JURA ARKHANGELSKI

Mr. JURA ARKHANGELSKI

orcid

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
5 January 2021 in IEEE Transactions on Industry Applications
Verified
Journal Article
Closed access
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
Mahamadou Abdou-TankariMahamadou Abdou-Tankari
Mahamadou Abdou-Tankari
GLGilles Lefebvre
Gilles Lefebvre
This study deals with the urban microgrid energy management that is dedicated to individual and collective self-consumption by providing flexibility to the distribution grid (DG). The proposed urban community microgrid is interconnected to a DG and it consists of an association of centralized storage units (called community energy storage system), community intermittent renewable generation, and intelligent energy management system (EMS). One of the main advantages of urban microgrid is that, in case of faults in the DG, it can cut existing interconnections and continue to supply the responsible community in the island mode. In this study, the developed urban microgrid EMS is based on the predictive control management through the day-ahead optimal power flow (DA-OPF) strategy. The main contributions of this work can be defined by two points. The first point is related to a development of the DA-OPF strategy for the urban microgrid based on the intelligent deep learning data forecasting and the mixed-integer nonlinear programming optimization methods. The second point concerns a development of an optimization function integrating the concept of ancillary services of DG flexibility. Experimental results and economical evaluation are presented in this article. By using the proposed strategies, it results in an important electricity price reduction for the considered urban microgrid, compared to a conventional distribution system and basic operation schemes.

7 April 2020 in Energies
Verified
Journal Article
Open access
PDF file
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
PSPierluigi Siano
Pierluigi Siano
Mahamadou Abdou-TankariMahamadou Abdou-Tankari
Mahamadou Abdou-Tankari
GLGilles Lefebvre
Gilles Lefebvre
In this paper, an innovative method for managing a smart-community microgrid (SCM) with a centralized electrical storage system (CESS) is proposed. The method consists of day-ahead optimal power flow (DA–OPF) for day-ahead SCM managing and its subsequent evaluation, considering forecast uncertainties. The DA–OPF is based on a data forecast system that uses a deep learning (DL) long short-term memory (LSTM) network. The OPF problem is formulated as a mathematical mixed-integer nonlinear programming (MINLP) model. Following this, the developed DA–OPF strategy was evaluated under possible operations, using a Monte Carlo simulation (MCS). The MCS allowed us to obtain potential deviations of forecasted data during possible day-ahead operations and to evaluate the impact of the data forecast errors on the SCM, and that of unit limitation and the emergence of critical situations. Simulation results on a real existing rural conventional community endowed with a centralized community renewable generation (CCRG) and CESS, confirmed the effectiveness of the proposed operation method. The economic analysis showed significant benefits and an electricity price reduction for the considered community if compared to a conventional distribution system, as well as the easy applicability of the proposed method due to the CESS and the developed operating systems.

5 May 2026 in Energies
Verified
Journal Article
Open access
PDF file
Rakibul HasanRakibul Hasan
Rakibul Hasan
Majdi MansouriMajdi Mansouri
Majdi Mansouri
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
Mahamadou Abdou-TankariMahamadou Abdou-Tankari
Mahamadou Abdou-Tankari
Accurate forecasting of energy consumption in sensor-rich environments remains challenging due to strong inter-sensor dependencies, temporal variability, and heterogeneous sensor behavior. This paper proposes a lightweight and interpretable multi-sensor fusion framework for short-term energy consumption forecasting. The heterogeneous sensor dataset is first preprocessed to handle missing values, outliers, and temporal misalignment, followed by synchronization of the multivariate signals on a common timeline to enable consistent learning. The proposed framework systematically investigates multiple strategies for exploiting information from synchronized multi-sensor data without performing explicit feature elimination or time-lag engineering. In particular, three fusion paradigms are considered: (i) Early Fusion, where all sensor measurements are jointly used as input features for a multivariate regression model; (ii) Late Fusion, where individual sensor predictors are trained independently and their outputs are combined using reliability-based weighting; and (iii) an attention-inspired fusion strategy, in which adaptive weights are assigned to sensor-level predictions based on their predictive reliability estimated from training errors and normalized via a softmax function. In addition, classical machine learning models including Random Forest (RF), Support Vector Regression (SVR), and Gradient Boosting (GB) are evaluated under the same experimental conditions to provide a consistent benchmark. Experimental results on a real-world building energy monitoring dataset consisting of nine heterogeneous sensors demonstrate that multi-sensor fusion approaches consistently improve forecasting performance compared to single-model baselines. Among the evaluated strategies, Late Fusion provides stable performance across strongly correlated loads, while the attention-inspired fusion strategy exhibits improved robustness when handling sensors with varying predictive reliability. To ensure robustness and reproducibility, results are reported using multiple chronological validation splits, with performance evaluated in terms of RMSE, MAE, and R2 along with statistical measures including standard deviation and confidence intervals. The proposed framework provides a practical balance between predictive accuracy, interpretability, and computational efficiency, making it suitable for smart building energy management and real-world deployment scenarios.

