基于神经网络的电力通信网故障量化计算方法  

Neural Network-based Method to Quantitatively Calculate Electric Communication Network Faults

作　　者：俞浩[1] YU Hao(State Grid Taizhou Power Supply Company,Taizhou 225300,China)

机构地区：[1]国网泰州供电公司,江苏泰州225300

出　　处：《电工技术》2024年第5期205-207,210,共4页Electric Engineering

摘　　要：当前的电力通信网在故障诊断方面只能进行定性的故障诊断与分析,无法实现复杂通信环境下定量的故障诊断与分析,因此将神经网络方法应用于电力通信网的故障诊断中,以实现定量的故障诊断与分析。首先,确定电力通信网故障信息量化计算指标,构建电力通信网的故障危害性评价体系,并引入专家评估系统;通过专家评估系统处理电力通信网上的故障信息;引入神经网络方法,对电力通信网中的权重参数进行训练,构建基于神经网络的电力通信网故障定量分析模型,实现对故障的量化分析与处理。通过算例测试表明,所提方法能有效计算出电力通信网故障等级信息及定量报警值,与传统方法相比更先进,可进行大规模推广应用。China's current electric communication network can only carry out qualitative rather than quantitative fault diagnosis and analysis under complex communication environment.Therefore this paper studies the application of neural network method to fault diagnosis of electric communication network,attempting to realize quantitative fault diagnosis and analysis.First the quantitative calculation index of fault information of electric communication network is determined,the fault hazard evaluation system is constructed,and the expert evaluation system is introduced.Then the fault information is processed by the expert evaluation system,and neural network method is adopted in training of weight parameters,enabling modeling of neural network-based quantitative fault analysis and realizing quantitative fault analysis.The proposed method is verified by case test effective in calculating fault level information and quantitative alarm value of electric communication network,and thus more advanced than conventional methods and worthy of large scale application.

关 键 词：神经网络 电力通信网 定量计算 

分 类 号：TM727[电气工程—电力系统及自动化]