基于改进灰狼算法优化BP神经网络的短时交通流预测模型

BP Neural Network Model for Short-time Traffic Flow Forecasting Based on Transformed Grey Wolf Optimizer Algorithm

准确的短时交通流预测是交通控制和交通诱导的依据.提出一种基于改进灰狼算法(TGWO)优化BP神经网络的短时交通流预测模型(TGWO-BP),有效提高短时交通流预测精度.针对标准灰狼算法(GWO)收敛速度慢,容易陷入局部极值的问题,提出一种自适应递减的收敛因子,使灰狼算法区分全局搜索和局部搜索;改进灰狼个体的位置更新公式,引入惯性权重,调节惯性权重大小使灰狼算法具有跳出局部极值的能力;对比分析TGWO-BP、GWOBP、PSO-BP、BP这4种短时交通流预测模型,结果显示,TGWO-BP的短时交通流预测模型误差为10.03%,达到较好的预测精度.

Accurate short-time traffic flow forecasting is the basis of traffic control and traffic induction. In this paper, a short time traffic flow forecasting model(TGWO-BP) is proposed based on transformed grey wolf optimizer algorithm(TGWO) and BP neural network, which can effectively improve the accuracy of short-time traffic flow forecast. Firstly, due to the drawbacks that the standard gray wolf algorithm converges slowly and tends to fall into the local extremum, an adaptive decreasing convergence factor is proposed, so that the grey wolf algorithm can distinguish the global search from the local search. Secondly, the position renewal formula of the gray wolf individual is improved by introducing the inertial weight. By adjusting the size of the inertial weight, the grey wolf algorithm has the ability to jump out of the local extremum. Finally, four short-time traffic flow forecasting models of TGWO-BP, GWO-BP, PSO-BP and BP are constructed, and the results show that the error of the short-time traffic flow forecasting model of TGWO-BP is 10.03%, and the accuracy of the prediction is better.