Issam ATTOUI

Dans un contexte industriel en pleine transformation, la maintenance prédictive joue un rôle crucial pour garantir la performance et la sécurité des équipements. Ce projet vise à développer des techniques de diagnostic intelligent pour les machines tournantes, intégrant des approches innovantes basées sur l'analyse vibratoire, l’apprentissage profond et l'intelligence artificielle. Cette combinaison technologique permettra de détecter, localiser et classifier de manière proactive les défauts des systèmes mécaniques, tout en favorisant une interconnectivité accrue des processus industriels. Grâce aux évolutions récentes, ce projet cherche à réduire les coûts de maintenance en optimisant l'équilibre entre l'entretien préventif et correctif. En intégrant les capacités de l’apprentissage profond, il sera possible de surveiller en temps réel les machines et d'ajuster rapidement les stratégies d'entretien. Les solutions développées seront adaptées aux besoins spécifiques de l'industrie algérienne, soutenant ainsi sa transition vers des pratiques plus efficaces. Le projet s’articule autour de l’élaboration de méthodes numériques avancées pour l'acquisition et le traitement des signaux vibratoires. L'intégration de l’apprentissage profond et de l'intelligence artificielle permettra une gestion proactive et intelligente de la santé des machines, offrant ainsi des solutions innovantes pour optimiser les processus industriels.

Signal-based automatic condition monitoring techniques are one of the effective ways to ensure the operational safety of modern industrial systems. Currently, the main challenge is to increase their effectiveness under noisy environment and with limited labeled data. In this article, a novel fast and automatic signal-based fault diagnosis procedure that does not require any kinds of machine learning algorithms is proposed. This procedure is based on the measurement of the similarity in the frequency domain between the collected data and a small amount of labeled reference signals (LRSs). The LRSs are obtained under various known operation conditions using fast Fourier transform (FFT) algorithm. The proposed approach is suitable for a real-time implementation and capable of successfully overcoming the challenge of condition monitoring of rotating machines under noisy environment and with limited labeled data. The merits, fastness, and also robustness against noise of the proposed technique are demonstrated experimentally through different practical applications, as well as compared to state-of-the-art procedures on a database of vibration signals measured under a variety of machine health and working conditions.

In modern diagnostic approaches, the key step consists in generating the features related to fault type and severity. In fact, the generated features should be able to help the classifier to determine the health condition of the monitored system based on the measured signal. In this paper, in order to make an effective diagnosis about the rolling-element bearing failure, novel generated features that can maintain the physical meaning of the extracted vibration signal, while identifying its relationship to rolling bearing damage, are proposed using a wrapper model. For this purpose, based only on the Most Impulsive Frequency Bands (MIFBs) of the measured vibration signals for many bearing conditions, 33 feature parameters are proposed. Using a wrapper scheme, these parameters can be reduced until a set of them are found improving the efficiency of the diagnostic approach. The effectiveness of the proposed predictive features is analyzed by comparing it with some related works using many testing data for several bearing conditions. The experimental results reveal that the proposed procedure has obtained a high level of accuracy of 99.83%.