• ▶ 2008-2024年被中国情报信息研究所评价中心评为“中国科技核心期刊”
  • ▶ 2019-2024年连续三届被中国科学院文献情报中心中国科学引文数据库CSCD(核心库)收录
  • ▶ 2021、2023年入编北京大学图书馆《中文核心期刊要目总览》
  • ▶ 2020-2024连续四年入选《科技期刊世界影响力指数(WJCI)报告》
樊广利, 曹红业, 徐 晋.基于HJ-1A CCD影像和ELM模型的太湖叶绿素a预测研究水资源与水工程学报[J].,2020,31(5):16-22
基于HJ-1A CCD影像和ELM模型的太湖叶绿素a预测研究
Prediction of chlorophyll a in Taihu Lake based on HJ-1A CCD imagery and ELM model
  
DOI:10.11705/j.issn.1672-643X.2020.05.03
中文关键词:  叶绿素a预测  HJ-1A CCD影像  极限学习机(ELM)  内陆湖泊  太湖
英文关键词:prediction of chlorophyll a  HJ-1A CCD imagery  extreme learning machine (ELM)  inland lake  Taihu Lake
基金项目:
作者单位
樊广利1,2, 曹红业3, 徐 晋2 (1.西北大学 城市与环境学院 陕西 西安 710127 2.西京学院 土木工程学院陕西 西安 710123 3.长安大学 地质工程与测绘学院 陕西 西安 710064) 
摘要点击次数: 1376
全文下载次数: 519
中文摘要:
      以典型二类水体——太湖为例,基于环境一号遥感影像,构建了基于 ELM模型的叶绿素a浓度预测模型,将预测结果与传统的BP人工神经网络和支持向量机 SVM进行了比较。研究结果表明:ELM模型预测值与实测值之间的R2高达0.911 4,而BP和SVM模型的R2分别为0.366 3和0.744 8,均方根误差RMSE由BP模型和SVM模型的3.728 8 μg/L和2.132 4 μg/L降为ELM模型的1.327 0 μg/L, ELM模型的平均相对误差MRE=2.65%,小于BP模型的6.59%和SVM模型的3.89%;与其他两种方法相比,ELM模型反演太湖水体叶绿素a浓度精度更高,ELM模型参数选择简单,可以显著提高模型的学习速度,不易陷入局部最优值,具有更好的泛化性能;ELM模型可以有效地应用于内陆水体叶绿素a浓度的预测。
英文摘要:
      Taking Taihu Lake as an example, a prediction model of chlorophyll a concentration based on extreme learning machine (ELM) model was constructed for the first time using the remote sensing images of Environment 1. The prediction results were then compared with those of the traditional BP artificial neural network and support vector machine (SVM) model. It was found that R2 of ELM model reached as high as 0.911 4; however it only reached 0.366 3 of BP and 0.744 8 of SVM model. Root mean squared error (RMSE) of BP model and SVM model reduced from 3.728 8 μg/L and 2.132 4 μg/L to 1.3270 μg/L of ELM model. The mean relative error (MRE) of ELM model was 2.65%, which was lower than that of BP model (6.59%) and SVM model (3.89%). Compared with these two models, the ELM model has higher accuracy in the back analysis of chlorophyll a concentration in Taihu Lake. The parameter selection of ELM model is simple, which can significantly improve the learning speed of the model. This model has better generalization performance and it is not prone to local optimum. The experiment results show that ELM model can be effectively applied to the prediction of chlorophyll a concentration in inland lakes.
查看全文  查看/发表评论  下载PDF阅读器
关闭