Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System
Commercially available low-cost air quality sensors have low accuracy. The improved accuracy of low-cost PM<sub>2.5</sub> sensors allows the use of low-cost sensor systems to reasonably investigate PM<sub>2.5</sub> emissions from industrial activities or to accurately estimat...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-10-01
|
| Series: | Atmosphere |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2073-4433/12/10/1306 |
| _version_ | 1797515271920943104 |
|---|---|
| author | Donggeun Park Geon-Woo Yoo Seong-Ho Park Jong-Hyeon Lee |
| author_facet | Donggeun Park Geon-Woo Yoo Seong-Ho Park Jong-Hyeon Lee |
| author_sort | Donggeun Park |
| collection | DOAJ |
| description | Commercially available low-cost air quality sensors have low accuracy. The improved accuracy of low-cost PM<sub>2.5</sub> sensors allows the use of low-cost sensor systems to reasonably investigate PM<sub>2.5</sub> emissions from industrial activities or to accurately estimate individual exposure to PM<sub>2.5</sub>. In this work, we developed a new PM<sub>2.5</sub> calibration model (HybridLSTM) by combining a deep neural network (DNN) optimized in calibration problems and a long short-term memory (LSTM) neural network optimized in time-dependent characteristics to improve the performance of conventional calibration algorithms of low-cost PM sensors. The PM<sub>2.5</sub> concentrations, temperature and humidity by low-cost sensors and gravimetric-based PM<sub>2.5</sub> measuring instrument were sampled for a sufficiently long time. The proposed model was compared with benchmarks (multiple linear regression model (MLR), DNN model) and low-cost sensor results. The gravimetric measurements were used as reference data to evaluate sensor accuracy. For root-mean-square error (RMSE) for PM<sub>2.5</sub> concentrations, the proposed model reduced 41–60% of error when compared with the raw data of low-cost sensors, reduced 30–51% of error when compared with the MLR model and reduced 8–40% of error when compared with the MLR model. R<sup>2</sup> of HybridLSTM, DNN, MLR and raw data were 93, 90, 80 and 59%, respectively. HybridLSTM showed the state-of-the-art calibration performance for a low-cost PM sensor. In other words, the proposed ML model has state-of-the-art calibration performance among the tested calibration algorithms. |
| first_indexed | 2024-03-10T06:44:05Z |
| format | Article |
| id | doaj.art-22130b49ae074c718c085a1581862e02 |
| institution | Directory Open Access Journal |
| issn | 2073-4433 |
| language | English |
| last_indexed | 2024-03-10T06:44:05Z |
| publishDate | 2021-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Atmosphere |
| spelling | doaj.art-22130b49ae074c718c085a1581862e022023-11-22T17:25:32ZengMDPI AGAtmosphere2073-44332021-10-011210130610.3390/atmos12101306Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring SystemDonggeun Park0Geon-Woo Yoo1Seong-Ho Park2Jong-Hyeon Lee3Research Institute of Environmental Health and Safety, 410, Jeongseojin-ro, Seo-gu, Incheon 404-844, KoreaResearch Institute of Environmental Health and Safety, 410, Jeongseojin-ro, Seo-gu, Incheon 404-844, KoreaResearch Institute of Environmental Health and Safety, 410, Jeongseojin-ro, Seo-gu, Incheon 404-844, KoreaResearch Institute of Environmental Health and Safety, 410, Jeongseojin-ro, Seo-gu, Incheon 404-844, KoreaCommercially available low-cost air quality sensors have low accuracy. The improved accuracy of low-cost PM<sub>2.5</sub> sensors allows the use of low-cost sensor systems to reasonably investigate PM<sub>2.5</sub> emissions from industrial activities or to accurately estimate individual exposure to PM<sub>2.5</sub>. In this work, we developed a new PM<sub>2.5</sub> calibration model (HybridLSTM) by combining a deep neural network (DNN) optimized in calibration problems and a long short-term memory (LSTM) neural network optimized in time-dependent characteristics to improve the performance of conventional calibration algorithms of low-cost PM sensors. The PM<sub>2.5</sub> concentrations, temperature and humidity by low-cost sensors and gravimetric-based PM<sub>2.5</sub> measuring instrument were sampled for a sufficiently long time. The proposed model was compared with benchmarks (multiple linear regression model (MLR), DNN model) and low-cost sensor results. The gravimetric measurements were used as reference data to evaluate sensor accuracy. For root-mean-square error (RMSE) for PM<sub>2.5</sub> concentrations, the proposed model reduced 41–60% of error when compared with the raw data of low-cost sensors, reduced 30–51% of error when compared with the MLR model and reduced 8–40% of error when compared with the MLR model. R<sup>2</sup> of HybridLSTM, DNN, MLR and raw data were 93, 90, 80 and 59%, respectively. HybridLSTM showed the state-of-the-art calibration performance for a low-cost PM sensor. In other words, the proposed ML model has state-of-the-art calibration performance among the tested calibration algorithms.https://www.mdpi.com/2073-4433/12/10/1306machine learningdeep learningcalibrationair qualitylow-cost sensorsexposure assessment |
| spellingShingle | Donggeun Park Geon-Woo Yoo Seong-Ho Park Jong-Hyeon Lee Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System Atmosphere machine learning deep learning calibration air quality low-cost sensors exposure assessment |
| title | Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System |
| title_full | Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System |
| title_fullStr | Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System |
| title_full_unstemmed | Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System |
| title_short | Assessment and Calibration of a Low-Cost PM<sub>2.5</sub> Sensor Using Machine Learning (HybridLSTM Neural Network): Feasibility Study to Build an Air Quality Monitoring System |
| title_sort | assessment and calibration of a low cost pm sub 2 5 sub sensor using machine learning hybridlstm neural network feasibility study to build an air quality monitoring system |
| topic | machine learning deep learning calibration air quality low-cost sensors exposure assessment |
| url | https://www.mdpi.com/2073-4433/12/10/1306 |
| work_keys_str_mv | AT donggeunpark assessmentandcalibrationofalowcostpmsub25subsensorusingmachinelearninghybridlstmneuralnetworkfeasibilitystudytobuildanairqualitymonitoringsystem AT geonwooyoo assessmentandcalibrationofalowcostpmsub25subsensorusingmachinelearninghybridlstmneuralnetworkfeasibilitystudytobuildanairqualitymonitoringsystem AT seonghopark assessmentandcalibrationofalowcostpmsub25subsensorusingmachinelearninghybridlstmneuralnetworkfeasibilitystudytobuildanairqualitymonitoringsystem AT jonghyeonlee assessmentandcalibrationofalowcostpmsub25subsensorusingmachinelearninghybridlstmneuralnetworkfeasibilitystudytobuildanairqualitymonitoringsystem |