<p>One of humanity’s most important problems is air pollution, which plays a significant role in health problems and climate change. This issue has intensified due to the surge in automobile use, industrial emissions, fuel consumption in transportation, and energy production. Consequently, air pollution forecasting has become essential. Because of the vast and diverse datasets collected from the control centers of air pollution, the forecasting of air pollution has emerged as a significant area of study, especially when Long Short-Term Memory (LSTM) neural network models are employed. These models are capable of spotting unique long-term patterns present in air pollution data. These models combine traditional statistical and machine learning techniques and often fail to deliver accurate predictions due to challenges such as noisy data and suboptimal hyperparameter configurations. An optimized LSTM representation is critical to accurately predicting pollution levels for multiple contaminants. By putting forth a model that combines the Genetic Algorithm and LSTM, this study tackles the problem of choosing the optimal hyperparameters for LSTM. The suggested approach forecasts the pollution levels for four important pollutants <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41810_2025_343_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="46" /> </InlineMediaObject> <EquationSource Format="TEX">\(PM_{2.5}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>P</mi> <msub> <mi>M</mi> <mrow> <mn>2.5</mn> </mrow> </msub> </mrow> </math></EquationSource> </InlineEquation>, <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41810_2025_343_Article_IEq2.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="38" /> </InlineMediaObject> <EquationSource Format="TEX">\(NO_x\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>N</mi> <msub> <mi>O</mi> <mi>x</mi> </msub> </mrow> </math></EquationSource> </InlineEquation>, <i>CO</i>, and <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="41810_2025_343_Article_IEq3.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="43" /> </InlineMediaObject> <EquationSource Format="TEX">\(PM_{10}\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>P</mi> <msub> <mi>M</mi> <mn>10</mn> </msub> </mrow> </math></EquationSource> </InlineEquation> for the following day by optimizing hyperparameters. The suggested model outperforms standard LSTM models and traditional machine learning regarding efficiency and prediction accuracy.</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

Air Pollution Forecasting Using Genetic Algorithm and LSTM Deep Learning Technique

  • Umang Rastogi,
  • Sushil Kumar,
  • Satya Prakash Maurya,
  • Akhilesh Kumar Yadav

摘要

One of humanity’s most important problems is air pollution, which plays a significant role in health problems and climate change. This issue has intensified due to the surge in automobile use, industrial emissions, fuel consumption in transportation, and energy production. Consequently, air pollution forecasting has become essential. Because of the vast and diverse datasets collected from the control centers of air pollution, the forecasting of air pollution has emerged as a significant area of study, especially when Long Short-Term Memory (LSTM) neural network models are employed. These models are capable of spotting unique long-term patterns present in air pollution data. These models combine traditional statistical and machine learning techniques and often fail to deliver accurate predictions due to challenges such as noisy data and suboptimal hyperparameter configurations. An optimized LSTM representation is critical to accurately predicting pollution levels for multiple contaminants. By putting forth a model that combines the Genetic Algorithm and LSTM, this study tackles the problem of choosing the optimal hyperparameters for LSTM. The suggested approach forecasts the pollution levels for four important pollutants \(PM_{2.5}\) P M 2.5 , \(NO_x\) N O x , CO, and \(PM_{10}\) P M 10 for the following day by optimizing hyperparameters. The suggested model outperforms standard LSTM models and traditional machine learning regarding efficiency and prediction accuracy.