CNN-LSTM Models Research Articles

Monthly climate prediction using deep convolutional neural network and long short-term memory

Climate change affects plant growth, food production, ecosystems, sustainable socio-economic development, and human health. The different artificial intelligence models are proposed to simulate climate parameters of Jinan city in China, include artificial neural network (ANN), recurrent NN (RNN), long short-term memory neural network (LSTM), deep convolutional NN (CNN), and CNN-LSTM. These models are used to forecast six climatic factors on a monthly ahead. The climate data for 72 years (1 January 1951–31 December 2022) used in this study include monthly average atmospheric temperature, extreme minimum atmospheric temperature, extreme maximum atmospheric temperature, precipitation, average relative humidity, and sunlight hours. The time series of 12 month delayed data are used as input signals to the models. The efficiency of the proposed models are examined utilizing diverse evaluation criteria namely mean absolute error, root mean square error (RMSE), and correlation coefficient (R). The modeling result inherits that the proposed hybrid CNN-LSTM model achieves a greater accuracy than other compared models. The hybrid CNN-LSTM model significantly reduces the forecasting error compared to the models for the one month time step ahead. For instance, the RMSE values of the ANN, RNN, LSTM, CNN, and CNN-LSTM models for monthly average atmospheric temperature in the forecasting stage are 2.0669, 1.4416, 1.3482, 0.8015 and 0.6292 °C, respectively. The findings of climate simulations shows the potential of CNN-LSTM models to improve climate forecasting. Climate prediction will contribute to meteorological disaster prevention and reduction, as well as flood control and drought resistance.

Prediction of IGBT Gate Oxide Layer's Performance Degradation Based on MultiScaleFormer Network.

Insulated gate bipolar transistors (IGBTs) are widely used in power electronic devices, and their health prediction problems have attracted much attention in the field of power electronic equipment health management. The performance degradation of IGBT gate oxide is one of the most important failure modes. In order to analyze this failure mechanism and the ease of implementation of a monitoring circuit, the gate leakage current of IGBTs was selected as the fault precursor parameter for the degradation of their gate oxide performance, and feature selection and fusion were carried out by using time domain characteristic analysis, grayscale correlation, Mahalanobis distance, Kalman filter, and other methods. Thus, a health indicator was obtained to characterize the degradation of IGBT performance, which was used to indicate the degree of aging of the IGBT gate oxide layer. In this paper, we propose an improved degradation prediction model called MultiScaleFormer, inspired by advanced design ideas of the iTransformer network architecture, combined with the health parameters of IGBTs to construct a degradation prediction model for the IGBT gate oxide layer. MultiScaleFormer showed the highest fitting accuracy compared with the Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), Support Vector Regression (SVR), Gaussian Process Regression (GPR), CNN-LSTM, and Transformer models in our experiment. The mean absolute error (MAE) of the MultiScaleFormer prediction was as low as 0.0087. Extraction of the health indicator and the construction and verification of the degradation prediction model were carried out on the dataset released by the NASA-Ames Laboratory. These results demonstrate the feasibility of the gate leakage current as a fault precursor parameter for IGBT gate oxide failure, and the feasibility and accuracy of the MultiScaleFormer prediction model for IGBT performance degradation.

A Hybrid Model of Conformer and LSTM for Ocean Wave Height Prediction

This study proposes a hybrid model (Conformer-LSTM) based on Conformer and Long Short-Term Memory networks (LSTM) to overcome the limitations of existing techniques and enhance the accuracy and generalizability of wave height predictions. The model combines the advantages of self-attention mechanisms and convolutional neural networks. It captures global dependencies through multi-head self-attention and utilizes convolutional layers to extract local features, thereby enhancing the model’s adaptability to dynamic changes in time series. The LSTM component handles long-term dependencies, optimizing the coherence and stability of predictions. Additionally, an adaptive feature fusion weight network is introduced to further improve the model’s recognition and utilization efficiency of key features. Experimental data come from the National Oceanic and Atmospheric Administration buoy data, covering wave height, wind speed, and other data from key maritime areas. Evaluation metrics include Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R2), ensuring a comprehensive assessment of model performance. The results show that the Conformer-LSTM model outperforms traditional LSTM, CNN, and CNN-LSTM models at multiple sites, confirming its potential in wave height prediction.

