Browsing by Author "Kouadri, Abdelmalek (supervisor)"

Now showing 1 - 7 of 7

Design of a multivariate alarm delay-timers system for a comment rotary kiln
(2019) Belkacem, Rayane; Rabia, Amina; Kouadri, Abdelmalek (supervisor)
Alarm systems play an important role in industrial processes in the prevention of major accidents or disasters that lead to casualties and economical losses. In practice, however, nuisance alarms interfere with operators’ judgment. A good thresholding system should be able to distinguish between normal and abnormal situations of sensors measurements. Accomplishing this became a very challenging task since various techniques already exist. In this thesis, a novel technique is used to enhance the performance of alarm systems. This technique is mainly based on modeling the process using the principal component analysis (PCA). Once the model is validated; Instead of the conventional control limits, a combined index is used with delay-timers to enhance the alarm system’s performance. Simulated and industrial examples are discussed to prove the effectiveness of the proposed method in designing industrial alarm systems. This method proved its efficiency to fulfill high alarm performance.
Hidden markov model-based approach for process monitoring
(2019) Benabdallah, Mounir; Lounaouci, Mohamed Lamine; Kouadri, Abdelmalek (supervisor)
Hidden Markov Models (HMMs) are a popular and ubiquitous tool for modelling a large range of time series data. It has been applied successfully to various complex problems, being especially effective with those requiring a huge amount of measured data, such as pattern recognition in speech, handwriting and even facial recognition. Since Fault detection and diagnosis is an important problem in process engineering recent studies are focusing on developing new techniques which are more accurate, sensitive to small faults, with no time delay and can monitor multi-mode process ef- fectively. In order to satisfy these requirements, a huge data are needed and a suitable model to process these data is the HMM. The main objective of this work is to develop novel HMM-based approach to diagnose various operating modes of a process includ- ing Bayesian methods for mode selection. The mode in this work refers to process operational statuses such as normal or abnormal operating conditions.
Hierarchical clustering technique for lung diseases
(2022) Koufi, Inas; Kouadri, Abdelmalek (supervisor)
Each year, pulmonary diseases are the underlying cause of death worldwide. The process of detecting lung diseases can be time-consuming and error-prone. Such errors can be expensive and aff ect patients’ lives. Accuracy and fast diagnosis are therefore crucial. Due to its high clinical impact and remaining challenges, medical image analysis has become a broad and active area of research in recent decades where various machine learning methods have been developed to assist in medical diagnosis. These machine learning models often use neural networks as a tool of image manip- ulation, feature extraction, and classifi cation and clustering techniques. Our proposed solution consists of using an alternative approach. In our study, distance and similarity measures have been applied to medical images hierarchical clustering in order to deter- mine their usability and accuracy in detecting lung diseases. Three indices were covered in this work in order to cluster chest X-rays: Euclidean distance, cosine distance, and Jensen-Shannon divergence. Each metric proposed has been applied to and tested on CheXphoto dataset with seven labels. Promising results were obtained with an accuracy range of 61.6% to 81.2% of correct predictions. Therefore, the proposed methods have a good application prospect and promotion value.
Intelligent fault diagnosis scheme for Plant-wide processes
(2021) Goutas, Firas; Aouaneche, Farouk; Kouadri, Abdelmalek (supervisor)
For fault detection and diagnosis in complex systems, model based methods are not very practical. Instead, data-driven techniques are widely used especially machine learning methods such as classification and k-mean clustering. However to our knowledge, hierarchical clustering technique based on KLD is not explored in this field. This study suggests a new approach to build an unsupervised model for fault diagnosis using hierarchical clustering with the statistical multivariate technique Kullback-Leibler divergence as an index to compute the dissimilarity degree between the different data distributions. These datasets are preprocessed through the principal component analysis (PCA) which allows to reduce the dimensionality and generates a principal and a residual subspace that can both be used to train our model which can be visualized in a dendrogram. In order to produce the optimal model, we set various CPVs to obtain different numbers of retained components, hence different models. The proposed method was applied to plant-wide Tennessee Eastman process to test the accuracy and the elapsed time of our algorithm. The results show the effectiveness of our optimal model with an accuracy of 86.36% of correct predictions. Keywords: Fault Diagnosis (FD), Machine Learning (ML), Hierarchical Clustering, Kullback- Leibler Divergence (KLD), Principal Component Analysis (PCA), Dissimilarity Index, Tennessee Eastman Process.
