基于动态自组织网络的多路口交通流智能预测系统  

Multi-Junction Traffic Flow Intelligent Prediction System Based on Dynamic Self-Organizing Network

作　　者：王慧[1] WANG Hui(Henan Polytechnic Institute,Henan Nanyang 473000,China)

机构地区：[1]河南工业职业技术学院,河南南阳473000

出　　处：《新一代信息技术》2019年第16期16-20,共5页New Generation of Information Technology

基　　金：基于动态自组织网络的智能车际信息融合系统(项目编号:182102210036)。

摘　　要：为使多路口交通流预测更加准确,避免交通事故、交通拥堵现象的出现,提出基于动态自组织网络的多路口交通流智能预测系统。将复杂的神经网络分解开来,在预测网络框架中,设置智能的网络参数,以此为前提对待预测数据进行选择、交叉处理,完成基于动态自组织神经网络的预测环境搭建。在此基础上,通过确定多路口交通流动情况,计算路口车辆汇总率,达到智能预测多路口交通流的目的,完成基于自组织神经网络的多路口交通流智能预测方法的搭建。模拟多路口交通环境环境设计对比实验结果表明,与传统预测方法相比,应用新型预测方法后,拥堵事故发生几率降低至35%左右,有效预测了多路口交通情况。In order to make the multi-junction traffic flow prediction more accurate and avoid the occurrence of traffic accidents and traffic congestion,a multi-junction traffic flow intelligent prediction system based on dynamic self-organizing network is proposed.The complex neural network is decomposed,and intelligent network parameters are set in the prediction network framework.As a premise,the prediction data is selected and cross processed,and the prediction environment construction based on the dynamic self-organizing neural network is completed.On this basis,by determining the multi-junction traffic flow,calculating the summary rate of intersection vehicles,and achieving the purpose of intelligently predicting multi-junction traffic flow,the intelligent multi-junction traffic flow prediction method based on self-organizing neural network is completed.The simulation experiment results of simulated multi-junction traffic environment design show that compared with the traditional forecasting method,after applying the new forecasting method,the probability of congestion accidents is reduced to about 35%,effectively predicting multi-junction traffic conditions.

关 键 词：自组织神经网络 多路口 交通流预测 智能参数 交叉处理 深度条件 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]