福海叶绿素含量的人工神经网络反演模型

The Neural Network Model on Chlorophyll Concentration in the Fuhai Lake

在圆明园福海同时进行光谱测量与叶绿素测定。使用传统经验统计算法、BP人工神经网络和径向基函数(RadialBasisFunction,RBF)网络,分别建立光谱值反演叶绿素浓度模型。结果表明,径向基函数网络具有结构自适应确定、输出不依赖初始权值、速度快、结果可靠、能充分利用光谱信息等优良特性,得到了相当好的拟合结果,是一种值得推广的预测叶绿素浓度的神经网络模型。

The spectral reflectance and concentration of chlorophyll-a are measured instantaneously in the Fuhai Lake.The authors apply empirical algorithms, BP network and RBF network to simulate and predict the chlorophyll concentration by the spectral reflectance.RBF network has such advantages as independence of the output on initial weight value, adaptation for determining the construction, rapidness, and reliance, using spectra information sufficiently.In conclusion, RBF network is a worthwhile popularized neural network model on simulation and prediction of the chlorophyll concentration.