M. MEZAACHE Hatem

MAA

Directory of teachers

Department

Departement of ELECTRONICS

Research Interests

Renewable Energy, Signal Processing, Control.

Contact Info

University of M'Sila, Algeria

Email : hatem.mezaache@univ-msila.dz

On the Web:

  • Google Scholar N/A
  • ResearchGate
    ResearchGate N/A
  • ORCID N/A
  • SC
    Scopus N/A

Recent Publications

2024-12-23

Enhanced Short-Term Wind Speed Forecasting Using Weather Inputs, Multistage Decomposition, and a Combination of Deep Learning Techniques

As an environmentally friendly renewable energy source, wind energy has garnered considerable attention over the past decades. However, the inconsistency and instability of wind speed can impact the security and stability of large-scale wind power grids. To enhance wind speed prediction accuracy and reduce its volatility and unpredictable nature, multistage decomposition techniques such as ICEEMDAN (Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise) and VMD (Variational Mode Decomposition) are increasingly utilized. These methods optimize wind speed forecasting by leveraging meteorological data.
In this study, we applied this model to predict wind speed in a city located in southern Algeria and evaluated its performance against other models, including: ICEEMDAN-VMD–LSTM, ICEEMDAN-VMD–BiLSTM, ICEEMDAN-VMD–BiGRU and ICEEMDAN-VMD–OP. Based on our experiments in this research, combination methods relying on the Out-Performance (OP) technique produced more accurate forecasts and better captured the real variations in wind speed series.
The model's short-term forecasting accuracy was assessed using statistical methods based on various indicators, such as the coefficient of determination (R²), RMSE, MAPE, and MABE. The final forecasted results were compared with the actual input dataset, which included simulation outputs derived from predictive models, using a scatter plot. The proposed combined model demonstrated excellent predictive performance, stability, and relevance in short-term wind speed forecasting, achieving a superior R² value of 99.11%.
Citation

M. MEZAACHE Hatem, (2024-12-23), "Enhanced Short-Term Wind Speed Forecasting Using Weather Inputs, Multistage Decomposition, and a Combination of Deep Learning Techniques", [international] 5thInternational Conference on Scientific and Academic Research , Konya- Turkey

Read Full Paper
2024-12-18

The Effects of Weather Inputs on Predictionof SolarRadiation and their Importance for Saharan Agriculture

This study examines the influence of solar energy prediction on Saharan agriculture, highlighting the crucial role of accurate forecasting in enhancing agricultural practices in arid regions. By leveraging advanced forecasting models that integrate machine learning and weather data analysis, Saharan desert farmers can optimize resource allocation, irrigation schedules, and crop management strategies based on the anticipated availability of solar energy. Improved solar energy forecasts allow farmers to plant more efficiently, reduce water consumption, and increase crop yields by aligning agricultural activities with expected sunlight levels. Additionally, reliable solar energy forecasts facilitate the integration of renewable energy sources into agricultural operations, promoting sustainability and resilience to climate variability, by evaluating the impact of solar irradiation forecasts on Saharan agriculture.
This work investigates the influence of meteorological parameters on solar irradiation forecasting using deep learning models applied to the Msila site. To achieve this, several Deep Learning models were implemented and compared, including Long Short-Term Memory (LSTM) recurrent neural networks, Convolutional Neural Networks (CNN), and Multilayer Perceptron (MLP). These models use local meteorological data collected at the Msila site to predict solar irradiation over short-term periods. Forecasts were made with a 15-minute time resolution over a one-year period. To obtain the best forecasting results, we utilized three of the most effective deep learning methods, as demonstrated in our past studies, namely LSTM, CNN, and MLP.
Citation

M. MEZAACHE Hatem, (2024-12-18), "The Effects of Weather Inputs on Predictionof SolarRadiation and their Importance for Saharan Agriculture", [national] 1er Séminaire National sur l'Agriculture Saharienne à l'Ere des Energies Renouvelables , Adrar

Read Full Paper
2024-12-15

Comparative Analyses of Forecasting Methods for Renewable Energy Across Forecast Horizons

Sustainable energy sources like solar and wind speed provide an economically efficient source of energy. Prediction of the output of renewable energy plays a crucial role in shaping decisions concerning electrical system operation and management. Forecasting precision in renewable energy output is essential to ensuring the reliability and stability of the grid, as well as for mitigating risks and minimizing costs within the energy market and power systems. Various statistical techniques were developed to predict solar radiation and wind speed for this purpose and there are two types approaches commonly used: Deep learning and artificial Neural network (ANN). This work propose the used of three statistical methods based in Elman Recurrent Neural Network (ERNN), Long Short Term Memory (LSTM) and Support Vector Machine (SVM) to forecast the output data in different forecasting horizons. Four evaluation deferent metric are used: Forecast skill (FS), Root mean square error (RMSE), Mean Absolute Error (MAE) and coefficient of determination ( R2). These metrics confirm the robustness and accuracy of the LSTM model, validated by its RMSE, MAE, FS, and R² values for both sites. These performances demonstrate the effectiveness of LSTM in capturing temporal patterns, with significant implications for weather forecasting and renewable energy applications.
Citation

