Single Long Short-term Memory Research Articles

Enhancing wave energy farm efficiency: Eigen-stacking ensemble framework

The significant wave height (SWH) plays a pivotal role across diverse domains, ranging from the assessment of wave energy potential to the optimization of marine operations and the advancement of ocean engineering techniques. Understanding the spatial and temporal distribution patterns of SWH data among buoy network stations holds utmost importance for various applications. This research introduces a novel methodology for accurate spatiotemporal SWH prediction across multiple stations within buoy monitoring networks. Leveraging eigen SWH time series enables the identification of dataset patterns, feature extraction, and dimensionality reduction. Examining eigen time series provides insights into SWH dynamics, allowing exclusive use for prediction across all stations. By incorporating this eigen SWH time series as the sole input into a stacking ensemble model, we introduce a highly efficient and accurate framework for SWH prediction. Notably, the methodology achieves success in forecasting hourly SWH values along the western United States coastline, utilizing a stacking ensemble technique for enhanced accuracy. Evaluation metrics confirm outstanding performance, with CE and R2 values surpassing 0.95 and 0.93, respectively. Further, a single Long Short-Term Memory (LSTM) model is incorporated as a benchmark to assess the effectiveness of the stacking ensemble model for significant wave height (SWH) prediction, with results demonstrating that the ensemble model outperforms the single LSTM in terms of accuracy and robustness. Consequently, the introduced stacking ensemble model promises autonomy in forecasting spatial-temporal SWH patterns, extending its applicability within ocean engineering and scientific research.

Enhancing stock market predictions via hybrid external trend and internal components analysis and long short term memory model

When it comes to financial decision-making, stock market predictability is extremely important since it offers valuable information that may guide investment strategies, risk management, and portfolio allocation overall. Traditional methods often fail to accurately predict stock prices due to their complexity and inability to handle non-linear and non-stationary patterns in market data. To address these issues, this study introduces an innovative model that combines the External Trend and Internal Components Analysis decomposition method (ETICA) with the Long Short-Term Memory (LSTM) model, aiming to enhance stock market predictions for S&P 500, NASDAQ, Dow Jones, SSE and SZSE indices. Through rigorous testing across various training data proportions and epoch settings, our findings reveal that the proposed hybrid model outperforms the single LSTM model, delivering significantly lower Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values. This enhanced precision reduces prediction errors, underscoring the model’s robustness and reliability. The superior performance of the ETICA-LSTM model highlights its potential as a powerful financial forecasting tool, promising to transform investment strategies, optimize risk management, and enhance portfolio performance.

EMD combined with ensemble of machine learning predictors for foreign exchange rate forecasting

Forecasting foreign exchange rates is a critical financial challenge. In this paper, we build on recent trends and address the limitations of prior research by proposing a novel approach. Our method combines empirical mode decomposition (EMD) with ensemble of machine learning predictors in foreign exchange rate forecasting. To demonstrate that our proposed method (called EMD-ML) is effective, we used the new approach to forecast six foreign exchange rate time series at a specific time. The first experiment was implemented to compare the proposed forecasting model EMD–LSTM, which combines empirical mode decomposition (EMD) with ensemble of Long Short-Term Memory (LSTM) models, and the single LSTM model. The results indicate that the proposed EMD–LSTM model is more effective than the single LSTM. Besides, to aim at comparing deep-learning models against shallow machine learning models in combination with the EMD decomposition, the second experiment compared EMD-LSTM with the approach which combines EMD with an ensemble of k-nearest neighbors’ predictors (called EMD-KNN method) and the results of the second experiment show that EMD-LSTM cannot outperform EMD-KNN in foreign exchange rates forecasting.

