Veuillez utiliser cette adresse pour citer ce document :
Titre: Identification des paramètres d’amélioration du départ de la BMX Race.
Auteur(s): MEKHEZZEM, Réda
Mots-clés: Machine Learning, data science, BMX, Linear regression, supervised learning.
apprentissage automatique, science de données, BMX, régression linéaire, apprentissage supervisé.
Date de publication: 13-jan-2020
Editeur: 04-05-2021
Référence bibliographique: salle des thèses
Collection/Numéro: BFST2535;
Résumé: This graduation project is a part of the multidisciplinary work carried out jointly by the IDD team of the LIAS laboratory and the RoBioSS team of the PPRIME Institute, which aim to improve the departure of the BMX Race drivers. The IDD team specializes in data processing whereas the RoBioSS team is specialized in the field of biomechanics. It offers a hardware solution (Cranks sensors, cameras, test workshop for pilots) to collect data related to the departure of the elite drivers of the French team. The RoBioSS team proposes biomechanical models to represent these departures in order to optimize them afterwards. However this solution does not facilitate the identification of the parameters responsible for the improvement of the departure and this is because of the complexity of the proposed biomechanical models. In order to simplify its models and to identify the relevant parameters impacting the departure of the BMX Race, the PPRIME institute called on the LIAS laboratory to work on new models based on a different field which is Machine learning. Machine learning is a computer process that aims to derive a set of rules from a dataset to build new knowledge. This process has been successfully applied in different areas, such as old sales analysis systems for predicting customer behavior and weather forecasts. This project aims to propose a simplified model but predictive of performance based on machine learning techniques. The purpose of my work is to study the initial data provided by the RoBioSS team, to design a solution compatible with this type of data and to test the algorithms on these same data.
Collection(s) :Master MID

Fichier(s) constituant ce document :
Fichier Description TailleFormat 
Mekhezzem.pdfCD4,79 MBAdobe PDFVoir/Ouvrir

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