基于BP神经网络模型的荔枝树叶面积测定方法

为了准确、快速地测定荔枝树叶面积，设计了一个BP神经网络模型，输入参数为叶片长度和叶片最大宽度，输出参数为叶面积。用LI-3000A型叶面积仪测量所得到的样本数据对网络进行训练，测试样本的网络输出与网络目标的相关系数达0.99609，网络模型是有效的。用训练后的网络模型对10组未参加建模的样本数据进行叶面积测定，误差平方和为1.2929，优于回归方程法的2.511。训练好的BP神经网络模型可以在不破坏叶片的情况下，简单、快速、经济地测定大量的荔枝树叶片面积。

BP neural network model for the measurement of the leaf area of litchi

In order to measure leaf area of Litchi, a BP neural network model was designed, whose input parameters are leaf length and leaf maximum width, and output parameter is leaf area. The sample data, which were obtained by measuring the leaves of Litchi using LI-3000A leaf area instrument, were employed to train the neural network model. The R-square of regression function between output and target of neural network model for testing samples is 0.99609. It indicates that the neural network model is valid. The trained neural network model was applied to measure the areas of ten pieces of leaves respectively, which had not been used to establish the neural network model, the sum squared error of measurement is 1.2929, better than the sum square error of regression function, which is 2.0795. The trained neural network model could be applied to measure numerous leaf area of litchi simply, quickly and economically, and need not destroy the measured leaves.