DCMOGADES: Distributed cooperation model of multi-objective genetic algorithm with distributed scheme. (English) Zbl 1176.90553

Abraham, Ajith (ed.) et al., Computational intelligence and applications. ISDA 2002, 2nd international workshop on intelligent systems design and applications, Atlanta, GA, USA, August 7–8, 2002. Atlanta, GA: Dynamic Publishers (ISBN 0-9640398-0-X). 155-160 (2002).
To find design variables that minimize or maximize values of objective functions is called an optimization problem. Optimization problems are often found in real world problems, such as structural design problems, job shop scheduling problems, adjustment problems of control, prediction of protein tertiary structure problems and so on. In many real world problems, we often found that there is not only one objective but also many objectives. These optimization problems are called multi-objective optimization problems. Usually, since there is a trade-off relation ship among the objectives, it is difficult to minimize/maximize all values of the objectives. Therefore, the solutions that are dominant to the other solutions are important and these solutions are called Pareto optimal solutions. To find Pareto-optimum solutions is one of the goals of multi-objective optimization problems.
Genetic Algorithm (GA) that simulates species’ heredity and evolution is one of optimization algorithms. GA is a multi-point searching method and a stochastic searching method. Usually, when the derivatives of the problems cannot be derived or the problems are discrete, a conventional gradient method is very difficult to apply to the problems. On the other hand, GA is very easy to apply to several types of problems. At the same time, GA might find Pareto optimum solutions at one trial, since GA is a multi-point searching method. Therefore, GA is very powerful tool to find Pareto-optimum solutions. In this few years, several new algorithms that can find good Pareto-optimum solutions with small calculation cost have been developed. Those are roughly divided into two categories; those are the methods using Pareto explicitly and implicitly. NSGAII and SPEA2 are the typical algorithm of the method using Pareto explicitly. MOGADES is in the category of the method using Pareto implicitly. Good Pareto-optimum solutions should have the following characteristics; Solutions should be close to the real Pareto front, solutions should not be concentrated but widespread and solutions should have the optimum solutions of every single objective function. In this study, to derive the good Pareto optimum solutions, two algorithms are combined and illustrated. The proposed algorithm is called Distributed Cooperation model of Multi-Objective Genetic Algorithm with Environmental Scheme (DCMOGADES). DCMOAGADES is applied to some test functions and the effectiveness of the proposed algorithm is illustrated.
For the entire collection see [Zbl 0997.00023].


90C29 Multi-objective and goal programming
90C59 Approximation methods and heuristics in mathematical programming
68T05 Learning and adaptive systems in artificial intelligence