M. MEZAACHE Hatem, (2024-12-15), "Comparative Analyses of Forecasting Methods for Renewable Energy Across Forecast Horizons", [national] Algerian Journal of Renewable Energy and Sustainable Development , University Ahmed Draya- Adrar- ALGERIA

Read Full Paper
2024-12-08

Hybrid Machine Learning Model Combining SVR Kernels and Meteorological Inputs for Enhanced Short-Term Solar Power Forecasting

The prediction of solar energy production is essential for effective integration into the energy mix, as this source is a pillar of renewable energy production. It ensures optimized grid management, stabilizes energy supply, and maximizes the use of solar energy. This work proposes an innovative approach for short-term solar power forecasting, based on machine learning techniques, specifically multivariate-input Support Vector Regressions (SVR) utilizing meteorological data. The Pearson Correlation Coefficient (PCC) is used to measure the correlation between each input feature and solar energy, guiding the selection of relevant variables. Three types of kernels are explored: linear, polynomial, and radial basis function (RBF). Each of these kernels is trained and evaluated individually to predict solar power. Subsequently, the predictions obtained by these three models are combined using a method called Out Performance (OP), which leverages the strengths of each kernel to achieve a more accurate final prediction. Experimental results show that the proposed hybrid approach is a robust and efficient solution, significantly improving the accuracy of solar power prediction compared to individual SVR models and other commonly used prediction techniques.
Citation

M. MEZAACHE Hatem, (2024-12-08), "Hybrid Machine Learning Model Combining SVR Kernels and Meteorological Inputs for Enhanced Short-Term Solar Power Forecasting", [national] 1st National Conference On Artificiel Intelligence And Information , Relizane- ALGERIA

Read Full Paper
2024-11-18

Solar Radiation and Wind Speed Prediction using an Ensemble of Deep Learning Methods for Different Horizons

Sustainable energy sources like solar and wind speed provide an economically efficient source of energy. Prediction the output of renewable energy plays a crucial role in shaping decisions concerning the operation and management of power systems. Precision in forecasting renewable energy output is essential for ensuring the reliability and stability of the grid, as well as for mitigating risks and minimizing costs within the energy market and power systems. Various statistical methods have been employed to forecast solar irradiation and wind speed for this purpose and there are two types approaches commonly used: Deep learning and artificial Neural network (ANN). This work propose the used of three types of statistical methods based in Elman Recurrent Neural Network (ERNN), Long Short Term Memory (LSTM) and Support Vector Machine (SVM) to forecast the output data in deferent horizon step. Four evaluation deferent metric are used: Forecast skill (FS), Root mean square error (RMSE) , Mean Absolute Error (MAE) and coefficient of determination ( R2) which indicate that the outperform model based on LSTM with R2= 0.9963 for Tamanrasset site and 0.9976 for Colorado site in horizon of 1 step ahead
Citation

M. MEZAACHE Hatem, (2024-11-18), "Solar Radiation and Wind Speed Prediction using an Ensemble of Deep Learning Methods for Different Horizons", [national] The Second Conference on Advances in Computational Intelligence, Systems and Networking , Bouira- ALGERIA

Read Full Paper
2024-11-09

A Multi-Stage Hybrid Model for Short-Term Solar Photovoltaic Power Forecasting An Approach Using Dimensionality Reduction and fused machine learning techniques case study: Msila city

In recent years, the forecasting of energy output from photovoltaic panels has become a critical factor in optimizing the efficiency of energy systems. Recent advancements in the integration of renewable energy into the power grid have underscored the growing need for accurate and reliable solar power production methods. This paper presents a new innovative multi-stage hybrid model for solar power prediction, combining a dimensionality reduction approach, a set of machine learning techniques, and a fusion strategy to enhance the performance of our model. First, a dimensionality reduction of the real data is performed using PCA (Principal Component Analysis) to simplify complex datasets while preserving their essential features. Next, the new output data sets are forecasted in the second stage using different machine learning techniques such as Bidirectional Long Short-Term Memory (BiLSTM) networks, Convolutional Neural Networks (CNN), and Extreme Learning Machine Artificial Neural Networks (ELMNN). In the subsequent step, the predictions generated by the three models are fused using the LSR (Least Squares Regression) method to obtain a more accurate final prediction. The proposed approach was applied to a specific site located in Msila. To evaluate our model, we used various criteria such as R-squared, RMSE, MAPE, and MABE. These criteria were compared with the results obtained from individual predictors based on machine learning techniques to demonstrate the improvement in the accuracy of our model for short-term forecasting of solar power.
Citation

