Search Results (191)

The air-conditioner noise in a rehabilitation room can seriously affect the mental state of patients. However, the existing single-layer active noise control (ANC) filters may fail to attenuate the complicated harmonic noise, and the deep recursive ANC method may fail to work in real time. To solve the problem, in a bending-pipe model, a new hybrid adaptive self-loading filtered-x least-mean-square (ASL-FxLMS) and convolutional neural network-gate recurrent unit (CNN-GRU) network is proposed. At first, based on the recursive GRU translation core, an improved CNN-GRU network with multi-head attention layers is proposed. Especially for complicated harmonic noises with more or fewer frequencies than harmonic models, the attenuation performance will be improved. In addition, its structure is optimized to decrease the computing load. In addition, an improved time-delay estimator is applied to improve the real-time ANC performance of CNN-GRU. Meanwhile, an adaptive self-loading FxLMS algorithm has been developed to deal with the uncertain components of complicated harmonic noise. Moreover, to achieve balance attenuation, robustness, and tracking performance, the ASL-FxLMS and CNN-GRU are connected by a convex combination structure. Furthermore, theoretical analysis and simulations are also conducted to show the effectiveness of the proposed method. Full article

Silent speech recognition (SSR) enables communication without vocalization by interpreting biosignals such as electromyography (EMG) and electroencephalography (EEG). Most existing SSR systems rely on high-density, non-wearable sensors and focus primarily on isolated word recognition, limiting their practical usability. This study presents a wearable SSR system capable of accurate sentence-level recognition using single-channel EMG and EEG sensors with real-time wireless transmission. A moving window-based few-shot learning model, implemented with a Siamese neural network, segments and classifies words from continuous biosignals without requiring pauses or manual segmentation between word signals. A novel sensor fusion model integrates both EMG and EEG modalities, enhancing classification accuracy. To further improve sentence-level recognition, a statistical language model (LM) is applied as post-processing to correct syntactic and lexical errors. The system was evaluated on a dataset of four military command sentences containing ten unique words, achieving 95.25% sentence-level recognition accuracy. These results demonstrate the feasibility of sentence-level SSR using wearable sensors through a window-based few-shot learning model, sensor fusion, and ML applied to limited simultaneous EMG and EEG signals. Full article

Gas-phase chemistry has been identified as a major computational bottleneck in both the forward and adjoint modes of chemical transport models (CTMs). Although previous studies have demonstrated the potential of deep learning models to simulate and accelerate this process, few studies have examined the applicability and performance of these models in adjoint sensitivity analysis. In this study, a deep learning emulator for gas-phase chemistry is developed and trained on a diverse set of forward-mode simulations from the Community Multiscale Air Quality (CMAQ) model. The emulator employs a residual neural network (ResNet) architecture referred to as FiLM-ResNet, which integrates Feature-wise Linear Modulation (FiLM) layers to explicitly account for photochemical and non-photochemical conditions. Validation within a single timestep indicates that the emulator accurately predicts concentration changes for 74% of gas-phase species with coefficient of determination (R²) exceeding 0.999. After embedding the emulator into the CTM, multi-timestep simulation over one week shows close agreement with the numerical model. For the adjoint mode, we compute the sensitivities of ozone (O₃) with respect to O₃, nitric oxide (NO), nitrogen dioxide (NO₂), hydroxyl radical (OH) and isoprene (ISOP) using automatic differentiation, with the emulator-based adjoint results achieving a maximum R² of 0.995 in single timestep evaluations compared to the numerical adjoint sensitivities. A 24 h adjoint simulation reveals that the emulator maintains spatially consistent adjoint sensitivity distributions compared to the numerical model across most grid cells. In terms of computational efficiency, the emulator achieves speed-ups of 80×–130× in the forward mode and 45×–102× in the adjoint mode, depending on whether inference is executed on Central Processing Unit (CPU) or Graphics Processing Unit (GPU). These findings demonstrate that, once the emulator is accurately trained to reproduce forward-mode gas-phase chemistry, it can be effectively applied in adjoint sensitivity analysis. This approach offers a promising alternative approach to numerical adjoint frameworks in CTMs. Full article

