论文部分内容阅读
提出一种新的神经网络——高维多输入层神经网络,给出了这种神经网络的结构图及其学习算法.这种神经网络可看作是BP神经网络将一部分输入结点移至其某些隐层上的结果,因此具有较少的连接权;又因为这种神经网络可以依照生产流程安排除第一层以外各输入层的位置,只要位置安排合适,这种神经网络可达到比BP神经网络好的效果.文章将这神经网络按加工工序用于热轧产品质量的建模中,并将结果与BP神经网络的结果进行比较,事实表明,由于连接权值的减少,这种神经网络具有较快的学习速度,在同样的时间内可以达到比BP神经网络更好的学习效果.
A new neural network-high-dimensional multi-input layer neural network is proposed, and the structure of this neural network and its learning algorithm are given. This neural network can be regarded as a BP neural network to move a part of input nodes to As a result of some hidden layers, it has less connection rights. Because this neural network can arrange the positions of input layers other than the first layer according to the production process, this neural network can reach Which is better than BP neural network.This paper uses this neural network to process the quality of hot rolled products according to the manufacturing process and compares the result with the result of BP neural network.The fact shows that due to the decrease of connection weight, Neural network with a faster learning speed, in the same time can reach better than the BP neural network learning.