摘要
针对经典粗糙集只能处理连续型数据的不足,引入邻域粗糙集理论对铝合金焊接接头疲劳寿命影响因素进行分析,建立既能处理符号型又能处理数值型属性变量的统一邻域粗糙集模型。运用前向贪心和后向剪枝相结合的算法进行属性约简,获取影响接头疲劳寿命的核心因素集,并定量计算各影响因素的权重。实验结果表明:所建立的邻域粗糙集模型不需要先验知识,可从疲劳样本数据本身出发得到铝合金焊接接头疲劳寿命各影响因素的客观评定,能够选择少量的特征。
To sole the problem that classical rough set theory can only deal with continuous data, neighborhood rough set theory is employed to analyze the factors that influence the fatigue life of the welded joints. A unified neighborhood rough set model is proposed, which can handle both symbolic and numeric attributes. Attribute reduction algorithm combining forward greedy algorithm and backward pruning algorithm is used to obtain the key factors and to quantitatively calculate the weights of the factors influencing the fatigue life of the welded joints. Experiment results of aluminum alloy welded joints show that the proposed neighborhood rough set model can choose a small number of features and obtain an objective evaluation of various influence factors from the sample data of the aluminum alloy welded joints without any priori knowledge.
出处
《吉林大学学报(工学版)》
EI
CAS
CSCD
北大核心
2017年第6期1848-1853,共6页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金项目(51175054)
辽宁省自然科学基金项目(2015020169)
关键词
材料合成与加工工艺
邻域粗糙集
焊接接头
疲劳寿命
material synthesis and processing technology
neighborhood rough set
welded joints
fatigue life