27 May 2025
Verified
Conference Paper
Closed access
AHAurelien Hazan
Aurelien Hazan
Mahamadou Abdou-TankariMahamadou Abdou-Tankari
Mahamadou Abdou-Tankari
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
Energy management, for example in building smart grids, often leads to an optimization problem, continuous or combinatorial, such as a linear or quadratic program (LP/QP), depending on the choice of cost function. High-level power-flow software toolboxes usually rely on classical algorithms, such as interior point methods for LP. Recent progresses in neural networks (NN) research evidenced that NN could be used to learn to optimize in place of classical solvers, and might in some cases bring superior speed or performance. In this article we discuss several alternatives available in a fast-growing field of research, their advantages and drawbacks, and give an illustrative example with simple OPF problems.

24 March 2026 in Artificial Intelligence Research and Applications
Verified
Journal Article
Open access
PDF file
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
Rakibul HasanRakibul Hasan
Rakibul Hasan
Mahamadou Abdou-TankariMahamadou Abdou-Tankari
Mahamadou Abdou-Tankari
GLGilles Lefebvre
Gilles Lefebvre
Accurate 24-hour energy forecasting in smart buildings remains a challenging real-world problem due to the highly non-linear, dynamic, and heterogeneous nature of IoT sensor data. This study extensively expands upon previous research that demonstrated the effectiveness of Bi-LSTM as a univariate predictor. To address existing gaps, we make two primary contributions: (1) developing a multivariate forecasting framework that incorporates eight diverse IoT sensor streams, including HVAC (heat pump) energy consumption and indoor environmental comfort metrics, and (2) benchmarking advanced attention-based architectures, such as the Temporal Fusion Transformer (TFT), optimized for interpretable, high-performance multidimensional time-series forecasting. Using real-world datasets from a multifunctional smart building in France (Pôle Culturel), experiments were conducted over approximately four months (October 23, 2024, to February 27, 2025) at 10-minute intervals. The preprocessing pipeline includes timestamp harmonization, Akima interpolation for non-linear data gaps, outlier correction, and feature scaling. Using MAE, RMSE, MAPE, and R2 as evaluation metrics, the experiments compare a multivariate Bi-LSTM against standard models such as LSTM, GRU, and the TFT. The results indicate that the multivariate Bi-LSTM (R2 = 0.9656) significantly reduces errors compared to univariate approaches, accurately capturing daily trends and peak loads. Furthermore, while the TFT effectively models complex multivariate dependencies (R2 = 0.8130), it demonstrates that attention-based models require specific architectural tuning for medium-scale IoT datasets, providing practical guidance for sustainable, data-driven energy management.

4 April 2026
Verified
Thesis
Closed access
JAJURA ARKHANGELSKI
JURA ARKHANGELSKI
Ces dernières années, l’agrégation des sources d’énergie distribuées et l’émergence d’autoconsommateurs individuels et collectifs sur un nœud de puissance rendent les flux énergétiques multidirectionnels. L’optimisation par les Agrégateurs se base sur une exploitation de données, d’outils d’aide à la décision ainsi que des services de flexibilité. Les travaux de thèse portent sur l’étude de cette problématique de gestion de la flexibilité des réseaux électriques par l’intégration des microréseaux d’autoconsommation, notamment urbains. La gestion de la flexibilité est réalisée par une stratégie de gestion prédictive originale développée au cours de ces travaux. Elle associe les méthodes classiques d’optimisation des flux de puissance (OPF) basées sur la programmation non linéaire à nombres mixtes (MILNP), les algorithmes d’apprentissage neuronal récurrent à longue mémoire à court terme (LSTM) et des critères d’évaluation dynamique et d’aide à la décision. La principale fonction multivariable de coût à minimiser est basée sur l’estimation prédictive, la veille pour le lendemain (prévision « J+1 »), du prix marginal moyen localisé aux différents nœuds du réseau électrique. Les algorithmes gèrent les négociations entre l’agrégateur et les usagers du réseau électrique, analysent de manière itérative les différents scénarios de services système de flexibilité et valident le choix de la configuration optimale pour le lendemain. Des simulations basées sur les algorithmes de Monte-Carlo ainsi que des validations expérimentales ont été réalisées. Elles ont permis de tester et améliorer la stratégie de gestion énergétique proposée.