Thinking sewing machines for intelligent garment manufacture

Quantitative fabric‐needle‐sewing machine interactions at different speeds have been used to construct qualitative rules mapping fabric properties to optimum sewing machine settings for the next generation of “intelligent sewing machines”, using model free estimation. The inference procedures of fuzzy logic have been implemented in a neural network to allow for optimization of output membership functions and subsequently, self‐learning. The technique is successfully applied to industrial lockstitch and overlock sewing machines. Optimum settings were achieved under static and dynamic machine conditions from the properties of difficult fabrics and compensation for mishandling by the operator over the speed range of the sewing machine.