蛙跳算法的改进及其应用研究

发布时间：2018-03-23 07:04

本文选题：蛙跳算法　切入点：支持向量回归机　出处：《新疆大学》2017年硕士论文　论文类型：学位论文

【摘要】：蛙跳算法(Shuffled Frog Leaping Algorithm,SFLA)自提出以来,引起学者的广泛关注,并在部分工程领域得到了成功的应用。SFLA算法在解决高维问题时收敛速度较为缓慢且容易陷入局部最优,为了改善SFLA算法的搜索性能,本文通过对其内部寻优结构分析,提出了几种改进的SFLA算法,并应用于氧化还原电位(Oxidation Reduction Potential,ORP)预测以及作业车间调度优化问题中。本文的主要研究工作如下:1)为了提高SFLA在求解高维问题时的收敛速度以及避免陷入局部最优,提出了基于局部寻优策略改进的ISFLA(Improved Shuffled Frog Leaping Algorithm)和与教学算法(Teaching-Learning-Based Optimization Algorithm)相结合的TLBO-SFLA两种改进算法。在ISFLA算法的寻优过程中,分别引入混沌序列对种群初始化,扩大了初始种群的搜索范围;引入粒子群的局部更新策略,加强了种群内部的信息交流;引入反向学习,增加了算法搜索后期解的多样性,降低了陷入局部最优的概率。在TLBO-SFLA算法中,则是将每个子种群看作一个班级来进行学习,最优个体“老师”通过“教”的方式提升班级的整体水平,学生之间的相互学习过程可以更好的实现差异化学习。选取测试函数对所提两种改进算法进行测试,实验结果表明所提改进算法收敛速度更快寻优精度更高。2)ORP作为细菌活性的重要评价指标,对ORP的精准预测有利于实现对氧化提金过程关键参数的及时调控。为了实现ORP的预测,建立了支持向量回归机模型对ORP进行预测,并选用改进的蛙跳算法对预测模型的关键参数进行优化,以达到较高的预测精度。选取新疆某金矿的实际生产数据建立预测模型,结果表明,基于改进蛙跳算法优化的支持向量回归机ORP预测精度更高。3)提出了可用于求解作业车间调度优化的基于工序编码的OSFLA(Optimized Shuffled Frog Leaping Algorithm)算法,选取标准的调度问题进行仿真测试,实验结果表明,OSFLA算法不仅可以求出最优解,而且搜索速度更快。
[Abstract]:Since it was put forward, the leapfrog Frog Leaping algorithm has attracted wide attention of scholars, and it has been successfully applied in some engineering fields. The convergence rate of .SFLA algorithm is slow and it is easy to fall into local optimum when solving the problem of high dimension. In order to improve the search performance of SFLA algorithm, this paper proposes several improved SFLA algorithms by analyzing its internal optimization structure. And it is applied to the oxidation-reduction potential Reduction potential orp prediction and job shop scheduling optimization problem. The main work of this paper is as follows: 1) in order to improve the convergence speed of SFLA in solving high dimensional problems and avoid falling into local optimum. Two improved algorithms, ISFLA(Improved Shuffled Frog Leaping algorithm based on local optimization strategy and TLBO-SFLA algorithm combined with Teaching-Learning-Based Optimization algorithm, are proposed. In the process of ISFLA optimization, chaotic sequences are introduced to initialize the population. The search range of initial population is enlarged, the local updating strategy of particle swarm is introduced, and the information exchange within population is strengthened. The diversity of the late solution of search algorithm is increased by introducing reverse learning. The probability of falling into local optimum is reduced. In the TLBO-SFLA algorithm, each subgroup is treated as a class to learn, and the optimal individual "teacher" improves the overall level of the class by "teaching". The process of mutual learning between students can better achieve differential learning. Select the test function to test the proposed two improved algorithms. The experimental results show that the improved algorithm has faster convergence speed and higher precision. 2ORP is an important evaluation index of bacterial activity. The accurate prediction of ORP is helpful to realize the timely control of the key parameters in the process of oxidizing gold extraction. In order to realize the prediction of ORP, The support vector regression model is established to predict ORP, and the improved leapfrog algorithm is used to optimize the key parameters of the prediction model to achieve a higher prediction accuracy. The actual production data of a gold mine in Xinjiang are selected to establish the prediction model. The results show that the ORP prediction accuracy of SVM based on improved leapfrog algorithm is higher. 3) A OSFLA(Optimized Shuffled Frog Leaping algorithm which can be used to solve job shop scheduling optimization is proposed. The standard scheduling problem is selected for simulation test. The experimental results show that the OSFLA algorithm can not only find the optimal solution, but also search faster.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【参考文献】