A Short-Term Prediction Model of PM2.5 Concentration Based on Deep Learning and Mode Decomposition Methods

Based on a set of deep learning and mode decomposition methods, a short-term prediction model for PM2.5 concentration for Beijing city is established in this paper. An ensemble empirical mode decomposition (EEMD) algorithm is first used to decompose the original PM2.5 timeseries to several high- to low-frequency intrinsic mode functions (IMFs). Each IMF component is then trained and predicted by a combination of three neural networks: back propagation network (BP), long short-term memory network (LSTM), and a hybrid network of a convolutional neural network (CNN) + LSTM. The results showed that both BP and LSTM are able to fit the low-frequency IMFs very well, and the total prediction errors of the summation of all IMFs are remarkably reduced from 21 g/m3 in the single BP model to 4.8 g/m3 in the EEMD + BP model. Spatial information from 143 stations surrounding Beijing city is extracted by CNN, which is then used to train the CNN+LSTM. It is found that, under extreme weather conditions of PM2.5 <35 g/m3 and PM2.5 >150 g/m3, the prediction errors of the CNN + LSTM model are improved by ~30% compared to the single LSTM model. However, the prediction of the very high-frequency IMF mode (IMF-1) remains a challenge for all neural networks, which might be due to microphysical turbulences and chaotic processes that cannot be resolved by the above-mentioned neural networks based on variable–variable relationship.

Download Full-text

Ultra Short-term Prediction of Pole Coordinates via Combination of Empirical Mode Decomposition and Neural Networks

Artificial Satellites ◽

10.1515/arsa-2016-0013 ◽

2016 ◽

Vol 51 (4) ◽

pp. 149-161

Author(s):

Yu Lei ◽

Danning Zhao ◽

Hongbing Cai

Keyword(s):

Neural Networks ◽

Empirical Mode Decomposition ◽

High Frequency ◽

Short Term ◽

Pass Filter ◽

Term Prediction ◽

Mode Decomposition ◽

Short Period ◽

The Future ◽

Short Term Prediction

Abstract It was shown in the previous study that the increase of pole coordinates prediction error for about 100 days in the future is mostly caused by irregular short period oscillations. In this paper, the ultra short-term prediction of pole coordinates is studied for 10 days in the future by means of combination of empirical mode decomposition (EMD) and neural networks (NN), denoted EMD-NN. In the algorithm, EMD is employed as a low pass filter for eliminating high frequency signals from observed pole coordinates data. Then the annual and Chandler wobbles are removed a priori from pole coordinates data with high frequency signals eliminated. Finally, the radial basis function (RBF) networks are used to model and predict the residuals. The prediction performance of the EMD-NN approach is compared with that of the NN-only solution and the prediction methods and techniques involved in the Earth orientation parameters prediction comparison campaign (EOP PCC). The results show that the prediction accuracy of the EMD-NN algorithm is better than that of the NN-only solution and is also comparable with that of the other existing prediction method and techniques.

Download Full-text

Short-term prediction of UT1-UTC by combination of the grey model and neural networks

Advances in Space Research ◽

10.1016/j.asr.2016.10.030 ◽

2017 ◽

Vol 59 (2) ◽

pp. 524-531 ◽

Cited By ~ 7

Author(s):

Yu Lei ◽

Min Guo ◽

Dan-dan Hu ◽

Hong-bing Cai ◽

Dan-ning Zhao ◽

...

Keyword(s):

Neural Networks ◽

Grey Model ◽

Short Term ◽

Term Prediction ◽

Short Term Prediction

Download Full-text

Very short-term prediction of torrential rains using polarimetric phased-array radar (MP-PAWR) and deep neural networks

10.1117/12.2598915 ◽

2021 ◽

Author(s):

Philippe Baron ◽

Hiroshi Hanado ◽

Dong-Kyun Kim ◽

Seiji Kawamura ◽

Takeshi Maesaka ◽

...

