Publications Scientifiques

Permanent URI for this communityhttps://dspace.univ-boumerdes.dz/handle/123456789/10

Browse

Start date: 2020Subject: search.filters.subject.Machine learning

Search Results

Now showing 1 - 10 of 26
  • No Thumbnail Available
    Item
    Offline Arabic handwritten character recognition: from conventional machine learning system to deep learning approaches
    (2022) Faouci, Soumia; Gaceb, Djamel; Haddad, Mohammed
    Researchers have made great strides in the area of Arabic handwritten character recognition in the last decades especially with the fast development of deep learning algorithms. The characteristics of Arabic manuscript text pose several problems for a recognition system. This paper presents a conventional machine learning system based on the extraction of a set of preselected features and an SVM classifier. In the second part, a simplified convolutional neural network (CNN) model is proposed, which is compared to six other CNN models based on the pre-trained architectures. The suggested methods were tested using three databases: two versions of the OIHACDB dataset and the AIA9K dataset. The experimental results show that the proposed CNN model obtained promising results, as it is able to recognise 94.7%, 98.3%, and 95.6% of the test set of the three databases OIHACDB-28, OIHACDB-40, and AIA9K, respectively.
  • No Thumbnail Available
    Item
    Using Machine Learning Algorithms for the Analysis and Modeling of the Rheological Properties of Algerian Crude Oils
    (Taylor and Francis Ltd., 2024) Souas, Farid; Oulebsir, Rafik
    Our research described in this report investigated the rheological behavior of crude oils from the Tin Fouye Tabankort oil field in Southern Algeria, focusing on their viscosity under varying temperatures (10 °C–50 °C). The results show that the oils exhibited non-Newtonian shear-thinning behavior at low shear rates, with the viscosity decreasing as the temperature was increased. At higher shear rates, the Herschel–Bulkley model accurately described the oils’ transition to Newtonian behavior. Machine learning models, including CatBoost, LightGBM, and XGBoost, were trained on the experimental data to predict the viscosity, with CatBoost and XGBoost showing superior performance. We suggest these findings are valuable for improving the efficiency of oil transportation and processing.
  • No Thumbnail Available
    Item
    Development of an expert-informed rig state classifier using naive bayes algorithm for invisible loss time measurement
    (Springer Nature, 2024) Youcefi, Mohamed Riad; Boukredera, Farouk Said; Ghalem, Khaled; Hadjadj, Ahmed; Ezenkwu, Chinedu Pascal
    The rig state plays a crucial role in recognizing the operations carried out by the drilling crew and quantifying Invisible Lost Time (ILT). This lost time, often challenging to assess and report manually in daily reports, results in delays to the scheduled timeline. In this paper, the Naive Bayes algorithm was used to establish a novel rig state. Training data, consisting of a large set of rules, was generated based on drilling experts’ recommendations. This dataset was then employed to build a Naive Bayes classifier capable of emulating the cognitive processes of skilled drilling engineers and accurately recognizing the actual drilling operation from surface data. The developed model was used to process high-frequency drilling data collected from three wells, aiming to derive the Key Performance Indicators (KPIs) related to each drilling crew’s efficiency and quantify the ILT during the drilling connections. The obtained results revealed that the established rig state excelled in automatically recognizing drilling operations, achieving a high success rate of 99.747%. The findings of this study offer valuable insights for drillers and rig supervisors, enabling real-time visual assessment of efficiency and prompt intervention to reduce ILT.
  • No Thumbnail Available
    Item
    Enhancing Porosity Prediction in Reservoir Characterization through Ensemble Learning: A Comparative Study between Stacking, Bayesian Model Optimization, Boosting, and Random Forest
    (Slovnaft VURUP a.s, 2024) Youcefi, Mohamed Riad; Alshokri, Ayman Inamat; Boussebci, Walid; Ghalem, Khaled; Hadjadj, Asma
    Accurate estimation of porosity is a critical factor in reservoir characterization. This study aims to enhance porosity prediction through the implementation and comparison of various stacking ensemble learning strategies. A dataset comprising 273 points, which consists of well logs and core measurements, was collected from two wells for model development. Four base learners, including Support Vector Regression (SVR), Multi-Layer Perceptron (MLP), Random Forest Regression (RFR), and XGBoost, were trained on this dataset. These models were then integrated using multiple stacking ensemble techniques, such as weighted averaging, Bayesian model averaging, and RFR as a meta-learner. Meta-learners were trained on predictions from the base learners, generated through cross-validation on leave-out data. Performance evaluations of both base and meta learners were conducted on a separate testing dataset using statistical and graphical error analysis. Results indicate that all learners demonstrated robust performance, with weighted averaging outperforming other strategies on testing data. The stacking ensemble approach, particularly through weighted averaging, effectively improved base learner performance on testing data by leveraging individual model strengths and mitigating weaknesses. The findings of this study are valuable for geoscientists and reservoir engineers in achieving accurate reservoir characterization and facilitating exploration activities.
