Noise Decomposition Research Articles

Fault Feature Extraction Method for Rolling Bearings Based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise and Variational Mode Decomposition

Due to the difficulty in dealing with non-stationary and nonlinear vibration signals using the single decomposition method, it is difficult to extract weak fault features from complex noise; therefore, this paper proposes a fault feature extraction method for rolling bearings based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and variational mode decomposition (VMD) methods. CEEMDAN was used to decompose the signal, and the signal was then screened and reconstructed according to the component envelope kurtosis. Based on the kurtosis of the maximum envelope spectrum as the fitness function, the sparrow search algorithm (SSA) was used to perform adaptive parameter optimization for VMD, which decomposed the reconstructed signal into several IMF components. According to the kurtosis value of the envelope spectrum, the optimal component was selected for an envelope demodulation analysis to realize fault feature extraction for rolling bearings. Finally, by using open data sets and experimental data, the accuracy of envelope kurtosis and envelope spectrum kurtosis as a component selection index was verified, and the superiority of the proposed feature extraction method for rolling bearings was confirmed by comparing it with other methods.

A Short-term solar irradiance forecasting modelling approach based on three decomposition algorithms and Adaptive Neuro-Fuzzy Inference System

In this study, a case study of four Indian cities i.e. Ajmer, Jaipur, Jodhpur and Kota in the state of Rajasthan are considered wherein 30 min ahead data have been obtained via the data site of the National Institute of Wind Energy and Wind Resource (NIWE) on which a proposed Global Horizontal Irradiance (GHI) prediction technique for all seasons is applied. Here, data has been pre-processed using three different signal decomposition algorithms in parallel i.e. Empirical Mode Decomposition (EMD), Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Ensemble Empirical Mode Decomposition (EEMD). Further, based on Pearson's Correlation Coefficient (PCC), corresponding IMFs & corresponding Residual obtained using the three decomposition algorithms are compared amongst each other respectively and that corresponding IMFs & corresponding Residual are chosen for signal reconstruction which are having highest correlation coefficient values. Thereafter, these selected IMFs and Residual from each algorithm are combined to form single input. In this way, three inputs formed from three decomposition algorithms based on the PCC values are fed to Adaptive Neuro-Fuzzy Inference System (ANFIS) for solar irradiance forecasting. The proposed technique shows significantly higher accurate results with less than 2% MAPE for different seasons considered at the site locations considered. Further, it terms of performance, the proposed technique is found to be independent of site location.

A parsimonious ensemble with optimal deep learning and secondary decomposition for short-term wind speed forecasting

Wind energy needs to be used efficiently, which depends heavily on the accuracy and reliability of wind speed forecasting. However, the volatility and nonlinearity of wind speed make this difficult. In volatility and nonlinearity reduction, we sequentially apply complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and variational mode decomposition (VMD) to secondarily decompose the wind speed data. This framework, however, requires effectively modeling multiple uncertainty components. Eliminating this limitation, we integrate crow search algorithm (CSA) with deep belief network (DBN) to generate a unified optimal deep learning system, which not only eliminates the influence of multiple uncertainties, but also only adopts DBN as a predictor to realize parsimonious ensemble. Two experiments demonstrate the superiority of this system.

A novel hierarchical power allocation strategy considering severe wind power fluctuations for wind-storage integrated systems

The development of renewable energy dominated power system and the integration of large-scale wind-storage energy system are inevitable trends of the future society. This paper developed a new methodology of hierarchical power allocation strategy considering wind power fluctuation. A novel idea is presented that fully considers the fluctuation and energy proportion. Based on the Quartile Method (QM) and Hampel-Savitzky Golay (HSG) strategy, a layered-based method that considers the volatility interval to effectively overcome the limitations of a single strategy caused by severe volatility interval is proposed. Then, an improved power division method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Dynamic Mode Decomposition (DMD) is proposed where CEEMDAN is used to decompose the Intrinsic Mode Function (IMF) of the remaining fluctuations. Furthermore, according to the required setting of energy proportion, DMD strategy is used for power fitting of the hybrid energy storage system (HESS). Through experimental verification, the hierarchical strategy proposed in this paper can effectively reduce the probability of severe volatility interval or distortion interval and ensure the sequence in the ideal interval. Moreover, the proposed CEEMDAN-DMD strategy can realize the power division according to the energy proportion demand. In model verification, the volatility improvement ζλ1min = 85.4766%. Set the energy demand ≥ 80%, the fitting energy proportion accounts for 87.7533%, which outperforms the traditional power division method. In addition, it is demonstrated that the theoretical method proposed in this paper is reliable and promising.

