Improvement of Parallal-Distributed Genetic-Algorithm VLSI55

遗传算法是一种受生命进化启发的概率搜索和优化技术。它是通用的，适用于各种各样的问题，因此一直受到人们的关注。为了加速遗传算法的进程，有两种最有前途的技术：并行化和VLSI实现。我们在国内率先设计了GA-VLSI，并率先开发了GA-VLSI的并行和分布式设计。本研究通过引入集群间迁移的动态适应技术，进一步改进了这种并行分布式GA-VLSI。在过去的工作站网络软件实现研究中，该技术被证明是有效的。这是世界上第一次尝试。该研究涉及硬件设计、并行和分布式计算、动态优化和问题求解等多种背景，因此需要各个领域的最新技术。我们研究了每个领域，如硬件设计方法和动态分布式计算，以便将它们集成到我们的研究中。然后，完成了系统设计、电路设计、硬件仿真与评估、逻辑综合。我们从中获得了很好的结果。

Genetic Algorithms (GA) is a probabilistic technique for search and optimization which was inspired from life evolution. It is general purpose, and applicable to wide variety of problems, so it has been collecting attentions. To accelerate GA process, two techniques are the most promising : parallelization and VLSI implementation. We have designed GA-VLSI in a domestic pioneering research, and first developed parallel and distributed design of GA-VLSI.This research is to improve this parallel and distributed GA-VLSI further by introducing a dynamic adaptation technique for inter-cluster migration. This technique was proved to effective in our past research in software implementation on top of workstation network. This is the first attempt in the world.The research has a diverse background of hardware design, parallel and distributed computation, dynamic optimization and problem solving, so we needed the newest technique in each area. We investigated each area, such as hardware design methodologies and dynamic distributed computation, so as to integrate them into our research. Then, we achieved system design, circuit design, hardware simulation and evaluation, and logic synthesis. We obtained a good results out of them.