The current research to find near-optimum solution(s) explores in a small population, which is coined as Micro-Genetic Algorithms (μGAs), with some genetic operators. Just as in the Simple-Genetic Algorithms (SGAs), the μGAs work with encoding population and are implemented serially. The major difference between SGAs and μGAs is how to make reproductive plan for more better searching strategy due to the population choice. This paper is conducted to implement hybrid μGAs in order to achieve fast searching for more better evolution and associated cost evaluation in global solution space. To achieve this implementation, the Air-Borne Selection (ABS) for a new reproductive plan is developed as a new strategic conception for hybrid μGAs. It is shown that the general μGAs implementation reaches a near-optimal region much earlier than the SGAs implementation. The superior performance of the general μGAs is demonstrated with two kinds of hard combinatorial optimisation problems, which are Travelling Salesman Problem (TSP) and cutting path planning in nesting. And then, the superior performance of the hybrid μGAs is demonstrated for two types of nesting problems.
Simple Genetic Algorithms (SGAs) have been appeared to hold a lot of promise as general purpose adaptive search procedures and researched as an useful tools for many function optimisation problems. SGAs work with a serially implemented encoding population of size N (normally, from 50 to 200) in a generation to generation evolution based on reproductive plan, which consists of selection, crossover and mutation. The selection is conducted depending on the fitness of the individual strings compared to the population fitness average. The most popular selection strategies currently in use are the roulette wheel selection and stochastic remainder selection. Therefore, one of main drawback of SGAs is the computational time penalty involved in evaluating the cost functions for large populations, generation after generation.
