Defesa de Dissertação de Mestrado – Denise Albertazzi Gonçalves – 6/5/2019
Defesa de Dissertação de Mestrado | |
Aluna | Denise Albertazzi Gonçalves |
Orientador | Prof. Marcelo Ricardo Stemmer, Dr. – DAS/UFSC |
Data | 6/5/2019 (segunda-feira) – 10h00
Sala PPGEAS I (piso superior) |
Banca | Prof. Marcelo Ricardo Stemmer, Dr. – DAS/UFSC (presidente);
Prof. Maurício Edgar Stivanello, Dr. – IFSC; Prof. Milton Pereira, Dr. – EMC/UFSC; Prof. Jomi Fred Hübner, Dr. – DAS/UFSC. |
Título | A Convolutional Neural Network Approach on Bead Geometry Estimation for a Laser Cladding System |
Abstract: Laser cladding is a complex manufacturing process which requires fine-tuning to achieve the desired geometry. In order to further understand the process, an automated method for clad bead final geometry estimation on a laser cladding system is proposed. To do so, convolutional neural network architectures were developed. They receive the camera image and process parameters as inputs, yielding width and height of the clad beads as outputs. The optical monitoring system’s hardware was updated as well. The results of the network’s performances show coefficients of determination between the target and the estimated values above 0.95 for each frame on the best cases, while the error means among all clad beads get to as little as 5 μm. Those results take the laboratory one step further into closed loop control for this process. |