作者：杨向飞¹, 赵子涵¹, 郑中¹, 向博¹, 吴跃康²

作者单位：1. 国网四川省电力公司, 四川 成都 610041;
2. 电子科技大学 电力系统广域测量与控制四川省重点实验室, 四川 成都 611731

关键词：智能变电站;神经网络专家系统;检修测试安措;检修边界;自动生成安措

摘要：

由于智能变电站中设备耦合关系复杂，现有的依靠人工拟票的检修测试安全措施主要依靠检修测试人员的经验，受测试人员水平的限制，容易漏项错项，引发检修过程中的安全事故。对此，该文提出基于神经网络专家系统的自动生成检修测试安措新方法，构建智能变电站检修测试任务中的通用规则模型，并利用神经网络推理计算改善现有专家系统的局限性，实现了自动生成优化的检修测试方案，提出明确的操作计划及设备的陪停措施。该方法可用于指导变电站检修工作人员正确开具设备检修测试安措票，以此降低人为因素造成的设备误动作或检修事故发生的概率。

Research on automatic generation of safety measures for intelligent substation maintenance and testing based on neural network and expert system
YANG Xiangfei¹, ZHAO Zihan¹, ZHENG Zhong¹, XIANG Bo¹, WU Yuekang²
1. State Grid Sichuan Electric Power Company, Chengdu 610041, China;
2. The Power System Wide-area Measurement and Control Sichuan Provincial Key Laboratory, University of Electronic Science and technology of China, Chengdu 611731, China
Abstract: Because the coupling relationship of equipment in smart substation is complex. The existing safety measures of maintenance and testing formulate by manual ticket is mainly rely on the people's professional knowledge of maintenance and testing, which is easy to missing items and lead to safety accidents in the process of maintenance. In view of this, this paper proposes a new method of automatic generation for security measures of maintenance and testing based on expert system, this paper also established general rule models in maintenance and testing tasks of smart substations, and used neural network to exceed the limitation of expert system to realize automatic generation of optimized test schemes and certain operation plan and equipment stopping measures. This method can be used to instruct substation maintenance staff to formulating security tickets for equipment maintenance and testing correctly, so as to reduce the probability of equipment misoperation or maintenance accidents caused by human factors.
Keywords: smart substation;neural network expert system;safety measures for maintenance and testing;maintenance boundary;automatic generation of safety measures
2019, 45(11):120-125  收稿日期: 2018-12-07;收到修改稿日期: 2019-03-17
基金项目:
作者简介: 杨向飞(1975-),男,四川成都市人,高级工程师,硕士,主要研究方向为电力系统继电保护
参考文献
[1] 肖燕. 新一代智能变电站信息流架构设计[J]. 中国电机工程学报, 2016, 36(5):1245-1251
[2] 韩祯祥, 文福拴. 人工智能及其在电力系统中的应用:从专家系统到人工神经元网络[J]. 电力系统自动化, 1991(3):5-15
[3] 杨兵, 丁辉, 罗为民, 等. 基于知识库的变压器故障诊断专家系统[J]. 中国电机工程学报, 2002, 22(10):121-124
[4] 苏盛, LI K K, CHAN W L, et al. 广域电流差动保护区划分专家系统[J]. 电网技术, 2005, 29(3):55-58
[5] 杨东升, 鲁观娜, 丁恒春. 电能表自动检定流水线模糊故障诊断专家系统[J]. 电测与仪表, 2017, 54(7):94-96
[6] 周志华, 陈世福. 神经网络集成[J]. 计算机学报, 2002, 25(1):1-8
[7] 厉劼翀, 周宁, 吕彬. 基于神经网络、模糊理论的变压器油中溶解气体诊断专家系统[J]. 电网技术, 2006, 30(s1):9-12
[8] 徐晓春, 汤同峰, 赵雨田, 等. 智能变电站二次设备检修安措规则数据库设计与实现[J]. 电力系统保护与控制, 2017, 45(2):111-116
[9] 李明, 韩学山, 王勇, 等. 变电站状态检修决策模型与求解[J]. 中国电机工程学报, 2012, 32(25):196-202
[10] 李永丽, 李致中, 杨维. 继电保护装置可靠性及其最佳检修周期的研究[J]. 中国电机工程学报, 2001, 21(6):63-65
[11] 倪益民, 杨宇, 樊陈, 等. 智能变电站二次设备集成方案讨论[J]. 电力系统自动化, 2014, 38(3):194-199