Scholarlink

<p>Traffic noise pollution is a major environmental and health concern in urban areas, causing stress, sleep disturbance, and cardiovascular issues. Accurate prediction of traffic noise is essential for effective noise control and urban planning. This study aims to develop a hybrid, computationally efficient framework for mid-block urban traffic noise prediction. The proposed model integrates Confidence Interval–based Wild Horse Optimization (CI-WHO) for optimal feature selection and an Improved Artificial Neural Network optimised with the African Vulture Optimisation Algorithm (IANN-AVOA) for accurate prediction. An Adaptive Min–Max scaler and Random Over-Sampling are used to preprocess and balance the dataset. Field data were collected from 15 monitoring stations in Kanpur, India, covering residential, commercial, industrial, and silent zones, with variables including traffic flow, speed, and meteorological conditions. Experimental results using Python 3.10 show that the proposed CI-WHO+IANN-AVOA model achieves 97.65% accuracy, 98.10% sensitivity, 94.46% specificity, and a 96.76% F-measure, outperforming CNN, LSTM, ANN, and RNN models. The framework demonstrates excellent generalisation and computational efficiency, offering a reliable approach for real-time traffic noise prediction and supporting sustainable urban noise management.</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

Urban traffic noise prediction using CI-WHO and IANN-AVOA optimization

  • 期刊论文

  • 发表日期：

    2026年3月10日

查看全文
  • Saurabh Upadhyay,
  • Manoranjan Parida,
  • Brind Kumar,
  • Praveen Kumar

摘要

Traffic noise pollution is a major environmental and health concern in urban areas, causing stress, sleep disturbance, and cardiovascular issues. Accurate prediction of traffic noise is essential for effective noise control and urban planning. This study aims to develop a hybrid, computationally efficient framework for mid-block urban traffic noise prediction. The proposed model integrates Confidence Interval–based Wild Horse Optimization (CI-WHO) for optimal feature selection and an Improved Artificial Neural Network optimised with the African Vulture Optimisation Algorithm (IANN-AVOA) for accurate prediction. An Adaptive Min–Max scaler and Random Over-Sampling are used to preprocess and balance the dataset. Field data were collected from 15 monitoring stations in Kanpur, India, covering residential, commercial, industrial, and silent zones, with variables including traffic flow, speed, and meteorological conditions. Experimental results using Python 3.10 show that the proposed CI-WHO+IANN-AVOA model achieves 97.65% accuracy, 98.10% sensitivity, 94.46% specificity, and a 96.76% F-measure, outperforming CNN, LSTM, ANN, and RNN models. The framework demonstrates excellent generalisation and computational efficiency, offering a reliable approach for real-time traffic noise prediction and supporting sustainable urban noise management.