基于改进SSD的输电线路防振锤缺陷检测

Defect detection of transmission line damper based on improved SSD

随着人工智能的发展,图像识别技术逐渐应用于智能电网无人机巡检的防振锤缺陷检测中。但是由于防振锤体积很小,像素信息少,目前的检测效果还不够理想。为了解决该问题,提出构建一种改进的SSD(Single Shot MultiBox Detector)深度学习神经网络模型去进行防振锤缺陷检测。在特征提取阶段,加入超大卷积核去大大提升网络的感受野;在预测回归阶段,引入SIOU目标框损失函数去提高网络的收敛性。实验结果表明,改进模块能提高SSD网络的检测精度,改进SSD模型的mAP为96.9%,优于现阶段其他主流目标检测模型,由此验证了所提方法的准确性和实用性。

With the development of artificial intelligence,image recognition technology has been gradually applied to the defect detection of damper in UAV patrol inspection of smart grid.However,due to the small volume of damper and the lack of pixel information,the current detection effect is not ideal.In order to solve this problem,an improved SSD(Single Shot MultiBox Detector)deep learning neural network model is proposed to detect the defects of damper.In the feature extraction stage,super convolution kernel is added to greatly enhance the receptive field of the network;In the prediction and regression stage,SIOU target frame loss function is introduced to improve the convergence of the network.The experimental results show that the improved module can improve the detection accuracy of SSD network,and the mAP of the improved SSD model is 96.9%,which is superior to other mainstream target detection models at this stage,thus verifying the accuracy and practicality of the proposed method.