  • No Thumbnail Available
    Item
    Modeling wax disappearance temperature using robust white-box machine learning
    (Elsevier Ltd, 2024) Nait Amar, Menad; Zeraibi, Noureddine; Benamara, Chahrazed; Djema, Hakim; Saifi, Redha; Gareche, Mourad
    Wax deposition is one of the major operational problems encountered in the upstream petroleum production system. The deposition of this undesirable scale can cause a variety of challenging problems. In order to avoid the latter, numerous parameters associated with the mechanism of wax deposition should be determined precisely. In this study, a new smart correlation was proposed for the accurate prediction of Wax disappearance temperature (WDT) using a robust explicit-based machine learning (ML) approach, namely gene expression programming (GEP). The correlation was developed using comprehensive experimental measurements. The obtained results revealed the promising degree of accuracy of the suggested GEP-based correlations. In this context, the newly-introduced correlations provided excellent statistical metrics (R2 = 0.9647 and AARD = 0.5963 %). Furthermore, performance of the developed correlation outperformed that of many existing approaches for predicting WDT. In addition, the trend analysis performed on the outcomes of the proposed GEP-based correlations divulged their physical validity and consistency. Lastly, the findings of this study provide a promising benefit, as the newly developed correlations can notably improve the adequate estimation of WDT, thus facilitating the simulation of wax deposition-related phenomena. In this context, the proposed correlations can supply the effective management of the production facilities and improvement of project economics since the provided correlation is a simple-to-use decision-making tool for production and chemical engineers engaged in the management of organic deposit-related issues.
  • No Thumbnail Available
    Item
    Intelligent multi-fault identification and classification of defective bearings in gearbox
    (SAGE Publications Inc., 2024) Damou, Ali; Ratni, Azeddine; Benazzouz, Djamel
    Bearing faults in gearbox systems pose critical challenges to industrial operations, needing advanced diagnostic techniques for timely and accurate identification. In this paper, we propose a new hybrid method for automated classification and identification of defective bearings in gearbox systems with identical rotating frequencies. The method successfully segmented the signals and captured specific frequency components for deeper analysis employing three distinct signal processing approaches, ensemble empirical mode decomposition EEMD, wavelet packet transform WPT, empirical wavelet transform EWT. By decomposing vibration signals into discrete frequency bands using WPT, relevant features were extracted from each sub-band in the time domain, enabling the capturing of distinct fault characteristics across various frequency ranges. This extensive set of features is then served as inputs for machine learning algorithm in order to identify and classify the defective bearing in the gearbox system. Random forest RF, decision tree DT, ensemble tree ET classifiers showcased a notable accuracy in classifying different fault types and their localizations. The new approach shows the high performance of the diagnostic gearbox with a minimum of accuracy (Min = 99.95 %) and higher stability (standard deviation = 0.1).
  • No Thumbnail Available
    Item
    Machine learning-based Shapley additive explanations approach for corroded pipeline failure mode identification
    (Elsevier Ltd, 2024) Ben Seghier, Mohamed El Amine; Mohamed, Osama Ahmed; Ouaer, Hocine
    Rapid failure mode identification of oil and gas pipelines can prevent catastrophic consequences, improve fast intervention and enhance the design safety of these critical systems. This paper proposes explainable-based machine learning models using to determine the failure mode of corroded pipelines as a function of geometric configurations, material properties, and corrosion defect details. To determine the best identification model, this study examined eight machine learning models, including Nave Bayes, K-Nearest Neighbors, Decision Tree, Random Forest, Adaptive Boosting, Extreme Gradient Boosting, Light Gradient Boosting Machine, and Category Boosting, based on a comprehensive experimental database for steel pipelines with various corrosion/crack defect configurations. Furthermore, the Shapley additive explanations approach is utilized to rank the input variables for failure mode identification and explains the machine learning model predicting a specific failure mode for a given sample. In identifying the failure mode of corroded pipelines, the proposed Extreme Gradient Boosting model indicated the highest accuracy in term of performance evaluation compared to all other proposed models. In addition, the model-explanation findings show that the important parameter influencing the failure mechanism of corroded pipelines is the depth of corrosion defects followed by the pipeline wall thickness. The proposed framework is adaptable enough to allow further use of experimental results for having new insights.