Enhancing COVID-19 detection through integrated CNN-LSTM models: Addressing the challenge of varied radiological presentations

In an era where COVID-19's varied radiological presentations challenge diagnostic precision, this paper introduces a novel fusion of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models to refine detection capabilities. The paper capitalizes on an expansive radiography database, utilizing lung segmentation and data augmentation to bolster the training set. Advanced CNN models extract salient features, which are then classified through meticulously optimized machine learning algorithms. The synergy between diverse CNN architectures and Bayesian Optimization markedly enhances hyperparameter tuning, culminating in a significant uplift in diagnostic accuracy. The integration of LSTM models further allows for nuanced severity assessments, paving the way for more personalized and effective pandemic management. This research heralds a new frontier in medical diagnostics, promising a leap in the accuracy of pandemic detection.

Improving Bus Passenger Flow Prediction Using Bi-LSTM Fusion Model and SMO Algorithm

Bus passenger flow prediction integrates data analytics and modelling techniques to forecast the number of passengers using bus services, incorporating historical usage patterns, demographics, weather, and events for optimal scheduling and resource allocation. The Bi-LSTM fusion model enhances accuracy by processing past and future features simultaneously, leveraging bidirectional LSTM layers and an attention mechanism to capture temporal dependencies. This approach not only refines insights crucial for urban mobility challenges like traffic management and demand forecasting but also improves route planning and service efficiency. The SMO algorithm initializes with a diverse spider monkey population exploring solution spaces. Through local and global leader phases, it iteratively updates positions based on fitness and probabilistic selections, maintaining a balance between exploration and exploitation. Perturbation-based updates in the local leader phase ensure adaptability, preventing premature convergence, while the global leader phase guides towards better solutions, enhancing efficiency in complex optimization tasks and promoting dynamic adaptation. In Dataset 1, the proposed model achieved a training time of 137 seconds, slightly longer than HA (115s), SARIMA (112s), GRU (123s), and ST-ResNet (113s). It demonstrated superior accuracy at 89%, surpassing HA (66%), SARIMA (68%), GRU (63%), DeepST (78%), and ST-ResNet (84%). In Dataset 2, the model exhibited the lowest RMSE, MAE, and MAPE%, indicating superior predictive accuracy over SVR, CNN, GCN, LSTM, and CONV LSTM models. These findings validate the proposed model's effectiveness in enhancing predictive capabilities for transit forecasting, underscoring its potential to optimize urban mobility and transportation management strategies significantly.

A Photovoltaic Prediction Model with Integrated Attention Mechanism

Solar energy has become a promising renewable energy source, offering significant opportunities for photovoltaic (PV) systems. Accurate and reliable PV generation forecasts are crucial for efficient grid integration and optimized system planning. However, the complexity of environmental factors, including seasonal and daily patterns, as well as social behaviors and user habits, presents significant challenges. Traditional prediction models often struggle with capturing the complex nonlinear dynamics in multivariate time series, leading to low prediction accuracy. To address this issue, this paper proposes a new PV power prediction method that considers factors such as light, air pressure, wind direction, and social behavior, assigning different weights to them to accurately extract nonlinear feature relationships. The framework integrates long short-term memory (LSTM) and gated recurrent units (GRU) to capture local time features, while bidirectional LSTM (BiLSTM) and an attention mechanism extract global spatiotemporal relationships, effectively capturing key features related to historical output. This improves the accuracy of multi-step predictions. To verify the feasibility of the method for multivariate time series, we conducted experiments using PV power prediction as a scenario and compared the results with LSTM, CNN, BiLSTM, CNN-LSTM and GRU models. The experimental results show that the proposed method outperforms these models, with a mean absolute error (MAE) of 12.133, root mean square error (RMSE) of 14.234, mean absolute percentage error (MAPE) of 2.1%, and a coefficient of determination (R2) of 0.895. These results indicate the effectiveness and potential of the method in PV prediction tasks.

