摘要
为了应对复杂工业场景下数据分析与处理所面临的挑战,提出一种新的模糊分类方法。该方法结合无监督聚类和有监督优化,对数据结构进行一种非线性变换,最终构建出具有更高分类性能的模糊分类器。首先,利用特征权重对原始数据的特征部分加权后进行无监督模糊聚类,构建初步的模糊分类器;其次,利用原始标签信息计算分类准确率作为优化目标,通过有监督优化模块优化算法获取最优特征权重;最后,通过最优特征权重对原始数据进行非线性变换,构建最优的模糊分类器。具体选用模糊C均值聚类算法与粒子群优化算法组合解释了所提模糊分类方法的构建与实现原理。实验采用多个不同领域的数据集进行训练与测试,结果表明所提方法具有良好的分类性能与泛化能力。
To cope with the challenge of data analysis and processing in complex industrial scenarios,a novel fuzzy classification method was proposed.The method combined the advantages of unsupervised clustering and supervised optimization techniques to perform a nonlinear transformation on the data structure and ultimately construct a fuzzy classifier with higher classification performance.The feature weights were weighted on the original feature part,and the unsupervised fuzzy clustering was performed to construct a preliminary classifier.Then,the classification accuracy calculated based on the original label information was used as the optimization objective,and the optimal feature weights were finally optimized through the supervised optimization module.Finally,the optimal feature weights were used to perform a nonlinear transformation on the original data to construct the optimal fuzzy classifier.Specifically,the proposed fuzzy classification method was clarified by a combined use of the fuzzy C-means clustering algorithm and the particle swarm optimization algorithm.Experimental results using multiple datasets from different fields demonstrated that the proposed method had better classification performance and generalization ability than general fuzzy classification methods.
作者
唐孝安
周宇
杨建新
张强
TANG Xiaoan;ZHOU Yu;YANG Jianxin;ZHANG Qiang(School of Management,Hefei University of Technology,Hefei 230009,China;Information Center,China North Industries Group Corporation,Beijing 100089,China)
出处
《计算机集成制造系统》
北大核心
2026年第3期1153-1162,共10页
Computer Integrated Manufacturing Systems
基金
国家自然科学基金资助项目(72101075,72171069,92367206)
工业装备质量大数据工信部重点实验室资助项目(2023-IEQBD-02)。
关键词
模糊C均值
粒子群优化
非线性变换
分类
fuzzy C-means
particle swarm optimization
nonlinear transformation
classification
