基于改进的双并联动态过程神经网络的旅游需求预测

Tourism demand forecast based on the improved double parallel dynamic process neural network

为克服前向过程神经网络收敛速度慢、精度低的问题,提出了一种改进的双并联动态过程神经网络,对于给定的全连接的过程神经网络,通过优化其连接权值和网络结构,删除冗余连接使之成为部分连接的过程神经网络系统,并给出了基于正交基函数展开的学习算法,从而降低了计算的成本。改进的双并联动态过程神经网络应用于旅游预测问题,结果表明其预测精度能够满足工程需要。

To solve the problems of slow convergence speed and low accuracy of the feedforward process neural networks, this paper presents an improved double parallel dynamic process neural network. A given fully connected process neural network may become a process neural network system with partial connection by optimizing connection weight matrix and network structure and deleting redun- dancy conneetion. The learning algorithm based on orthogonal basis function is also given, which reduces the calculating cost. The pro- posed prediction method is utilized in tourism demand forcast and the test results show the prediction precision can meet the requirement of engineering.