摘要
冷缩中间接头是整个配网电缆中最薄弱的环节,一旦地下积水,接头长时间浸泡后,很容易受潮,影响了电缆的可靠供电。针对这一问题,设计一种基于寿命数据分析的配网电缆冷缩中间接头受潮诊断模型。采集介质损耗因子、电容量、局部放电量三种寿命表征参数作为后期诊断的基础,利用灰狼算法计算每种特征参数的重要性系数,并与三种寿命表征参数标准化数值相乘,实现特征融合。利用深度置信网络构建受潮诊断模型,计算每种受潮程度的发生概率,将最大值对应的受潮程度标签作为诊断结果,实现配网电缆冷缩中间接头受潮诊断。结果表明:所研究诊断模型的Dice相似系数最大,靠近1,说明该模型的诊断结果更接近实际结果,诊断准确性更高。
The cold shrink intermediate joint is the weakest link in the entire distribution network cable.Once the underground water is accumulated and the joint is soaked for a long time,it is easy to be affected by moisture,which affects the reliable power supply of the cable.In order to solve this problem,a diagnosis model based on the analysis of life data is designed for the moisture in the cooling contraction intermediate joints of distribution network cables.Collect three life characterization parameters,including dielectric loss factor,capacitance and partial discharge,as the basis for later diagnosis.Use the gray wolf algorithm to calculate the importance coefficient of each characteristic parameter,and multiply them with the standardized values of the three life characterization parameters to achieve feature fusion.The depth confidence network is used to build the moisture diagnosis model,calculate the probability of occurrence of each moisture degree,and take the moisture degree label corresponding to the maximum value as the diagnosis result,so as to realize the moisture diagnosis of the cooling contraction intermediate joint of the distribution network cable.The results show that the Dice similarity coefficient of the diagnostic model studied is the largest,close to 1,indicating that the diagnostic results of the model are closer to the actual results,and the diagnostic accuracy is higher.
作者
周幸
皮云霞
李敏
马馨兰
ZHOU Xing;PI Yunxia;LI Min;MA Xinlan(Southern Power Grid Digital Grid Research Institute Co.,Ltd.,Guangzhou 510600,China)
出处
《自动化与仪器仪表》
2023年第8期71-74,共4页
Automation & Instrumentation
基金
基于物资实物编码的智慧供应链应用研究项目(00HK0000001458)。
关键词
寿命数据
配网电缆
冷缩中间接头
置信网络
受潮诊断模型
life data
distribution network cable
cold shrink intermediate joint
confidence network
damping diagnosis model