M. MEZAACHE Hatem, (2024-11-09), "A Multi-Stage Hybrid Model for Short-Term Solar Photovoltaic Power Forecasting An Approach Using Dimensionality Reduction and fused machine learning techniques case study: Msila city", [national] SECOND NATIONAL CONFERENCE OF MATERIALS SCIENCES AND RENEWABLE ENERGY , RELIZANE, ALGERIA

Read Full Paper
2024-10-13

Impact of Meteorological Inputs on Solar Irradiation Forecasting Using Deep Learning Models - Application to the Msila Site

Solar irradiation and hydroelectricity are among the main sources of renewable energy, used for many years in the industrial system of the Msila station. Solar irradiation is essential for energy production in solar power plants and is influenced by several environmental factors such as relative humidity, temperature, wind speed, and atmospheric pressure. The variability of these parameters makes forecasting the amount of electricity generated from solar sources particularly complex. This variability can cause significant fluctuations in solar energy production, thus complicating the efficient planning and management of energy resources. To overcome these challenges, precise forecasting of solar irradiation is indispensable. It not only helps optimize production and reduce operational costs but also facilitates grid integration, supports project planning, and ensures more effective resource management. This work explores the impact of meteorological parameters on solar irradiation forecasting using deep learning models applied to the Msila site. To achieve this objective, we implemented and compared several Deep Learning models, including Long Short-Term Memory (LSTM) recurrent neural networks, Convolutional Neural Networks (CNN), and Multilayer Perceptron (MLP). These models use local meteorological data collected at the Msila site to predict solar irradiation over short-term periods. Forecasts are made with a 15-minute time resolution over a one-year period. To refine our proposed system, two evaluation methods are used: a statistical evaluation based on three criteria—Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R²)—as well as a graphical method using scatter plots to compare actual data with the predicted results for different models. The results demonstrate that the LSTM model stands out for its ability to capture the sequential nature of meteorological data, allowing it to provide more accurate forecasts than other evaluated systems.
Citation

M. MEZAACHE Hatem, (2024-10-13), "Impact of Meteorological Inputs on Solar Irradiation Forecasting Using Deep Learning Models - Application to the Msila Site", [national] 1er Séminaire National : Eau, Environnement et Energies Renouvelables , M'sila- ALGERIA

Read Full Paper
2024-04-17

A Set of Forecasting Methods to Predict Solar Irradiance and Wind Speed for Different Horizons

Sustainable energy sources like solar and wind speed provide an economically efficient source of energy. Prediction of the output of renewable energy plays a crucial role in shaping decisions concerning electrical system operation and management. Forecasting precision in renewable energy output is essential to ensuring the reliability and stability of the grid, as well as for mitigating risks and minimizing costs within the energy market and power systems. Various statistical techniques were developed to predict solar radiation and wind speed for this purpose and there are two types approaches commonly used: Deep learning and artificial Neural network (ANN). This work propose the used of three statistical methods based in Elman Recurrent Neural Network (ERNN), Long Short Term Memory (LSTM) and Support Vector Machine (SVM) to forecast the output data in different forecasting horizons. Four evaluation deferent metric are used: Forecast skill (FS), Root mean square error (RMSE), Mean Absolute Error (MAE) and coefficient of determination ( R2). These metrics confirm the robustness and accuracy of the LSTM model, validated by its RMSE, MAE, FS, and R² values for both sites. These performances demonstrate the effectiveness of LSTM in capturing temporal patterns, with significant implications for weather forecasting and renewable energy applications.
Citation

M. MEZAACHE Hatem, (2024-04-17), "A Set of Forecasting Methods to Predict Solar Irradiance and Wind Speed for Different Horizons", [national] 1er Conférence National sur l'application d'intelligence A rtificiel le et le Développement Durable , EL-BAYATH

Read Full Paper
2023-10-26

A Very Short Term Photovoltaic Power Forecasting Model using LDA Method and Deep Learning based on Multivariate Weather datasets

