Optimising the Fibre-to-Yarn Production Process: Finding a Blend of Fibre Qualities to Create an Optimal Price/Quality Yarn.

An important aspect of the fibre-to-yarn production process is the quality and price of the resulting yarn. The yarn should have optimal product characteristics, while maintaining as low a price as possible. Early optimisation models of the fibre-to-yarn process, based on neural networks and genetic algorithms, were severely limited in their potential applications as they generated unrealistic (ideal) conditions for the process. In this paper, a method is presented to model and optimise the fibre-to-yarn production process which avoids the aforementioned problems. A neural network is used to model the process, with the machine settings and fibre quality parameters as input and yarn tenacity and elongation as output. A constrained optimisation algorithm is used afterwards to optimise the blend of fibre qualities to obtain the best yarns. This results in an optimal price-yarn quality surface where each point corresponds with a set of blend coefficients and machine settings. Furthermore, constraints can easily be adjusted to correspond to real-life production environments.