摘要
针对复合材料结构健康监测问题,文章提出一种基于神经网络的仿生光纤-复合材料智能结构载荷定位方法。基于复合材料健康监测系统试验平台,利用MATLAB软件,对载荷数据样本进行训练建立BP神经网络载荷定位模型。研究讨论样本数量、隐含层节点数、训练函数等参数对神经网络模型的影响,结果表明,当神经网络的隐含层节点数为6、训练样本数为300、训练函数为trainlm时,建立的光纤智能结构载荷定位模型的预测误差最小,为0.27cm。当去掉边缘检测点后,误差减小为0.09 cm。该方法实现了对复合材料结构上任意位置的载荷定位,且精度较高,该研究具有重要的应用价值。
Aiming at the health monitoring problem of composite structures,this paper proposes a bionic fiber-composite intelligent structure load location method based on neural network.Based on the composite material health monitoring system test platform,the BP neural network load location model was established by training the load data samples with MATLAB software.The influence of sample number,hidden layer node number,training function and other parameters on the neural network model is discussed.The results show that when the hidden layer node number of the neural network is 6,the number of training samples is 300,and the training function is trainLM,the average error of the established load location model of the fiber intelligent structure is minimum,which is 0.27cm.When the edge detection points are removed,the prediction error is reduced to 0.09 cm.This method can locate the load at any position on the composite structure with high precision,which has important application value.
作者
陈缪琪
蒋雨轩
沈令斌
CHEN Miaoqi;JIANG Yuxuan;SHEN Lingbin(Jinling University of Science and Technology,School of Network and Communication Engineering,Nanjing 211169,China)
出处
《长江信息通信》
2022年第7期26-29,共4页
Changjiang Information & Communications
基金
江苏省自然科学基金青年项目(No.BK20190112)
金陵科技学院高层次人才启动项目(No.jit-b-201814,jit-b-202012)
2020年金陵科技学院校级科研基金孵化项目(No.jit-fhxm-2002)
金陵科技学院2021年校级大学生创新创业训练计划项目(省级重点项目:NO.202113573011Z)
金陵科技学院2021年校级“科教融合”项目(NO.2021KJRH21)。
