Multi-objective Optimization of Yarn Characteristics Using Evolutionary Algorithms: A Comparative Study

摘要

In cotton spinning industries, attainment of the most desired yarn characteristics mainly depends on different parameters of the ring or rotor spinning process. Thus, it is often required to determine the optimal parametric settings of a spinning process with the help of some optimization tools. In this paper, two multi-response optimization problems are considered and subsequently solved using four popular evolutionary algorithms, i.e. artificial bee colony algorithm, ant colony optimization algorithm, particle swarm optimization algorithm and non-dominated sorting genetic algorithm-II for searching out the global optimal settings of ring and rotor spinning processes. As the process parameters’ settings derived using single response optimization solutions are often impractical to maintain, it is always recommended to set them based on the results of multi-response optimization techniques. It is observed that among these four algorithms, particle swarm optimization excels over the others with respect to the derived optimal solution, consistency of the solution and convergence speed. The developed scatter diagrams also help in investigating the effects of changing values of different process parameters on various yarn qualities.