Air Quality Prediction Modeling Research Articles

Median-KNN Regressor-SMOTE-Tomek Links for Handling Missing and Imbalanced Data in Air Quality Prediction

The Air Quality Index (AQI) dataset contains information on measurements of pollutants and ambient air quality conditions at certain location that can be used to predict air quality. Unfortunately, this dataset often has many missing observations and imbalanced classes. Both of these problems can affect the performance of the prediction model. In particular, predictions for the minority class are very important because inaccurate predictions can be fatal or cause big losses. Moreover, the missing data may lead to biased results. This paper proposes the single imputation of the median and the multiple imputations of the k-Nearest Neighbor (KNN) regressor to handle missing values of less than or equal to 10% and more than 10%, respectively. At the same time, the SMOTE-Tomek Links address the imbalanced class. These proposed approaches to handle both issues are then used to assess the air quality prediction of the India AQI dataset using Naive Bayes (NB), KNN, and C4.5. The five treatments show that the proposed method of the Median-KNN regressor-SMOTE-Tomek Links is able to improve the performance of the India air quality prediction model. In other words, the proposed method succeeds in overcoming the problems of missing values and class imbalance.

Air Quality Index Prediction

We address air index forecast by inferring, using deep and machine learning algorithms, the concentration per hour of air pollutants (such ozone, Particulate Matter (2.5), and Sulphur). DL is one of the most popular methods, may efficiently apply optimization algorithms to train a model on large amounts of information and data. The air quality prediction model is trained using time series data. In our approach, the sequence module makes use of deep CNN, and time series data is progressively fed into the CNN model for training. In CNN, there are many different functioning layers, including RNN, and LSTM. The CNN layer can be utilized to successfully extract the features that are sequential of time series. Advanced features outperform normal features when analyzing data of time series. CNN's pooling layer oversees down sampling. CNN is good at forecasting air quality, according to test data. Key Words: AQP, CNN, PM, RNN, DL

Simulation Evaluation of the Contribution of Typical PM2.5 Pollution and Trans-regional Transport to Ningbo Pollution in the Yangtze River Delta

The weather research and forecasting model (WRF) and the nested air quality prediction model system (NAQPMS) were used to simulate a typical heavy PM2.5 pollution event in the Yangtze River Delta (YRD) on January 10, 2015. The results showed that NAQPMS simulation accurately reproduced the development process of heavy PM2.5 pollution in the YRD. External regional pollution transport was a significant factor, primarily through three channels:the inland channel (Henan-Anhui-Jiangsu), coastal channel (Hebei-Shandong-Jiangsu), and ocean channel (Hebei-Shandong-Huanghai Sea-Jiangsu). The source apportionment of estimating the contribution across regions with Ningbo as the receptor showed that the coastal channel was the main channel in the pollution occurrence stage, and the large transport contribution was from Jiangsu (36.51%). The inland and ocean channels were more important in the heavy pollution stage, and the main external regional contributions were from Zhejiang North (25.55%), Anhui (12.33%), and the ocean (16.92%). Pollution dissipation was mainly affected by the ocean channel, and the main external contribution regions were Jiangsu (14.74%) and the sea (31.43%).

Unexpected high contribution of in-cloud wet scavenging to nitrogen deposition induced by pumping effect of typhoon landfall in China

Atmospheric nitrogen deposition has large eco-environmental effects such as ocean acidification, eutrophication in coastal areas. However, knowledge of the source and the pathway of N deposition in coastal areas is limited, especially during tropical storms, hindering the accurate quantification of how anthropogenic activities influence the ocean ecosystem. In this study, the Nested Air Quality Prediction Modeling System was used to investigate the wet deposition of N induced by typhoon Hagupit over eastern coastal China from an in- and below-cloud process perspective. Our results reveal for the first time an enhancement mechanism of N deposition related to the ‘pumping effect’ of the typhoon. Different from the non-typhoon conditions, air pollutants in the typhoon-affected regions were pumped into the higher altitudes and deposited via the in-cloud scavenging process in the moving path of the typhoon-affected regions. This study updates our understanding of the source–receptor relationship on atmospheric wet deposition caused by tropical cyclones.