In ultra-wideband (UWB) radio fuzes, the signal processing unit’s capability to rapidly and accurately extract target characteristics under battlefield conditions directly determines detonation precision and reliability. Escalating electronic warfare creates complex electromagnetic environments that compromise UWB fuze reliability through false alarms and missed detections. This study pioneers a novel signal processing architecture. The framework integrates: (1) fixed-parameter Least Mean Squares (LMS) front-end filtering for interference suppression; (2) One-Dimensional Convnlutional Neural Network (1D-CNN) recognition trained on One-Dimensional Conditional Generative Adversarial Network (1D-CGAN)-augmented datasets. Validated on test samples, the system achieves 0% false alarm/miss detection rates and 97.66% segment recognition accuracy—representing a 5.32% improvement over the baseline 1D-CNN model trained solely on original data. This breakthrough resolves energy-threshold detection’s critical vulnerability to deliberate jamming while establishing a new technical framework for UWB fuze operation in contested spectra. Full article

Deepfake audio refers to speech that has been synthetically generated or altered through advanced neural network techniques, often with a degree of realism sufficient to convincingly imitate genuine human voices. As these manipulations become increasingly indistinguishable from authentic recordings, they present significant threats to security, undermine media integrity, and challenge the reliability of digital authentication systems. In this study, a robust detection framework is proposed, which leverages the power of self-supervised learning (SSL) and attention-based modeling to identify deepfake audio samples. Specifically, audio features are extracted from input speech using two powerful pretrained SSL models: HuBERT-Large and WavLM-Large. These distinctive features are then integrated through an Attentional Multi-Feature Fusion (AMFF) mechanism. The fused features are subsequently classified using a NeXt-Time Delay Neural Network (NeXt-TDNN) model enhanced with Efficient Channel Attention (ECA), enabling improved temporal and channel-wise feature discrimination. Experimental results show that the proposed method achieves a 0.42% EER and 0.01 min-tDCF on ASVspoof 2019 LA, a 1.01% EER on ASVspoof 2019 PA, and a pooled 6.56% EER on the cross-channel ASVspoof 2021 LA evaluation, thus highlighting its effectiveness for real-world deepfake detection scenarios. Furthermore, on the ASVspoof 5 dataset, the method achieved a 7.23% EER, outperforming strong baselines and demonstrating strong generalization ability. Moreover, the macro-averaged F1-score of 96.01% and balanced accuracy of 99.06% were obtained on the ASVspoof 2019 LA dataset, while the proposed method achieved a macro-averaged F1-score of 98.70% and balanced accuracy of 98.90% on the ASVspoof 2019 PA dataset. On the highly challenging ASVspoof 5 dataset, which includes crowdsourced, non-studio-quality audio, and novel adversarial attacks, the proposed method achieves macro-averaged metrics exceeding 92%, with a precision of 92.07%, a recall of 92.63%, an F1-measure of 92.35%, and a balanced accuracy of 92.63%. Full article

