Mold design is one of the most important activities in the injection molding process. It is a complex task which affects several downstream processes including mold construction, quality of part produced, and cost of mold manufacture. Various factors such as part dimensions, number of undercuts, parting line, cavity detail, tolerances, and number of cavities per mold have been found influencing the complexity of a mold design. This paper demonstrates the application of a backpropagation neural network, running on a parallel computer, to evalaute the complexity level of a mold. The outputs from the network are classified into three levels: easy, moderate, and difficult. Ten part samples have been used to determine the ability of the network in classifying the levels of mold complexity.
展开▼