The AirGAM 2022r1 air quality trend and prediction model

Abstract. This paper presents the AirGAM 2022r1 model – an air quality trend and prediction model developed at the Norwegian Institute for Air Research (NILU) in cooperation with the European Environment Agency (EEA) over 2017–2021. AirGAM is based on nonlinear regression GAMs – generalised additive models – capable of estimating trends in daily measured pollutant concentrations at air quality monitoring stations, discounting for the effects of trends and time variations in corresponding meteorological data. The model has been developed primarily for the compounds NO2, O3, PM10, and PM2.5. Meteorological input data consist of temperature, wind speed and direction, planetary boundary layer height, relative and absolute humidity, cloud cover, and precipitation over the period considered. The exact set of meteorological variables used in the model depends on the compound selected for analysis. In addition to meteorological variables introduced in the model as covariates, i.e. explanatory variables for the concentration levels, the model also incorporates time variables such as the day of the week, day of the year, and overall time, which is related to the model's trend term. The trend analysis is performed at each station separately. Thus, the model only considers the temporal features of concentrations and meteorology at a station, rather than any spatial correlations or dependencies between stations. AirGAM is implemented using the R language for statistical computing and, in particular, the GAM package mgcv. In the model, meteorological and time covariates are represented and estimated as smooth nonlinear functions of the corresponding variables. Thus, the trend term is defined and estimated as a smooth nonlinear function of time over the period selected for analysis. Once fitted to training data, the model may be used as a prediction tool capable of predicting air pollutant concentrations for new sets of meteorological and time data which are not in the training set – e.g. for cross-validation or forecasting purposes. The model does not explicitly use emissions or background concentrations – these are sought to be implicitly represented through the estimated nonlinear relations between meteorology, time, and concentrations. In addition to meteorology-adjusted trends, the program also produces unadjusted trends – i.e. trends based on the same regression set-up but only including the time covariates. Both types of trends can be output in the same run, making it possible to compare them. Ideally, the meteorology-adjusted trend will show the trend in concentration mainly due to changes in emissions or physicochemical processes not induced by changes in meteorology. AirGAM has been developed and tested primarily in trend studies based on measurement data hosted by the EEA, including the AirBase data (before 2013) and the Air Quality e-Reporting (AQER) data from 2013 and onwards. Still, the model is general and could be applied in other regions with other input data. The EEA data provide daily or hourly surface measurements at individual monitoring stations in Europe. For input meteorological data, we extract time series from the gridded meteorological re-analysis (ERA5) provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) for each monitoring station. The paper presents results with the model for all AirBase/AQER stations in Europe from the latest EEA trend study for 2005–2019.

Balanced Spider Monkey Optimization with Bi-LSTM for Sustainable Air Quality Prediction

A reliable air quality prediction model is required for pollution control, human health monitoring, and sustainability. The existing air quality prediction models lack efficiency due to overfitting in prediction model and local optima trap in feature selection. This study proposes the Balanced Spider Monkey Optimization (BSMO) technique for effective feature selection to overcome the local optima trap and overfitting problems. The air quality prediction data were collected from the Central Pollution Control Board (CPCB) from four cities in India: Bangalore, Chennai, Hyderabad, and Cochin. Normalization is performed using Min-Max Normalization and fills the missing values in the dataset. A Convolutional Neural Network (CNN) is applied to provide deep representation of the input dataset. The BSMO technique selects the relevant features based on the balancing factor and provides the relevant features for the Bi-directional Long Short-Term Memory (Bi-LSTM) model. The Bi-LSTM model provides the time series prediction of air quality for four cities. The BSMO model obtained higher feature selection performance compared to existing techniques in air quality prediction. The BSMO-BILSTM model obtained 0.318 MSE, 0.564 RMSE, and 0.224 MAE, whereas Attention LSTM reached 0.699 MSE, 0.836 RMSE, and 0.892 MAE. Our solution may be of particular interest to various governmental and non-governmental institutions focused on maintaining high Quality of Life (QoL) on the local or state level.

