Machine learning for the identification and classification of pieces in industrial manufacturing

In the world of industry, there is an increasing demand for machines, maybe not more sophisticated but more intelligent, capable of automating as many functions as possible, because the more tasks a machine can do, the more time for other things a human has. One possibility of increasing a machine’s...

Descripción completa

Detalles Bibliográficos
Autor: García Gómez, Raúl
Tipo de recurso: tesis de maestría
Fecha de publicación:2024
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/414900
Acceso en línea:https://hdl.handle.net/2117/414900
Access Level:acceso abierto
Palabra clave:Neural networks (Computer science)
Machine learning
Computer network protocols
Xarxes neuronals (Informàtica)
Aprenentatge automàtic
Protocols de xarxes d'ordinadors
Àrees temàtiques de la UPC::Enginyeria electrònica
Àrees temàtiques de la UPC::Informàtica
Descripción
Sumario:In the world of industry, there is an increasing demand for machines, maybe not more sophisticated but more intelligent, capable of automating as many functions as possible, because the more tasks a machine can do, the more time for other things a human has. One possibility of increasing a machine’s intelligence is the addition of new technologies like Machine Learning and IoT. Using as an example a piece classification problem, in this master's thesis, convolutional neural networks (CNN) models have been designed for the identification and classification of images and the transmission of its result to an industrial machine. Different systems for sending data to an industrial machine have been planned and created through the use of communication protocols such as Modbus TCP, OPC UA or MQTT. The Python programming language has been used with tools such as communication libraries for the aforementioned protocols and the Machine Learning development framework Tensorflow. At the end of this work, convolutional neural network model that identified and classified parts was created, and this information could be sent to an industrial machine that has one of the communication protocols named before.