Maissa Taktak

Ms. Maissa Taktak

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
13 May 2025 in Energies
Verified
Journal Article
Open access
PDF file
Maissa TaktakMaissa Taktak
Maissa Taktak
Faouzi DerbelFaouzi Derbel
Faouzi Derbel
Accurate energy consumption forecasting is essential for efficient power grid management, yet existing deep learning models struggle with the multi-scale nature of energy consumption patterns. Contemporary approaches like LSTM and GRU networks process raw time series directly, failing to distinguish between distinct frequency components that represent different physical phenomena in household energy usage. This study presents a novel methodological method that systematically decomposes energy consumption signals into low-frequency components representing gradual trends and daily routines and high-frequency components capturing transient events, such as appliance switching, before applying predictive modeling. Our approach employs computationally efficient convolution-based filters—uniform and binomial—with varying window sizes to separate these components for specialized processing. Experiments on two real-world datasets at different temporal resolutions (1 min and 15 min) demonstrate significant improvements over state-of-the-art methods. For the Smart House dataset, our optimal configuration achieved an R² of 0.997 and RMSE of 0.034, substantially outperforming previous models with R² values of 0.863. Similarly, for the Mexican Household dataset, our approach yielded an R² of 0.994 and RMSE of 13.278, compared to previous RMSE values exceeding 82.488. These findings establish frequency decomposition as a crucial preprocessing step for energy forecasting as it significantly improve the prediction in smart grid applications.

24 August 2025
Verified
Conference Paper
Closed access
Maissa TaktakMaissa Taktak
Maissa Taktak
KTKhaled Taouil
Khaled Taouil
AFAhmed Fakhfakh
Ahmed Fakhfakh
FDFaouzi Derbel
Faouzi Derbel
Energy consumption forecasting remains a critical challenge in power grid management, with existing methods struggling to balance computational complexity with prediction accuracy. Here, we present a novel hybrid approach that decomposes power consumption signals into frequency components before applying deep learning, addressing the limitations of current approaches in capturing both long-term trends and rapid fluctuations. By separating low-frequency trends from high-frequency components using moving average filtration, and processing each through specialized Long Short-Term Memory (LSTM) networks, our method achieves superior prediction accuracy compared to traditional single-stream approaches. The frequency decomposition plays a crucial role in improving predictive performance by allowing the model to focus separately on persistent consumption patterns and transient variations, reducing noise interference, and enhancing feature extraction. Testing on two distinct datasets – Smart House and Mexican Household – demonstrates the robustness of the methodology, yielding Root Mean Square Error (RMSE) values of 0.0132 and 0.0195, respectively. Our findings establish frequency decomposition as an effective strategy for energy forecasting, enabling better capture of both short- and long-term dependencies, with immediate applications in smart grid management and energy optimization systems.

17 September 2024 in IEEE Sensors Letters
Verified
Journal Article
Closed access
Maissa TaktakMaissa Taktak
Maissa Taktak
MBMohamed Khalil Baazaoui
Mohamed Khalil Baazaoui
IKIlef Ketata
Ilef Ketata
SSSalwa Sahnoun
Salwa Sahnoun
AFAhmed Fakhfakh
Ahmed Fakhfakh
Faouzi DerbelFaouzi Derbel
Faouzi Derbel
Precise distance estimation is essential in various fields, influencing customary aspects from daily activities to advanced research. In wireless Sensor Networks(WSNs) accurate distance estimation is crucial for different applications such as localization, energy efficiency, dynamic routing, coverage optimization, etc. In this paper, we strive to assess distance accurate estimation across various technologies, including a sub-GHz low-power, low-data-rate radio chip, and the ultra-wideband transceiver. We utilize a combination of time of flight(ToF), link quality metrics (LQM), and machine learning (ML) techniques to elucidate the strengths and limitations of each technology.

20 December 2025
Verified
Conference Paper
Closed access
Maissa TaktakMaissa Taktak
Maissa Taktak
AFAhmed Fakhfakh
Ahmed Fakhfakh
FDFaouzi Derbel
Faouzi Derbel

15 April 2026
Verified
Conference Paper
Closed access
Maissa TaktakMaissa Taktak
Maissa Taktak
AFAhmed Fakhfakh
Ahmed Fakhfakh
FDFaouzi Derbel
Faouzi Derbel
Multi-horizon energy forecasting remains a fundamental challenge for smart grid management, with the optimal modeling strategy for different forecast horizons remaining an open question. Effective energy management requires accurate predictions from 6 hours to 24 hours in advance to enable proactive load balancing, generator dispatch scheduling, and demand response activation. This paper compares two distinct approaches: Multi-Resolution Models (MRM) that train specialized models for each forecast horizon using resampled data matching the prediction time scale, and Multi-Horizon (MH) models that employ fixed 1 hour temporal resolution with multi-step-ahead prediction. We evaluate both strategies using GRU networks with binomial frequency decomposition on datasets collected over 14 months from household energy consumption in northeastern Mexico. Results demonstrate that resolution adaptation provides substantial advantages: MRM achieves 63-89% lower RMSE across all horizons compared to MH, with R2 values above 0.53 even at 24-hour horizons versus 0.22 for MH. These findings indicate that matching data granularity to forecast horizon substantially improves prediction accuracy for smart grid applications requiring multi-temporal forecasting.

28 October 2025
Verified
Conference Paper
Closed access
Maissa TaktakMaissa Taktak
Maissa Taktak
KTKhaled Taouil
Khaled Taouil
AFAhmed Fakhfakh
Ahmed Fakhfakh
FDFaouzi Derbel
Faouzi Derbel
Energy consumption forecasting at hourly granularity is particularly challenging for sustainable energy management systems, as it represents long-term consumption pattern as well as short-term volatility. This paper introduces an optimized frequency decomposition-based method of hourly energy forecasting. By separating low-frequency and high-frequency components of hourly power consumption signals using a moving average filter, and then applying specialized Long Short-Term Memory (LSTM) networks to each component, accurate predictions of energy consumption patterns can be achieved. The primary innovation in this work is the rigorous systematic optimization of the lookback parameter with thorough evaluation based on $R^{2}$, RMSE, and MAE performance measures. Two distinct datasets—Smart House and Mexican Household—have been employed in an attempt to cross-validate the paper and have established the efficacy of the optimized approach with $R^{2}$ values of 0.959 and 0.991, respectively. The results show that a lookback of 120 hours for Smart House and 144 hours for Mexican Household data were optimal to capture weekly energy consumption trends adequately.