基于深度学习SCNet的钢筋混凝土锈蚀裂缝识别

Identification of Corroded Cracks in Reinforced Concrete Based on Deep Learning SCNet Model

为提高钢筋混凝土锈蚀裂缝检测分类的效率和精度,提出了一种基于深度学习卷积神经网络(Convolutional Neural Network,CNN)的钢筋混凝土锈蚀裂缝识别模型SCNet(Steel CorrosionNet).首先通过原始数据采集和数据增强构建了39000张图片的裂缝数据集,然后利用TensorFlow学习框架和Python构建神经网络模型并进行训练测试,根据模型的训练精度和测试精度进行网络结构和网络参数的优化,最终将SCNet识别模型与两种传统检测方法进行对比.结果表明:文中所建立的SCNet三分类神经网络模型达到了96.8%的分类准确率,可以有效识别分类钢筋混凝土锈蚀裂缝,并且具有较高的准确率和可测性;在图像数据有阴影、扭曲等噪声干扰的条件下,两种传统检测方法已不能达到理想的分类效果,SCNet模型仍能表现出相对稳定的分类性能.

In order to improve the efficiency and accuracy of corroded cracks detection and classification in rein⁃forced concrete,a corroded cracks identification model Steel Corrosion Net(SCNet)based on deep learning Convolu⁃tional Neural Network(CNN)is proposed.A data set of 39000 crack figures is firstly built by original data collection and data enhancement,a SCNet three-classification neural network model is then built and tested by TensorFlow learning framework and Python.According to the training and testing accuracies of the model,the structure and pa⁃rameters of the SCNet network model are optimized and the result of the SCNet is compared with two traditional test⁃ing methods.The result shows that the SCNet model established in this paper achieves the classification accuracy of 96.8%,which means the SCNet model can effectively identify and classify the corroded cracks in reinforced concrete with high accuracy and measurability.Under the conditions of noise interference such as shadows and distortions,those two traditional testing methods fail to ideally classify,whereas the SCNet model shows a relatively stable classi fication performance.