Fault diagnosis method for hydropower unit via the incorporation of chaotic quadratic interpolation optimized deep learning model

Hydro-turbine fault diagnosis is crucial for hydropower plants’ safe and stable operation. This paper proposes a deep learning model for hydro-turbine fault diagnosis based on chaotic quadratic interpolation optimization (CQIO). Chaotic mapping optimizes the initial population of the QIO algorithm by introducing randomness and diversity, improving the algorithm’s performance and stability. The CQIO is used to calculate the optimal hyperparameter combinations for CNN-LSTM models. It can improve the model’s stability and reduce computational resource consumption. The results show that the fault accuracy of the proposed method reaches 96.7% and 93.6%, respectively, which is higher than the CNN, LSTM, CNN-LSTM, and QIO-CNN-LSTM models. Notably, the diagnostic accuracy about impact faults exceeds that of wear faults, with the latter exhibiting an augmented diagnostic accuracy as sediment increases.

Enhancing brain tumor detection: integrating CNN-LSTM and CNN-BiLSTM models for efficient classification in MRI images

Enhancing brain tumor detection: integrating CNN-LSTM and CNN-BiLSTM models for efficient classification in MRI images

Optimized Ensemble Learning Models Based on Clustering and Hybrid Deep Learning for Wireless Intrusion Detection

Machine learning (ML) and deep learning (DL) are used in numerous fields, particularly to develop effective intrusion detection systems (IDS). Existing wireless network IDS, which rely on a single ML algorithm and have limitations. These include a high rate of false positives, difficulties in recognizing distinct attack patterns, and a high acquisition cost for annotated training datasets. However, hostile threats are always evolving, networks need a smart security solution. In comparison to other ML approaches, DL algorithms are more successful in intrusion detection. This paper presents a DL based ensemble model that combines Multi-verse through Chaotic Atom Search Optimization (MCA) for preprocessing, which eliminates unsolicited/recurrent information in the dataset. The process of optimized feature selection uses Principal Component Analysis (PCA), Chaotic Manta-ray Foraging Optimizations (CMFO), and a grounded grouping method to partition the optimized feature dataset into k-diverse clusters. The recommended model then stacks Support Vector Machine (SVM) as the ensemble model’s meta-learner classifier, pre-training the hybrid DL prototypes using the optimized feature dataset cluster. The CNN-LSTM and CNN-GRU models, which integrate Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), are the hybrid DL prototype’s key components. The suggested model’s performance has been enhanced and compared to six ML techniques: NB, SVM, J48, RF, MLP, and kNN models, utilizing measures such as accuracy, precision, recall, and F-measure. The public can access the Aegean Wi-Fi Intrusion Dataset (AWID) which is used for evaluating the recommended model and is outperformed the contemporary models in the literature.

Weighted voting ensemble of hybrid CNN-LSTM Models for vision-based human activity recognition

Weighted voting ensemble of hybrid CNN-LSTM Models for vision-based human activity recognition

Deep Learning Approaches for Water Stress Forecasting in Arboriculture Using Time Series of Remote Sensing Images: Comparative Study between ConvLSTM and CNN-LSTM Models

Irrigation is crucial for crop cultivation and productivity. However, traditional methods often waste water and energy due to neglecting soil and crop variations, leading to inefficient water distribution and potential crop water stress. The crop water stress index (CWSI) has become a widely accepted index for assessing plant water status. However, it is necessary to forecast the plant water stress to estimate the quantity of water to irrigate. Deep learning (DL) models for water stress forecasting have gained prominence in irrigation management to address these needs. In this paper, we present a comparative study between two deep learning models, ConvLSTM and CNN-LSTM, for water stress forecasting using remote sensing data. While these DL architectures have been previously proposed and studied in various applications, our novelty lies in studying their effectiveness in the field of water stress forecasting using time series of remote sensing images. The proposed methodology involves meticulous preparation of time series data, where we calculate the crop water stress index (CWSI) using Landsat 8 satellite imagery through Google Earth Engine. Subsequently, we implemented and fine-tuned the hyperparameters of the ConvLSTM and CNN-LSTM models. The same processes of model compilation, optimization of hyperparameters, and model training were applied for the two architectures. A citrus farm in Morocco was chosen as a case study. The analysis of the results reveals that the CNN-LSTM model excels over the ConvLSTM model for long sequences (nine images) with an RMSE of 0.119 and 0.123, respectively, while ConvLSTM provides better results for short sequences (three images) than CNN-LSTM with an RMSE of 0.153 and 0.187, respectively.