Additive manufacturing (AM) enables the production of complex components but often results in poor surface quality due to its layer-by-layer deposition process. To improve surface finish, postprocessing methods like abrasive flow machining (AFM) are necessary. However, conventional AFM struggles with achieving uniform polishing in intricate regions, especially at internal corners. This study proposes a liquid metal-driven abrasive flow (LM-AF) strategy designed for polishing complex internal channels in AM parts. By combining experimental and numerical simulations, the research investigates surface roughness variations, particularly focusing on the Sa (Arithmetic Average Surface Roughness) parameter. Experimental results show that conventional AFM leaves significant roughness at internal corners compared to adjacent areas. To address this, a hybrid GA-NN-GA (Genetic Algorithm–Neural Network-Genetic Algorithm) optimization model was developed. The model uses a neural network to predict Sa based on key parameters, with genetic algorithms applied for training and optimization. The optimal process parameters identified include a NaOH concentration of $1\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{L}$, a voltage of 50 V, abrasive concentration of $10\mathrm{\%}$, and a frequency of 428.3 Hz. With these parameters, LM-AF significantly reduced roughness at internal corners of flow channels, achieving uniformity with Sa values reduced from $25.365 \mathsf{\mu }\mathrm{m}$ to $15.780 \mathsf{\mu }\mathrm{m}$, from 22.950 $\mathsf{\mu }\mathrm{m}$ to $15.718 \mathsf{\mu }\mathrm{m}$, and from $10.933 \mathsf{\mu }\mathrm{m}$ to $10.055 \mathsf{\mu }\mathrm{m}$, outperforming traditional AFM methods. Full article

Quartz is a widely used mineral in dosimetric and geochronological applications due to its stable luminescence properties under ionizing radiation. This study presents an artificial neural network (ANN)-based approach to predict the optically stimulated luminescence (OSL) decay curves of quartz extracted from Mediterranean beach sand samples in Turkey. Experimental OSL signals were obtained from quartz samples irradiated with beta doses ranging from 0.1 Gy to 1034.9 Gy. The dataset was used to train ANN models with three different learning algorithms: Levenberg–Marquardt (LM), Bayesian Regularization (BR), and Scaled Conjugate Gradient (SCG). Forty-seven decay curves were used for training and three for testing. The ANN models were evaluated based on regression accuracy, training–validation–test performance, and their predictive capability for low, medium, and high doses (1 Gy, 72.4 Gy, 465.7 Gy). The results showed that BR achieved the highest overall regression (R = 0.99994) followed by LM (R = 0.99964) and SCG (R = 0.99820), confirming the superior generalization and fits across all dose ranges. LM performs optimally at low-to-moderate doses, and SCG delivers balanced yet slightly noisier predictions. The proposed ANN-based method offers a robust and effective alternative to conventional kinetic modeling approaches for analyzing OSL decay behavior and holds considerable potential for advancing luminescence-based retrospective dosimetry and OSL dating applications. Full article

While horizontal roughing filters (HRFs) remain widely acclaimed for their exceptional efficiency in water treatment, especially in developing countries, they are inherently susceptible to clogging, which necessitates timely maintenance interventions. Conventional methods for managing clogging in HRFs typically involve evaluating filter head loss coefficients against established water quality standards. This study utilizes artificial neural network (ANN) for the prediction of clogging duration and effluent turbidity in HRF equipment. The ANN was configured with two outputs, the clogging duration and effluent turbidity, which were predicted concurrently. Effluent turbidity was modeled to enhance the network’s learning process and improve the accuracy of clogging prediction. The network steps of the iterative training process of ANN used different types of input parameters, such as influent turbidity, filtration rate, pH, conductivity, and effluent turbidity. The training, in addition, optimized network parameters such as learning rate, momentum, and calibration of neurons in the hidden layer. The quantities of the dataset accounted for up to 70% for training and 30% for testing and validation. The optimized structure of ANN configured in a 4-8-2 topology and trained using the Levenberg–Marquardt (LM) algorithm achieved a mean square error (MSE) of less than 0.001 and R-coefficients exceeding 0.999 across training, validation, testing, and the entire dataset. This ANN surpassed models of scaled conjugate gradient (SCG) and obtained a percentage of average absolute deviation (%AAD) of 9.5. This optimal structure of ANN proved to be a robust tool for tracking the filter clogging duration in HRF equipment. This approach supports proactive maintenance and operational planning in HRFs, including data-driven scheduling of backwashing based on predicted clogging trends. Full article