Interval-valued PCA-based approach for fault detection in dynamic systems
(2022) Hellati, Sami Oussama; Kouari, Abdellah Anis; Kouadri, Abdelmalek (supervisor)
Fault detection and diagnosis is an important domain in modern process engineering, where principal component analysis (PCA) is one of its powerful data-driven techniques. The use of PCA in dynamic systems will approximate the dynamic behavior with a static one, which is not convenient. To address this issue, one of the most well-known approaches is the use of time- lag-shifted data; this approach is known as dynamic principal component analysis (DPCA). However, DPCA is still not an optimal solution due to the eﬀect of uncertainties on the model parameters, which will lead to drifts and aﬀect the performance of the model. In this disser- tation, a new approach is proposed to overcome this issue by including uncertainties in the modeling phase, which will ensure a safe interval for the data to ﬂuctuate. This approach is called interval-valued dynamic principal component analysis (IV-DPCA). To test the perfor- mance of IV-DPCA, real data obtained from a cement manufacturing plant were used to build and test the PCA, DPCA, and IV-DPCA models, then the three models were compared to each other in terms of false alarm rate (FAR), missed alarms rate (MDR), and detection time delay (DTD).
Observer-based active fault tolerant control of a three tank system (DTS-200)
(2022) Taib, Mohamed Amine; Kouadri, Abdelmalek (supervisor)
Multi Input Multi Output (MIMO) systems theory is concerned with complex dynamical systems with multiple input(control) variables ur that regulate multiple output variables ym . In general, a MIMO system is a collection of independent or interconnected Single Input Single Output (SISO) subsystems that can be conditionally represented. The system’s separate subsystems are referred to as direct (or forward path) channels, and each one establishes a connection between the appropriate scalar input and output. This project deals with nonlinear multivariable system control; the work will begin with linearizing the model around a suitable equilibrium point and extracting the system’s transfer function matrix representation. The RGA will be used to select the optimal pairings, and two observers will be designed to estimate the states and decouple disturbance and noise. The system will be infl uenced by two types of faults under the scope of Fault Tolerant Control: actuator faults and sensor faults, both abrupt and random faults are estimated. Finally, P and PI controllers are used to design the system’s regulation and tracking mode responses in order to improve system performance, particularly settling time. A simulation under Matlab Simulink will be carried on to simulate and control the system.
Reduced kernel PCA Technique for fault
(2021) Boulmerdj, Anes; Bounaas, Alaeddine; Kouadri, Abdelmalek (supervisor)
Fault detection and diagnosis is an important problem in process engineering. It is the central component of abnormal event management (AEM) which has attracted a lot of attention recently. This thesis discuses different classes of FDD approaches for process monitoring. In addition, it presents main results of fault detection and diagnosis in a cement manufacturing plant using three monitoring techniques. The techniques are based on multivariate statistical analysis and a threshold strategy. The process is statistically modeled using Principle Component Analysis (PCA), kernel PCA and new proposed reduced KPCA to cope with the computational problem introduced by KPCA. The proposed RKPCA method consists on reducing the number of observations in a data matrix using a proposed algorithm based on fractal dimension. The Hotelling’s T², Q in addition to the new proposed index called the combined statistic f are used as fault indicators for testing PCA, KPCA and the suggested approach RKPCA carried out using the cement rotary kiln system. The three methods are compared to in terms of False Alarms Rate (FAR), Missed Alarms Rate (MDR), Detection Time Delay (DTD) and the cost function (J). The obtained results demonstrate the effectiveness of the proposed technique in reducing the number of observations from 768 to 11, leading to an 11x11 kernel matrix instead of 768x768, hence, diminishing computational time and storage requirement. Moreover, it has effectively detected the different types of faults when using statistical indices.