https://etalpykla.vilniustech.lt/handle/123456789/159401 2026-04-11T21:33:44Z https://etalpykla.vilniustech.lt/handle/123456789/159673 Advanced Battery Management for Electric Vehicles: A Deep Dive into Estimation Techniques Based on Deep Learning for the State of Health and State of Charge of Lithium-Ion Batteries El Fallah, Saad; Kharbach, Jaouad; Vanagas, Jonas; Vilkelytė, Živilė; Tolvaišienė, Sonata; Ikmel, Ghita The precision of state of charge (SoC) prediction prediction of the SoC is necessary to avoid deep discharging and remains an important challenge in the field of electric vehicles and overcharging, which can damage batteries and shorten their life. the energy storage industry. The SoC is the percentage of energy The degradation of lithium-ion batteries is a key area of research, available in a battery relative to its total capacity. A precise as these type of batteries are extensively employed in many sectors, in particular electric vehicles, electronic devices and renewable energies. Evaluation systems based on deep learning, like Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), are becoming increasingly popular for their ability to process complex data and make accurate predictions. As lithium-ion batteries degrade over time, their reliability in storing and delivering energy is diminishing. To effectively monitor and manage this degradation, researchers are turning to evaluation systems based on deep learning. These approaches enable the prediction of battery state of health (SoH) and SoC, making it easier to optimize battery use and extend battery life. This article presents various techniques for predicting the SoH and SoC of batteries to evaluate the degradation of cells. 2024-01-01T00:00:00Z https://etalpykla.vilniustech.lt/handle/123456789/159672 Empowering Industrial Energy Management: Advancing Short-Term Load Forecasting with LSTM and CNN Deep Learning Models - Insights from a Moroccan Case Study Boumais, Khaoula; Messaoudi, Fayçal; Lagnaoui, Saloua; El Fallah, Saad; Udris, Dainius Self-consumption of electricity plays an important role in the energy transition and using green, sustainable energy sources for industrial self-sufficiency and electricity bills, meeting part of their own energy needs and even generating 20% of the annual surplus that could be sold to the grid. This study aims to forecast short-term load spanning between 2022 and 2023, employing deep learning models, specifically Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. Various performance metrics have been used to evaluate and compare the accuracy of these two models, including mean squared error, root mean squared error, mean absolute error, and mean absolute deviation. Results reveal LSTM's superior performance over CNN, with LSTM demonstrating adeptness in capturing underlying patterns while CNN tends to learn noise, leading to a divergence in performance metrics between training and validation data. The findings underscore the significance of LSTM for accurate load forecasting and suggest the inclusion of additional hyperparameter optimization to enhance the reliability of short-term load predictions, distinct from previous studies. 2024-01-01T00:00:00Z https://etalpykla.vilniustech.lt/handle/123456789/159671 Adaptive Methods for Kernel Initialization of Convolutional Neural Network Model Applied to Plant Disease Classification Lagnaoui, Saloua; Boumais, Khaoula; El Fallah, Saad; En-Naimani, Zakariae; Haddouch, Khalid; Matuzevičius, Dalius Convolutional Neural Networks are instrumental in artificial intelligence, especially in image processing, where their ability to autonomously learn hierarchical features has led to significant breakthroughs. However, the success of these models is intricately tied to the judicious choice of hyperparameters, which include the configuration of convolutional layers, activation functions, and kernel initialization methods. This research explores kernel initialization methods in Convolutional Neural Network models, seven diverse initialization methods (Glorot Uniform, Ones initialization, Zero initialization, Constant initialization, Random initialization, HeNormal initialization, and Orthogonal initialization) are comprehensively compared. The primary objective is to showcase the sensitivity of Convolutional Neural Networks to these various initialization techniques. The study not only aims to reveal the nuanced impact of kernel initialization but also introduces an adaptive method to enhance model performance. By delving into the intricacies of initialization methods, this research contributes to the improvement of Convolutional Neural Networks effectiveness, especially in critical applications like Plant Disease classification. 2024-01-01T00:00:00Z https://etalpykla.vilniustech.lt/handle/123456789/159670 Memristor-Based Phase-Frequency Detector for Phase-Locked Loop Applications Elashkar, Nahla; Ibrahim, Ghada; Aboudina, Mohamed; Fahmy, Hossam; Hussein, Ahmed The present paper introduces and illustrates a novel, straightforward Phase-Frequency Detection (PFD) circuit based on two memristor components. This PFD approach can represent the phase or frequency difference between two sinusoidal inputs as a DC signal. As a result, the suggested method does away with the Low Pass Filter (LPF) block in the Phase-Locked Loop (PLL) structure, resulting in a reduction in the dissipated power and overall system area and an increase in the PLL efficiency when employed in many applications such as the communications module of body implants. First, the linear dopant drift memristor model is used to derive a closed-form analytical solution for the proposed PFD concept. Later, simulations for the proposed PFD circuit are carried out utilizing the more realistic nonlinear dopant drift memristor model to verify the validity of the expected findings despite considering all known non-idealities of actually realized memristor devices. The emergence of this circuit will encourage the investigation and development of the first memristor-based PLL system, which can be constructed directly using this Phase detector cascaded by a memristor-based Voltage Controlled Oscillator (VCO), with the latter being extensively researched in the literature. 2024-01-01T00:00:00Z