Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction
Air pollution has threatened human life in many countries worldwide due to human activity, industrialization, and urbanization over the past few decades. In air forecasting, particulate matter (PM2.5) is a significant health concern. Thus, PM2.5 concentrations must be accurately predicted to protect...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
University of Belgrade
2024-01-01
|
| Series: | Yugoslav Journal of Operations Research |
| Subjects: | |
| Online Access: | https://doiserbia.nb.rs/img/doi/0354-0243/2024/0354-02432400024A.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832581974328344576 |
|---|---|
| author | Aradhyamatada Swamy Rohitha U.M. |
| author_facet | Aradhyamatada Swamy Rohitha U.M. |
| author_sort | Aradhyamatada Swamy |
| collection | DOAJ |
| description | Air pollution has threatened human life in many countries worldwide due to human activity, industrialization, and urbanization over the past few decades. In air forecasting, particulate matter (PM2.5) is a significant health concern. Thus, PM2.5 concentrations must be accurately predicted to protect communities from air pollution. This work aims to increase air quality forecasting by predicting their quality. The significant achievement of this work was the design of a new FS (Feature selection) and prediction method for air quality. Circle Chaotic Map Tuna Swarm Optimization (CCMTSO) and FCNN-LSTM (Fully Convolutional Neural Network - Long short-term term Memory) based DL (Deep Learning) have been used to select features and estimate air quality prediction. The FCNN-LSTM algorithm is generated by CCMTSO using previous information from the target station and nearby stations with chosen attributes. The FCNN model uses geographical features to filter out pollution components, meteorological circumstances, and station interactions. Using the training set, the network is trained until convergence once the model's structure has been established. The suggested approach outperforms competing systems regarding the accuracy of PM2.5 prediction and effectiveness in extracting spatiotemporal data. Three metrics are employed to assess the efficiency of the proposed framework: Root Mean Squared Error (RMSE), coefficient of determination (R2), and Mean Absolute Error (MAE). The findings demonstrate that the results achieved by the proposed system are 7.214, 13.437, and 0.961 for MAE, RMSE, and R2 at a batch size of 128. Utilizing LSTM and FCNN, this algorithm can extract the temporal and spatial components of the information with good precision and reliability. |
| format | Article |
| id | doaj-art-9ae24dc490d6461fbfdfb61a52d9a250 |
| institution | Kabale University |
| issn | 0354-0243 1820-743X |
| language | English |
| publishDate | 2024-01-01 |
| publisher | University of Belgrade |
| record_format | Article |
| series | Yugoslav Journal of Operations Research |
| spelling | doaj-art-9ae24dc490d6461fbfdfb61a52d9a2502025-01-30T06:47:14ZengUniversity of BelgradeYugoslav Journal of Operations Research0354-02431820-743X2024-01-0134466968610.2298/YJOR2402016024A0354-02432400024ACircle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality predictionAradhyamatada Swamy0https://orcid.org/0009-0002-9490-7852Rohitha U.M.1https://orcid.org/0009-0007-6242-0050Department of Electronics & Communication, Proudha Devaraya Institute of Technology, Hosapete, Visvesveraya Technological University, Belagavi, IndiaDepartment of Electronics & Communication, Proudha Devaraya Institute of Technology, Hosapete, Visvesveraya Technological University, Belagavi, IndiaAir pollution has threatened human life in many countries worldwide due to human activity, industrialization, and urbanization over the past few decades. In air forecasting, particulate matter (PM2.5) is a significant health concern. Thus, PM2.5 concentrations must be accurately predicted to protect communities from air pollution. This work aims to increase air quality forecasting by predicting their quality. The significant achievement of this work was the design of a new FS (Feature selection) and prediction method for air quality. Circle Chaotic Map Tuna Swarm Optimization (CCMTSO) and FCNN-LSTM (Fully Convolutional Neural Network - Long short-term term Memory) based DL (Deep Learning) have been used to select features and estimate air quality prediction. The FCNN-LSTM algorithm is generated by CCMTSO using previous information from the target station and nearby stations with chosen attributes. The FCNN model uses geographical features to filter out pollution components, meteorological circumstances, and station interactions. Using the training set, the network is trained until convergence once the model's structure has been established. The suggested approach outperforms competing systems regarding the accuracy of PM2.5 prediction and effectiveness in extracting spatiotemporal data. Three metrics are employed to assess the efficiency of the proposed framework: Root Mean Squared Error (RMSE), coefficient of determination (R2), and Mean Absolute Error (MAE). The findings demonstrate that the results achieved by the proposed system are 7.214, 13.437, and 0.961 for MAE, RMSE, and R2 at a batch size of 128. Utilizing LSTM and FCNN, this algorithm can extract the temporal and spatial components of the information with good precision and reliability.https://doiserbia.nb.rs/img/doi/0354-0243/2024/0354-02432400024A.pdfair-pollutioncircle chaotic map tuna swarm optimization (ccmtso)deep learningforecastingfcnn-lstm (fully convolutional neural network - longshort term memory) |
| spellingShingle | Aradhyamatada Swamy Rohitha U.M. Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction Yugoslav Journal of Operations Research air-pollution circle chaotic map tuna swarm optimization (ccmtso) deep learning forecasting fcnn-lstm (fully convolutional neural network - longshort term memory) |
| title | Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction |
| title_full | Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction |
| title_fullStr | Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction |
| title_full_unstemmed | Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction |
| title_short | Circle chaotic map tuna swarm optimization (CCMTSO) based feature selection and deep learning approach for air quality prediction |
| title_sort | circle chaotic map tuna swarm optimization ccmtso based feature selection and deep learning approach for air quality prediction |
| topic | air-pollution circle chaotic map tuna swarm optimization (ccmtso) deep learning forecasting fcnn-lstm (fully convolutional neural network - longshort term memory) |
| url | https://doiserbia.nb.rs/img/doi/0354-0243/2024/0354-02432400024A.pdf |
| work_keys_str_mv | AT aradhyamatadaswamy circlechaoticmaptunaswarmoptimizationccmtsobasedfeatureselectionanddeeplearningapproachforairqualityprediction AT rohithaum circlechaoticmaptunaswarmoptimizationccmtsobasedfeatureselectionanddeeplearningapproachforairqualityprediction |