In Textile Industry, Yarn is produced with help of cotton fibre on Spinning Machine. The quality of yarn is determined by cotton fibres used. Hence, it is important to predict the fibre properties for required ‘customer defined yarn’ with less cost. Various researches developed fibre to yarn models which determine yarn properties from fibre specifications. If the produced yarn quality is not satisfactory, entire process needs to be re-executed with new set of raw materials. Therefore, prediction of fibre properties according to required yarn is must. This process is ‘Reverse Engineering’.
This paper studies predictability and profitability using innovative computational model. It is carried out by hybrid approach involving Artificial Neural Network (ANN), Genetic Algorithm (GA) and Linear Programming (LP) Methodology. ANN is developed to predict multi-property fibre from required yarn. GA searches optimal fibre properties from available fibre stock. ANN acts as fitness function to GA. GA is also in use to predict whether required yarn can be spunned using available fibre stock. Result of GA is provided to LP Model which is used to decide proportionality of cotton bled and cost optimization. The result shows that the accuracy of proposed integrated approach is higher than Industrial Standards.
Back-propogation algorithm, genetic algorithm, optimization
