基于人工神经网络的交通流预测方法综述

Review of Traffic Flow Prediction Methods Based on Artificial Neural Network

交通流预测对于交通流诱导、交通管控、交通规划等具有重要参考价值,但准确、实时、快速预测未来交通流量一直是个难题。当前,神经网络已有几十种不同的模型,不同的预测模型都有其不同的优缺点。本研究针对基于神经网络的交通流预测研究现状进行综述,首先简要综述几种常用的神经网络(BP神经网络、RBF神经网络及小波神经网络)在交通流预测中的研究及应用效果,分析了3种人工神经网络在交通流预测中的应用场景及网络模型。其次,从优化要素、优化方法及优化效果3个方面总结和分析了基于神经网络的优化方法及应用效果;最后对基于人工神经网络的交通流量预测优化方向提出建议。

Traffic flow prediction has important reference value for traffic flow guidance,traffic control and traffic planning,but accurate,real-time and fast prediction of future traffic flow is always a difficult problem.At present,there are dozens of different neural network prediction models,which have their own advantages and disadvantages.The current research status of neural network based traffic flow forecasting is summarized.First the research and application effects of several commonly used neural networks(BP neural network,RBF neural network and wavelet neural network)in traffic flow prediction are briefly reviewed,and the application scenarios of 3 artificial neural networks in traffic flow prediction are analyzed Second.the optimization methods and application effects based on neural network are summarized and analyzed from the aspects of optimization elements,optimization methods and optimization effects.Finally,the optimization direction of traffic flow prediction based on artificial neural network is suggested.