论文部分内容阅读
BP 神经网络是人工神经网络中应用最广泛的一种多层前馈神经网络。针对它容易陷入局部极小值及隐层节点大多利用经验试凑来确定的缺点，本文提出了一种基于蚁群算法的BP神经网络结构及参数优化方法，利用蚁群算法的全局寻优能力克服BP神经网络存在的不足。最后，将该方法用于短时交通流预测，实验结果表明：利用蚁群算法优化神经网络是有效的，预测结果也有较高精度。“,”BP neural network is the most widely used multilayer feedforward artificial neural networks, however,it is vulnerable to be trapped in local minimum and there is no systematic method to determine the number of hidden layer nodes thus usually done empirically. This paper introduces a method to optimize the structure and parameters of BP neural network which integrates ant colony algorithm with BP neural network to overcome shortcomings of traditional BP neural networks. The proposed method has been applied in short-term traffic flow forecasting. Simulation results demonstrate that the new BP neural network based on ant colony algorithm is more effective and can provide higher precision in traffic flow forecasting.