作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2024, Vol. 50 ›› Issue (6): 367-376. doi: 10.19678/j.issn.1000-3428.0067946

• 开发研究与工程应用 • 上一篇    下一篇

智能网联汽车中联合驾驶风格的交通流数据有效性分析

高家豪1, 胡创业1, 丁男1,2, 刘战东1   

  1. 1. 新疆师范大学计算机科学技术学院, 新疆 乌鲁木齐 830054;
    2. 大连理工大学工业装备智能控制与优化教育部重点实验室, 辽宁 大连 116024
  • 收稿日期:2023-06-28 修回日期:2023-07-28 发布日期:2023-10-30
  • 通讯作者: 丁男,E-mail:dingnan@dlut.edu.cn E-mail:dingnan@dlut.edu.cn
  • 基金资助:
    新疆维吾尔自治区自然科学基金(2021D01E20);国家自然科学基金(62072071,62262066)。

Validity Analysis of Traffic-Flow Data for Combined Driving Style in Intelligent Connected Vehicle

GAO Jiahao1, HU Chuangye1, DING Nan1,2, LIU Zhandong1   

  1. 1. College of Computer Science and Technology, Xinjiang Normal University, Urumqi 830054, Xinjiang, China;
    2. Key Laboratory of Intelligent Control and Optimization for Industrial Equipment, Dalian University of Technology, Dalian 116024, Liaoning, China
  • Received:2023-06-28 Revised:2023-07-28 Published:2023-10-30

摘要: 在智能网联汽车(ICV)中,提高驾驶数据的有效性是提升车辆安全性的基石。只有准确的、可靠的驾驶数据才能为车辆的安全性提供可靠的依据和支持。与传统的异常分析相比,ICV数据有效性分析面临数据异常的多样性(传感器异常、驾驶行为、恶意篡改等)。如何将车辆自身数据特征、驾驶风格和交通流特征相结合,提供有效的数据异常检测方法,已成为智能网联汽车中新的问题。针对ICV系统,采用结合驾驶风格和交通流理论的方法,设计基于粒子群优化的TE-PSO-SVM数据有效性检测算法,实现驾驶数据的有效检测。首先定义驾驶风格识别系数Rad,设计驾驶风格量化模型;其次建立交通流模型,将车辆状态数据与驾驶风格和交通流理论相结合,通过长短期记忆(LSTM)网络对车辆速度预测;最后通过TE-PSO-SVM算法进行数据有效性检测。由于ICV数据的多样性,单一模型在多类型异常混合并存的场景中检测精度仍有局限,利用多个模型的优势构建模型池,并提出基于强化学习的模型选择(RLBMS)算法。在真实数据集highD上的实验结果表明,在不同噪声环境下,TE-PSO-SVM算法模型的F1值相比于传统SVM模型平均提升约8.1个百分点;RLBMS算法模型在不同噪声环境下的F1值相比于模型池中检测率最高的算法平均提高约1.7个百分点,明显提高了数据有效性检测的准确率。

关键词: 智能网联汽车, 驾驶风格, 交通流理论, 粒子群优化算法, 强化学习, 有效性分析

Abstract: In an Intelligent Connected Vehicle (ICV), improving the effectiveness of driving data is crucial for improving vehicle safety. Only accurate and reliable driving data can provide a reliable basis and support for vehicle safety. Compared with the conventional anomaly analysis, the analysis of ICV data validity is accompanied by various data anomalies (e.g., sensor anomalies, driving behavior, and malicious tampering). Researchers of intelligent networked vehicles are attempting to identify a method to combine the vehicle's data characteristics, driving style, and traffic-flow characteristics to provide an effective data-anomaly detection method. For an ICV system, a TE-PSO-SVM data-validity detection algorithm based on particle swarm optimization is designed by combining driving style and traffic-flow theory to effectively detect driving data. First, the driving-style recognition coefficient Rad is defined and the driving-style quantitative model is designed. Second, a traffic-flow model is established, which combines the driving style and traffic-flow theory with vehicle-state data to predict vehicle speed via the Long Short-Term Memory (LSTM) network. Finally, the TE-PSO-SVM algorithm is used to detect the validity of the data. Owing to the diversity of ICV data, the detection accuracy of a single model is limited in scenarios where multiple types of anomalies coexist. To fully utilize the advantages of multiple models, a model pool is constructed, and a Reinforcement Learning-Based Model Selection(RLBMS) algorithm is proposed. Experiments on real dataset highD show that the F1 value of the TE-PSO-SVM algorithm model is 8.1 percentage points higher than that of the conventional SVM model under different noise environments. Compared with the result of the algorithm with the highest detection rate in the model pool, the F1 value of the RLBMS algorithm model in different noise environments is approximately 1.7 percentage points higher on average, which further improves the accuracy of data-validity detection.

Key words: Intelligent Connected Vehicle(ICV), driving style, traffic flow theory, Particle Swarm Optimization(PSO) algorithm, reinforcement learning, validity analysis

中图分类号: