基于AE-BP模型的杨木胶合板应力损伤识别  

Identification of stress damage in poplar plywood based on AE-BP model

作　　者：刘佳[1] 于孟言 高珊[1] 陈昱龙 冯蔓萱 杜鑫宇 LIU Jia;YU Mengyan;GAO Shan;CHEN Yulong;FENG Manxuan;DU Xinyu(College of Civil Engineering and Transportation,Northeast Forestry University,Harbin 150040,Heilongjiang,China)

机构地区：[1]东北林业大学土木与交通学院,黑龙江哈尔滨150040

出　　处：《中南林业科技大学学报》2024年第4期169-179,共11页Journal of Central South University of Forestry & Technology

基　　金：国家自然科学基金面上项目(32071685);国家自然科学基金青年项目(31600453)。

摘　　要：【目的】利用声发射(AE)技术对应力损伤全过程中的杨木胶合板进行无损检测,并利用BP神经网络对AE检测结果进行识别,以提高胶合板损伤检测精度。【方法】以市场占有量较高的托盘用杨木胶合板作为研究对象,在联合AE和应力损伤试验过程中,提取6个AE特征参数,利用声发射RA-AF联合分析法区分杨木胶合板产生裂纹的类型,采用K-均值聚类分析方法确定损伤演化程度和AE特征参数之间的对应关系,利用BP神经网络建立损伤识别模型,并对识别网络进行测试训练。【结果】AE信号幅度和上升时间可有效地表征杨木胶合板应力损伤从微裂纹萌生、产生宏观裂纹至完全断裂的损伤演化过程;通过RA-AF联合分析发现:杨木胶合板在应力损伤试验第一阶段主要为剪切破坏损伤,第二、三阶段主要为拉伸破坏损伤;通过K-均值聚类分析发现损伤类型与AE峰值频率之间的存在较强对应关系,可有效的表征不同的损伤类型:在31 kHz内为基体开裂,在31~100 kHz内为脱胶分层,大于100 kHz为纤维断裂;构建AE-BP神经网络模型对应力损伤类型训练样本的拟合优度是95.94%,测试集的拟合优度是98.89%,模型总拟合优度是96.51%,网络训练效果较好。【结论】在应力承载AE监测过程中,通过构建AE-BP模型,可对杨木胶合板产生的未知损伤进行有效检测并准确识别。【Objective】Acoustic emission(AE)is used to detect the damage of poplar plywood in the whole process of stress damage,and BP neural network is applied to identify the results of AE,so as to improve the damage detection accuracy of plywood.【Method】Poplar plywood used in pallets with high market share was taken as the research object.During the joint AE and stress damage test,six AE characteristic parameters were extracted,the crack types of plywood were distinguished by acoustic emission RAAF joint analysis method,and the corresponding relationship between damage evolution degree and AE characteristic parameters was determined by K-means clustering analysis method.The damage identification model was established by BP neural network,and the identification network was trained by test.【Result】AE signal amplitude and rise time effectively characterized the evolution process of stress damage from microcrack initiation,macroscopic crack to complete fracture;Through RA-AF analysis,it was found that in the first stage of bending test,the main damage of poplar plywood is shear failure.In the second and third stages,the main damage was tensile failure;Based on cluster analysis,it was found that there was a strong corresponding relationship between damage types and AE Peak frequencies,the different damage patterns could be effectively characterized:matrix cracking within 31 kHz,debond and delamination within 31-100 kHz,and fiber fracture above 100 kHz;The AE-BP neural network model could identify the fuzzy damage types with the goodness of fit of the training samples was 95.94%,the goodness of fit of the test set was 98.89%,and the total goodness of fit of the model was 96.51%,and the network training was more effective.【Conclusion】The damage types of poplar plywood during AE monitoring can be effectively detected and accurately identified by constructing AE-BP model.

关 键 词：杨木胶合板 声发射 BP神经网络 损伤识别 

分 类 号：S781.3[农业科学—木材科学与技术]