Encoder-Decoder Model for Forecast of PM2.5 Concentration per Hour

Yan, Lianshan; Wu, Yaowen; Yan, Luqi; Zhou, Min

doi:10.1109/ic3.2018.00020

Cited by 19 publications

(5 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the experiments, the authors present results for seven combinations without explaining why these combinations are selected. L. Yan et al use the E-D model to predict PM2.5 in [31]. The authors use all other features, including the monthly average PM2.5 concentration, daily average PM2.5 concentration, PM10 concentration, AQI, SO2, CO, NO2, O3, average temperature, humidity, pressure, and wind speed per hour Comparison between models using Hanoi dataset with features selected by GA per day.…”

Section: Related Workmentioning

confidence: 99%

PM2.5 Prediction Using Genetic Algorithm-Based Feature Selection and Encoder-Decoder Model

Nguyen

et al. 2021

IEEE Access

View full text Add to dashboard Cite

The concentration of fine particulate matter (PM2.5), which represents inhalable particles with diameters of 2.5 micrometers and smaller, is a vital air quality index. Such particles can penetrate deep into the human lungs and severely affect human health. This paper studies accurate PM2.5 prediction, which can potentially contribute to reducing or avoiding the negative consequences. Our approach's novelty is to utilize the genetic algorithm (GA) and an encoder-decoder (E-D) model for PM2.5 prediction. The GA benefits feature selection and remove outliers to enhance the prediction accuracy. The encoder-decoder model with long short-term memory (LSTM), which relaxes the restrictions between the input and output of the model, can be used to effectively predict the PM2.5 concentration. We evaluate the proposed model on air quality datasets from Hanoi and Taiwan. The evaluation results show that our model achieves excellent performance. By merely using the E-D model, we can obtain more accurate (up to 53.7%) predictions than those of previous works. Moreover, the GA in our model has the advantage of obtaining the optimal feature combination for predicting the PM2.5 concentration. By combining the GA-based feature selection algorithm and the E-D model, our proposed approach further improves the accuracy by at least 13.7%.

show abstract

Section: Related Workmentioning

confidence: 99%

PM2.5 Prediction Using Genetic Algorithm-Based Feature Selection and Encoder-Decoder Model

Nguyen

et al. 2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…CNN-LSTM model (APNet) is proposed for PM2.5 concentration prediction (Huang and Kuo, 2018). Manifold learning is used for feature extraction before DNN (Xie, 2017), convolution recurrent neural networks (D-CRNN) (Zhao and Zettsu, 2019), deep uncertainty learning (Zhang, 2017), deep belief network (Xie, 2017), encoder-decoder model (Yan et al, 2018) are all used for predicting PM2.5 concentration. Most of the study and existing techniques have been used for forecasting PM2.5 pollutant but the recent study shows that there is a growth in NO 2 concentration in the atmosphere which needs to be detected and early measures should be taken to overcome it.…”

Section: Related Workmentioning

confidence: 99%

NO<SUB align="right">2 pollutant concentration forecasting for air quality monitoring by using an optimised deep learning bidirectional GRU model

SonawaniShilpa¹,

PatilKailas²,

Chumchu

2021

IJCSE

View full text Add to dashboard Cite

Air pollution is the most crucial environmental problem to be handled as it has adverse effects on human health, agriculture and climate. Considering its devastating effects, this study is to estimate and monitor nitrogen dioxide (NO 2 ) pollutant concentration. A novel deep learning bidirectional gated recurrent units (GRUs) model is proposed. It is evaluated for its performance with other models like timeseries methods, sklearn machine learning regression methods, AUTOML frameworks and all advanced and hybrid deep learning techniques. The model is evaluated and optimised for the number of features, number of neurons, number of look backs and epochs. It is implemented on real-time dataset of Pune city in India. We also propose a system to monitor concentration of pollutants for buildings or houses in smart city environments which could be helpful to residents, the government and central authorities to prevent excessive pollution levels to avoid adverse effects.

show abstract

“…In addition to deep uncertainty [33], Manifold Learning [34], and Gated Recurring Unit (GRU) [35] have recently been used and work well on air pollution forecasting. Moreover, the manifold learning method, deep belief network [34] , and encoder-decoder model [36] have been also widely used for air pollution forecasting. Both traditional and modern methods can investigate and forecast API trends.…”

Section: Related Workmentioning

confidence: 99%

A Novel One-Dimensional CNN with Exponential Adaptive Gradients for Air Pollution Index Prediction

et al. 2020

View full text Add to dashboard Cite

Air pollution is one of the world’s most significant challenges. Predicting air pollution is critical for air quality research, as it affects public health. The Air Pollution Index (API) is a convenient tool to describe air quality. Air pollution predictions can provide accurate information on the future pollution situation, effectively controlling air pollution. Governments have expressed growing concern about air pollution due to its global effect on human health and sustainable growth. This paper proposes a novel forecasting model using One-Dimensional Deep Convolutional Neural Network (1D-CNN) and Exponential Adaptive Gradients (EAG) optimization to predict API for a selected location, Klang, a city in Malaysia. The proposed 1D-CNN–EAG exponentially accumulates past model gradients to adaptively tune the learning rate and converge in both convex and non-convex areas. We use hourly air pollution data over three years (January 2012 to December 2014) for training. Parameter optimization and model evaluation was accomplished by a grid-search with k-folds cross-validation. Results have confirmed that the proposed approach achieves better prediction accuracy than the benchmark models in terms of Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and the Correlation Coefficient (R-Squared) with values of 2.036, 2.354, 4.214 and 0.966, respectively, and time complexity.

show abstract

Encoder-Decoder Model for Forecast of PM2.5 Concentration per Hour

Cited by 19 publications

References 8 publications

PM2.5 Prediction Using Genetic Algorithm-Based Feature Selection and Encoder-Decoder Model

PM2.5 Prediction Using Genetic Algorithm-Based Feature Selection and Encoder-Decoder Model

NO<SUB align="right">2 pollutant concentration forecasting for air quality monitoring by using an optimised deep learning bidirectional GRU model

A Novel One-Dimensional CNN with Exponential Adaptive Gradients for Air Pollution Index Prediction

Contact Info

Product

Resources

About