基于卷积神经网络方法的起重机械电气系统原理图识别初步研究

Preliminary Study on Schematic Diagram Recognition of Lifting Machinery Electrical System Based on Convolution Neural Network Method

特种设备检验检测人员在起重机械电气系统原理图审查中,效率低,容易出现审查疏忽等问题,故本文研究了卷积神经网络在电气系统原理图规范化审核中的应用。在初步研究中,确定了电气元器件符号样本,图像和标签值的对应;针对样本量较少的情况,对已有样本图片进行处理以增加样本量;基于Python的程序资源库,对图像样本进行规范化前处理,得到输入规格统一的图像数据;最后通过基于Python的TensorFlow框架,构建了卷积神经网络识别模型。

Low efficiency of special equipment inspection and detection personnel in the review of the schematic diagram of the electrical system of hoisting machinery,which was prone to the negligence of the review and other potential safety problems,the application of convolutional neural network in the standardized review of the schematic diagram of the electrical system has been studied.In the preliminary study,the correspondence between symbol samples,images and label values of electrical components was determined;For the case of small sample size,increased the sample size by processing the existing sample images;Through the program resource library based on Python,the image samples were normalized and preprocessed to obtain the image data with unified input specifications;Finally,through the TensorFlow framework based on Python,a convolutional neural network recognition model was constructed.