Development of two-phase neural network-genetic algorithm hybrid model in modeling damage evolution in roll forming of aluminum sheet

• Published in: International journal of material forming Vol. 6; no. 4; pp. 423 - 436
• Main Authors: Park, Hong-Seok, Anh, Tran-Viet
• Format: Journal Article
• Language: English
• Published: Paris Springer Paris 01.12.2013
• Subjects: CAE) and Design; Computational Intelligence; Computer-Aided Engineering (CAD; Engineering; Machines; Manufacturing; Materials Science; Mechanical Engineering; Original Research; Processes; Roll forming; ANN-GA hybrid model; Damage evolution
• ISSN: 1960-6206; 1960-6214
• DOI: 10.1007/s12289-012-1096-5

In this paper a two-phase artificial neural network-genetic algorithm (ANN-GA) hybrid model has been developed for the modeling and prediction of the damage evolution in the roll forming (RF) process of aluminum sheet metal, as a function of process parameters. The multilayer perceptron is used to build the network while the genetic algorithm (GA) is employed to optimize the network structure in the modeling phase. In detail, the number of hidden layer, hidden neurons, weights and biases of the network are optimized by GA to minimize the error between predicted values and actual results. After the modeling phase the optimization of parameters is carried out in the optimization phase to minimize the damage in the aluminum sheet during the forming process. In this work the experimental data used for training and verifying the network is obtained automatically by the integration between CAD-CAE. As a result, the predicted results are validated with the actual values and a good agreement is observed. Moreover, the parametric study also is performed to find the relative influences of process parameters on the damage evolution. It is proven that the hybrid model is the powerful tool for modeling and predicting such a highly nonlinear problem as the damage evolution in RF process. The developed two-phase ANN – GA hybrid model is a new approach that can bring benefits to the forming industry by predicting and preventing the failure at the design stage, as well as improving efficiently the product's quality by optimizing the process parameters.