基于知识图谱的校园分布式光伏发电站设备故障自动化检测系统  

Automatic fault detection system of campus distributed photovoltaic power station based on knowledge graph

作　　者：成志鑫 曾跞 唐骁 吴秋实 司风琪[2] 周建新[2] CHENG Zhixin;ZENG Luo;TANG Xiao;WU Qiushi;SI Fengqi;ZHOU Jianxin(Huaneng Nanjing Jinling Power Generation Co.,Ltd.,Nanjing 210000,China;School of Energy and Environment,Southeast University,Nanjing 210000,China)

机构地区：[1]华能南京金陵发电有限公司,江苏南京210000 [2]东南大学能源与环境学院,江苏南京210000

出　　处：《电子设计工程》2025年第1期86-90,共5页Electronic Design Engineering

基　　金：2022年中国华能集团有限公司科技项目(HNKJ22-H119)。

摘　　要：针对校园分布式光伏发电自动化控制系统在故障检测中存在分类检测率低、耗时长等问题,设计了一种基于知识图谱的检测系统。系统硬件包括STM32F103RCT6型单片机、VMS-300AL型红外传感器、温度传感器、电压及电流传感器,通信模块利用SPI总线与单片机相连接;在系统软件层面,选择了自顶向下的知识图谱构建方式,并给出了故障知识的数据模式图,基于DNN网络模型训练输出集的特征提升系统分类检测能力。实验结果显示,所提出检测系统设计的故障分类检测能力较强,针对于训练集和故障集的检测率分别为99.53%和99.37%,检测耗时较少。Aiming at the problems of low classification detection rate and long time⁃consuming in fault detection of campus distributed photovoltaic power generation automation control system,a detection system based on knowledge map is designed.The system hardware includes STM32F103RCT6 single chip microcomputer,VMS-300AL infrared sensor,temperature sensor,voltage and current sensor,and the communication module is connected with the single chip microcomputer by SPI bus.On the system software level,the top⁃down knowledge map construction mode is selected,and the data pattern diagram of fault knowledge is given.Based on DNN network model,the characteristics of output set are trained to improve the system classification and detection ability.The experimental results show that the fault classification detection ability of the proposed detection system is strong,and the detection rates for the training set and the fault set are 99.53%and 99.37%respectively,and it takes less time to detect.

关 键 词：知识图谱 分布式光伏发电 自动化检测 LSTM DNN 

分 类 号：TN02[电子电信—物理电子学]