Research on Denoising Method for Hydroelectric Unit Vibration Signal Based on ICEEMDAN-PE-SVD.

Vibration monitoring and analysis play a crucial role in the fault diagnosis of hydroelectric units. However, accurate extraction and identification of fault features from vibration signals are challenging because of noise interference. To address this issue, this study proposes a novel denoising method for vibration signals based on improved complementary ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), permutation entropy (PE), and singular value decomposition (SVD). The proposed method is applied for the analysis of hydroelectric unit sway monitoring. Firstly, the ICEEMDAN method is employed to process the signal and obtain several intrinsic mode functions (IMFs), and then the PE values of each IMF are calculated. Subsequently, based on a predefined threshold of PE, appropriate IMFs are selected for reconstruction, achieving the first denoising effect. Then, the SVD is applied to the signal after the first denoising effect, resulting in the SVD spectrum. Finally, according to the principle of the SVD spectrum and the variation in the singular value and its energy value, the signal is reconstructed by choosing the appropriate reconstruction order to achieve the secondary noise reduction effect. In the simulation and case analysis, the method is better than the commonly used wavelet threshold, SVD, CEEMDAN-PE, and ICEEMDAN-PE, with a signal-to-noise ratio (SNR) improvement of 6.9870 dB, 4.6789 dB, 8.9871 dB, and 4.3762 dB, respectively, and where the root-mean-square error (RMSE) is reduced by 0.1426, 0.0824, 0.2093 and 0.0756, respectively, meaning that our method has a better denoising effect and provides a new way for denoising the vibration signal of hydropower units.

GRU Neural Network Based on CEEMDAN–Wavelet for Stock Price Prediction

Stock indices are considered to be an important indicator of financial market volatility in various countries. Therefore, the stock market forecast is one of the challenging issues to decrease the uncertainty of the future direction of financial markets. In recent years, many scholars attempted to use different conventional statistical and deep learning methods to predict stock indices. However, the non-linear financial noise data will usually cause stochastic deterioration and time lag in forecast results, resulting in existing neural networks that do not demonstrate good prediction results. For this reason, we propose a novel framework to combine the gated recurrent unit (GRU) neural network with the complete ensemble empirical mode decomposition of adaptive noise (CEEMDAN) to predict the stock indices with better accuracy, in which the wavelet threshold method is especially used to denoise high-frequency noises in the sub-signals to exclude noise interference for future data predictions. Firstly, we choose representative datasets collected from the closing prices of S&P500 and CSI 300 stock indices to evaluate the proposed GRU-CEEMDAN–wavelet model. Additionally, we compare the improved model to the traditional ARIMA and several modified neural network models using different gate structures. The result shows that the mean values of MSE and MAE for GRU based on CEEMDAN–wavelet are the smallest by significance analysis. Overall, we found that our model could improve prediction accuracy and alleviates the time lag problem.

An efficient lightweight neural network using BiLSTM-SCN-CBAM with PCA-ICEEMDAN for diagnosing rolling bearing faults

This study proposes an efficient rolling bearing fault diagnosis model of a hybrid neural network with a lightweight attention mechanism. Firstly, to achieve the low complexity of deep learning (DL) computation, data reduction and denoising are performed by sparse convolutional network (principal component analysis and improved complete ensemble empirical modal decomposition of adaptive noise), then processed data is imported to the hybrid neural network model with convolutional block attention module. The bi-directional long short-term memory and sparse convolutional networks are used in the backbone of the model. A lightweight, generalized attention mechanism is introduced to the last layer of the model for enhancing feature learning, which can further improve the diagnostic accuracy and efficiency. Compared with existing DL fault diagnosis models, In simulating the most realistic cross-conditions and cross-platform conditions, which leads to the random nature of fault generation and makes model diagnosis more complex, the proposed method still maintains less running time and excellent diagnostic accuracy. Finally, the experimental results fully prove that the model has reliable robust and efficient, and it achieves the best balance of diagnostic accuracy and diagnostic efficiency of the hybrid DL model.