Zoonotic outbreak risk prediction with long short-term memory models: a case study with schistosomiasis, echinococcosis, and leptospirosis

BackgroundZoonotic infections, characterized with huge pathogen diversity, wide affecting area and great society harm, have become a major global public health problem. Early and accurate prediction of their outbreaks is crucial for disease control. The aim of this study was to develop zoonotic diseases risk predictive models based on time-series incidence data and three zoonotic diseases in mainland China were employed as cases.MethodsThe incidence data for schistosomiasis, echinococcosis, and leptospirosis were downloaded from the Scientific Data Centre of the National Ministry of Health of China, and were processed by interpolation, dynamic curve reconstruction and time series decomposition. Data were decomposed into three distinct components: the trend component, the seasonal component, and the residual component. The trend component was used as input to construct the Long Short-Term Memory (LSTM) prediction model, while the seasonal component was used in the comparison of the periods and amplitudes. Finaly, the accuracy of the hybrid LSTM prediction model was comprehensive evaluated.ResultsThis study employed trend series of incidence numbers and incidence rates of three zoonotic diseases for modeling. The prediction results of the model showed that the predicted incidence number and incidence rate were very close to the real incidence data. Model evaluation revealed that the prediction error of the hybrid LSTM model was smaller than that of the single LSTM. Thus, these results demonstrate that using trending sequences as input sequences for the model leads to better-fitting predictive models.ConclusionsOur study successfully developed LSTM hybrid models for disease outbreak risk prediction using three zoonotic diseases as case studies. We demonstrate that the LSTM, when combined with time series decomposition, delivers more accurate results compared to conventional LSTM models using the raw data series. Disease outbreak trends can be predicted more accurately using hybrid models.

Hybrid Model for Multistep-Ahead Rainfall Forecast in Northeast India: A Comparative Study

Abstract Among all hydrometeorological parameters, rainfall strongly correlates with hydrometeorological disasters. The rainfall forecast process remains challenging due to the nonlinear, nonstationary nature and multiscale variability of rainfall. Moreover, the unique microclimate in different regions further complicates the forecasting process. This study proposes a hybrid model employing multivariate singular spectrum analysis (MSSA) and long short-term memory (LSTM) for multistep-ahead hourly rainfall forecasting in urban areas of northeast India. The model is trained and evaluated using high-resolution (12 km) hourly meteorological data from the Indian Monsoon Data Assimilation and Analysis (IMDAA) dataset for Guwahati (plain) and Aizawl (hilly) regions from 2015 to 2019. The hybrid model outperforms the single LSTM model in both plain and hilly regions, with an average percentage gain of 47.99% and 43.88% for symmetric mean absolute percentage error (SMAPE) and root-mean-square error (RMSE) in the case of the Guwahati dataset and 84.59% and 82.27% in the case of the Aizawl dataset, respectively. The performance of the LSTM model significantly improves as the zero values in the observed data are eliminated after reconstruction by MSSA. This enables the model to discern essential patterns and relationships in the data, which leads to more accurate forecasts. However, the hybrid model underestimates the rainfall, which can be tackled by hypertuning the parameters. The study highlights the importance of considering the interplay between rainfall and meteorological parameters for accurate rainfall forecasting in urban areas. The proposed MSSA–LSTM model can be used as a decision support tool for urban planning and disaster management. Significance Statement This study addresses a critical need in multistep-ahead hourly rainfall forecasting in urban areas, with a focus on the unique conditions of northeast India. Our research has identified the presence of red noise in the rainfall data, shedding light on the complexities of the underlying rainfall patterns. Furthermore, we delve into the intricate interplay between rainfall and meteorological parameters, providing valuable insights into the factors influencing rainfall dynamics. Notably, our study underscores the region-specific challenges in rainfall forecasting. While the hybrid model demonstrates reasonable accuracy in the plains, its performance in the hilly region falls short of expectations. This highlights the nuanced nature of rainfall prediction in areas with varying topography and emphasizes the need for tailored forecasting approaches.

Heteroscedasticity effects as component to future stock market predictions using RNN-based models.