Early-stage degradation trajectory prediction for lithium-ion batteries: A generalized method across diverse operational conditions

Accurate early prediction of the degradation trajectory of lithium-ion batteries (LIBs) can accelerate battery development, production, and design optimization. However, existing early-stage prediction methods for degradation trajectory prediction face challenges in dealing with insufficient data and the generalization under different operational conditions. To address this issue, a novel transfer learning based data-driven method is proposed, which integrates convolutional neural network (CNN) and long short-term memory (LSTM) neural networks. Additionally, we incorporate a temporal attention (TA) mechanism to selectively focus on informative capacity fade patterns and leverage Bayesian Optimization (BO) for hyper-parameters optimization. The proposed method is validated using experimental degradation data from six 5Ah low-temperature batteries (−20 °C, −10 °C) and public datasets. Results demonstrate high accuracy leveraging merely 10% of initial battery data. For the low-temperature aging experimental data, the best prediction result proposed is 0.025 Ah root mean square error (RMSE) and 0.019 Ah mean absolute error (MAE). Compared to baseline CNN-LSTM models, our framework achieved an average reduction of 62.6% in RMSE for early battery capacity trajectory prediction, while maintaining high accuracy across diverse operational conditions. The proposed framework puts forward a promising solution toward enabling adaptive data-driven capacity monitoring under different practical application conditions.

On Stock Volatility Forecasting under Mixed-Frequency Data Based on Hybrid RR-MIDAS and CNN-LSTM Models

Most of the deep-learning algorithms on stock price volatility prediction in the existing literature use data such as same-frequency market indicators or technical indicators, and less consider mixed-frequency data, such as macro-data. Compared with the traditional model that only inputs the same-frequency data such as technical indicators and market indicators, this study proposes an improved deep-learning model based on mixed-frequency big data. This paper first introduces the reserve restricted mixed-frequency data sampling (RR-MIDAS) model to deal with the mixed-frequency data and, secondly, extracts the temporal and spatial features of volatility series by using the parallel model of CNN-LSTM and LSTM, and finally utilizes the Optuna framework for hyper-parameter optimization to achieve volatility prediction. For the deep-learning model with mixed-frequency data, its RMSE, MAE, MSLE, MAPE, SMAPE, and QLIKE are reduced by 18.25%, 14.91%, 30.00%, 12.85%, 13.74%, and 23.42%, respectively. This paper provides a more accurate and robust method for forecasting the realized volatility of stock prices under mixed-frequency data.

CNN-BiLSTM: A Novel Deep Learning Model for Near-Real-Time Daily Wildfire Spread Prediction

Wildfires significantly threaten ecosystems and human lives, necessitating effective prediction models for the management of this destructive phenomenon. This study integrates Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM) modules to develop a novel deep learning model called CNN-BiLSTM for near-real-time wildfire spread prediction to capture spatial and temporal patterns. This study uses the Visible Infrared Imaging Radiometer Suite (VIIRS) active fire product and a wide range of environmental variables, including topography, land cover, temperature, NDVI, wind informaiton, precipitation, soil moisture, and runoff to train the CNN-BiLSTM model. A comprehensive exploration of parameter configurations and settings was conducted to optimize the model’s performance. The evaluation results and their comparison with benchmark models, such as a Long Short-Term Memory (LSTM) and CNN-LSTM models, demonstrate the effectiveness of the CNN-BiLSTM model with IoU of F1 Score of 0.58 and 0.73 for validation and training sets, respectively. This innovative approach offers a promising avenue for enhancing wildfire management efforts through its capacity for near-real-time prediction, marking a significant step forward in mitigating the impact of wildfires.

