针对复合材料层合板的基体裂纹损伤与脱层损伤的不易区分辨识的问题，采用Lamb波对层合板进行损伤检测，对接收到的传感信号进行特征提取与筛选，创新性地引入加权核双子支持向量基（weighted kernels -twin support vector machine，WK-TWSVM）的机器学习方法对基体裂纹与脱层损伤进行自动分类识别。为了进一步提高损伤辨识精度，采用简化粒子群优化（simple particle swarm optimization,SPSO）算法对WK-TWSVM的核函数权值及模型参数进行了寻优处理，并与其他粒子群优化算法就行了分析比较。试验分析结果表明，基于Lamb波的SPSO-WK-TWSVM复合材料层合板损伤辨识方法能够对复合材料层合板基体裂纹与脱层损伤进行准确的自动识别，识别精度明显高于其他TWSVM优化算法及传统的机器学习方法。
Abstract
Aiming at the problem of it being difficult to distinguish matrix crack damage and delamination damage of composite laminates, Lamb wave was used to detect damages of composite laminates, feature extraction and screening of the received sensing signals were conducted, and the machine learning method based on weighted kernels-twin support vector machine (WK-TWSVM) was introduced innovatively to automatically classify and identify matrix crack and delamination damage. In order to further improve the accuracy of damage recognition, the simple particle swarm optimization (SPSO) algorithm was used to optimize kernel functions’ weights and model parameters, and SPSO algorithm was compared with other PSO algorithms. Test results showed that the damage identification method based on SPSO-WK-TWSVM and Lamb wave can automatically identify matrix crack and delamination damage of composite laminates, and its recognition accuracy is obviously higher than those of other TWSVM optimization algorithms and traditional machine learning methods.
关键词
复合材料层合板 /
Lamb波 /
损伤分类辨识 /
简化粒子群优化 /
双子支持向量基
{{custom_keyword}} /
Key words
composite laminate /
Lamb wave /
damage classification identification /
simple particle swarm optimization (SPSO) algorithm /
weighted kernels-twin support vector machine (WK-TWSVM)
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］陈雪峰, 杨志勃, 田绍华, 等. 复合材料结构损伤识别与健康监测展望［J］. 振动、测试与诊断, 2018, 38(1): 1-10.
CHEN Xuefeng, YANG Zhibo, TIAN Shaohua, et al. Prospects for damage identification and health monitoring of composite structures ［J］. Vibration. Testing and Diagnosis, 2018, 38(1): 1-10.
［2］SANDERS D R, KIM Y I， STUBBS N. Nondestructive evaluation of damage in composite structures using modal parameters［J］. Experimental Mechanics, 1992, 32(3): 240-251.
［3］HAO Z, YANG Z B, YU S, et al. Wave propagation of laminated composite plates via GPU-based wavelet finite element method［J］. Science China Technological Sciences, 2017, 60(6): 832-843.
［4］刘小峰, 杨康俊, 柏林. 基于 Lamb 波递归量化分析的复合材料裂纹损伤定征研究［J］. 振动与冲击, 2019, 38(10): 250-255.
LIU Xiaofeng, YANG Kangjun, BO Lin. Crack damage investigation of composite materials based on the Lamb wave and recursive quantitative analysis［J］. Journal of Vibration and Shock, 2019, 38 (10): 250-255.
［5］LARROSA C, LONKAR K， CHANG F K. In situ damage classification for composite laminates using Gaussian discriminant analysis［J］. Structural Health Monitoring, 2014, 13(2): 190-204.
［6］WILSON C L， FU K C. Monitoring fatigue-induced transverse matrix cracks in laminated composites using built-in acousto-ultrasonic techniques［J］. Structural Health Monitoring, 2016, 15(3): 335-350.
［7］华生明. 结合小波分析和BP神经网络的复合材料损伤检测技术研究［D］. 南京: 南京航空航天大学, 2015.
［8］王峰林, 王长龙, 胡永江. 基于改进LMS算法的复合材料超声检测缺陷识别［J］. 军械工程学院学报, 2013, 25(5): 66-69.
WANG Fenglin, WANG Changlong， HU Yongjiang. Recognition of composite materials flaws within ultrasonic testing based on improved LMS algorithm ［J］. Journal of Ordnance Engineering College, 2013, 25 (5): 66-69.
［9］王明. 基于光纤传感和EMD分解的复合材料损伤监测研究［D］. 武汉: 武汉理工大学, 2012.
［10］崔建国, 张善好, 于明月, 等. 基于GRNN-ELM的飞机复合材料结构损伤识别［J］. 南京航空航天大学学报, 2017, 49(4): 468-473.
CUI Jianguo, ZHANG Shanhao, YU Mingyue, et al. Damage identification of aircraft composite structures based on GRNN-ELM ［J］. Journal of Nanjing University of Aeronautics and Astronautics, 2017, 49 (4): 468-473.
［11］HUANG H J, WEI X X， ZHOU Y Q. Twin support vector machines: A survey ［J］. Neurocomputing, 2018, 300: 34-43.
［12］丁世飞. 孪生支持向量机：理论、算法与拓展［M］. 北京:科学出版社, 2017: 16-31.
［13］TOYAMA N, YASHIRO S, TAKATSUBO J, et al. Stiffness evaluation and damage identification in composite beam under tension using Lamb waves［J］. Acta Materialia, 2005, 53(16): 4389-4397.
［14］吴福林. 孪生支持向量机模型选择问题研究［D］. 徐州: 中国矿业大学, 2015.
［15］PEDERSEN M E H, CHIPPERFIELD A J. Simplifying particle swarm optimization［J］. Applied Soft Computing, 2010, 10(2): 618-628.
［16］WANG J J, WU L, KONG J, et al. Maximum weight and minimum redundancy: A novel framework for feature subset selection［J］. Pattern Recognition, 2013, 46(6):1616-1627.
［17］GLINSKY G V, GLINSKII A B, STEPHENSON A J, et al. Gene expression profiling predicts clinical outcome of breast cancer［J］. Nature, 2002, 415(6871): 530-536.
［18］BISHOP C M. Neural networks for pattern recognition［M］. Oxford: Oxford University Press, 1995.
［19］LI K S, MA Z Y, FU P, et al. Quantitative evaluation of surface crack depth with a scanning laser source based on particle swarm optimization-neural network［J］. NDT & E International, 2018, 98: 208-214.
［20］NIE S K, WANG Y J, XIAO S L, et al. An adaptive chaos particle swarm optimization for tuning parameters of PID controller ［J］. Optim. Control Appl. Meth., 2017, 38: 1091-1102.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(2106 KB)