A contrastive learning based universal representation for time series forecasting

Time series forecasting has wide applications in our daily lives, such as meteorological warnings and decision-making. However, traditional supervised models do not perform well on forecasting tasks due to the lack of annotated training data available in real time series. Recently, researchers have proposed self-supervised methods, especially the contrastive learning approach, which can build a universal representation framework for downstream tasks through data augmentation and contrastive loss. Although it alleviates the shortage of labeled data, the data augmentation approach directly transferred from computer vision is not appropriate for the time domain due to noise vectors and unrelated variables that may interfere with the accuracy of representation. In this paper, we propose a novel time series forecasting model based on disentangled seasonal-trend representation named ACST. It employs an improved cycle generative adversarial data augmentation method to generate samples close to real data for contrastive loss. Moreover, we apply gated residual networks and a noise decomposition module to reduce the impact of different noise vectors and feature variable weights on the results. Extensive experiments show that ACST achieves an average improvement of 26.8% on six benchmarks.

Analysis of fully discrete mixed finite element scheme for stochastic Navier–Stokes equations with multiplicative noise

This paper is concerned with stochastic incompressible Navier–Stokes equations with multiplicative noise in two dimensions with respect to periodic boundary conditions. Based on the Helmholtz decomposition of the multiplicative noise, semi-discrete and fully discrete time-stepping algorithms are proposed. The convergence rates for mixed finite element methods based time-space approximation with respect to convergence in probability for the velocity and the pressure are obtained. Furthermore, with establishing some stability and using the negative norm technique, the partial expectations of the $$H^1$$ and $$L^2$$ norms of the velocity error are proved to converge optimally.

Runoff prediction of lower Yellow River based on CEEMDAN–LSSVM–GM(1,1) model

Accurate medium and long-term runoff forecasts play a vital role in guiding the rational exploitation of water resources and improving the overall efficiency of water resources use. Machine learning is becoming a common trend in time series forecasting research. Least squares support vector machine (LSSVM) and grey model (GM(1,1)) have received much attention in predicting rainfall and runoff in the last two years. “Decomposition-forecasting” has become one of the most important methods for forecasting time series data. Complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) decomposition method has powerful advantages in dealing with nonlinear data. Least squares support vector machine (LSSVM) has strong nonlinear fitting ability and good robustness. Gray model (GM(1,1)) can solve the problems of little historical data and low serial integrity and reliability. Based on their respective advantages, a combined CEEMDAN–LSSVM–GM(1,1) model was developed and applied to the runoff prediction of the lower Yellow River. To verify the reliability of the model, the prediction results were compared with the single LSSVM model, the CEEMDAN–LSSVM model and the CEEMDAN–support vector machines (SVM)–GM(1,1). The results show that the combined CEEMDAN–LSSVM–GM(1,1) model has a high accuracy and the prediction results are better than other models, which provides an effective prediction method for regional medium and long-term runoff prediction and has good application prospects.

A spatiotemporal model for PM2.5 prediction based on the K‐Core idea and label distribution

AbstractWith the increasingly severe problem of PM2.5 environmental pollution, the threat to human health is gradually increasing. Therefore, accurate prediction of PM2.5 concentration is of great significance to the healthy life of human beings. To make up for the deficiencies of previous studies on PM2.5 concentration prediction, a spatiotemporal model (Spatiotemporal prediction model of label distribution, LDSPM) for PM2.5 concentration prediction based on the K‐Core algorithm concept and label distribution learning was proposed. Leveraging K‐Core ideas and the label distribution support vector regression model of the label distribution paradigm, the influence weight of each meteorological factor on PM2.5 concentration in each piece of data was obtained with the decomposition of meteorological factors using the complete ensemble empirical mode decomposition of adaptive noise. Using a long short‐term memory neural network to predict each decomposed signal and obtain the forecast data of meteorological factors. Finally, according to the expected weight and meteorological factor data, a particle swarm optimization extreme learning machine is used to train the prediction, and the predicted value of PM2.5 is obtained. The experimental results show that the forecasting model performs better than other combined and single forecasting models. It provides new directions and ideas for PM2.5 concentration prediction.

