不同模型在南宁市年降水量预测中的应用比较  

Comparison of Different Models in Annual Precipitation Prediction in Nanning City

作　　者：杜懿[1,2,3] 麻荣永 赵立亚[1,2,3] DU Yi;MA Rongyong;ZHAO Liya(College of Civil Engineering and Architecture,Guangxi University,Nanning 530004,China;Guangxi Key Laboratory of Disaster Prevention and Engineering Safety,Guangxi University,Nanning 530004,China)

机构地区：[1]广西大学土木建筑工程学院,广西南宁530004 [2]广西大学工程防灾与结构安全教育部重点实验室,广西南宁530004 [3]广西大学广西防灾减灾与工程安全重点实验室,广西南宁530004

出　　处：《人民珠江》2018年第4期1-4,共4页Pearl River

基　　金：国家自然科学基金项目(51369005);广西防灾减灾与工程安全重点实验室系统性研究项目(2013ZDX04)

摘　　要：以南宁市1961—2015年降水量资料为基础,通过建立灰色GM(1,1)模型、滑动平均改进的GM(1,1)模型以及改进的BP神经网络模型来对南宁市年降水量进行预测研究。结果表明,以上模型均适用于南宁市的年降水量预测,模型精度分别达到了92.2%、96.6%和95.1%。其中以改进的灰色模型与BP神经网络模型预测效果最佳,平均相对误差均控制在了5%以内,充分表明预测模型的适用性与合理性。该研究为城市降水量预测提供了一种新途径,对城市水资源合理规划具有一定的指导意义。In this paper,GM(1,1)model,improved GM(1,1)model by moving average and improved BP neural network model were used to predict the annual precipitationin Nanning city based on the precipitation data from 1961 to 2015.The results showed that the above models were applicable to the annual precipitation predictionin Nanning city,and the models accuracy reached 92.2%,96.6%and 95.1%,respectively.The improved GM(1,1)model and BP neural network model had the best predictive effect,and the average relative error was under 5%,which fully indicated the applicability and rationality of the prediction model.This study provided a new way for city precipitation prediction and had guiding significance for city water resources planning and management.

关 键 词：年降水量 预测研究 灰色模型 滑动平均 神经网络 

分 类 号：TV121.[水利工程—水文学及水资源]