Heteroscedasticity effects are useful for forecasting future stock return volatility. Stock volatility forecasting provides business insight into the stock market, making it valuable information for investors and traders. Predicting stock volatility is a crucial task and challenging. This study proposes a hybrid model that predicts future stock volatility values by considering the heteroscedasticity element of the stock price. The proposed model is a combination of Generalized Autoregressive Conditional Heteroskedasticity (GARCH) and a well-known Recurrent Neural Network (RNN) algorithm Long Short-Term Memory (LSTM). This proposed model is referred to as GARCH-LSTM model. The proposed model is expected to improve prediction accuracy by considering heteroscedasticity elements. First, the GARCH model is employed to estimate the model parameters. After that, the ARCH effect test is used to test the residuals obtained from the model. Any untrained heteroscedasticity element must be found using this step. The hypothesis of the ARCH test yielded a p-value less than 0.05 indicating there is valuable information remaining in the residual, known as heteroscedasticity element. Next, the dataset with heteroscedasticity is then modelled using an LSTM-based RNN algorithm. Experimental results revealed that hybrid GARCH-LSTM had the lowest MAE (7.961), RMSE (10.466), MAPE (0.516) and HMAE (0.005) values compared with a single LSTM. The accuracy of forecasting was also significantly improved by 15% and 13% with hybrid GARCH-LSTM in comparison to single LSTMs. Furthermore, the results reveal that hybrid GARCH-LSTM fully exploits the heteroscedasticity element, which is not captured by the GARCH model estimation, outperforming GARCH models on their own. This finding from this study confirmed that hybrid GARCH-LSTM models are effective forecasting tools for predicting stock price movements. In addition, the proposed model can assist investors in making informed decisions regarding stock prices since it is capable of closely predicting and imitating the observed pattern and trend of KLSE stock prices.

Forecasting ionospheric TEC using least squares support vector machine and moth-flame optimization methods in China

The total electron content (TEC) of the ionosphere is an important parameter to describe the ionosphere, and it is a great significance to monitor and predict it accurately. In this paper, a hybrid ionospheric TEC prediction model based on the least squares support vector machine (LSSVM) and the Moth-Flame Optimization (MFO) algorithm is proposed. The parameters of the LSSVM model are optimized by the MFO algorithm. We use observation data of 15 GNSS stations from the Crustal Movement Observation Network of China (CMONOC) to extract ionospheric TEC from 2012 to 2019. The ionospheric TEC is forecasted using solar and geomagnetic activity indices in both the low solar activity year (2019) and the high solar activity year (2015). The results show that the prediction performance of the MFO_LSSVM model is significantly better than that of the IRI model, SVM model, and LSSVM model. Compared with the other three models, there are more stable prediction results in the low and high solar activity years. At the same time, the predicted value of the MFO_LSSVM model has a good correlation with the measured value, and it also has good prediction potential in areas with active geomagnetic activity. The comparison with the Long Short-Term Memory (LSTM) model shows that the MFO_LSSVM model has better performance than the single LSTM model. In conclusion, the MFO_LSSVM model can accurately predict ionospheric TEC in China, and has better accuracy than traditional long-term and short-term models.

A Weighted Feature Fusion Model for Unsteady Aerodynamic Modeling at High Angles of Attack

Unsteady aerodynamic prediction at high angles of attack is of great importance to the design and development of advanced fighters. In this paper, a weighted feature fusion model (WFFM) that combines the state-space model and neural networks is proposed to build an unsteady aerodynamic model for the precise simulation and control of post-stall maneuvers. In the proposed model, the influences of the physical model on neural networks are considered and adjusted by introducing a standardization layer and a new weighting method. A long short-term memory (LSTM) network is used to fuse two mappings: one from flight states to aerodynamic loads, and the other from low-fidelity data to high-fidelity data. Data from wind tunnel oscillation experiments at high angles of attack using a new kind of wire-driven parallel robot and the traditional tail support are used for verifying the proposed aerodynamic model. The output of the WFFM is also compared with predictions from other models, such as the state-space model, single LSTM model, and feature fusion model not including a feature weighting layer. Results demonstrate improved accuracy of the proposed model in the interpolation and extrapolation tests. Furthermore, the WFFM is applied to the flight simulation of F-16 with different control inputs. Compared with conventional models, the WFFM shows improved accuracy and better generalization capability.

