【作者】 张睿；

【导师】 刘冰； 吕会军；

【作者基本信息】 长春工业大学 ， 工程硕士（专业学位）， 2021， 硕士

【摘要】 群智能优化算法是现阶段用于解决最优解优化问题的重要方法之一,被广泛应用于各种现实生活中存在的实际问题中。正余弦优化算法(Sine Cosin Algorithm,SCA)是新颖的群智能算法,是基于正弦和余弦三角函数特性而开发的算法,主要用于解决全局优化问题。本文介绍了一种新颖的正弦余弦算法改进方法,可以提高解的利用能力,并减少经典SCA搜索方程中存在的多样性的溢出,所提出的算法被称为ISCA。该算法的关键特征是将交叉算子与单个解决方案的最佳状态相结合,并整合了自学习、全局搜索机制以及贪婪选择机制。为了评估ISCA算法的性能,在一组经典函数和多个工程问题以及多级阈值图像分割问题中对ISCA算法进行了测试。测试结果表明:(1)在经典23个经典函数测试结果中,本研究改进的ISCA算法在迭代过程中的探索和开发能力优于原有的SCA算法。新算法具有寻优精度高、需要计算量小、收敛速度快以及鲁棒性强的特点,可以很好的解决各经典函数问题;(2)在五个复杂非线性约束优化工程问题测试结果中。本研究改进的ISCA算法在五个经典工程问题中的求解表现均优于对比算法。证明了本研究改进的ISCA算法在求解复杂非线性约束优化问题的可行性。也为使用本研究改进的ISCA算法进行多目标优化奠定基础;(3)在多级阈值图像分割问题测试结果中。将改进的ISCA阈值分割方法在一组基准图像上进行了测试,并在结果部分设计了两组实验进行了比较。基于统计分析、收敛行为分析和性能指数分析的实验结果,ISCA算法的可以完全胜任研究中提到的多级阈值图像分割任务。证明了该算法在本研究问题中的有效性、准确性和鲁棒性。综上所述,在经典函数、工程问题和多级阈值分割的数值实验和分析表明,所提出的算法(ISCA)可以有效地解决现实生活中的优化问题。更多还原

【Abstract】 One of the most important methods for solving optimal optimization problems is the swarm intelligence algorithm,which is widely used in a variety of real-life problems.The Sine Cosin Algorithm(SCA)is a novel group intelligence algorithm,which is based on the properties of sine and cosine trigonometric functions and is mainly used to solve global optimisation problems.This paper presents a novel improvement to the Sine Cosine Algorithm that improves solution utilisation and reduces the overflow of diversity present in the classical SCA search equations,the proposed algorithm is referred to as ISCA.the key features of the algorithm are the combination of crossover operators with the best state of a single solution and the integration of self-learning,global search mechanisms as well as greedy selection mechanisms.To evaluate the performance of the ISCA algorithm,the ISCA algorithm was tested on a set of classical functions and several engineering problems as well as a multi-level threshold image segmentation problem.The test results show that(1)the improved ISCA algorithm of this study outperforms the original SCA algorithm in terms of exploration and exploitation ability during iterations in the classical 23 classical functions test results.The new algorithm has the characteristics of high accuracy in seeking,small amount of computation required,fast convergence and robustness,and can solve each classical function problem well;(2)In the test results of five complex nonlinear constrained optimization engineering problems.The improved ISCA algorithm outperforms the comparison algorithm in solving all five classical engineering problems.This demonstrates the feasibility of the improved ISCA algorithm in solving complex non-linear constrained optimization problems.It also lays the foundation for multi-objective optimization using the improved ISCA algorithm in this study;(3)In the test results of the multi-level threshold image segmentation problem.The improved ISCA thresholding method was tested on a set of benchmark images,and two sets of experiments were designed in the results section for comparison.Based on the experimental results of statistical analysis,convergence behaviour analysis and performance index analysis,the ISCA algorithm’s can fully perform the multi-level thresholding image segmentation task mentioned in the study.The effectiveness,accuracy and robustness of the algorithm in this research problem are demonstrated.In summary,numerical experiments and analysis on classical functions,engineering problems and multilevel threshold segmentation show that the proposed algorithm(ISCA)can effectively solve real-life optimisation problems.更多还原