Shadow-Enhanced Self-Attention and Anchor-Adaptive Network for Video SAR Moving Target Tracking

Video synthetic aperture radar (Video SAR) has drawn much attention because it can continuously observe and track the moving target. Rather than tracking the target directly, it is better to track its shadow because the shadow has no location shift, and the back-scattering characteristic is stable. However, most current shadow tracking methods not only suffer from false alarms because their discrimination capacities are not good enough but also suffer from missed detection because the feature extraction capacities are limited under complicated environment. Therefore, we propose a shadow-enhanced self-attention and anchor-adaptive network (SE-SA-AAN) to achieve accurate moving target tracking for Video SAR. Firstly, the pre-processing technique sparse low-rank noise decomposition (SLRND) is proposed for enhancing shadows’ salience to facilitate subsequent feature extraction. Secondly, the transformer self-attention mechanism (TSAM) is embedded in the parameters-shared backbone in the feature extraction network to concentrate on regions of interests for suppressing clutter interference. Then, the representative features are input to the detector and tracker. The detector adds the semantic guided anchor-adaptive mechanism (SGAAM) to obtain optimized anchors that match the shadows’ location and shape in each frame. Meanwhile, the tracker applies a Siamese network to achieve trajectory tracking for each shadow. Based on the detection and tracking results, a data association is applied to achieve moving targets tracking. Finally, experiments on Sandia National Laboratories (SNL) data demonstrate that SE-SA-AAN outperforms the state-of-the-art methods FairMOT, TransTrack and Centertrack by 6.4%, 7.8% and 8.3% multiple object tracking accuracy (MOTA) separately.

Prediction of gas consumption based on LSTM-BPNN hybrid model

ABSTRACT The economic benefits of iron and steel enterprises require the support of Byproduct gas prediction. A novel hybrid model for data pre-processing is proposed to ensure the stability and accuracy of gas prediction. First, each component obtained from Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) decomposition is processed using Sample Entropy (SE), and the members are denoised and restructured by singular spectrum analysis (SSA). Then, the restructured data are predicted by the Back Propagation Neural Network (BPNN) and Long Short-Term Memory (LSTM) and weighted by Particle Swarm Optimization (PSO). Finally, the data are integrated to obtain the final results. The novel hybrid model makes allowances for the linear and nonlinear characteristics of the series and successfully overcomes the limitations of individual models to get accurate and stable prediction results. The results show that the prediction accuracy is improved by 20% on average by adopting the novel hybrid model.

Hybrid wind speed forecasting using ICEEMDAN and transformer model with novel loss function

Wind energy technologies have been investigated extensively due to worldwide environmental challenges and rising energy demand. Therefore, accurate and reliable wind speed forecasts are essential for large scale wind power integration. But seasonal and stochastic winds make forecasting difficult. Hence, this study proposes a novel loss function-based hybrid deep learning architecture for the wind speed forecasting (WSF). The proposed hybrid model is developed using the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) decomposition method for the denoising wind speed data, and transformer network (TRA) with a novel Kernel MSE loss function (NLF) for the WSF (NLF-ICEEMDAN-TRA). As the MSE loss function is sensitive to outliers and fails to identify the non-linear characteristics in wind speed data, a novel kernel MSE loss function has been used for the training of the transformer network. Wind speed data from two wind farms located in Block Island and the Gulf Coast have been used to validate the effectiveness of the proposed hybrid model. Because current WSF methods performance declines as the time ahead rises, the proposed hybrid model is verified using eight time horizons: 5 min, 10 min, 15 min, 30 min, 1 h, 2 h, 24 h and 48 h ahead WSF. To investigate the performance of proposed hybrid model in WSF, six individual WSF models, six hybrid WSF models, and six NLF based hybrid WSF models are employed for comparative analysis. The experimental results demonstrated that the proposed hybrid model achieved the best results for all eight time horizon WSF for both wind farm sites, with a significant improvement.

