山东科学 ›› 2022, Vol. 35 ›› Issue (3): 72-81.doi: 10.3976/j.issn.1002-4026.2022.03.009

• 交通运输 • 上一篇    下一篇

基于集群智能的高速公路交通状态识别算法

曾朝晖1(),王江锋1,*(),教欣萍1,熊慧媛1,龚希志1,2   

  1. 1.北京交通大学 交通运输学院 综合交通运输大数据应用技术交通运输行业重点实验室,北京 100044
    2.河南省交通规划设计研究院股份有限公司,河南 郑州450000
  • 收稿日期:2021-04-28 出版日期:2022-06-20 发布日期:2022-06-10
  • 通信作者: 王江锋 E-mail:19120944@bjtu.edu.cn;wangjiangfeng@bjtu.edu.cn
  • 作者简介:曾朝晖(1997—),男,硕士研究生,研究方向为智能交通。E-mail: 19120944@bjtu.edu.cn
  • 基金资助:
    国家重点研发计划(2019YFF0301403);国家自然科学基金(61973028);河南省交通厅科技项目(2020G3)

Highway traffic state recognition based on swarm intelligence

ZENG Zhao-hui1(),WANG Jiang-feng1,*(),JIAO Xin-ping1,XIONG Hui-yuan1,GONG Xi-zhi1,2   

  1. 1. The MOT Key Laboratory of Transport Industry of Big Data Application Technologies for Comprehensive Transport, School of Traffic and Transportation,Beijing Jiaotong University, Beijing 100044, China
    2. Henan Transportation Planning and Design Institute Co., Ltd., Henan 450000, China
  • Received:2021-04-28 Online:2022-06-20 Published:2022-06-10
  • Contact: Jiang-feng WANG E-mail:19120944@bjtu.edu.cn;wangjiangfeng@bjtu.edu.cn

摘要:

针对传统交通状态识别算法仅考虑交通参数个体特征差异而存在识别率较低的问题,引入集群智能概念,提出了既考虑交通参数个体特征差异,又考虑个体参数所蕴含的群体特征差异性的高速公路交通状态识别算法。由于模糊C均值聚类算法(fuzzy C-means algorithm,FCM)在交通状态识别泛化能力上存在收敛缓慢的不足,基于反向学习策略以及鲸鱼优化算法(whale optimization algorithm,WOA),考虑个体交通参数所蕴含的集群行为增强了交通状态初始聚类中心种群的多样性,设计了一种具有良好的全局搜索能力集群智能的高速公路交通状态识别算法,融合了反向学习、WOA和FCM算法,克服了FCM识别算法容易陷入局部最优的局限。实证分析结果表明,所提出的交通状态识别算法具有良好的识别效果,准确率达到92%,且收敛速度较FCM算法更快。

关键词: 高速公路, 交通状态, 模式识别, 鲸鱼算法, 模糊聚类, 反向学习

Abstract:

To solve the problem of low recognition rate in traditional traffic state recognition algorithms that only consider the differences in individual characteristics of traffic parameters, the concept of swarm intelligence is introduced. A highway traffic state recognition algorithm that considers the differences in individual characteristics of traffic parameters as well as the differences in group characteristics contained in individual parameters is proposed. Because the fuzzy C-means algorithm (FCM) has the disadvantage of slow convergence in the generalization ability of traffic state recognition, the opposition-based learning strategy and whale optimization algorithm (WOA) are adopted. Considering the individual traffic parameters that contain the cluster behavior to enhance the diversity of the initial clustering center population of the traffic state, a swarm intelligence-based highway traffic state recognition algorithm OWF with good global search ability is designed, which incorporates reverse learning along with the WOA and FCM algorithms and overcomes the problem that the FCM recognition algorithm is prone to local optima. Empirical analysis results show that the proposed traffic state recognition algorithm has a good recognition effect, with an accuracy rate of 92%, and the convergence speed is greater than that of the FCM algorithm.

Key words: :highway, traffic state, pattern recognition, WOA, FCM, opposition-based learning

中图分类号: 

  • U491