基于双注意力机制优化CNN架构的GIS局部放电模式识别

PD pattern recognition for GIS based on CNN architecture optimized by the double attention mechanism

GIS局部放电模式识别是其绝缘状态评估的重要部分,为对GIS局部放电信号进行准确、高效识别,提出一种基于双注意力机制优化CNN的GIS局部放电信号模式识别方法。首先,搭建GIS局部放电测试平台,并在GIS气室内人为设置4种典型缺陷,分别通过超高频和超声波检测法对不同缺陷局部放电信号进行采集;然后,基于二者数据特点分别进行数据预处理,并构建由超高频局部放电谱图图像特征和超声信号格拉米角场密度分布组成的特征空间;最后,通过双注意力机制优化的卷积神经网络对输入图像进行深层特征提取,由网络尾端的Softmax分类器进行结果预测。结果表明:融合多特征参数算法能够达到97.57%左右的识别准确率,高于单一特征识别率;同时在同一网络输入时,双注意力机制优化的卷积神经网络在算法识别率、训练速度和鲁棒性等方面均优于常见网络算法。

The PD pattern recognition is an important part of the insulation state evaluation of GIS.For the purpose of the accurate and efficient identification of PD signals,a new method of PD signal pattern recognition is proposed based on CNN optimized by the dual attention mechanism in this paper.Firstly,the GIS PD test platform is built,and four typical defects are set up in GIS chamber.The PD signals of different defects are collected by the UHF and ultrasonic detection respectively.Then,the data preprocessing is carried out based on the characteristics of the data obtained by methods mentioned above respectively.The feature space composed by the image features of UHF PD spectrum and the gram angle field density distribution of ultrasonic signal are constructed.Finally,the input image is extracted through the method of convolutional neural network optimized by double attention mechanism,and the results are predicted by a softmax classifier at the end of the network.It is shown that 97.57%recognition accuracy can be achieved by the fusion algorithm,which is higher than the recognition rate considering the single feature.The convolutional neural network optimized by the double attention mechanism is superior to the common algorithm in the aspects of the recognition rate,training speed and robustness.