Genetic Algorithms Encoding Study and a Sufficient Convergence Condition of GAs
Meng, Q.C.,Feng, T.J.,Chen, Z.,Zhou, C.J.
DOI: https://doi.org/10.1109/icsmc.1999.814168
1999-01-01
Abstract:This paper studies the encoding techniques of genetic algorithms and a sufficient convergence condition on genetic encoding in genetic algorithms is presented. Some new categories of genetic codes are defined, such as uniform code, bias code, tri-sector code and symmetric codes, etc, and they are applied in some problem optimizations and a robotic problem solution. These codes have found their application in developing some special and powerful genetic algorithms. For example, based on symmetric code theory, new genetic strategy, GASC: Genetic Algorithm with Symmetric Code, is developed. In the paper, some key definitions on encoding are given out, such as living-block, dead-block, link, fix-link, living-population, as well as some operations on genetic, such as bit-transposition, bit-and, population-and, member-and, etc. Some of our research shows that genetic encoding techniques have a very important influence on the performance of genetic algorithms. The convergence speed of genetic algorithms with some specially developed codes will be much faster than conventional genetic algorithms. That is very significant for finding more applications of genetic algorithms, as, in many cases, genetic algorithm applications are limited by their convergence speed