In-event surrogate model training for regional transmission tower-line system dynamic responses prediction in realistic hurricane wind fields

Surrogate models have shown improved accuracy in predicting infrastructure responses during dynamic loadings. However, training a surrogate model for complex loading inputs across the entire hazard region remains challenging. This study provides insight into the training of surrogate models to estimate the responses of transmission tower-line structures in a coupled high-dimensional and high-resolution wind field and presents innovative methods for addressing these challenges. Four data- and physics-based spatial-temporal decoupling sampling methods are employed and cross-compared to obtain the most representative in-event wind profiles for training the surrogate model. Long Short-Term Memory (LSTM) is utilised as the surrogate model framework to predict the dynamic responses of the structure during the 2017 Hurricane Harvey. The accuracy and robustness of two transmission tower-line structure configuration surrogate models are validated by comparing the predictions with finite element analyses by using randomly distributed temporal and geospatial wind profiles throughout the hurricane. Finally, a single LSTM surrogate model is developed, trained by applying the full reference wind speed range of Hurricane Harvey for the regional-scale structural performance evaluation of the transmission tower-line system. The results demonstrate that the proposed surrogate model training methodology is general and can be applied to regional-scale structural performance evaluations.

Fault Diagnosis of Fuel Cells by a Hybrid Deep Learning Network Fusing Characteristic Impedance

Establishing an accurate fault diagnosis system for the proton exchange membrane (PEM) fuel cell is essential for ensuring stable performance and delaying degradation. In this paper, a novel fault diagnosis method fusing characteristic impedance for the PEM fuel cell based on a hybrid deep learning network by combing the residual network (ResNet) and long short-term memory (LSTM) is proposed. Specifically, the characteristic impedance that can reflect internal dynamics loss is fused as the feature input alongside other commonly vehicular measurement signals and decoded to form a feature matrix. The feature matrix is then transferred to realize 25 categories of fault detection, including different degrees of membrane drying, flooding, and air starvation. The results showed the accuracy of ResNet-LSTM can reach 99.632% with a good balance of computational burden and is higher than that of a single LSTM, ResNet, and convolutional neural network, as well as traditional machine learning methods because such the hybrid structure can make full use of feature learning capability of ResNet and the time series analysis capability of LSTM. Moreover, the proposed hybrid framework is further validated and compared under different proportions of training samples, noise levels, and input signals.

Hybrid deep learning models for time series forecasting of solar power

Forecasting solar power production accurately is critical for effectively planning and managing renewable energy systems. This paper introduces and investigates novel hybrid deep learning models for solar power forecasting using time series data. The research analyzes the efficacy of various models for capturing the complex patterns present in solar power data. In this study, all of the possible combinations of convolutional neural network (CNN), long short-term memory (LSTM), and transformer (TF) models are experimented. These hybrid models also compared with the single CNN, LSTM and TF models with respect to different kinds of optimizers. Three different evaluation metrics are also employed for performance analysis. Results show that the CNN–LSTM–TF hybrid model outperforms the other models, with a mean absolute error (MAE) of 0.551% when using the Nadam optimizer. However, the TF–LSTM model has relatively low performance, with an MAE of 16.17%, highlighting the difficulties in making reliable predictions of solar power. This result provides valuable insights for optimizing and planning renewable energy systems, highlighting the significance of selecting appropriate models and optimizers for accurate solar power forecasting. This is the first time such a comprehensive work presented that also involves transformer networks in hybrid models for solar power forecasting.

An Empirical Mode Decomposition-Based Hybrid Model for Sub-Hourly Load Forecasting