Membrane transporters play a crucial role in any cell. Identifying the substrates they translocate across membranes is important for many fields of research, such as metabolomics, pharmacology, and biotechnology. In this study, we leverage recent advances in deep learning, such as amino acid sequence embeddings with protein language models (pLMs), highly accurate 3D structure predictions with AlphaFold 2, and structure-encoding 3Di sequences from FoldSeek, for predicting substrates of membrane transporters. We test new deep learning features derived from both sequence and structure, and compare them to the previously best-performing protein encodings, which were made up of amino acid k-mer frequencies and evolutionary information from PSSMs. Furthermore, we compare the performance of these features either using a previously developed SVM model, or with a regularized feedforward neural network (FNN). When evaluating these models on sugar and amino acid carriers in A. thaliana, as well as on three types of ion channels in human, we found that both the DL-based features and the FNN model led to a better and more consistent classification performance compared to previous methods. Direct encodings of 3D structures with Foldseek, as well as structural embeddings with ProstT5, matched the performance of state-of-the-art amino acid sequence embeddings calculated with the ProtT5-XL model when used as input for the FNN classifier. Full article

Artificial neural networks (ANNs) adjust to the underlying behavior in the dataset using a training rule or optimizer. The most popular first-and second-order optimizers, Adam (AD) and Levenberg–Marquardt (LM), were compared with the aim of predicting extreme flash floods of a runoff-dominated hydrological system. A fully connected multilayer perceptron with a shallow structure was used to reduce complexity and limit overfitting. The inputs of the ANN were determined by rainfall–water level cross-correlation analysis. For each optimizer, the hyperparameters of the ANN were selected using a grid search and the cross-validation score on a novel criterion (PERS PEAK) mixing the persistency (PERS) and the quality of flood-peak restitution (PEAK). For an extreme and unseen event used as a test set, LM outperformed AD by 25% on all performance criteria. The peak water level of this event, 66% greater than that of the training set, was predicted by 92% after more training iterations were done by the LM optimizer. This shows that the ANN can predict beyond the ranges of the training set, given the right optimizer. Nevertheless, the LM training time was up to five times longer than that of AD during grid search. Full article

This paper presents a simulation-based investigation into control strategies for mitigating the consequences of cyber-assault on the steering systems of the Maritime Autonomous Surface Ships (MASS). The study focuses on two simulation experiments conducted within the Simulink/MATLAB environment, utilizing the catamaran “Nymo” MASS mathematical model to represent vessel dynamics. Cyber-attacks are modeled as external disturbances affecting the rudder control signal, emulating realistic interference scenarios. To assess control resilience, two configurations are compared during a representative turning maneuver to a specified heading: (1) a Proportional–Integral–Derivative (PID) regulator augmented with a Least Mean Squares (LMS) adaptive filter, and (2) a Nonlinear Autoregressive Moving Average with Exogenous Input (NARMA-L2) neural network regulator. The PID and LMS configurations aim to enhance the disturbance rejection capabilities of the classical controller through adaptive filtering, while the NARMA-L2 approach represents a data-driven, nonlinear control alternative. Simulation results indicate that although the PID and LMS setups demonstrate improved performance over standalone PID in the presence of cyber-induced disturbances, the NARMA-L2 controller exhibits superior adaptability, accuracy, and robustness under adversarial conditions. These findings suggest that neural network-based control offers a promising pathway for developing cyber-resilient steering systems in autonomous maritime vessels. Full article

This study aimed to train and validate two artificial neural network (ANN) models, one with four hidden layers and the other with five hidden layers, to predict the daily photovoltaic power output of a 20 Kw photovoltaic power plant with bifacial photovoltaic modules with tilt angle variation from 0° to 90° in two Moroccan cities, Ouarzazate and Oujda. To validate the two proposed models, photovoltaic power data calculated using the System Advisor Model (SAM) software version 2023.12.17 were employed to predict the average daily power of the photovoltaic plant for December, utilizing MATLAB software Version R2020a 9.8, and for the tilt angles corresponding to the latitudes of the two cities studied. The results differ from one model to another according to their mean absolute error (MAE), root mean squared error (RMSE), and coefficient of determination (R²) values. The artificial neural network model with five hidden layers obtained better results with a R² value of 0.99354 for Ouarzazate and 0.99836 for Oujda. These two proposed models are trained using the Levenberg Marquardt (LM) optimizer, which is proven to be the best training procedure. Full article