Enhancing Air Quality Prediction Accuracy Using Hybrid Deep Learning

PM2.5 (Particulate Matter) and PM10 are the most common pollutants, and the increasing of concentration in the air will threaten people’s health. The machine learning method has recently been of particular interest to many researchers due to its effectiveness in air quality prediction models. Many solutions employing deep learning-based techniques such as Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and hybrid CNN-LSTM models to enhance air quality prediction accuracy have been developed. This paper proposes a hybrid Encoder STM model for predicting the next day to the next five days’ PM2.5 and PM10 concentrations in Hanoi. Additionally, we proposed five extended features to increase the accuracy of prediction. Then other models, namely the LSTM model and the Bidirectional LSTM model, are also considered for PM2.5 and PM10 concentration prediction. Our results show that the proposed approaches outperform other state-of-the-art deep learning-based methods on both Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) due to low error and the small number of features.

A Bidirectional Long Short-Term Neural Network Model to Predict Air Pollutant Concentrations: A Case Study of Tehran, Iran

Air pollution is a major challenge of new civilized world, which has intensified as a result of industrialization, urbanization expansion, and rapid traffic growth and increasing anthropogenic activities. Air pollutants can cause serious toxicological impacts on human health and the environment. Therefore, knowledge of pollutant concentrations can be used as key information in pollution control programs and policies. The main aim of this research was to provide a new hybrid model for short-term air quality prediction. Air pollution is the presence of pollutants in the air such as sulphur dioxide (SO2), particle substances (PM), nitrogen oxides (NOX), and ozone (O3). Sulfur dioxide (SO2), particulate matter with particle size less than 10 microns (PM10), particulate matter (PM2.5), nitrogen dioxide (NO2), carbon monoxide (CO), and ozone (O3) in the air all have an impact on the Air Quality Index (AQI). The AQI's variation is extremely nonlinear and complex. One of the most desired features of an AQI prediction model is the ability to predict the AQI level that is dangerous to the population. To increase forecast accuracy, an air quality prediction model based on deep recurrent neural networks and the BiLSTM model is created in this research. AQIs (O3, PM2.5, PM10, NO2, SO2, CO) prediction has been examined among several air pollutants. The data on O3, PM2.5, PM10, NO2, SO2, and CO concentrations are average daily reports from the Tarbiat Modares station in Tehran's 6th district, collected between March 2018 and June 2019. The station is located at 35.71751 latitude and 51.385909 longitudes. The proposed model was compared with Shallow Learning and MLP neural networks which had better and more reliable results over the aforementioned neural network in predicting the concentration of pollutants. The regression coefficient obtained to predict the concentrations of pollutants O3, PM2.5, PM10, NO2 and SO2 was 0.87, 0.62, 0.84, 0.67, 0.75 and 0.72, respectively. The results showed that the Bi-LSTM deep neural network model is a reliable method to predict hourly air pollutant concentrations.

Quantifying the Source Contributions to Poor Atmospheric Visibility in Winter over the Central Plains Economic Region in China

The Central Plains Economic Region (CPER) is one of the most polluted regions in China. Air pollution has caused visibility degradation due to the light extinction of fine particles (PM2.5). However, the source of light extinction and visibility degradation is still unclear. In this study, the nested air quality prediction model system coupled with an online tracer-tagging module has been used to quantify the contribution of emission sectors and regions to visibility degradation. The light extinction coefficients were well reproduced over CPER. The results showed that resident-related emissions, traffic and industry were the main sectors of visibility degradation over CPER, contributing 55~62%, 10~28%, and 9~19%, respectively. The contribution of local emissions and regional transport was also investigated, and the results showed that regional transport dominated the light extinction (56~68%), among which transport within Henan province contributes significantly (12~45%). Sensitivity tests showed that the reduction in the resident-related sector was more effective than that of the industry sector. Emission control of 40% in resident-related, industry, and traffic sectors over the whole region can achieve the goal of good visibility. This study will provide scientific suggestions for the control strategies development to mitigate visibility degradation over CPER.

