Veuillez utiliser cette adresse pour citer ce document :
Affichage complet
Élément Dublin CoreValeurLangue
dc.contributor.authorBazi, Rabah-
dc.contributor.otherRechak, Saïd, Directeur de thèse-
dc.contributor.otherBenkedjouh, Tarak, Directeur de thèse-
dc.descriptionThèse de Doctorat : Génie Mécanique : Alger, École Nationale Polytechnique : 2022fr_FR
dc.description.abstractIn this study we have developed two new models for the automatic monitoring of the state of health of cutting tools. This method consists of estimating; validate and calculate their residual life (RUL). The method uses monitoring data provided by sensors (Force, Acceleration AND AE), and is based on two main stages: the first stage online and the second stage offline. In the first model, the raw signals provided by the sensors are first processed to extract useful information through the use of continuous wavelet transform (CWT), source separation "SCA" and the Tagichi-Mahal system. -Anubis (MTS); and in the second model, variational mode decomposition (VMD) and (CNN-BLSTM) were applied. Data processing after source separation is the subject of the development of the health indicator. The prediction of the residual life allows the determination of the accuracy and the various parameters of the prognosis like the RMSE, which allow to classify and validate the robustness of this model. The proposed method can be implemented on real data in order to determine the wear of the cutting tools during the machining process of the different materials.fr_FR
dc.subjectTool Wear Monitoringfr_FR
dc.subjectData Extraction and Reductionfr_FR
dc.subjectHealth Indicatorfr_FR
dc.subjectRemaining useful Lifefr_FR
dc.subjectDeep Learningfr_FR
dc.subjectExtreme Learningfr_FR
dc.titleDiagnosis and prognosis of cutting tools based on blind sources separation : application to millingfr_FR
Collection(s) :Département Génie Mécanique

Fichier(s) constituant ce document :
Fichier Description TailleFormat 
BAZI.Rabah.pdfD00162212.58 MBAdobe PDFVoir/Ouvrir

Tous les documents dans DSpace sont protégés par copyright, avec tous droits réservés.