This paper presents a decentralized energy management system (EMS) based on Multilayer Perceptron Artificial Neural Networks (MLP-ANNs) and a Modified Cheetah Optimizer (MCO) to account for uncertainty in renewable generation and load demand. The proposed framework applies an MLP-ANN with Levenberg–Marquardt (LM) training for high-precision forecasts of photovoltaic/wind generation, ambient temperature, and load demand, greatly outperforming traditional statistical methods (e.g., time-series analysis) and resilient backpropagation (RP) in precision. The new MCO algorithm eliminates local trapping and premature convergence issues in classical optimization methods like Particle Swarm Optimization (PSO) and Genetic Algorithms (GAs). Simulations on a test microgrid verily demonstrate the advantages of the framework, achieving a 26.8% cost-of-operation reduction against rule-based EMSs and classical PSO/GA, and a 15% improvement in forecast accuracy using an LM-trained MLP-ANN. Moreover, demand response programs embodied in the system reduce peak loads by 7.5% further enhancing grid stability. The MLP-ANN forecasting–MCO optimization duet is an effective and cost-competitive decentralized microgrid management solution under uncertainty. Full article

Climate change continues to pose significant challenges to global water security, with dams being particularly vulnerable to hydrological cycle alterations. This study investigated the climate-based impact on the hydrology of the Doğancı dam, located in Bursa, Turkey, using feed-forward neural networks (FNNs). The modeling used meteorological parameters as inputs. The employed FNN comprised one input, hidden, and output layer. The efficacy of the models was evaluated by comparing the correlation coefficients (R), mean squared errors (MSE), and mean absolute percentage errors (MAPE). Furthermore, two training algorithms, namely Levenberg-Marquardt and resilient backpropagation, were employed to determine the algorithm that yields more accurate output predictions. The findings of the study showed that the model using air temperature, solar radiation, solar intensity, evaporation, and evapotranspiration as predictors for the water budget and water level of the Doğancı dam exhibited the lowest MSE (0.59) and MAPE (1.31%) and the highest R (0.99) compared to other models under LM training. The statistical analysis determined no significant difference (p > 0.05) between the Levenberg and Marquardt and resilient backpropagation training algorithms. However, a visual interpretation revealed that the Levenberg-Marquardt algorithm outperformed the resilient backpropagation, yielding lower errors, higher correlation values, and faster convergence for the models tested in this study. The novelty of this study lies in the use of certain meteorological inputs, particularly snow depth, for dam inflow forecasting, which has seldom been explored. Moreover, this study compared two widely used ANN training algorithms and applied the modeling framework to a region of strategic importance for Turkey’s water security. This study highlights the effectiveness of ANN-based modeling for hydrological forecasting and determining climate-induced impacts on water bodies such as dams and reservoirs. Full article

In order to accurately evaluate the operating status of secondary equipment in smart substations, this paper establishes a secondary equipment status evaluation index system and proposes a secondary equipment status evaluation method based on multi-model fusion ensemble learning according to the differences of multiple machine learning algorithms as learners. The method consists of a two-layer structure. First, the original data is divided, and the divided data is used to perform k-fold verification on several base learners in the first layer. Then, the fully connected cascade (FCC) neural network in the second layer is used to fuse multiple base learners, and the Levenberg–Marquardt (LM) algorithm is used to train the FCC neural network so that the model converges quickly and stably. Simulation experimental analysis shows that the accuracy of secondary equipment status assessment of the proposed method is 98.71%, which can effectively evaluate the operating status of secondary equipment and provide guidance for the maintenance of smart substation systems and secondary equipment. Full article