Sub-hourly load forecasting can provide accurate short-term load forecasts, which is important for ensuring a secure operation and minimizing operating costs. Decomposition algorithms are suitable for extracting sub-series and improving forecasts in the context of short-term load forecasting. However, some existing algorithms like singular spectrum analysis (SSA) struggle to decompose high sampling frequencies and rapidly changing sub-hourly load series due to inherent flaws. Considering this, we propose an empirical mode decomposition-based hybrid model, named EMDHM. The decomposition part of this novel model first detrends the linear and periodic components from the original series. The remaining detrended long-range correlation series is simplified using empirical mode decomposition (EMD), generating intrinsic mode functions (IMFs). Fluctuation analysis is employed to identify high-frequency information, which divide IMFs into two types of long-range series. In the forecasting part, linear and periodic components are predicted by linear and trigonometric functions, while two long-range components are fitted by long short-term memory (LSTM) for prediction. Four forecasting series are ensembled to find the final result of EMDHM. In experiments, the model’s framework we propose is highly suitable for handling sub-hourly load datasets. The MAE, RMSE, MARNE, and R2 of EMDHM have improved by 20.1%, 26.8%, 22.1%, and 5.4% compared to single LSTM, respectively. Furthermore, EMDHM can handle both short- and long-sequence, sub-hourly load forecasting tasks. Its R2 only decreases by 4.7% when the prediction length varies from 48 to 720, which is significantly lower than other models.

Ultra-Short-Term Wind Power Prediction Based on eEEMD-LSTM

The intermittent and random nature of wind brings great challenges to the accurate prediction of wind power; a single model is insufficient to meet the requirements of ultra-short-term wind power prediction. Although ensemble empirical mode decomposition (EEMD) can be used to extract the time series features of the original wind power data, the number of its modes will increase with the complexity of the original data. Too many modes are unnecessary, making the prediction model constructed based on the sub-models too complex. An entropy ensemble empirical mode decomposition (eEEMD) method based on information entropy is proposed in this work. Fewer components with significant feature differences are obtained using information entropy to reconstruct sub-sequences. The long short-term memory (LSTM) model is suitable for prediction after the decomposition of time series. All the modes are trained with the same deep learning framework LSTM. In view of the different features of each mode, models should be trained differentially for each mode; a rule is designed to determine the training error of each mode according to its average value. In this way, the model prediction accuracy and efficiency can make better tradeoffs. The predictions of different modes are reconstructed to obtain the final prediction results. The test results from a wind power unit show that the proposed eEEMD-LSTM has higher prediction accuracy compared with single LSTM and EEMD-LSTM, and the results based on Bayesian ridge regression (BR) and support vector regression (SVR) are the same; eEEMD-LSTM exhibits better performance.

Long-term prediction of runoff of Heiher River in China based on EMD-LSTM networks

Due to the influence of global climate change and human activities, the time series data of river runoff become complex and non-stationary. These properties make sequence prediction difficult and with low accuracy. In order to improve the prediction accuracy, Empirical Mode Decomposition (EMD) was introduced to Long Short-Term Memory (LSTM) networks in this paper. The EMD decomposed a non-stationary time series into multiple components. We trained an LSTM network for each component, and added their predictions to obtain the final forecasting value of original sequence. At last, the EMD-LSTM model was applied to the annual runoff sequence in the upper Heihe River. By comparing with single LSTM, it shows that the EMD-LSTM has higher accuracy for the long-term prediction of river runoff.

A hybrid deep learning approach to improve real-time effluent quality prediction in wastewater treatment plant

Wastewater treatment plant (WWTP) operation is usually intricate due to large variations in influent characteristics and nonlinear sewage treatment processes. Effective modeling of WWTP effluent water quality can provide valuable decision-making support to facilitate their operations and management. In this study, we developed a novel hybrid deep learning model by combining the temporal convolutional network (TCN) model with the long short-term memory (LSTM) network model to improve the simulation of hourly total nitrogen (TN) concentration in WWTP effluent. The developed model was tested in a WWTP in Jiangsu Province, China, where the prediction results of the hybrid TCN-LSTM model were compared with those of single deep learning models (TCN and LSTM) and traditional machine learning model (feedforward neural network, FFNN). The hybrid TCN-LSTM model could achieve 33.1 % higher accuracy as compared to the single TCN or LSTM model, and its performance could improve by 63.6 % comparing to the traditional FFNN model. The developed hybrid model also exhibited a higher power prediction of WWTP effluent TN for the next multiple time steps within eight hours, as compared to the standalone TCN, LSTM, and FFNN models. Finally, employing model interpretation approach of Shapley additive explanation to identify the key parameters influencing the behavior of WWTP effluent water quality, it was found that removing variables that did not contribute to the model output could further improve modeling efficiency while optimizing monitoring and management strategies.

