A neural network system for the prediction of process parameters in pressure die casting

In this work an artificial intelligent neural network system is developed to generate the process parameters for the pressure die casting process. The scope of this work includes analysing a physical model of the pressure die casting filling stage based on the governing equations of die cavity filling, and the collection of feasible casting data for the training of the network through the use of simulation package MELTFLOW and also from experts in the die casting industry. The multi-layer feed-forward network is trained with data collected directly from the industry using MATLAB application tool box. In this work the neural network is developed using three different training algorithms; namely the error back-propagation algorithm, the momentum and adaptive learning algorithm, and the Levenberg–Mrquardt approximation algorithm. It is found that the Levenberg–Mrquardt approximation algorithm is the preferred method for this application, as it reduces the sum-squared error to a small value. The accuracy of the network developed is tested by comparing the data generated from the network with that from an expert from a local die casting industry. It has been realised that with the use of this system the selection of process parameters becomes much simpler to even a novice user without prior knowledge of die casting process and optimisation techniques.