摘要
降雨量预测对于水资源的管理非常重要,可以帮助决策者提前做出应对举措,降低灾情发生时带来的经济损失和人员伤亡。同时,降雨量预测对人们的日常生活、出行等也有着非常重要参考意义。通过分类回归树算法构建两个预测降雨量的模型,然后通过粒子群算法对模型中的参数进行优化。此外,为解决原算法不具备处理数据流问题的能力,根据dsCART算法的思想,对原算法生成决策树的过程做出了调整,使其具有增量学习的能力,提高其在气象信息系统中的实用性。最终通过实验验证了该改进方法的可行性、有效性。
Rainfall forecasting is very important for water resources management,which can help decision makers to take measures in advance to reduce the economic losses and casualties caused by disasters.At the same time,rainfall forecasting is of great significance to people's daily life,travel and so on.In this paper,two models for predicting rainfall are constructed by means of the classified regression tree algorithm,and the parameters in the models are optimized by means of the particle swarm optimization(PSO).In addition,the process of generating decision tree on the basis of original algorithm is adjusted according to the thought of the dsCART algorithm,which makes it have the ability of incremental learning and improve the practicability in the meteorological information system,so as to solve the problem that the original algorithm does not have the ability to deal with data flow.Finally,the feasibility and effectiveness of the improved method are verified with the experiments.
作者
李正方
杜景林
周芸
LI Zhengfang;DU Jinglin;ZHOU Yun(Nanjing University of Information Science&Technology,Nanjing 210044,China)
出处
《现代电子技术》
北大核心
2020年第2期133-137,141,共6页
Modern Electronics Technique
基金
国家自然科学基金(41575155)
关键词
降雨量预测
CART算法
粒子群优化算法
增量学习
性能评价
实验验证
rainfall forecast
CART algorithm
particle swarm optimization algorithm
incremental learning
performance evaluation
experimental verification