DeepFogAQ: A fog-assisted decentralized air quality prediction and event detection system

Air pollution is a pervasive environmental and public health concern, prompting the imperative development of predictive systems to facilitate proactive interventions. The complexity of predicting air pollution arises from intricate factors, including the accumulation of pollutants, traffic dynamics, and industrial emissions. Traditional methodologies, reliant on historical or real-time data analysis, often encounter limitations in providing comprehensive and accurate solutions to this multifaceted problem. To address the existing problem, we propose a novel fog-assisted decentralized air quality prediction and event detection system (DeepFogAQ) for managing air pollution of future cities. We integrate Deep Learning (DL), Fog Computing (FC), Complex Event Processing (CEP), and virtualization technologies within the architecture of DeepFogAQ. Specifically, for predicting pollutant concentrations, we employ Transformers, CNN-LSTM, GRU, and RFR models. Additionally, we construct Fog and Cloud layers based on container-based virtualization technology. To demonstrate the feasibility of the system, the developed ML/DL models were run on DeepFogAQ and alarm levels for future air quality were derived. In this way, both the success of the prediction models and the validity of the architecture were ensured. Experimental results showed that Transformers is the most successful model in air quality prediction and event detection. As a result, the proposed DeepFogAQ architecture has the potential to offer a powerful alternative to decision-makers to solve the air pollution problem with its decentralized, scalable, and fault-tolerant structure.

Refining Short-Term Power Load Forecasting: An Optimized Model with Long Short-Term Memory Network

Short-term power load forecasting involves the stable operation and optimal scheduling of the power system. Accurate load forecasting can improve the safety and economy of the power grid. Therefore, how to predict power load quickly and accurately has become one of the urgent problems to be solved. Based on the optimization parameter selection and data preprocessing of the improved Long Short-Term Memory Network, the study first integrated particle swarm optimization algorithm to achieve parameter optimization. Then, combined with convolutional neural network, the power load data were processed to optimize the data and reduce noise, thereby enhancing model performance. Finally, simulation experiments were conducted. The PSO-CNN-LSTM model was tested on the GEFC dataset and demonstrated stability of up to 90%. This was 22% higher than the competing CNN-LSTM model and at least 30% higher than the LSTM model. The PSO-CNN-LSTM model was trained with a step size of 1.9×10^4, the relative mean square error was 0.2345×10^-4. However, when the CNN-LSTM and LSTM models were trained for more than 2.0×10^4 steps, they still did not achieve the target effect. In addition, the fitting error of the PSOCNN-LSTM model in the GEFC dataset was less than 1.0×10^-7. In power load forecasting, the PSOCNN- LSTM model's predicted results had an average absolute error of less than 1.0% when compared to actual data. This was an improvement of at least 0.8% compared to the average absolute error of the CNNLSTM prediction model. These experiments confirmed that the prediction model that combined two methods had further improved the speed and accuracy of power load prediction compared to traditional prediction models, providing more guarantees for safe and stable operation of the power system.

Research on electric vehicle charging load prediction method based on spectral clustering and deep learning network

With the increasing prominence of environmental and energy issues, electric vehicles (EVs) as representatives of clean energy vehicles have experienced rapid development in recent years, and the charging load has also exhibited statistical characteristics. Accurate prediction of EV charging load is crucial to improve grid load dispatch and intelligent level. However, current research on EV charging load prediction still faces challenges such as data reliability, complexity and variability of charging behavior, uncertainty, and lack of standardization methods. Therefore, this paper proposes an electric vehicle charging load prediction method based on spectral clustering and deep learning network (SC-CNN-LSTM). Firstly, to address the insufficient amount of EV charging load data, this paper proposes to use Monte Carlo simulation to sample and simulate historical load data. Then, in order to identify the internal structure and patterns of charging load, the sampled and simulated dataset is clustered using spectral clustering, dividing the data into different clusters, where each cluster represents samples with similar charging load characteristics. Finally, based on the different sample features of each cluster, corresponding CNN-LSTM models are constructed and trained and predict using the respective data. By modifying the model parameters, the prediction accuracy of the model is improved. Through comparative experiments, the proposed method in this paper has significantly improved prediction accuracy compared to traditional prediction methods without clustering, thus validating the effectiveness and practicality of the method.

AI-powered COVID-19 forecasting: a comprehensive comparison of advanced deep learning methods.

