Scholarlink

Highway icing poses a significant threat to traffic safety, necessitating an accurate and timely forecasting system to mitigate associated risks. This paper involved the collection and analysis of comprehensive data sets encompassing temperature, humidity, wind speed, and road icing conditions across diverse meteorological scenarios. The effect of temperature, humidity, and wind speed on ice condensation was tested by linear regression model, respectively. And the comprehensive effect was tested by adopting multiple regression model. Results shown that the accuracy rate reaches 77% by using multi-factor discriminant analysis. To further validate the model’s practical utility, the experimental data were simulated using the Bayesian discriminant theory model. A total of 1,530 ice-covering records should have been generated in the prediction, and 1,236 ice-covering records were actually generated, with a forecast accuracy of 80.7%, and the false alarm and omission rates of the system were 18% and 3%, respectively. This study, therefore, holds substantial theoretical significance and practical implications for enhancing road safety in icy conditions.

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

Study on Highway Ice Condensation Prediction Based on Bayesian Discriminant Model

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yilong Liu,
  • Jianlong Li,
  • Huaijun Peng

摘要

Highway icing poses a significant threat to traffic safety, necessitating an accurate and timely forecasting system to mitigate associated risks. This paper involved the collection and analysis of comprehensive data sets encompassing temperature, humidity, wind speed, and road icing conditions across diverse meteorological scenarios. The effect of temperature, humidity, and wind speed on ice condensation was tested by linear regression model, respectively. And the comprehensive effect was tested by adopting multiple regression model. Results shown that the accuracy rate reaches 77% by using multi-factor discriminant analysis. To further validate the model’s practical utility, the experimental data were simulated using the Bayesian discriminant theory model. A total of 1,530 ice-covering records should have been generated in the prediction, and 1,236 ice-covering records were actually generated, with a forecast accuracy of 80.7%, and the false alarm and omission rates of the system were 18% and 3%, respectively. This study, therefore, holds substantial theoretical significance and practical implications for enhancing road safety in icy conditions.