There is an ever-increasing demand for manufacturing industry to become more robust, flexible and responsive and to be more competitive through greater efficiency. At the same time design problems are becoming more complex. In order to succeed under such circumstances, problems have to be solved within a reasonable time with or without exact results.
Traditional methods often employed to solve complex optimization problems in the design and manufacturing industry tend to preclude elaborate exploration of the search space, often resulting in suboptimal solutions. Advances in evolutionary computation (EC) are generating considerable interest for solving these problems. EC is proving robust in delivering global optimal solutions and helps to resolve limitations encountered in traditional methods. EC harnesses the power of natural selection to turn computers into optimization tools. Most industrial processes are large scale, high dimensional, non-linear and uncertain. While many issues have been addressed in recent research efforts, limitations to the applications of these techniques to the fields of design and manufacturing still exist, restricting realistic solutions from being achieved. Current challenges include complexity of the search space, the nature of constraints, multiple objectives, and search within integrated qualitative/quantitative space.
We have invited people from academia and industry to submit papers on their recent research experience in the application of evolutionary computation to the fields of design and manufacturing. We received over 35 submissions. After a rigorous review, 14 papers have been accepted. Each tackles the application of particular novel evolutionary approaches to manufacturing and design problems. The initial idea was to bring forward all new approaches of EC with applications in design and manufacturing. In fact, the submissions which came through were mainly about either a novel approach applied to a well known manufacturing problem or an established method exploited for solving real manufacturing and design problems.
We have selected high quality papers dealing with differential evolution, particle swarm optimization, ant colony optimization, and the other evolutionary algorithms extended and empowered with statistical and experience-based information techniques. The application of these methods has been widely discussed and examined for various design and manufacturing problems. Allahverdi and Al-Anzi have proposed an evolutionary method for a rather theoretical problem with any scheduling technique, called ‘dominance relations of schedules’. In the second paper, the same authors have used particle swarm optimization and taboo search for a two-stage scheduling problem with setup time inclusive makespan. Another paper tackles abstract problems with novel methods is authored by Tasgetiren et al. They have applied a number of hybrid algorithms; each combines variable neighbourhood searches with particle swarm optimization and differential evolution methods—the problem tackled is a single machine scheduling problem with tardiness measure. Likewise, Pitakaso et al. have tackled a planning problem, namely the capacitated lot sizing problem, with an ant colony optimization algorithm. They used test problems to verify their methods. Also, Yigit et al. have solved larger size location problems with evolutionary meta-heuristics, where the problem can be counted as one of the main abstract problem types to be extended to real applications in logistics. Yu and Ram have also done exclusive research on dynamic job shop scheduling in an environment where routs are flexible and setups are dependent on the sequences. They have used multi agent approaches with biological inspiration to examine such highly interesting problems.
Kwong et al. have used a novel genetic algorithm based on analysis of variance for training artificial neural networks so as to simulate a real manufacturing process of the electronic industry, while Su and Yang have applied two phased metaheuristics to control the yield of a real manufacturing process. Dimopulous has suggested a manufacturing cell design optimized from a multi-objective point of view by using a recently developed genetic programming method. Song et al. have proposed a discrete evolution strategy for scheduling production of complex make-to-order products where each consists of multiple levels of product structure. Karen et al. have developed a Taguchi method based genetic algorithm for optimizing design of real automotive parts. Prestwich et al. have studied a local search algorithm for real template design for the printing industry. Yousef and El-Maraghy have studied and discussed modelling and optimization of multi level reconfigurable manufacturing systems with genetic algorithms. Finally, Afzal and Meeran developed a neural network based feature recognition model to handle orthographic images, where they trained their neural net with a particular search algorithm.
As can be seen, our aims are completely successful with respect to tackling problems in reality and the advances of evolutionary algorithms, which are pretty mature and proven robust. Therefore, they can undoubtedly be applied to design and manufacturing problems.
We would like to acknowledge the valuable and immeasurable contributions of the reviewers. We would also like to take this opportunity to express our gratitude to Mr John Middle who supported us during every stage of this editorial work.
Mehmet E. Aydin
University of Bedfordshire Department of Computing and Information Systems Luton, UK
Mustafa Ozbayrak
Bahcesehir University Department of Industrial Engineering Istanbul, Turkey
Terence C. Fogarty
London South Bank University Faculty of Business Computing and Information Management London, UK
Guest Editors