Scholarlink

<p>Unsafe driving behavior significantly contributes to road accidents. This study introduces a safety evaluation and optimization framework designed for different driving styles. High-frequency driving data and positioning information were collected from urban road experiments, and five key indicators (average speed, speed fluctuation, acceleration range, speeding frequency, and speed change frequency) were selected for analysis. A weighted calculation method was employed to derive a safety score, which was integrated into a driving behavior evaluation model correlated with accident rates. Machine learning models, including neural networks, random forests, support vector machines (SVM), and XGBoost, were compared in terms of fitting performance, with SVM demonstrating the highest accuracy. To optimize driving behavior, fuzzy control was applied to regulate speed and acceleration, based on driving style and traffic conditions. The optimized safety scores showed substantial improvements of 1.6%, 10.81%, 18.22%, 31.98%, and 110.29% across different driving styles. These findings confirm the effectiveness of the proposed method in enhancing driving safety.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fuzzy Control-Based Safety Evaluation and Optimization of Driving Behavior

  • 期刊论文

  • 发表日期：

    2025年12月16日

查看全文
  • Weiqi Zhou,
  • Majun Fei,
  • Chaofeng Pan,
  • Dehua Shi,
  • Hai Zhao

摘要

Unsafe driving behavior significantly contributes to road accidents. This study introduces a safety evaluation and optimization framework designed for different driving styles. High-frequency driving data and positioning information were collected from urban road experiments, and five key indicators (average speed, speed fluctuation, acceleration range, speeding frequency, and speed change frequency) were selected for analysis. A weighted calculation method was employed to derive a safety score, which was integrated into a driving behavior evaluation model correlated with accident rates. Machine learning models, including neural networks, random forests, support vector machines (SVM), and XGBoost, were compared in terms of fitting performance, with SVM demonstrating the highest accuracy. To optimize driving behavior, fuzzy control was applied to regulate speed and acceleration, based on driving style and traffic conditions. The optimized safety scores showed substantial improvements of 1.6%, 10.81%, 18.22%, 31.98%, and 110.29% across different driving styles. These findings confirm the effectiveness of the proposed method in enhancing driving safety.