A New Genetic Algorithm to Solve Effectively Highly Multiobjective Problems: Poga

摘要: Since the pioneering work of Holland [14], Genetic Algorithms (GA) have become one of the most popular optimization techniques. The main reasons behind this success can be attributed to their effectiveness in exploring massive search spaces with little tendency to be deceptively attracted by local optima, and inherent capability of handling both real and discrete variables. Furthermore, they are relatively easy to code and are somewhat problem domain independent, even though prior knowledge and fine tuning are often required to achieve best performances. Broadly speaking, GA is a search procedure inspired by natural selection and genetics and it is based on the concept of “survival of the fittest”. Given an initial population of “individuals”, recombination and selection are repeatedly performed until a set of good enough solutions is found.The great potential of GA was indubitable and apparent to the scientific community, but it was only with Goldberg [13] that it became evident that they could be applied to a broad range of engineering problems. With Goldberg’s work, the transition between single-objective GA to multi-objective GA (MOGA) was first paved and a manifold of methodologies have been flourishing in the literature ever since.