Spatiotemporal prediction of air quality based on LSTM neural network

Seng, Dewen; Zhang, Qiyan; Zhang, Xuefeng; Chen, Guangsen; Chen, Xiyuan

doi:10.1016/j.aej.2020.12.009

Cited by 91 publications

(38 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…By varying the input batch size and recording the total average of the model performances, the proposed LSTM model was more accurate than the DAE model. Seng et al [29] used LSTM model to predict air pollutant data (PM 2.5 , CO, NO 2 , O 3 , SO 2 ) at 35 monitoring stations in Beijing. They proposed a comprehensive model called multi-output and multi-index of supervised learning (MMSL) based on spatiotemporal data of present and surrounding stations.…”

Section: Related Workmentioning

confidence: 99%

Optimising Deep Learning at the Edge for Accurate Hourly Air Quality Prediction

Wardana

Gardner

Fahmy

2021

Sensors

View full text Add to dashboard Cite

Accurate air quality monitoring requires processing of multi-dimensional, multi-location sensor data, which has previously been considered in centralised machine learning models. These are often unsuitable for resource-constrained edge devices. In this article, we address this challenge by: (1) designing a novel hybrid deep learning model for hourly PM2.5 pollutant prediction; (2) optimising the obtained model for edge devices; and (3) examining model performance running on the edge devices in terms of both accuracy and latency. The hybrid deep learning model in this work comprises a 1D Convolutional Neural Network (CNN) and a Long Short-Term Memory (LSTM) to predict hourly PM2.5 concentration. The results show that our proposed model outperforms other deep learning models, evaluated by calculating RMSE and MAE errors. The proposed model was optimised for edge devices, the Raspberry Pi 3 Model B+ (RPi3B+) and Raspberry Pi 4 Model B (RPi4B). This optimised model reduced file size to a quarter of the original, with further size reduction achieved by implementing different post-training quantisation. In total, 8272 hourly samples were continuously fed to the edge device, with the RPi4B executing the model twice as fast as the RPi3B+ in all quantisation modes. Full-integer quantisation produced the lowest execution time, with latencies of 2.19 s and 4.73 s for RPi4B and RPi3B+, respectively.

show abstract

Section: Related Workmentioning

confidence: 99%

Optimising Deep Learning at the Edge for Accurate Hourly Air Quality Prediction

Wardana

Gardner

Fahmy

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Sagar V Belavadi [15] used a scalable architecture to monitor and gather real-time air pollutant concentration data from wireless sensor network in various places and to forecast future air pollutants concentrations. Xiaotong Sun [16] established a spatio-temporal GRU-based (Gated Recurrent Units) prediction framework which takes the spatial information into consideration to predict PM2.5 concentrations in the hour scale. Zou Xiangyu [17] established STA-LSTM neural network based on LSTM, in which a STA (Spatio-Temporal Attention) mechanism was introduced to capture the relative influence of surrounding stations on the prediction area.…”

Section: Prediction On Spatio-temporal Factorsmentioning

confidence: 99%

“…The prediction models based on deep learning [9][10][11][12][13][14] can extract the features existing in the air quality data and can achieve higher prediction accuracy. Some methods [15][16][17][18][19][20][21][22][23][24][25] simulate the temporal and spatial dependence of air quality data at the same time. But widely-used machine learning methods often suffer from high variability in performance in different circumstances.…”

Section: Introductionmentioning

confidence: 99%

Air Quality Prediction Based on Integrated Dual LSTM Model

Chen

Guan

2021

IEEE Access

View full text Add to dashboard Cite

Air quality prediction is an important reference for meteorological forecast and air controlling, but over fitting often occurs in prediction algorithms based on a single model. Aiming at the complexity of air quality prediction, a prediction method based on integrated dual LSTM (Long Short-Term Memory) model was proposed in this paper. Firstly, the Seq2Seq (Sequence to Sequence) technology is used to establish a single-factor prediction model which can obtain the predicted value of each component in air quality data, independently. Each component of air quality is regarded as time series data in the forecasting process. Then, the LSTM model with attention mechanism is used as the multi-factor prediction model. The influencing factors of air quality, like the data of neighboring stations and weather data, are considered in the model. Finally, XGBoosting (eXtreme Gradient Boosting) tree is used to integrate two models. The final prediction results can be obtained by accumulating the predicted values of the optimal subtree nodes. Through evaluation and analysis using five evaluation methods, the proposed method has better performance in terms of error and model expression power. Compared with other various models, the precision of prediction data has been greatly improved in our model.

show abstract

“…The bijective gated recurrent unit is introduced in [ 14 ], which exploits recurrent auto-encoders to predict the next frame in some cases. A multi-output and multi-index of supervised learning [ 15 ] method with LSTM [ 11 ] is proposed for spatiotemporal prediction, which can model the long-term dynamics. In pursuit of alleviating gradient vanishing, convolutional LSTM extended by [ 6 , 7 ] introduces a zigzag memory flow and gradient highway unit (GHU).…”

Section: Literature Reviewmentioning

confidence: 99%

“…Its difference from the above-mentioned recurrent models is that MIM [ 18 ] applies differencing in memory transitions to transform the time-varying polynomial into a constant, which enables the deterministic component predictable. However, these methods [ 14 , 15 , 16 , 17 , 18 ] are still challenging to perform long-term prediction since excessive gate transitions would cause the loss of representations.…”

Section: Literature Reviewmentioning

confidence: 99%

Dual Memory LSTM with Dual Attention Neural Network for Spatiotemporal Prediction

Guan

2021

Sensors

View full text Add to dashboard Cite

Spatiotemporal prediction is challenging due to extracting representations being inefficient and the lack of rich contextual dependences. A novel approach is proposed for spatiotemporal prediction using a dual memory LSTM with dual attention neural network (DMANet). A new dual memory LSTM (DMLSTM) unit is proposed to extract the representations by leveraging differencing operations between the consecutive images and adopting dual memory transition mechanism. To make full use of historical representations, a dual attention mechanism is designed to capture long-term spatiotemporal dependences by computing the correlations between the current hidden representations and the historical hidden representations from temporal and spatial dimensions, respectively. Then, the dual attention is embedded into DMLSTM unit to construct a DMANet, which enables the model with greater modeling power for short-term dynamics and long-term contextual representations. An apparent resistivity map (AR Map) dataset is proposed in this paper. The B-spline interpolation method is utilized to enhance AR Map dataset and makes apparent resistivity trend curve continuous derivative in the time dimension. The experimental results demonstrate that the developed method has excellent prediction performance by comparisons with some state-of-the-art methods.

show abstract

Spatiotemporal prediction of air quality based on LSTM neural network

Cited by 91 publications

References 19 publications

Optimising Deep Learning at the Edge for Accurate Hourly Air Quality Prediction

Optimising Deep Learning at the Edge for Accurate Hourly Air Quality Prediction

Air Quality Prediction Based on Integrated Dual LSTM Model

Dual Memory LSTM with Dual Attention Neural Network for Spatiotemporal Prediction

Contact Info

Product

Resources

About