Pasar al contenido principal

DISEÑO Y CALIBRACION DE UN SISTEMA MODULAR DE INSPECCION INTELIGENTE PARA VERIFICACION Y CALIDAD EN LINEA 4.0

PID2021-125530OB-I00

•
Nombre agencia financiadora Agencia Estatal de Investigación
Acrónimo agencia financiadora AEI
Programa Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprograma Subprograma Estatal de Generación de Conocimiento
Convocatoria Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Año convocatoria 2021
Plan Estatal 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Centro beneficiario UNIVERSIDAD DE ZARAGOZA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Surface roughness prediction in turning processes for grey cast iron: A hybrid machine learning approach integrating infrared thermography

  • Aguado, Sergio
  • Pueo, Marcos
  • Acero, Raquel
  • Majarena, Ana Cristina
  • Santolaria, Jorge
Workpiece surface quality is a critical control parameter in machining processes, influencing functional performance, dimensional precision, and wear resistance. However, accurately predicting surface roughness is complex, often limited by the computational demands of traditional high-precision methods and the reliance of existing models solely on cutting parameters, hindering real-time monitoring. This research introduces a hybrid Artificial Neural Network (ANN) methodology specifically developed for predicting surface roughness of grey cast iron GG-25 workpieces machined in turning processes, a material previously unstudied in this context. The methodology integrates real-time infrared thermal measurements from multiple defined regions of interest (ROIs) within the tool-workpiece contact zone, along with cutting parameters. Experimental results demonstrated that feed rate (f) is the most significant cutting parameter (effect = 0.43) affecting surface quality, followed by its combination with cutting speed (Vc) (effect = −0.25) and cutting speed (effect = 0.18). Correlation and non-linear regression analyses revealed complex, often exponential relationships between temperature and surface roughness, showing temperature an upward trend as machining progressed. The developed ANN achieves a correlation coefficient (R) value of 0.99 both when predicting the roughness arithmetic mean deviation (Ra) parameter in training conditions and when using data from experiments not used in training (validations data). Moreover, the model reaches a correlation coefficient value of 0.85 (test data) under cutting conditions different from those used in experiments, demonstrating robustness, significantly outperforming Support Vector Regression (SVR). This model represents a highly potential tool for real-time online inspection.
Zaguán. Repositorio Digital de la Universidad de Zaragoza
Publicación Artículo científico (JournalArticle)

Thermal imaging–assisted neural network modeling for accurate tool wear prediction in dry turning of grey cast iron

  • Pueo, Marcos
  • Acero, Raquel
  • Aguado, Sergio
  • Majarena, Ana Cristina
  • Santolaria, Jorge
Accurate tool wear monitoring is crucial for machining efficiency and product quality, yet traditional statistical methods often fail to capture complex non-linear interactions. This study addresses these limitations by employing Artificial Intelligence (AI), specifically a Multi-Layer Perceptron (MLP) artificial neural network, for tool wear prediction. Experiments involved dry turning of grey cast iron under several cutting parameters, acquiring machining real-time temperature data with an infrared thermal camera and measuring tool flank wear (VB). Statistical analyses in this study revealed that, although correlations between tool wear and temperature exist, regression models exhibited high variability in predictive performance, with coefficients of determination (R2) ranging from 11.6% to over 95%. This underscores the inability of conventional approaches to fully model the complex relationships. The developed network, with hyperparameters optimized through Design of Experiments tool, however, demonstrated high predictive capabilities. For fixed cutting conditions, it achieved a correlation coefficient (R) of 0.999 and Mean Square Error (MSE) of 5.55E-05. For generalized prediction across a wider range of conditions, the optimal network configuration yielded an R of 0.998 and an MSE of 8.62E-05. To ensure consistency of the modeĺs performance and prevent over-parameterization, k-fold cross-validation and early stopping criterion were used. This work confirms infrared thermography’s reliability and justifies the use of neural networks to overcome the shortcomings of traditional statistical analyses. The AI-driven approach effectively models intricate variable relationships, offering a robust and precise solution for real-time tool wear monitoring and optimization in industrial machining.