进化计算的过去、现在与未来.pdf

“遗传算法/进化计算研究生学术论坛”论文集 进化计算的过去、现在与未来* 孙瑞祥 屈梁生 西安交通大学诊断与控制学研究所 摘 要 本文对“进化计算”的历史、现状与展望进行了系统地综述。进化计算是 20 世纪 90 年代兴起的一门模拟生物进化与遗传规律的计算学科。目前已发展成为与人工神经网络、 模糊逻辑相并列的智能计算支撑技术。进化计算最重要的应用是传统优化方法无法或难以解 决的科学与工程中的复杂优化问题。与传统优化方法相比，进化计算的优势在于全局优化性、 梯度信息不依赖性、简单易实施等。当前，进化计算主要由遗传算法、遗传编程、进化策略、 进化编程、DNA 计算、分子计算等不同的分支组成。文中简介了它们的基本原理、异同点和 适合应用的问题，以及目前存在的难题。对未来的重要研究方向也进行了展望。 关键词 进化计算 计算智能 遗传算法 遗传编程 进化策略 进化编程 Evolutionary Computation: Past, Present and Future SUN Rui-Xiang and QU Liang-Sheng Institute of Diagnostics and Cybernetics (IDC), Xi’an Jiaotong University Abstract This paper systematically reviews the history, the state-of-the-art, and the prospects of evolutionary computation(EC), a novel computing discipline rose in 1990’s. Roughly speaking, EC is motivated from the principles of biological evolution and genetics. Now, it has been developed as one of the underpinning technologies for computational intelligence (CI), being the partnerships with artificial neural networks(ANN) and fuzzy logic(FL). From the viewpoint of application, EC is mainly applied to optimize the complicated industrial problems, which can not be tackled well with the traditional optimization methods. EC possesses many advantages over the conventional optimization techniques such as global optimization, independence of gradient, ease of implementation, and so on. Contemporarily, EC consists of the following avenues: genetic algorithms, genetic programming, evolution strategies, evolutionary programming, DNA computing, and molecular computing, etc. The rationales, similarities and differences, the application problems, and the puzzles of EC