Predicting Temperature and Humidity in Roadway with Water Trickling Using Principal Component Analysis-Long Short-Term Memory-Genetic Algorithm Method

The heat dissipated from high geo-temperature underground surrounding rocks is the main heat source of working faces, while thermal water upwelling and trickling into the roadway will notably deteriorate the mine’s climate and thermal comfort. Predicting airflow temperature and relative humidity (RH) is conductive to intelligent control of air conditioning cooling and ventilation regulation. To accommodate this issue, an intelligent technique was proposed, integrating a genetic algorithm (GA) and long short-term memory (LSTM) based on rock temperature, inlet air temperature, water temperature, water flow rate, RH, and ventilation time. A total of 21 input features including over 200 pieces of data were collected from an independently developed modeling roadway to construct a dataset. Principal component analysis (PCA) was conducted to reduce features, and GA was used to tune the LSTM and PCA-LSTM architectures for best performance. The following research results were yielded. The proposed PCA-LSTM-GA model is more reliable and efficient than the single LSTM model or hybrid LSTM-GA model in predicting the air temperature Tfout and humidity RHout at the end of the water trickling roadway. The importance scores (ISs) indicate that Tfout is mainly influenced by the surrounding rock temperature (IS 0.661) and the inlet air temperature (IS 0.264). While RHout is primarily influenced by the rock temperature in the water trickling section (IS 0.577), the inlet air temperature (IS 0.187), and the trickling water temperature and flow rate (total IS 0.136), and it has an evident time effect. In addition, we developed relevant equipment and provided engineering practice methods to use the machine learning model. The proposed model, which can predict the mine microclimate, serves to facilitate coal and geothermal resource co-mining as well as thermal hazard control.

A Neural Network Weights Initialization Approach for Diagnosing Real Aircraft Engine Inter-Shaft Bearing Faults

The deep learning diagnosis of aircraft engine-bearing faults enables cost-effective predictive maintenance while playing an important role in increasing the safety, reliability, and efficiency of aircraft operations. Because of highly dynamic and harsh operating conditions of this system, such modeling is challenging due to data complexity and drift, making it difficult to reveal failure patterns. As a result, the objective of this study is dual. To begin, a highly structured data preprocessing strategy ranging from extraction, denoising, outlier removal, scaling, and balancing is provided to solve data complexity that resides specifically in outliers, noise, and data imbalance problems. Gap statistics under k-means clustering are used to evaluate preprocessing results, providing a quantitative estimate of the ideal number of clusters and thereby enhancing data representations. This is the first time, to the best of authors’ knowledge, that such a criterion has been employed for an important step in a preliminary ground truth validation in supervised learning. Furthermore, to tackle data drift issues, long-short term memory (LSTM) adaptive learning features are used and subjected to a learning parameter improvement method utilizing recursive weights initialization (RWI) across several rounds. The strength of such methodology can be seen by application to realistic, extremely new, complex, and dynamic data collected from a real test-bench. Cross validation of a single LSTM layer model with only 10 neurons shows its ability to enhance classification performance by 7.7508% over state-of-the-art results, obtaining a classification accuracy of 92.03 ± 0.0849%, which is an exceptional performance in such a benchmark.

