近红外光谱分析中的BP模型建立过程，其中在原有BP神经网络模型的基础上自己编程加入了近红外模型验证系数的计算，如R-SquareC.R-SquareP,RMSEC,RMSEP,RSD等。并加入循环语句，设定区间程序可以自动运行并判断最佳隐含层结点数，最佳学习速率，动量因子，训练次数等，同时加入一个验证校正模型实例。-The process of BP model in near infrared spectrum analysis, including the original BP neural network model on the basis of their programming joined the nir model validation coefficient calculation, such as R- SquareC. R- SquareP, RMSEC, RMSEP, RSD, etc. And join the loop statement, set interval program can be run automatically and determine the best hidden layer nodes, the optimal learning rate, momentum factor, the number of training, etc.