摘要
针对模糊神经网络算法处理高维耦合港口物流功能评价模型时出现不能较好地识别和利用信息、应用新知识的能力与学习新知识的能力不平衡等多重问题,研发出改良模糊神经网络算法。这种方法利用了改进的自适应交叉和变异算子,且将其应用于模拟退火机制对传统遗传算法进行改进,利用改进过的新算法所发出的属值进行自动调整,仿真验证所使用的软件是Matlab2016b,使高维耦合港口物流功能评价过程中泛化能力与学习能力失衡等问题得到有效解决,并在诸多方面具有显著优势,例如较强的稳定性和干扰抵抗能力,以及高效的搜索能力等。论文将研究对象确定为国内东北地区的某综合性港口,在研究过程中用于开发验证环境的平台是Eclipse,选择的分析方法是实证法,通过最终获得的结果可知,文中建立的模型能够对港口物流功能进行全面评价,在评价适应性、模型拟合度、并行搜索效率等方面具有明显优势。
Aiming at the multiple problems of fuzzy neural network algorithm,such as the poor recognition and utilization of information,the imbalance between the ability to apply new knowledge and the ability to learn new knowledge,when it deals with the high-dimensional coupling port logistics function evaluation model,an improved fuzzy neural network algorithm is developed.This method uses the improved adaptive crossover and mutation operator,and applies it to the simulated annealing mechanism to improve the traditional genetic algorithm,and automatically adjusts the membership value issued by the improved new algorithm. The software used in the simulation verification is Matlab2016b,which makes the problems such as the imbalance between generalization ability and learning ability in the process of high-dimensional coupling port logistics function evaluation. It can be solved effectively,and has significant advantages in many aspects,such as strong stability and interference resistance,and efficient search ability. In this paper,the research object is determined to be a comprehensive port in the northeast of China. Eclipse is the platform used to develop the verification environment in the research process,and the empirical method is chosen as the analysis method.Through the final results,we can see that the model established in this paper can comprehensively evaluate the port logistics function,and has the advantages of adaptability,model fitting,parallel search efficiency,etc obvious advantages.
作者
王瑞玺
尚东方
阳志文
WANG Ruixi;SHANG Dongfang;YANG Zhiwen(Tianjin Institute of Water Transportation Engineering,Ministry of Transportation,Tianjin 300456)
出处
《计算机与数字工程》
2020年第2期344-349,360,共7页
Computer & Digital Engineering
基金
国家自然科学基金项目“复杂造波平台下波浪数模与物模全息耦合模型研究”(编号:51509119)资助。
关键词
港口物流功能
改进模糊神经网络算法
改进遗传算法
评价模型
实证分析
port logistics function
improved fuzzy neural network algorithm
improved genetic algorithm
evaluation mod-el
empirical analysis