Chemistry‐Weather Interacted Model System GRAPES_Meso5.1/CUACE CW V1.0: Development, Evaluation and Application in Better Haze/Fog Prediction in China

AbstractThe Chinese Meteorology Administration chemistry model Chinese Unified Atmospheric Chemistry Environment (CUACE) is online integrated into the mesoscale operational numerical weather prediction (NWP) model (GRAPES_Meso5.1) and aerosol‐cloud‐radiation interaction is achieved to establish the first version (V1) of chemistry‐weather (CW) interacted model GRAPES‐Meso5.1/CUACE CW V1. The most polluted winter 2016–2017 is selected to study the meteorology impacts on haze/fog prediction, the impact of aerosol‐radiation, aerosol‐cloud and CW interaction (ARI, ACI, CWI) on haze/fog prediction and NWP. Single way model without CWI displays reasonable PM2.5 and visibility prediction in general. However, modeled PM2.5 peaks are underestimated and visibility valleys are overestimated during haze/fog pollution, the underestimation of relative humidity (RH) contributes major to this misestimation; CWI model cut the negative bias of PM2.5 peaks and the positive bias of visibility valleys. The improvement of 5 and 3 km low visibility by CWI during severe haze/fog period is more obvious than that of 10 km, which just compensates for the largest deficiency in low visibility prediction related with severe haze/fog by single way model; The NWP including sea level pressures, RH, temperature, wind speed are also improved by CWI from surface to upper troposphere; ARI contributes larger to the predicted PM2.5,visibility and NWP improvement than that of ACI, their relative contributions varies with model vertical height and the overlapping condition of cloud and aerosols. Due to the joint contribution of RH and PM2.5, CWI's improving on visibility is larger than PM2.5. This study illustrates the importance of including CWI in air quality prediction model.

Secondary Modeling of Air Quality Based on LSTM Cycle Neural Network

Based on the “primary prediction data” generated by the operation of the commonly used air quality prediction model WRF-CMAQ and the “measured data” obtained from the actual monitoring of the air quality monitoring stations, this paper establishes the air quality prediction constant value by comprehensively applying the long short-term memory (LSTM) neural network algorithm and other methods. Secondly, the meteorological conditions are reasonably classified, and the K-Means algorithm is used to eliminate the influence of the dimensional difference between the data. The original data is missing or abnormal. In this paper, the mixed interpolation method is proposed to deal with the missing data, and the isolated forest algorithm is used to deal with the abnormal data. Thirdly, three possible classification schemes are found by using the method of Elbow Method, and the correlation analysis and data visualization methods are used respectively. After discussing the three meteorological classification conditions, the optimal classification number is determined to be 4 categories. Finally, a model is established which is applicable to A, B and C, three monitoring points, and verify the rationality of the modeling and forecasting process and the robustness of the model.

Complementary ensemble empirical mode decomposition and independent recurrent neural network model for predicting air quality index

Establishing a scientific and effective air quality prediction model is of great scientific value and practical significance for protecting people’s health and promoting social harmony and stability. However, existing prediction models have certain shortcomings in various aspects. To address these shortcomings, this paper combines different methods to achieve better prediction accuracy. Outlier analysis is done for air quality index (AQI) using an isolation forest algorithm. An air quality prediction system that consists of data preprocessing, optimization, prediction, and modification is established. The complementary ensemble empirical mode decomposition (CEEMD), modified particle swarm optimization (MPSO) algorithm, independent recurrent neural network (IndRNN), and nonlinear correction strategy are employed for the prediction. We select five cities with different AQI as the experiment sites, and three experiments are designed to test the accuracy of the prediction model. The results reveal that (1) by using CEEMD data decomposition technology to deal with the non-stationarity and nonlinearity of the original data, the prediction accuracy of the original cyclic neural network model can be improved by about 15%. (2) The prediction system with the CEEMD-MPSO-IndRNN model as the core prediction module and with a nonlinear error correction strategy has good scalability and robustness for air quality prediction. (3) The performance of the air quality prediction model constructed with systematic thought is better than other comparable models, and it can effectively predict the time series data of different cities and frequencies.

