ABSTRACT
Yarn engineering has attained an accurate model predicting yarn and fibre properties status with the development of ANN. The studies in literature have shown clearly that the relationship prevailing between yarn parameters and fiber properties is non-linear and complex. The usual statistical technique still not capable of deciding the exact fiber properties required for a particular yarn property as well as how these properties affect the yarn properties. The advent of high-speed fiber-testing machines and development of powerful modelling tools such as artificial neural network (ANN) have provided a great stimulus in the yarn engineering research. The feasibility of yarn engineering can be verified by developing a yarn-to-fiber or reverse engineering model, using ANN. This concept is entirely different from the prevailing forward models, as it predicts exactly in reverse manner, viz; the properties of final yarn, by using the fiber properties as inputs. This model could handle numerous dependent and independent variables. Characteristics like generalisation, vigour against changes, etc., have made it a superior choice for modelling than others. A number of attempts are made by various researchers to engineer yarn quality by utilising back-propagation algorithm of artificial neural network. Microprocessor base classical linear programing approach was made use of in combination with ANN ensuring cost minimisation of optimised quality cotton fiber mix.
