摘要
以闽江河口为例,采用BP神经网络和支持向量机两种方法分别构建河口盐度预测模型,比较分析两种方法在河口盐度预测的精度和适用性。研究结果表明:(1)两种模型预测结果均能较好地实现河口盐度预测,支持向量机具有更好的泛化性能和适用性。(2)两种模型对低盐度都具有较好的预测精度,支持向量机在高盐度预测方面优势较为明显。(3)当样本数量较小时,支持向量机预测结果精度较好,两种模型的预测差异随着样本量增加逐渐减小。基于支持向量机河口盐度预测模型更适用于河口盐度预测。
In order to compare and analyze the accuracy and applicability of Neural network and support vector machine methods in estuary salinity prediction,this paper applied back propagation neural network(BPNN)and support vector machine(SVM)to construct estuary salinity prediction model respectively with Minjiang River estuary as example.The results show that:(1)The two models can successfully predict the estuary salinity with satisfactory precision and SVM has better generalization performance and applicability.(2)The two models have good accuracy in predicting low salinity,but SVM has strong advantage in predicting high salinity.(3)The prediction accuracy of SVM is better as the samples are small,but the difference between the two models decreases with the samples rise.The prediction model of estuary salinity based on SVM is more suitable for the prediction analysis of estuary salinity.
作者
方艺辉
陈兴伟
FANG Yihui;CHEN Xingwei(School of Information Engineer,Fujian Business University,Fuzhou 350506,China;School of Geographic Science,Fujian Normal University,Fuzhou 350007,China;Cultivation Base of State Key Laboratory Humid Subtropical Mountain Ecology,Fuzhou 350007,China)
出处
《水文》
CSCD
北大核心
2022年第5期51-55,共5页
Journal of China Hydrology
基金
福建省中青年教师教育科研项目(JAT190496)
国家自然科学基金资助项目(41877167)。
关键词
BP神经网络
支持向量机
数据驱动模型
盐度预测
咸潮入侵
闽江河口
back propagation neural network
support vector machine
data-driven model
salinity prediction
salt tide invasion
Minjiang River estuary