论文部分内容阅读
利用生物发酵过程运行数据为样本，对BP神经网络进行训练，考察了网络结构（隐含层数、隐含结点数议及学习率η和动量参数α对收敛精度与收敛速度的影响，提出了稳定区的概念和多输出网络分割的思想，并给出了得到较好收敛速度与精度的学习方案。
Using the data of biological fermentation process as training samples, the BP neural network was trained, and the influence of the network structure (the number of hidden layers, the number of hidden nodes, the learning rate η and the momentum parameter α on the convergence accuracy and convergence speed) The concept of stable region and the idea of multi-output network segmentation, and gives a learning solution to get better convergence speed and accuracy.