Artificial neural network modelling and optimization of elastic and an-elastic spring back in polymer parts produced through ISF

Traditional polymer processing techniques results in thermal degradation of the desired mechanical properties of a component. Incremental sheet forming (ISF), an innovative flexible cold forming process, can avoid this issue. However, this type of cold processing suffers from serious drawback of spring back due to elastic recovery. Therefore, process control is mandatory to achieve the desired part accuracy. This requires employment of a conducive set of process parameters, thereby turning this into an optimization problem. Using statistical techniques, though beneficial, does not result in an acceptable mean squared error (MSE). In this paper, we explore the use of machine learning methods for error prediction and parameter optimization. Particularly, we propose to use the combination of genetic algorithm and artificial neural network hybrid model for optimized set of parameters that result in a greedy minimized shape error. Experiments on Polypropylene sheet show that the proposed approach results in better optimization compared to statistical technique in minimizing the MSE.