| 刘俊领， 程晓颖.基于电阻抗断层成像的碳纤维增强复合材料定量化损伤研究[J].轻工机械,2024,42(5):35-43 |
| 基于电阻抗断层成像的碳纤维增强复合材料定量化损伤研究 |
| Quantitative Damage Study of Carbon Fiber Reinforced Polymeric Based on Electrical Impedance Tomography |
| |
| DOI：10.3969/j.issn.1005 2895.2024.05.005 |
| 中文关键词: 碳纤维增强复合材料 电阻抗断层成像 去噪卷积神经网络 检测概率 准静态压痕 |
| 英文关键词:carbon fiber reinforced polymeric EIT(electrical impedance tomography) DnCNNs(denoising convolutional neural networks） probability of detection QSI(quasi static indentation) |
| 基金项目:“纺织之光”应用基础研究项目：基于织物结构因素的环形编织增强复合材料结构功能一体化研究（J202103）。 |
|
| 摘要点击次数: 31 |
| 全文下载次数: 12 |
| 中文摘要: |
| 针对电阻抗断层成像（electrical impedance tomography，EIT）在反演的过程中存在的病态的、不适定性的特点以及检测结果难量化等问题，笔者提出通过去噪卷积神经网络（denoising convolutional neural networks，DnCNNs）以及搭建电阻抗断层成像损伤监测的检测概率（probability of detection，POD）函数来实现对碳纤维增强复合材料（carbon fiber reinforced polymeric，CFRP）高效准确的可视化检测和结果量化评估。借助EIDORS软件进行有限元划分，通过正问题求解来模拟时差数据，构建出不同形状的损伤重建图像；然后通过在去噪神经网络中使用单个残差单元来预测重构图像中的噪音映射，有效地去除重建图像中的伪影，从而实现准确高质量地检测图像；此外，基于统计处理定量分析碳纤维增强复合材料损伤。通过电阻抗断层成像来监测碳纤维增强复合材料的损伤衍化过程，借助准静态压痕（quasi static indentation，QSI）试验实时采集电阻抗断层成像数据和声发射事件，将损伤程度与电导率振幅作为检测概率函数中的缺陷参数，通过应用信号响应方法参数预测POD曲线，通过设置特定的POD阈值获得可检测性极限轮廓，从而实现通过电导率的振幅对损伤程度进行预测。结果表明：该方法不仅能对碳纤维增强复合材料进行高质量损伤图像重建，还可以定量化评价碳纤维增强复合材料的损伤程度，有利于具有自感知能力的复合材料在工业领域的应用。 |
| 英文摘要: |
| Electrical impedance tomography (EIT) had pathological and ill posed characteristics during the inversion process, as well as difficulties in quantifying the detection results.Denoising convolutional neural networks (DnCNNs) and the construction of a probability of detection (POD) function for damage monitoring by electrical impedance tomography were used to achieve efficient and accurate visual inspection and quantitative evaluation of results for carbon fiber reinforced polymeric (CFRP). With the help of electrical impedance tomography reconstruction software (EIDORS) different shapes of damage reconstruction images were constructed by finite element division and solving the positive problem to simulate the time difference data. The noise mapping in the reconstructed result was then predicted by using a single residual unit in the denoising neural network, which effectively removes artifacts from the reconstructed image, resulting in an accurate and high quality detection image. In addition to this, carbon fiber reinforced polymeric damage was quantitatively analyzed based on statistical processing.The damage evolution process of carbon fiber reinforced polymeric was monitored by electrical impedance tomography (EIT), and the EIT data and acoustic emission events were collected in real time with the help of quasi static indentation (QSI) test. Using the degree of damage and the amplitude of conductivity as defect parameters in the detection probability function, the POD curves were predicted by applying signal response method parameters.The results show that the above experimental methods and results can not only realize high quality damage image reconstruction of carbon fiber reinforced polymeric, but also quantitatively evaluate the damage degree of carbon fiber reinforced polymeric, which are conducive to the application of composites with self awareness in aerospace, transportation, energy and other fields. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
