based on MapReduc was built. Finally, the actual rainfall level simulation test was carried out. Simulation results show that compared with NBC and HWNN algorithm, WMNBC algorithm can forecast the rainfall level more accurately and improve the forecasting efficiency, which provides a new way for rainfall grade forecasting.

关 键 词：降雨等级 朴素贝叶斯分类 信息密度 属性赋权 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]