一种基于PCNN的电力设备故障区域提取方法

摘要

Aiming at the problem of fault diagnosis during detecting electrical equipment,pulse coupled neural network for infrared image segmentation was studied,and a novel segmentation approach was presented in this paper.Firsdy,a new dynamic threshold is set by using the neural pulse output and the activities.Meanwhile,a relationship between the parameters and characteristics of firing region of neurons is set to allow the neurons to produce the pulse output.And then,a non-parametric clustering rule is incorporated in the model to ensure that the captured neurons with brightness similarity to be pulsed together.The dynamic threshold is then updated since a terminal condition is provided.Finally,experimental results show the higher efficiency of our method for image segmentation in compared with traditional thresholding methods,normalized cuts and classic PCNN on real-world infrared images.%针对红外自动监控电力设备是否存在故障问题,结合脉冲耦合神经网络(PCNN)同步点火特性,提出一种基于PCNN的红外图像感兴趣区域提取方法.首先针对原始的动态阈值振荡问题,采用神经元点火信息构建新的动态阈值,并建立连接系数与点火区域信息之间的内在关系,从而使得神经元自适应地发生点火.为了进一步确保每一次迭代中所捕获的神经元与点火区域的相似性,在模型框架内融合了一种聚类规则,进而有效更新动态阈值,并给出了停止迭代的方法.实验表明,该提取区域方法性能优于传统阈值、normalized cuts以及经典PCNN模型等方法.