群智能算法及在确定模糊测度中的应用研讨.pdf

摘 要 摘 要 受到自然界动物群体行为的启发，出现了模拟动物群体自治行为的群智能算法。群 智能算法是一类算法的统称，具有代表性的算法有：粒子群算法、蚁群算法和人工鱼群 算法。群智能算法出现以来，很多学者开始关注和研究。该类算法被不断地应用于特定 的问题中，并取得了良好的效果。确定模糊测度问题是应用模糊测度和模糊积分的前提， 由于该问题可以转化为优化问题，所以可以采用软计算技术来解决。在本文中，我们针 对确定模糊测度这个具体问题，应用群智能算法来确定模糊测度，验证了群智能算法的 思想可以引入到这个具体问题中，并验证了群智能算法的优势：实现容易，具有很强的 健壮性等。另外，通过算法原理上和实验上的分析和比较，我们得出了粒子群算法、蚁 群算法和人工鱼群算法性能差异：粒子群算法具有较快的收敛能力；蚁群算法能够得到 更精确的解；人工鱼群算法兼具二者特性。 关键词 模糊测度 PSO 算法 ACO 算法 AFSA 算法 I Abstract Abstract Inspired by natural animal group behavior, swarm intelligent algorithms used to simulate animal behavior are presented. Swarm intelligent algorithms are referred to a class of algorithms. The representative algorithms are called particle swarm algorithm, ant colony algorithm and artificial fish warm algorithm. Since the emergence of swarm intelligence algorithms, many scholars have began to pay attention and study. These algorithms are applied to particular problems and achieve good results. How to determine fuzzy measure is a basic problem when applying fuzzy measure and fuzzy integral. This problem can be translated to an optimization, so it can be solved by soft computing techniques. In this paper, we use swarm intelligent algorithm to identify fuzzy measures from data. We verify the feasibility and the advantages: easy implement, good robustness, etc. what is more, through the analysis and comparison of the algorithms on theory and experiment, we conclude the performance differences of particle swarm algorithm, ant colony algorithm and artificial fish warm algorithm: particle swarm algorithm owns fast convergence; ant col