基于K均值算法-TOF聚类分析的复合材料板冲击损伤层析研究
Clustering analysis of time of flight(TOF)for tomographic evaluation of impact damage in composites based on K-means algorithm
林君 1杨乐辉 1赵金玲 1李念 1裘进浩 2赵建平 1常乐1
作者信息
- 1. 南京工业大学 机械与动力工程学院,南京 211816;南京工业大学 高端装备可靠性制造研究院,南京 211816
- 2. 南京航空航天大学 航空航天结构力学及控制全国重点实验室,南京 210016
- 折叠
摘要
碳纤维增强型复合材料(CFRP)中的冲击损伤是威胁压力容器结构安全的重要隐患,有必要对CFRP中的冲击损伤进行层析研究,促成CFRP复合材料损伤定量化分析.提出基于K均值算法对超声C扫的飞行时间(TOF)图像进行聚类分析,并结合TOF值确定不同簇所在层,对CFRP冲击损伤进行逐层定量化分析.同时,分析了冲击损伤的TOF图像,验证了基于K均值算法-TOF聚类分析的冲击损伤层析方法,并进一步通过水浸超声C扫试验获得多组试验数据,指出K均值聚类中簇的总数K为铺层总数的1.25倍即可满足损伤逐层分析,实现了不同铺层总数CFRP层合板冲击损伤的层析研究.TOF聚类分析方法可以直接对CFRP层合板各铺层的冲击损伤进行定量化,具有一定的工程意义.
Abstract
Impact damage in carbon fiber reinforced polymer(CFRP)poses a significant threat to the structural safety of pressure vessels.It is crucial to conduct a detailed analysis of impact damage in CFRP for the quantitative analysis of damage in CFRP composite materials.A clustering analysis based on the K-means algorithm was proposed for time of flight(TOF)images obtained from ultrasonic C-scans.The TOF values were used in conjunction with the K-means algorithm to determine the layers associated with different clusters,enabling a layer-by-layer quantitative analysis of impact damage in CFRP.The proposed method was validated by applying it to TOF images of impact damage.Furthermore,to realize layer-by-layer analysis of impact damage in CFRP laminates with different total layers,numerous experimental data obtained by immersion ultrasonic C-scan experiments were analyzed.It demonstrates that setting the number of clusters K in K-means algorithms to 1.25 times the total layers can satisfy the layer-by-layer analysis of the damage.The TOF cluster analysis method can directly quantify the impact damage of each layup of CFRP laminates,which has important significance in engineering applications.
关键词
碳纤维增强型复合材料(CFRP)/冲击损伤/K均值算法/TOF聚类分析Key words
carbon fibre reinforced polymer(CFRP)/impact damage/K-means algorithm/cluster analysis of TOF引用本文复制引用
基金项目
国家重点研发计划项目(2022YFC3004503)
国家重点攻关项目(2023ZY01001)
江苏省研究生科研与实践创新计划项目(SJCX23_0474)
出版年
2024
国家科技期刊平台
NSTL
万方数据