Modeling topological nature of gas–liquid mixing process inside rectangular channel using RBF-NN combined with CEEMDAN-VMD

In this work, signal processing techniques (i.e., complete ensemble empirical mode decomposition with adaptive noise-variational mode decomposition) and neural network algorithms (i.e., RBF neural network) are used for predicting the topological nature of gas-liquid mixtures in rectangular channels. Specifically, the indicator first Betti number is introduced to describe the topological structure of the gas-liquid mixing process. Firstly, the original topological nature time series were secondary decomposed by the complete ensemble empirical mode decomposition with adaptive noise and variational mode decomposition to reduce the randomness and volatility of the original signal. Then the decomposed signals are fed into RBF neural network for modeling and prediction. The important specific and quantitative results are that the prediction accuracy of topological nature time series can be improved by 0.46%∼5.84% with the hybrid model proposed. Moreover, the prediction accuracy of each working condition is more than 95%, and the prediction accuracy under C1 condition is the highest, reaching 97.93%. The conclusion from the above is that the hybrid model can improve the prediction accuracy and efficiency, reduce the number of experiments and shorten the design cycle. The gap it filled in the literature is that the prediction of gas–liquid mixing topological nature time series in rectangular channels. In addition, it is of great significance to engineering design in the fields of heat energy and chemical engineering.

Semidecoupled decomposition-based fractional-order variational model for low-light enhancement

Low-light enhancement is an important technique for improving image quality. This is because low-light enhancement is expected to improve image visibility while maintaining visual naturalness of the image. In recent years, many methods have been researched to enhance low-light images, including histogram-based, fusion-based, and learning-based methods. The most representative and widely used low-light image enhancement method is the so-called Retinex-based method. However, they tend to have many limitations. The limitations of the Retinex-based method are as follows. (1) Due to strong imaging noise or less-effective image decomposition, this results in a large number of artifacts in the enhanced results. (2) Although the first problem can be partially solved by exploring prior information, it often complicates the optimization process. (3) Small-magnitude details are often lost in enhanced results. To overcome these drawbacks, we propose a model called the fractional-order Retinex model. At the same time, Retinex images are decomposed in an effective semidecoupled way. More concretely, the illumination layer T is gradually estimated only with the observed image S based on the proposed variation model, whereas the reflectance layer R is jointly estimated by the intermediate T and S. Experimental results demonstrate the effectiveness of our method.

Prediction of CORS Water Vapor Values Based on the CEEMDAN and ARIMA-LSTM Combination Model

By relying on the advantages of a uniform site distribution and continuous observation of the Continuously Operating Reference Stations (CORS) system, real-time high-precision Global Navigation Satellite System/Precipitable Water Vapor (GNSS/PWV) data interpretation can be carried out to achieve accurate monitoring of regional water vapor changes. The study of the atmospheric water vapor content and distribution changes is the basis for the realization of rainfall forecasting and water vapor circulation research. Such research can provide data support for the effective forecasting of regional precipitation in megacities and the construction of a more sensitive flood prevention and warning system. Nowadays, a single model is often adopted for GNSS/PWV time series. This makes it challenging to match the high randomness characteristic of water vapor change. This study proposes a hybrid model that takes into account the linear and nonlinear aspects of water vapor data by using complete empirical mode decomposition (CEEMDAN) of adaptive noise, differential autoregressive integrated moving average (ARIMA), and the long-short-term memory network (LSTM). The CEEMDAN is used to decompose the water vapor data series. Then, the high- and low-frequency data are modeled separately, reducing the sequence’s complexity and non-stationarity. In selecting the prediction model, we use the ARIMA model for the high-frequency series and the ARIMA–GWO–LSTM ensemble model for the low-frequency sub-series and residual series. The model is verified using GNSS/PWV time series data collected at the Hong Kong CORS station in July 2021. The results show the following: (1) The LSTM model optimized by the grey wolf optimization algorithm (GWO) is comparable with the single LSTM model in the low-frequency sequence prediction process, and the error items are reduced by 30% after calculation. (2) During the process from CEEMDAN decomposition to the use of the combination model for prediction, the accuracy evaluation indexes of the station increase by more than 20%. The interpolation method can accurately determine the regional water vapor spatial variation, which is of practical significance for local rainfall forecasting. High-frequency data obtained by CEEMDAN decomposition demonstrate the dramatic changes in water vapor before and after the rainfall, which can provide ideas for improving the accuracy of rainfall forecasting.

