摘要： 针对城市交通流的复杂性、随机性、非线性等特点，利用遗传算法(GA)优化小波神经网络(WNN)，以克服传统神经网络收敛速度慢、易陷入局部最小点等缺陷，在此基础上建立基于GA-WNN的城市交通流预测模型。利用GA-WNN、GA-BP和WNN模型对南昌市南京西路交通流进行仿真预测，实验结果表明，GA-WNN模型的预测效果较好，相比GA-BP和WNN模型具有更高的预测精度和更快的收敛速度。

关键词: 交通流预测, 遗传算法, 小波神经网络, 预测模型

Abstract: Considering the characteristics of complexity, randomness and nonlinear in urban traffic volume, Wavelet Neural Network(WNN) is optimized by Genetic Algorithm(GA) to overcome the problems of slow network convergence rate and falling into local minimum which exist in traditional Neural Network(NN), and prediction model of urban traffic volume based on GA-WNN is established. Simulation predictions for Nanjing West Road in Nanchang City are conducted with GA-WNN, GA-BP and WNN models, whose results show that GA-WNN model has better prediction effect, higher prediction accuracy and faster convergence speed than GA-BP and WNN models.

Key words: traffic volume prediction, Genetic Algorithm(GA), Wavelet Neural Network(WNN), prediction model

中图分类号: 

• TP391