Scholarlink

Artificial intelligence algorithms undergo assessment to forecast environmental pollution levels from car exhaust emissions throughout Indore India. Predictions of air pollution levels must be accurate to develop proper mitigation plans since Indore currently undergoes fast urbanization while its vehicle population continues to grow. We designed various machine learning algorithms to predict PM2.5, NO2, and CO concentrations through evaluation of random forests and support vector regression and deep neural networks using historical data from air quality sensors and traffic patterns and meteorological factors. Experts tested several complex integrated deep learning systems composed of recurrent networks and convolutional structures for obtaining spatial and temporal patterns within the data input. The random forest model delivered superior prediction accuracy because its model performed with R2 values of 0.85 for PM2.5 and 0.82 for NO2 and 0.79 for CO. An evaluation of predictive features showed that traffic volume and temperature and wind speed proved to be the strongest indicators. The air quality estimations resulted from traffic expansion models and the models performed accurately with these predictions. This machine learning platform delivers useful evaluation tools for vehicular pollution management to both environmental agencies and urban planners working in Indore as the population density continues to increase.

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

Machine Learning Approaches for Predicting Air Quality-Impacts of Vehicular Emissions in Indore’s Expanding Urban Environment

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jitendra Jayant,
  • A. C. Tiwari,
  • Angur Bala Jayant

摘要

Artificial intelligence algorithms undergo assessment to forecast environmental pollution levels from car exhaust emissions throughout Indore India. Predictions of air pollution levels must be accurate to develop proper mitigation plans since Indore currently undergoes fast urbanization while its vehicle population continues to grow. We designed various machine learning algorithms to predict PM2.5, NO2, and CO concentrations through evaluation of random forests and support vector regression and deep neural networks using historical data from air quality sensors and traffic patterns and meteorological factors. Experts tested several complex integrated deep learning systems composed of recurrent networks and convolutional structures for obtaining spatial and temporal patterns within the data input. The random forest model delivered superior prediction accuracy because its model performed with R2 values of 0.85 for PM2.5 and 0.82 for NO2 and 0.79 for CO. An evaluation of predictive features showed that traffic volume and temperature and wind speed proved to be the strongest indicators. The air quality estimations resulted from traffic expansion models and the models performed accurately with these predictions. This machine learning platform delivers useful evaluation tools for vehicular pollution management to both environmental agencies and urban planners working in Indore as the population density continues to increase.