Scholarlink

Greenhouse gases such as Carbon Dioxide (CO2), Methane (CH4), and Ozone (O3) threaten human health and the environment by polluting the air. Previous research at Telkom University, Bandung, used the Backpropagation Artificial Neural Network (ANN) method to predict Particulate Matter 2.5 (PM2.5) concentrations. The results show Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) values of 8.32 μg/m3 and 37% at the GKU measurement station (a height of 35 m) and 12.49 μg/m3 and 15% at the Deli measurement station (height 15 m). The CO2, CH4, and O3 prediction parameters were combined at the Telkom University Landmark Tower (TULT) measuring station (70 m height) to optimize the prediction model using deep learning techniques to improve the system. The prediction model is assessed using the Long Short-Term Memory (LSTM) algorithm, which stores relevant information for long periods. This reduces RMSE values: 0.089923 for CO2, 0.060467 for CH4, and 0.036242 for O3. This improved LSTM model can estimate measured gas levels for the next hour, so it has the potential to be applied as an early warning system for air quality in Bandung as well as to provide information about GHG exposure to the public.

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

Greenhouse Gas Prediction Using LSTM Algorithm Based on Microsensor in Bandung City, Indonesia

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Andre Suwardana Adiwidya,
  • Tania Christiana Alexandra,
  • Michelle Kurniawan,
  • Annisa Zahwatul Ummi,
  • Maulana Fauzan Athalla Halinda,
  • Indah Cikal Al Gyfary Oktaviany,
  • Prichel Adisatya Kampong,
  • Irvin Judah Lalintia,
  • Vivian Lee,
  • Dini Rizqi Amalia,
  • Nabilah Indira Putra,
  • Lailatul Rohma,
  • Rahmat Awaludin Salam,
  • Indra Chandra

摘要

Greenhouse gases such as Carbon Dioxide (CO2), Methane (CH4), and Ozone (O3) threaten human health and the environment by polluting the air. Previous research at Telkom University, Bandung, used the Backpropagation Artificial Neural Network (ANN) method to predict Particulate Matter 2.5 (PM2.5) concentrations. The results show Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) values of 8.32 μg/m3 and 37% at the GKU measurement station (a height of 35 m) and 12.49 μg/m3 and 15% at the Deli measurement station (height 15 m). The CO2, CH4, and O3 prediction parameters were combined at the Telkom University Landmark Tower (TULT) measuring station (70 m height) to optimize the prediction model using deep learning techniques to improve the system. The prediction model is assessed using the Long Short-Term Memory (LSTM) algorithm, which stores relevant information for long periods. This reduces RMSE values: 0.089923 for CO2, 0.060467 for CH4, and 0.036242 for O3. This improved LSTM model can estimate measured gas levels for the next hour, so it has the potential to be applied as an early warning system for air quality in Bandung as well as to provide information about GHG exposure to the public.