期刊名称：Advances in Electrical and Computer Engineering
印刷版ISSN：1582-7445
电子版ISSN：1844-7600
出版年度：2015
卷号：15
期号：1
页码：23-32
DOI：10.4316/AECE.2015.01004
出版社：Universitatea "Stefan cel Mare" Suceava
摘要：Meta-heuristic techniques are powerful tools used to find an optimal solution for complex problems to which classical techniques find difficult to solve. The features among all the meta-heuristic techniques are the high amount of computational resources spent on their implementation and the computing effort generated on their execution. For this reason, many works have proposed their use on the base of software methodologies without achieving online or real-time performance. In the present work, two strategies that implement the Genetic Algorithms are presented by using the micro-population concept with the objective of reducing computational resources, increasing the heuristic search speed, and providing simplicity in its design. Both strategies are implemented in hardware architecture; the first, as a software strategy in a proprietary embedded processor, the second, as a hardware co-processor unit. In order to validate the proposed approaches, several tests to optimize a motion controller in a servo system are presented and compared with a classical tuning technique.