Photovoltaic (PV) system-generated solar energy has inconsistent and variable properties, which makes controlling electric power distribution and preserving grid stability extremely difficult. A photovoltaic (PV) system’s performance is profoundly affected by the amount of sunlight that reaches the solar cell, the season of the year, the ambient temperature, and the humidity of the air. Every renewable energy technology, sadly, has its problems. As a result, the system is unable to function at its highest or best level. To combat the unstable and intermittent performance of solar power output, it is essential to achieve a precise PV system output power. This work introduces a new approach to enhancing accuracy and extending the time range of very-short-term solar energy forecasting (15 min step ahead) by using multivariate time series inputs in different seasons. First, Linear Discriminat Analysis (LDA) is used to select the relevant factors from the mixed meteorological input data. Secondly, two very short-term deep learning prediction models, CNN and LSTM, are used to predict PV power for a shuffled and reduced database of weather inputs. Finally, the predicted outputs from the two models are combined using classification strategy. The proposed method is applied to one year of real data collected from a solar power plant located in southern Algeria, to demonstrate that this technique can improve the forecasting accuracy compared to other techniques, as determined by statistical analysis involving normalized root mean square error (NRMSE), mean absolute error (MAE), mean bias error (MBE), and determination coefficient. (R2).
Citation

M. MEZAACHE Hatem, (2023-10-26), "A Very Short Term Photovoltaic Power Forecasting Model using LDA Method and Deep Learning based on Multivariate Weather datasets", [national] Engineering Proceedings , Multidisciplinary Digital Publishing Institute (MDPI)

Read Full Paper
2018

Auto-Encoder with Neural Networks for Wind Speed Forecasting

The use of wind energy is progressively utilized to produce electrical energy. Wind energy is related to the variation of some atmospheric variables such as wind speed, wind direction, air density and pressure. Recently, numerous methods base on Artificial Intelligence techniques to forecast wind speed have been proposed in the literature. The present study proposes a new artificial intelligence approach to forecast wind speed time series, it is composed from two blocs: The first one is based on the use of a deep architecture. The Autoencoder which is a type of deep neural networks, utilized generally for Denoising, is used to diminish the dimensionality of the wind speed time series. In the second bloc of the proposed methodology, the Elman neural network is used to predict future values of wind speed, it is a type of recurrent neural networks that are very sensitive to historical variations. To evaluate our approach we used the following error indicators: Mean Absolute Bias Error (MABE), Root Mean Square Error (RMSE),Mean Absolute Percentage Error (MAPE)and correlation coefficient (R2). The obtained results are compared with Extreme Learning Machine neural networks.
Citation

M. MEZAACHE Hatem, (2018), "Auto-Encoder with Neural Networks for Wind Speed Forecasting", [international] International Conference on Electronics and Electrical Engineering ICEEE 2018 , Bouira, Algeria

Read Full Paper
2017

Wind Speed Forecasting with Autoencoder and Elman Neural Network

In recent years, the use of wind energy is increasingly used to produce electric power. Wind energy is related to the variation of specific atmospheric variables such as wind speed, wind direction, air density and pressure. Recently, several methods base on Artificial Intelligence techniques to forecast wind speed have been reported in the literature. The present study proposes a new artificial intelligence approach to forecast wind speed time series, it is composed from two blocs: The first one is based on the use of a deep architecture. The autoencoder which is a type of deep neural networks, used mainly for Denoising, is used to reduce the dimensionality of the wind speed time series. In the second bloc of the proposed methodology, the Elman neural network is used to predict future values of wind speed, it is a type of recurrent neural networks that are very sensitive to historical variations. To evaluate our approach we used the following error indicators: Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Normalized Mean Square Error (NMSE). The obtained results are compared with other neural networks architectures.
Citation

M. MEZAACHE Hatem, (2017), "Wind Speed Forecasting with Autoencoder and Elman Neural Network", [international] The 1st International Conference On Electronics And New echnologies (ICENT) , M'sila. Algeria

Read Full Paper
2016

Kernel Principal Components Analysis with Extreme Learning Machines for Wind Speed Prediction.

Nowadays, wind power and precise forecasting are of great importance for the development of modern electrical grids. In this paper we propose a prediction system for time series based on Kernel Principal Component Analysis (KPCA) and Extreme Learning Machine (ELM). To compare the proposed approach, three dimensionality reduction techniques were used: full space (50 variables), part of space (last four variables) and classical Principal Components Analysis (PCA). These models were compared using three evaluation criteria: mean absolute error (MAE), root mean square error (RMSE), and normalized mean square error (NMSE). The results show that the reduction of the original input space affects positively the prediction output of the wind speed. Thus, It can be concluded that the non linear model (KPCA) model outperform the other reduction techniques in terms of prediction performance.
Citation

M. MEZAACHE Hatem, (2016), "Kernel Principal Components Analysis with Extreme Learning Machines for Wind Speed Prediction.", [international] The 7th International Renewable Energy Congress , Hammamet, Tunisia.

Read Full Paper
← Back to Researchers List