基于膜系统的粒子群优化算法在产业集群演化中的研究与应用

发布时间：2018-01-15 22:25

本文关键词：基于膜系统的粒子群优化算法在产业集群演化中的研究与应用　出处：《山东师范大学》2014年硕士论文　论文类型：学位论文

【摘要】：产业集群，指的是某一类相关的企业大量聚集在某一特定区域的经济现象。对于集群内的企业来讲，通过企业的聚集获得了巨大的竞争优势，从而取得更好的发展与丰厚的经济效益。产业集群是一种基于自组织结构的经济现象。自组织结构的特征是自我适应和自我组织，这一点和产业集群不谋而合。同样，若将集群看作一个由众多企业和机构构成的系统，，那它也是一种自组织系统。产业集群形成过程中也是经由开放的耗散结构不断演化而来的。微粒群的寻优与产业集群的集聚具有相通性。产业集群的形成实际上是一个自组织的过程，粒子群优化算法是自组织算法，其寻优过程也是自组织的。若将集群中的企业视为粒子群优化算法中的粒子，集群所处的位置正是集群竞争力最大的位置，将其看作粒子群优化算法中的最优解的位置，那么产业集群的聚集过程可以视为粒子群的寻优过程。可见，微粒群的寻优与产业集群的集聚是相通的。基于PSO算法局限性的思考，并受到P系统的启发，本文提出了一种基于P系统的粒子群优化算法（P-PSO）。在本文中P系统中的膜有主膜和辅助膜之分，设立主膜一个，若干辅助膜。粒子被放入P系统之后，主膜内粒子与辅助膜内粒子进行合理分工，主膜内粒子负责“开发”（即在辅助膜内粒子的引导下，搜寻最优解的具体位置），“探索”任务是由辅助膜内的粒子来完成（即尽可能的遍历搜索空间，搜寻可能存在最优解的区域，为主膜内粒子的搜索提供引导）。其中，主膜内粒子与辅助膜粒子之间的信息交流由P系统中的交流规则来实现。为了达到探索与开发的目的，辅助膜内粒子需要保持较高的粒子活性，主膜内粒子要有精细化搜索的能力。对于新算法，我们借助常用的测试函数进行了检测，结果表明P-PSO算法具有很好性能。为了用粒子群优化算法来模拟产业集群的形成问题，我们将产业集群微粒群化。产业集群的竞争力值为粒子群优化算法中所求解的目标函数的值；产业集群的地理坐标为PSO算法中粒子搜索空间中的位置；集群内部企业之间肯定有“合作”与“竞争”，这可以通过PSO算法中“自我认知”部分和“社会”部分来实现。最后我们以山东汽车产业集群为例，运用P-PSO算法模拟集群内企业的聚集过程，通过实证分析，对汽车产业集群的发展进行了预测。
[Abstract]:Industrial cluster refers to the economic phenomenon that a certain kind of related enterprises gather in a certain specific area in large quantities. For the enterprises in the cluster, they obtain a huge competitive advantage through the agglomeration of enterprises. In order to achieve better development and rich economic benefits. Industrial cluster is an economic phenomenon based on self-organization structure. Self-organization structure is characterized by self-adaptation and self-organization, which coincides with industrial cluster. If a cluster is regarded as a system composed of many enterprises and institutions, it is also a self-organizing system. The formation of industrial cluster is actually a process of self-organization, particle swarm optimization algorithm is self-organization algorithm. If the enterprises in the cluster are regarded as particles in the particle swarm optimization algorithm, the location of the cluster is the most competitive position. Considering it as the location of the optimal solution in PSO, the aggregation process of industrial cluster can be regarded as the optimization process of PSO, which shows that the optimization of PSO is related to the agglomeration of industrial cluster. Based on the limitation of PSO algorithm, and inspired by P system. In this paper, a particle swarm optimization algorithm based on P system is proposed. In this paper, the membrane of P system is divided into main membrane and auxiliary membrane, and a main membrane is set up. After the particles were put into the P system, the particles in the main film and the particles in the auxiliary film were divided reasonably, and the particles in the main film were responsible for the "development" (that is, under the guidance of the particles in the auxiliary film). Searching for the specific location of the optimal solution, the "exploration" task is accomplished by the particles in the auxiliary film (that is, traversing the search space as much as possible, searching for the region where the optimal solution may exist. In order to achieve the purpose of exploration and development, the information exchange between the main film particles and the auxiliary membrane particles is realized by the communication rules in P system. The particle in the auxiliary film needs to keep high particle activity, and the particle in the main film should have the ability of fine searching. For the new algorithm, we use the commonly used test function to detect the new algorithm. The results show that the P-PSO algorithm has good performance. In order to simulate the formation of industrial clusters with particle swarm optimization (PSO) algorithm, we transform the PSO into industrial clusters. The competitiveness of industrial clusters is the value of the objective function solved by PSO. The geographical coordinate of industrial cluster is the position of particle search space in PSO algorithm. There must be "cooperation" and "competition" among enterprises in the cluster, which can be realized through the part of "self-cognition" and "society" in PSO algorithm. Finally, taking Shandong automobile industry cluster as an example, we use P-PSO algorithm to simulate the clustering process of enterprises in the cluster, and predict the development of automobile industry cluster through empirical analysis.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP18

【参考文献】