  • No Thumbnail Available
    Item
    Artificial Intelligence Technique in earth sciences for porosity prediction in shaly petroleum reservoir from geophysical well-logs data. Application to Hassi R'mel field, Algeria
    (MEACSE Publication, 2024) Aliouane, Leila; Ouadfeul, Sid-ali
    Machine learning techniques are becoming very popular in earth sciences, mainly in petroleum exploration and exploitation. Reservoir characterization using geophysical well-logs data analysis is commonly conducted and plays a central role in formation evaluation in petroleum domain. The most petrophysical parameters that describe the reservoir are the porosity, the permeability and the water saturation where the porosity is the main key. Using conventional methods, the estimation of the porosity is very difficult, mainly in shaly reservoirs where the presence of clay affects considerably, the porosity and the permeability. For that, we propose to accurately predict the porosity from geophysical recordings crossed the formation of wells using machine learning methods such as multi-layer neural network. The input layer are constituted by the petrophysical well-logs data and the output layer presented by one neuron corresponding to the predicted porosity. The training step of neural network machine (NNM) is processed using core data (CORPOR) by minimizing the root mean square error using Radial Basis Function algorithm (RBF). Once trained, the model is then applied to the target wells to predict porosity (PORRBF). The predicted porosity match the core values with good accuracy. This approach provides significantly a robust computation method and reduces dependency on prior domain knowledge
  • No Thumbnail Available
    Item
    A new methodology to predict the sequence of GFRP layers using machine learning and JAYA algorithm
    (Elsevier, 2023) Fahem, Noureddine; Belaidi, Idir; Oulad Brahim, Abdelmoumin; Capozucca, Roberto; Thanh, Cuong Le; Khatir, Samir; Abdel Wahab, Magd M.
    In this paper, the best stacking sequence using experimental tests of GFRP composites is investigated. The main objective of this work is to determine the main specification of GFRP composite material, which is represented by its physics-mechanical properties, weight, and cost, before performing a series of experimental tests based on various stacking sequences. Our methodology is divided into three stages. The first stage is characterized by extracting the bending data from mechanical tests of some GFRP composites. In the second stage, the validated numerical model is used to simulate numerous cases of stacking sequences. In the last stage, the extracted data is used to determine the parameters for different stacking sequences using an inverse technique based on ANN and JAYA algorithm. The results provide a good prediction of parameters as well as a good orientation to make decisions on the best GFRP stacking sequence to be used, according to the required specifications of the manufacturer.
  • No Thumbnail Available
    Item
    Classification of Left/Right Hand and Foot Movements from EEG using Machine Learning Algorithms
    (Institute of Electrical and Electronics Engineers Inc, 2023) Cherifi, Dalila; Berghouti, Baha Eddine; Boubchir, Larbi
    In recent years, there has been growing interest in utilizing Electroencephalography (EEG) data and machine learning techniques to develop innovative solutions for individuals with disabilities. The ability to accurately classify hands and foot motion based on EEG signals holds great potential for enabling individuals to regain control and functionality of their disabled parts, improving their quality of life and independence. Making a better solution than the traditional ones that often require physical contact or can be challenging to operate. In our study, we have focused on hands (right/left) and foot motion disabilities, using supervised Machine Learning algorithms for the classification of EEG data related to left/right hand and foot movements; aiming to reach accurate results that can contribute to providing a solution for people with this kind of motion disabilities. Three supervised machine learning algorithms are considered for the EEG classification, namely Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM), using Common Spatial Patterns (CSP) algorithm and logarithm of the variance (logvar) for feature extraction. In our experiments, we adopted these algorithms to classify the Motor Imagery EEG dataset for hands and foot movements given in BCI Competition IV. The data we used went through different steps before fitting into the models such as filtering, feature extraction, and discrimination. We achieved significant success in accurately classifying hand movements in the initial experiment, attaining an impressive classification accuracy of up to 97.5% with SVM and LDA. Furthermore, in the multi-classification task involving both hand (right/left) and foot movements, KNN and SVM classifiers yielded commendable results up to 87%. These models can be further used and developed, where a hardware implementation will be done as a further work for this study.