Keyword(s):

Neural Networks ◽

Deep Neural Networks ◽

Phased Array ◽

Short Term ◽

Term Prediction ◽

Phased Array Radar ◽

Short Term Prediction

Download Full-text

Changes in the ultra-low frequency wave field during the precursor phase to the Sichuan earthquake: DEMETER observations

Annales Geophysicae ◽

10.5194/angeo-31-1597-2013 ◽

2013 ◽

Vol 31 (9) ◽

pp. 1597-1603 ◽

Cited By ~ 10

Author(s):

S. N. Walker ◽

V. Kadirkamanathan ◽

O. A. Pokhotelov

Keyword(s):

Low Frequency ◽

Large Earthquake ◽

Short Term ◽

Frequency Wave ◽

Sichuan Earthquake ◽

Ulf Wave ◽

Precursor Phase ◽

Demeter Satellite ◽

Short Term Prediction ◽

Atmospheric Gravity

Abstract. Electromagnetic phenomena observed in association with increases in seismic activity have been studied for several decades. These phenomena are generated during the precursory phases of an earthquake as well as during the main event. Their occurrence during the precursory phases may be used in short-term prediction of a large earthquake. In this paper, we examine ultra-low frequency (ULF) electric field data from the DEMETER satellite during the period leading up to the Sichuan earthquake. It is shown that there is an increase in ULF wave activity observed as DEMETER passes in the vicinity of the earthquake epicentre. This increase is most obvious at lower frequencies. Examination of the ULF spectra shows the possible occurrence of geomagnetic pearl pulsations, resulting from the passage of atmospheric gravity waves generated in the vicinity of the earthquake epicentre.

Download Full-text

Predicting Station-Level Short-Term Passenger Flow in a Citywide Metro Network Using Spatiotemporal Graph Convolutional Neural Networks

ISPRS International Journal of Geo-Information ◽

10.3390/ijgi8060243 ◽

2019 ◽

Vol 8 (6) ◽

pp. 243 ◽

Cited By ~ 8

Author(s):

Yong Han ◽

Shukang Wang ◽

Yibin Ren ◽

Cheng Wang ◽

Peng Gao ◽

...

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

Traffic Management ◽

Deep Structure ◽

Prediction Errors ◽

Flow Volume ◽

Volume Data ◽

Short Term ◽

Passenger Flow ◽

Metro Network

Predicting the passenger flow of metro networks is of great importance for traffic management and public safety. However, such predictions are very challenging, as passenger flow is affected by complex spatial dependencies (nearby and distant) and temporal dependencies (recent and periodic). In this paper, we propose a novel deep-learning-based approach, named STGCNNmetro (spatiotemporal graph convolutional neural networks for metro), to collectively predict two types of passenger flow volumes—inflow and outflow—in each metro station of a city. Specifically, instead of representing metro stations by grids and employing conventional convolutional neural networks (CNNs) to capture spatiotemporal dependencies, STGCNNmetro transforms the city metro network to a graph and makes predictions using graph convolutional neural networks (GCNNs). First, we apply stereogram graph convolution operations to seamlessly capture the irregular spatiotemporal dependencies along the metro network. Second, a deep structure composed of GCNNs is constructed to capture the distant spatiotemporal dependencies at the citywide level. Finally, we integrate three temporal patterns (recent, daily, and weekly) and fuse the spatiotemporal dependencies captured from these patterns to form the final prediction values. The STGCNNmetro model is an end-to-end framework which can accept raw passenger flow-volume data, automatically capture the effective features of the citywide metro network, and output predictions. We test this model by predicting the short-term passenger flow volume in the citywide metro network of Shanghai, China. Experiments show that the STGCNNmetro model outperforms seven well-known baseline models (LSVR, PCA-kNN, NMF-kNN, Bayesian, MLR, M-CNN, and LSTM). We additionally explore the sensitivity of the model to its parameters and discuss the distribution of prediction errors.

Download Full-text