Soft computing technologies in quality control with applications to injection molding

An intelligent system, designed to determine parameter settings and minimize out-of-control situations in injection molding process, is presented. This intelligent system is intended to run through the injection molding process as well as during other manufacturing processes with similar characteristics as the aforementioned process.; A design of experiments is conducted to both determine the significant parameters that have a direct effect on the quality of a molded part and to provide information to collect sufficient data sets to be used in training and testing the intelligent system, which consists of a set of parallel Adaptive-Network-based Fuzzy Inference Systems (ANFIS). The inputs of the ANFIS structures are either the discrepancies between dimensional characteristics of molded parts and measured values regarding dimensional characteristics obtained from the design of experiments or the seriousness of appearance problems in molded parts. The output of each ANFIS structure is the magnitude of changes in positive or negative direction from the parameter settings to produce a part that meet required specifications (good part).; The intelligent system also consists of a series of feed forward neural networks. The inputs of the neural networks are appearance or dimensional characteristics of molded parts. The outputs of each neural network are the magnitude of changes in positive or negative direction from the parameter settings to produce a good part.; An application at an injection molding facility is presented that demonstrates the use of the designed intelligent system and its ability to generate optimal process parameters settings for the injection molding process. The neural networks and ANFIS structures run on-line satisfactorily providing the operator which parameter(s) to change if appearance or dimensional problems are present.