The coronavirus disease 2019 (COVID-19) pandemic continues to pose significant challenges to the public health sector, including that of the United Arab Emirates (UAE). The objective of this study was to assess the efficiency and accuracy of various deep-learning models in forecasting COVID-19 cases within the UAE, thereby aiding the nation's public health authorities in informed decision-making. This study utilized a comprehensive dataset encompassing confirmed COVID-19 cases, demographic statistics, and socioeconomic indicators. Several advanced deep learning models, including long short-term memory (LSTM), bidirectional LSTM, convolutional neural network (CNN), CNN-LSTM, multilayer perceptron, and recurrent neural network (RNN) models, were trained and evaluated. Bayesian optimization was also implemented to fine-tune these models. The evaluation framework revealed that each model exhibited different levels of predictive accuracy and precision. Specifically, the RNN model outperformed the other architectures even without optimization. Comprehensive predictive and perspective analytics were conducted to scrutinize the COVID-19 dataset. This study transcends academic boundaries by offering critical insights that enable public health authorities in the UAE to deploy targeted data-driven interventions. The RNN model, which was identified as the most reliable and accurate for this specific context, can significantly influence public health decisions. Moreover, the broader implications of this research validate the capability of deep learning techniques in handling complex datasets, thus offering the transformative potential for predictive accuracy in the public health and healthcare sectors.

EEG channel selection using Gramian Angular Fields and spectrograms for energy data visualization

Electroencephalography (EEG)-based brain-computer interfaces (BCIs) have a wide range of applications in affect recognition. The usage of irrelevant information channels when decoding brain activity from different regions can negatively impact the task at hand. Therefore, further research is needed to determine the ideal channels for detecting superior performances. In this study, deep learning models and 2-D image representations are used to assess the efficacy of EEG channels for subjective valence responses to energy data visualizations. The EEG signals are converted into spectrograms and Gramian Angular Field (GAF) images. A hybrid Convolution Neural Network (CNN)-Long-Short Term Memory (LSTM) model and LSTM network are used to extract feature sets from the converted images. These feature sets, after reducing their dimensionality using a principal component analysis (PCA) method, are fed into a boosting classifier (AdaBoost). The performance metrics of both models and 2-D representations are compared. The LSTM and CNN-LSTM models with GAFs and spectrograms achieve state-of-the-art accuracies. The CNN-LSTM model achieves the highest performance when using spectrograms for the F8 channel, while the GAF method performs relatively lower for the F3 channel. The CNN-LSTM model with the spectrogram method produces reliable results, and hence, this method is used for further analysis of the remaining channel pairs. This study suggests that using a single EEG channel is more effective than using multiple channels for recognizing emotions in energy data visualizations. The proposed methodology is an efficient way to select channels for this purpose.

Ultra-short-term operation situation prediction method of active distribution network based on convolutional neural network long short term memory

The large-scale and high-permeability access of distributed generations (DGs) is making the distribution networks develop into active distribution networks (ADNs). The increasing complexity of the structure of ADNs also increases the vulnerabilities for cyberattacks. To accurately predict the operation situation of ADNs no matter whether cyberattacks occurred, this paper proposes an ultra-short-term operation situation prediction method of ADNs based on CNN-LSTM. Firstly, an ADNs situation prediction index system is established, which can comprehensively evaluate the future operation situation of ADNs from different perspectives. Secondly, a multi-feature data reconstruction method based on the time sliding window algorithm is proposed, which is utilized for constructing 2-dimensional matrix and as input datasets for multiple independent CNN-LSTM models to predict each situation prediction index. Then, there are 8 independent CNN-LSTM models shared same input matrix to predict 8 ADNs situation prediction index value without interfering with each other. Among them, CNN is used to extract the spatial correlation feature vectors of ADNs historical operation data and meteorological factor data in ultra-short term, which are reconstructed in time series and used as the input data of LSTM. LSTM is used to extract long-term historical temporal relationship to obtain ADNs operation situation prediction index value. Finally, we utilize the fuzzy analytic hierarchy process to calculate the future operation situation of ultra-short-term ADNs. Taking a real ADNs with 33-node as an example, we compared our proposed methodology to other existing excellent machine learning-based models, the result demonstrates that our proposed prediction method exhibits outstanding performance in time consumption and prediction accuracy.