Uncertainties in the simulated intercontinental transport of air pollutants in the springtime from emission and meteorological inputs

The uncertainty in the quantitative estimates of long-range transport of pollutants in air quality models is large. In this study, based on the Global Nested Air Quality Prediction Modeling System (GNAQPMS-SM), using Latin hypercube sampling (LHS) method, we conducted a base simulation and an ensemble of 1° × 1° simulations in April 2017 to assess the uncertainty in model estimates of O3 and PM2.5 intercontinental transport from uncertainties in input parameters (emissions, temperature, wind field). The results show that the base simulation has biases in some regions, and when the uncertainty of input parameters is taken into account, the observations mostly lie within the uncertainty range of the 50 simulated group samples. The uncertainties of input parameters are reasonably quantified in terms of their propagation during computation in GNAQPMS-SM. The best performance occurs with the multiple linear regression (MLR) method, which is used to fit ensemble simulation results. The uncertainty range of monthly O3 transport from North America (NAM) to Europe (EUR) at a 1.2 km altitude is 4.91–5.44 ppb, from EUR to China is 0.78–0.96 ppb, and from China to NAM is 0.76–0.85 ppb. Correspondingly, the MLR result of each is 5.12 ppb, 0.80 ppb, and 0.77 ppb, respectively. The MLR result is thought to be closer to the actual transport amounts, and most of them lie within the uncertainty range. The appreciable differences between base simulation and MLR results reflect the uncertainties in the input parameters. Finally, taking O3 transport from China to Japan as an example, the maximum absolute value of the daily average Spearman correlation coefficient |ρ‾|max between input parameters and contributions is calculated to identify the key uncertainty sources in several regions of O3 transport to Japan. Meteorological and air pollutant observations and the performance of meteorological simulations should be strengthened.

Reconciling the total carbon budget for boreal forest wildfire emissions using airborne observations

Abstract. Wildfire impacts on air quality and climate are expected to be exacerbated by climate change with the most pronounced impacts in the boreal biome. Despite the large geographic coverage, there is limited information on boreal forest wildfire emissions, particularly for organic compounds, which are critical inputs for air quality model predictions of downwind impacts. In this study, airborne measurements of 193 compounds from 15 instruments, including 173 non-methane organics compounds (NMOG), were used to provide the most detailed characterization, to date, of boreal forest wildfire emissions. Highly speciated measurements showed a large diversity of chemical classes highlighting the complexity of emissions. Using measurements of the total NMOG carbon (NMOGT), the ΣNMOG was found to be 50 % ± 3 % to 53 % ± 3 % of NMOGT, of which, the intermediate- and semi-volatile organic compounds (I/SVOCs) were estimated to account for 7 % to 10 %. These estimates of I/SVOC emission factors expand the volatility range of NMOG typically reported. Despite extensive speciation, a substantial portion of NMOGT remained unidentified (47 % ± 15 % to 50 % ± 15 %), with expected contributions from more highly-functionalized VOCs and I/SVOCs. The emission factors derived in this study improve wildfire chemical speciation profiles and are especially relevant for air quality modelling of boreal forest wildfires. These aircraft-derived emission estimates were further linked with those derived from satellite observations demonstrating their combined value in assessing variability in modelled emissions. These results contribute to the verification and improvement of models that are essential for reliable predictions of near-source and downwind pollution resulting from boreal forest wildfires.