DETECTION MODEL FOR FAKE NEWS ON COVID-19 IN INDONESIA

Today, fake information has become a significant problem, exacerbated by the acceleration of access to information. The spread of fake information has a dangerous impact, especially regarding global health issues, for example COVID-19. People can access various resources to obtain information, including online sites and social media. One of the methods to control the spread of false information is detecting hoaxes. Many methods have been developed to identify hoaxes; most previous studies have focused on developing hoax detection methods using data from a single source in English. The present study is carried out to detect fake news in Indonesian language using multiple data sources, including traditional and social media in the context of COVID-19. The study uses Long Short-Term Memory (LSTM) and the Robustly Optimised Bidirectional Encoder Representations from Transformers Pre-Training Approach (RoBERTa). The LSTM approach is used to develop four different architectures that varied based on: (1) the use of text-only versus the use of both title and text; (2) the number of LSTM and dense layers; and (3) the activation function. The LSTM model with text-only data, a single LSTM layer and two dense layers, outperformed other LSTM architectures, achieving the highest accuracy of 92.17%. The LSTM models require a considerably short training time of 23–27 minutes for 3,847 articles and has a detection time of 3.8–4.1 ms per article. The RoBERTa classifiers outperformed all LSTM models with an accuracy of over 97% and a significantly better training time, with a margin of more than 50% compared to LSTM classifiers, although it had a slightly longer test time. Both LSTM and RoBERTa models outperformed the Naïve Bayes and SVM benchmark methods in terms of accuracy, precision, and recall. Therefore, this study shows that both LSTM and RoBERTa methods are reliable and can be reasonably implemented for real-time fake news detection.

Unified CNN-LSTM for keyhole status prediction in PAW based on spatial-temporal features

Despite the high efficiency of keyhole plasma arc welding (K-PAW), it still has several deficiencies, such as narrow welding parameter ranges, easily disturbed welding process and instability of welding quality, etc. It is a significant prerequisite to predict the keyhole/penetration status accurately for maintaining the welding process stability and improving the welding quality. Most researchers have focused on visual inspection techniques and convolutional neural networks (CNN), establishing a mathematical model of the correlation between weld pool images and penetration status. While CNN could extract the features of single weld pool images, it is difficult to predict the evolution trend of the incoming moments. In this paper, a novel model based on CNN and LSTM (long-short term memory) was developed to extract both the spatial and temporal features of topside weld pool images, and consequently the complex keyhole behaviors were predicted and described. The comparative study is carried out on different models, i.e. the single CNN model, the single LSTM model, and the CNN-LSTM model that integrates both spatial features of single images and temporal features of sequence. The keyhole initialization and establishing period, as the typical and critical welding scenario in full-penetration welding, is investigate and compared, resulting in a predicted value that is close to reality. Furthermore, ahead prediction of keyhole status was adopted, maintaining over 80% accuracy even when predicting keyhole behaviors 2 s into the future. Consequently, the unified CNN-LSTM model effectively improves the prediction accuracy of keyhole/penetration status, promising for intelligent K-PAW technology.

Classification of Epileptic Seizure Types Using Multiscale Convolutional Neural Network and Long Short-Term Memory

The accurate classification of seizure types using electroencephalography (EEG) signals plays a vital role in determining a precise treatment plan and therapy for epilepsy patients. Among the available deep network models, Convolutional Neural Networks (CNNs) are the most widely adopted models for learning and representing EEG signals. However, typical CNNs have high computational complexity, leading to overfitting problems. This paper proposes the design of two effective, lightweight deep network models; the 1D multiscale neural network (1D-MSCNet) model and the Long Short-term Memory (LSTM)-based compact CNN (EEG-LSTMNet) model. The 1D-MSCNet model comprises three modules: a spectral–temporal convolution module, a spatial convolution module, and a classification module. It extracts features from input EEG trials at multiple frequency/time ranges, identifying relationships between the spatial distribution of their channels. The EEG-LSTMNet model includes three convolutional layers, namely temporal, depthwise, and separable layers, a single LSTM layer, and two fully connected classification layers to extract discriminative EEG feature representations. Both models have been applied to the same EEG trials collected from the Temple University Hospital (TUH) database. Results revealed F1-score values of 96.9% and 98.4% for the 1D-MSCNet and EEG-LSTMNet, respectively. Based on the demonstrated outcomes, both models outperform related state-of-the-art methods due to their architectures’ adoption of 1D modules and layers that reduce the computational effort needed, solve the overfitting problem, and enhance classification efficiency. Hence, both models could be valuable additions for neurologists to help them decide upon precise treatments and drugs for patients depending on their type of seizure.