Minimum additional material volume prediction for preform product by using abductive network and Taguchi method

Purpose: To predict the minimum value of additional material volume for an acceptable preform product. To predict an acceptable preform product without shape defect such as unfilling in a closed-die forging operation. Design/methodology/approach: In order to reduce the number of experiments, an orthogonal array from the Taguchi’s experimental method will be utilized to design the process parameter combinations for database sets to promote the prediction precision. Also, in order to reduce the number of experiments to get the minimum additional material volume of preform, the abductive network is applied to synthesize the data sets obtained from the numerical simulation. Findings: The minimum additional material volume can be determined as 7.6% for an acceptable preform product in conjunction with the billet settle position, E, of 11.8 mm and the aspect ratio of width to height, B/H, of 1.4. Research limitations/implications: The Taguchi method can be used to narrow the ranges of process parameters for database sets which can promote the precision of abductive network to search for the the minimum additional material volume for an acceptable preform product. The abductive network is applied to synthesize the data sets obtained from the numerical simulation of the reduced ranges of the process parameters. Practical implications: The combination of the abductive network and Taguchi method can be used as a reference and guidance for the development of searching the minimum or maximum value of one of the process parameters, accompanying by the other suitable parameters. Originality/value: An assessment model of the closed-die forging process is developed using a neural network system and Taguchi method. Based on the developed neural network, the additional material volume of preform product, one of the forging process parameters can be minimum accompanying by the other suitable process parameters to get an acceptable product.