基于AI芯片的输电通道外破隐患边缘智能监测装置研制及应用

塔吊、挖掘机等外破隐患导致输电通道事故频繁发生，有效地检测输电通道周围的外破隐患对保障输电线路安全稳定的运行意义重大。该文以边缘智能芯片为基础，研制了一种输电通道外破隐患边缘智能检测装置，并提出了一种适用于计算资源有限前端装置的轻量化外破隐患识别方法。利用深度残差网络对输电通道图像进行视觉特征提取；利用候选区域生产网络RPN捕获外破隐患目标的候选区域，再用全卷积神经网络FCN进行外破隐患的目标分类与定位。以实际采集的输电通道图像构建成样本集，进行模型测试与验证，结果表明所提方法在边缘装置中表现出良好的适用性。

Development and Application of the Intelligent Monitoring Device for Hidden Danger Edge of the Transmission Channel Based on AI Chip

Tower cranes， excavators and other external damage hazards lead to frequent transmission channel accidents. Effectively detecting the external damage hazards around the transmission channel is of great significance to ensure the safe and stable operation of the transmission line. Therefore， based on the edge intelligent chip， an intelligent edge detection device for hidden dangers of the transmission channel is developed， and a lightweight hidden danger identification method suitable for the front-end device with limited computing resources is proposed. Firstly， the visual feature of the transmission channel image is extracted using the depth residual network， and secondly， the candidate area of the hidden danger target is captured using the candidate area production network RPN， furthermore the full convolution neural network FCN is used to classify and locate the hidden danger of the external damage. Finally， the actual collected transmission channel images are constructed into a sample set for model test and experimental verification. The experimental results show that the proposed method has good applicability in the edge device.