An air quality prediction model based on improved Vanilla LSTM with multichannel input and multiroute output

Long short-term memory (LSTM), especially vanilla LSTM (VLSTM), has been widely used in air quality prediction field. However, VLSTM has many more parameters, thereby making training slow and prediction performance unstable. The VLSTM network input data have not been selected for better efficiency. In this paper, we propose an air quality prediction model based on the improved VLSTM with multichannel input and multiroute output (IVLSTM-MCMR). The proposed model includes the IVLSTM and MCMR modules. The proposed IVLSTM module is developed by improving the VLSTM inner structure of VLSTM in order to reduce the number of parameters that help to accelerate the convergence. A new historical information usage approach is further proposed to obtain a stable training process. For the MCMR module, a multichannel data input model (MC) with an improved linear similarity dynamic time warping is introduced to choose the valid data as the input of IVLSTM. A multiroute output model (MR) is designed to integrate the results from MC, in which the results of different target stations with different features are output by different routes. We evaluate the proposed model with the collected data from Beijing, China, and the experimental results show that our model achieves improvements regarding the predication performance.

Deep learning-based gas-phase chemical kinetics kernel emulator: Application in a global air quality simulation case

The global atmospheric chemical transport model has become a key technology for air quality forecast and management. However, precise and rapid air quality simulations and forecast are frequently limited by the model’s computational performance. The gas-phase chemistry module is the most time-consuming module in air quality models because its traditional solution method is dynamically stiff. To reduce the solving time of the gas phase chemical module, we built an emulator based on a deep residual neural network emulator (NN) for Carbon Bond Mechanism Z (CBM-Z) mechanism implemented in Global Nested Air Quality Prediction Modeling System. A global high resolution cross-life multi-species dataset was built and trained to evaluate multi-species concentration changes at a single time step of CBM-Z. The results showed that the emulator could accelerate to approximately 300–750 times while maintaining an accuracy similar to that of CBM-Z module (the average correlation coefficient squared was 0.97) at the global scale. This deep learning-based emulator could adequately represent the stiff kinetics of CBM-Z, which involves 47 species and 132 reactions. The emulated ozone (O3), nitrogen oxides (NOx), and hydroxyl radical (OH) were consistent with those of the original CBM-Z module in different global regions, heights, and time. Our results suggest that data-driven emulations have great potential in the construction of hybrid models with process-based air quality models, particularly at larger scales.

Low-cost nature-inspired deep learning system for PM2.5 forecast over Delhi, India.

Air quality has a tremendous impact on India's health and prosperity. Air quality models are crucial tools for surveying and projecting air pollution episodes, which can be used to issue health advisories to take action ahead of time. Short-term increases in air pollution trigger many adverse health events; a fast, efficient, cost-effective, and reliable air quality prediction model would aid in minimizing the effect on health and prosperity. Deterministic models, on the other hand, are less robust in predicting the pollutant series since it is non-stationary and non-linear. Atmospheric chemistry models are computationally expensive and often rely on outdated emissions information. We propose a deep learning model in this study that integrates neural networks, fuzzy inference systems, and wavelet transforms to predict the most prominent air pollutant affecting Delhi, India i.e., PM2.5 (particulate matter of aerodynamic diameter less than or equal to 2.5 µm). We have included the main aspects of air quality models in this research i.e., less computational time (7min approximately using I5-1035G1, 1.19GHz processor), less resource-intensive (dependent only on the pollutant lagged values), and high spatial resolution (1km) for forecasting air quality three days ahead. The model predictions show a significant correlation coefficient lying in [0.96,0.98], [0.86,0.93], and [0.82,0.91] with Central Pollution Control Board (CPCB) monitored data at various sites in Delhi for one, two, and three days of forecast respectively.

Estimating Full-Coverage PM2.5 Concentrations Based on Himawari-8 and NAQPMS Data over Sichuan-Chongqing

