作者：张惠山1, 孟荣1, 付炜平1, 王昭雷1, 贾建东2
作者单位：1. 国网河北省电力有限公司检修分公司, 河北 石家庄 050000;
2. 华北电力大学电力工程系, 河北 保定 071003
关键词：长短时记忆网络;注意力机制;云模型;继电保护;可靠性预测
摘要：
继电保护装置可靠性是保证电力系统安全、稳定运行的基础，为准确预测继电保护装置的可靠性，构建基于长短时记忆（long-short term memory，LSTM）网络的继电保护装置可靠性预测模型。提出将注意力模型嵌入LSTM网络以进一步提高LSTM网络的预测准确度，建立基于云模型的继电保护装置劣化程度指标定量和定性评估之间不确定性转换模型。结果表明：嵌入注意力模型的LSTM网络比LSTM网络的预测准确度显著提高，具有较好的逼近能力和泛化能力，均方根误差为1.28%，平均绝对百分比误差为6.62%，当劣化程度指标大于40%时的相对误差控制在–5%~5%范围内。采用建立的云模型分析方法实现相同评级范围的样本劣化程度指标定性和定量联合评判，可为定性评级的模糊性和随机性提供定量评价指导。
Study on the reliability prediction of the relay protection device based on LSTM network
ZHANG Huishan1, MENG Rong1, FU Weiping1, WANG Zhaolei1, JIA Jiandong2
1. Maintenance Branch Company of the State Grid Hebei Electric Power Co., Ltd., Shijiazhuang 050000, China;
2. Department of Electrical Engineering, North China Electric Power University, Baoding 071003, China
Abstract: Reliability of the relay protection device is the basis for ensuring the safe and stable operation of the electric power system. In order to accurately predict the reliability of relay protection device, a reliability prediction model based on long-short term memory network is established. It proposes to embed the attention model into the LSTM network to further improve the prediction accuracy of LSTM networks, and establish an uncertainty conversion model between quantitative and qualitative evaluation of relay protection device degradation index based on cloud model. The results show that the prediction accuracy of LSTM network embedded in the attention model is significantly improved than that of the LSTM network. The LSTM network embedded in the attention model shows good approximation ability and generalization ability, and the root mean square error and the average absolute percentage error is 1.28% and 6.62%, respectively. The relative error is controlled within the range of -5% to 5% when the deterioration index is greater than 40%. The established cloud model analysis method is adopted to realize the qualitative and quantitative joint evaluation for sample degradation of the same rating range, which provides the quantitative evaluation guidance for the ambiguity and randomness of the qualitative rating.
Keywords: long-short term memory network;attention model;cloud model;relay protection;reliability prediction
2023, 49(5):164-170 收稿日期: 2021-08-16;收到修改稿日期: 2021-11-03
基金项目: 国家电网公司科技基金项目（kj2021-055）
作者简介: 张惠山(1976-),男,甘肃天水市人,教授级高级工程师,硕士,从事电力系统继电保护状态监测与诊断关键技术研究
参考文献
[1] 叶远波, 谢民, 陈晓东, 等. 高压继电保护装置的失效特性分析[J/OL]. 电测与仪表: 1-6[2021-02-26]. https://kns.cnki.net/kcms/detail/23.1202.TH.20210226.1425.006.html.
[2] 雷明, 陈一悰, 刘峰, 等. D5000继电保护设备在线监视及分析应用提升[J]. 电网技术, 2020, 44(3): 1197-1202
[3] 闵家新, 郎小毅, 李洋, 等. 基于泛在物联网的继电保护设备智能运维技术[J]. 科学技术与工程, 2021, 21(3): 1024-1029
[4] KIBRIA M G, NGUYEN K, VILLARDI G P, et al. Big data analytics, machine learning and artificial intelligence in next-generation wireless networks[J]. IEEE Access, 2018, 6(3): 32328-32338
[5] HUANG Q, JING S, LI J, et al. Smart substation: state of the art and future development[J]. IEEE Transactions on Power Delivery, 2017, 32(2): 1098-1105
[6] 顾朝敏, 李晓峰, 董驰, 等. 基于Lora的开关柜关键部位温度监测系统研究[J]. 中国测试, 2021, 47(3): 122-125
[7] 姜丹, 梁春燕, 吴军英, 等. 基于大数据分析的电力运行数据异常检测示警方法[J]. 中国测试, 2020, 46(7): 18-23
[8] 刘一民, 章家欢, 杨心平, 等. 基于宏观微观的继电保护设备评估体系构建方法[J]. 电网技术, 2020, 44(8): 3090-3096
[9] 李祖明, 孙仲民, 顾乔根, 等. 基于继电保护装置态势感知及辅助决策的智能运维系统[J]. 电力系统保护与控制, 2020, 48(19): 142-150
[10] 黄劼, 林鸿伟, 黄彦. 基于蚁群算法的继电保护系统可靠性分析[J]. 国外电子测量技术, 2017, 36(9): 104-108
[11] 何旭, 姜宪国, 张沛超, 等. 基于SVM的小样本条件下继电保护可靠性参数估计[J]. 电网技术, 2015, 39(5): 1432-1437
[12] 吴迪, 汤小兵, 李鹏, 等. 基于深度神经网络的变电站继电保护装置状态监测技术[J]. 电力系统保护与控制, 2020, 48(5): 81-85
[13] ZHANG S L, WANG Y X, LIU M Q, et al. Data-based line trip fault prediction in power systems using LSTM networks and SVM[J]. IEEE Access, 2018, 6(3): 7675-7686
[14] 马玉锴, 田亮. 基于证据理论与云模型的火电机组节能减排绩效综合评价[J/OL]. 华北电力大学学报(自然科学版): 1-8[2021-07-07]. https://kns.cnki.net/kcms/detail/13.1212.TM20210707.1621.002.html.
