摘要
BP神经网络是当前比较常用的人工神经网络,针对BP神经网络存在易陷入局部极小值、收敛速度慢等缺陷,将LM算法引入到BP神经网络中,以改进BP神经网络在预测时的训练过程,并利用轨道交通客流的时间序列对其有效性进行验证,结果证明该方法对轨道交通客流的短时预测有着更高的准确度和精度。
BP neural network algorithm is the currently common artificial neural network.Referring to the shortcomings existing in BP neural network,such as：local minimum,low convergence rate,this paper introduces LM algorithm to improve BP neural network algorithm,then trains the BP neural network prediction model and applies this method to the time sequence of urban rail transit passenger flow to perform validation.The result shows that the method has a higher accuracy and precision to the short-term prediction of urban rail transit passenger flow.
作者
王立政
朱从坤
WANG Li-zheng;ZHU Cong-kun(College of Civil Engineering,Suzhou University of Science and Technology,Suzhou 215011, China)
出处
《价值工程》
2018年第3期154-156,共3页
Value Engineering
关键词
轨道交通客流
短时预测
BP神经网络
LM算法
urban rail transit passenger flow
short-term prediction
BP neural network
LM algorithm