Fine particulate matter (PM2.5) has attracted extensive attention due to its harmful effects on humans and the environment. The sparse ground-based air monitoring stations limit their application for scientific research, while aerosol optical depth (AOD) by remote sensing satellite technology retrieval can reflect air quality on a large scale and thus compensate for the shortcomings of ground-based measurements. In this study, the elaborate vertical-humidity method was used to estimate PM2.5 with the spatial resolution 1 km and the temporal resolution 1 hour. For vertical correction, the scale height of aerosols (Ha) was introduced based on the relationship between the visibility data and extinction coefficient of meteorological observations to correct the AOD of the Advance Himawari Imager (AHI) onboard the Himawari-8 satellite. The hygroscopic growth factor (f(RH)) was fitted site-by-site and month by month (1–12 months). Meanwhile, the spatial distribution of the fitted coefficients can be obtained by interpolation assuming that the aerosol properties vary smoothly on a regional scale. The inverse distance weighted (IDW) method was performed to construct the hygroscopic correction factor grid for humidity correction so as to estimate the PM2.5 concentrations in Sichuan and Chongqing from 09:00 to 16:00 in 2017–2018. The results indicate that the correlation between “dry” extinction coefficient and PM2.5 is slightly improved compared to the correlation between AOD and PM2.5, with r coefficient values increasing from 0.12–0.45 to 0.32–0.69. The r of hour-by-hour verification is between 0.69 and 0.85, and the accuracy of the afternoon is higher than that of the morning. Due to the missing rate of AOD in the southwest is very high, this study utilized inverse variance weighting (IVW) gap-filling method combine satellite estimation PM2.5 and the nested air-quality prediction modeling system (NAQPMS) simulation data to obtain the full-coverage hourly PM2.5 concentration and analyze a pollution process in the fall and winter.

Air quality prediction models based on meteorological factors and real-time data of industrial waste gas

With the rapid economic growth, air quality continues to decline. High-intensity pollution emissions and unfavorable weather conditions are the key factors for the formation and development of air heavy pollution processes. Given that research into air quality prediction generally ignore pollutant emission information, in this paper, the random forest supervised learning algorithm is used to construct an air quality prediction model for Zhangdian District with industrial waste gas daily emissions and meteorological factors as variables. The training data include the air quality index (AQI) values, meteorological factors and industrial waste gas daily emission of Zhangdian District from 1st January 2017 to 30th November 2019. The data from 1st to 31th December 2019 is used as the test set to assess the model. The performance of the model is analysed and compared with the backpropagation (BP) neural network, decision tree, and least squares support vector machine (LSSVM) function, which has better overall prediction performance with an RMSE of 22.91 and an MAE of 15.80. Based on meteorological forecasts and expected air quality, a daily emission limit for industrial waste gas can be obtained using model inversion. From 1st to 31th December 2019, if the industrial waste gas daily emission in this area were decreased from 6048.5 million cubic meters of waste gas to 5687.5 million cubic meters, and the daily air quality would be maintained at a good level. This paper deeply explores the dynamic relationship between waste gas daily emissions of industrial enterprises, meteorological factors, and air quality. The meteorological conditions are fully utilized to dynamically adjust the exhaust gas emissions of key polluting enterprises. It not only ensures that the regional air quality is in good condition, but also promotes the in-depth optimization of the procedures of regional industrial enterprises, and reduces the conflict between environmental protection and economic development.

Air-Quality Prediction Based on the EMD–IPSO–LSTM Combination Model

Owing to climate change, industrial pollution, and population gathering, the air quality status in many places in China is not optimal. The continuous deterioration of air-quality conditions has considerably affected the economic development and health of China’s people. However, the diversity and complexity of the factors which affect air pollution render air quality monitoring data complex and nonlinear. To improve the accuracy of prediction of the air quality index (AQI) and obtain more accurate AQI data with respect to their nonlinear and nonsmooth characteristics, this study introduces an air quality prediction model based on the empirical mode decomposition (EMD) of LSTM and uses improved particle swarm optimization (IPSO) to identify the optimal LSTM parameters. First, the model performed the EMD decomposition of air quality data and obtained uncoupled intrinsic mode function (IMF) components after removing noisy data. Second, we built an EMD–IPSO–LSTM air quality prediction model for each IMF component and extracted prediction values. Third, the results of validation analyses of the algorithm showed that compared with LSTM and EMD–LSTM, the improved model had higher prediction accuracy and improved the model fitting effect, which provided theoretical and technical support for the prediction and management of air pollution.