Two-stage hybrid model for hydrological series prediction based on a new method of partitioning datasets

• A new method of dividing datasets based on PSO is proposed. • BNN incorporates Langevin dynamics in the parameter update step (LDBNN). • Integration of LDBNN, CEEMDAN, VMD and the proposed method yield the hybrid model. • The new data dividing method performs better than the traditional dividing method. • The hybrid model yields excellent point and interval predictions. Obtaining accurate point estimates and reliable interval prediction results for rainfall and runoff series is important to aid in water resource decision-making and planning management in a changing environment. In this paper, we propose a two-stage hybrid model (P-CVEE-LDBNN). The model uses the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and variational mode decomposition (VMD) methods for data preprocessing and a Bayesian neural network (LDBNN) based on Langevin dynamics as a prediction model. Then, a new dataset partitioning method based on a particle swarm optimization algorithm, which is different from the traditional dataset partitioning method, is applied. A one-stage hybrid model (T-CEE-LDBNN) that integrates the CEEMDAN method and the traditional method of partitioning datasets, a one-stage hybrid model (P-CEE-LDBNN) that integrates the CEEMDAN method and the new method of partitioning datasets, and the two-stage hybrid P-CVEE-LDBNN model were compared. These models were applied to monthly runoff and monthly precipitation series from seven hydrological and meteorological stations in the Yellow River basin. The mean absolute error (MAE), absolute root mean squared error (RMSE), Nash–Sutcliffe efficiency (NSE), and correlation coefficient (R) were used to evaluate the predictive ability of the models. The containment ratio (CR), the average bandwidth (IW), and the average asymmetry degree (S) were used to evaluate the interval prediction results of the models. The results show that (1) the subset obtained with the new method of data partitioning considering statistical properties is more favorable for model prediction. (2) The second decomposition approach based on VMD is a reliable method that can significantly improve the prediction accuracy of the final model. (3) Compared with traditional neural networks that can only obtain deterministic point prediction results, Bayesian methods can provide intervals with prediction results, making the results of the two-stage hybrid P-CVEE-LDBNN model highly reliable.

Extraction of multi-scale features enhances the deep learning-based daily PM2.5 forecasting in cities

Characterising the daily PM2.5 concentration is crucial for air quality control. To govern the status of the atmospheric environment, a novel hybrid model for PM2.5 forecasting was proposed by introducing a two-stage decomposition technology of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and variational mode decomposition (VMD); subsequently, a deep learning approach of long short-term memory (LSTM) was proposed. Five cities with unique meteorological and economic characteristics were selected to assess the predictive ability of the proposed model. The results revealed that PM2.5 pollution was generally more severe in inland cities (66.98 ± 0.76 μg m−3) than in coastal cities (40.46 ± 0.40 μg m-3). The modelling comparison showed that in each city, the secondary decomposition algorithm improved the accuracy and prediction stability of the prediction models. When compared with other prediction models, LSTM effectively extracted featured information and achieved relatively accurate time-series prediction. The hybrid model of CEEMDAN-VMD-LSTM achieved a better prediction in the five cities (R2 = 0.9803 ± 0.01) compared with the benchmark models (R2 = 0.7537 ± 0.03). The results indicate that the proposed approach can identify the inherent correlations and patterns among complex datasets, particularly in time-series analysis.

A novel hybrid model for six main pollutant concentrations forecasting based on improved LSTM neural networks

In recent years, air pollution has become a factor that cannot be ignored, affecting human lives and health. The distribution of high-density populations and high-intensity development and construction have accentuated the problem of air pollution in China. To accelerate air pollution control and effectively improve environmental air quality, the target of our research was cities with serious air pollution problems to establish a model for air pollution prediction. We used the daily monitoring data of air pollution from January 2016 to December 2020 for the respective cities. We used the long short term memory networks (LSTM) algorithm model to solve the problem of gradient explosion in recurrent neural networks, then used the particle swarm optimization algorithm to determine the parameters of the CNN-LSTM model, and finally introduced the complete ensemble empirical mode decomposition of adaptive noise (CEEMDAN) decomposition to decompose air pollution and improve the accuracy of model prediction. The experimental results show that compared with a single LSTM model, the CEEMDAN-CNN-LSTM model has higher accuracy and lower prediction errors. The CEEMDAN-CNN-LSTM model enables a more precise prediction of air pollution, and may thus be useful for sustainable management and the control of air pollution.