作者：朱雅魁1, 赵莎莎1, 李争2
作者单位：1. 国网河北省电力有限公司营销服务中心，河北 石家庄 050000;
2. 河北科技大学，河北 石家庄 050000
关键词：电能质量;对称点模式;栈式降噪自编码;暂态系统
摘要：
针对复杂电能质量扰动信号非平稳性和非线性导致的信号特征难以直接提取和识别的问题，该文提出一种基于自适应遗传算法优化（adaptive genetic algorithm，AGA）的栈式稀疏自编码器（stacked sparse autoencoder，SSAE）和对称点模式（symmetrized dot pattern，SDP）域转换的暂态电能质量扰动识别方法。首先，通过Matlab仿真随机生成6种单一扰动信号和9种复合扰动信号，通过SDP方法将原始时域扰动信号转换至极坐标域，实现扰动信号可视化并生成对应的扰动图谱，对扰动图谱进行参数优化；然后，基于Tensorflow开源框架搭建SSAE识别模型，并由AGA算法完成模型结构及其参数的优化，实现扰动图谱的深度特征提取与挖掘；最后，由末端分类器进行无监督学习分类，比较常见扰动识别方法的优劣。结果表明：该文提出的基于AGA-SSAE和SDP域转换的暂态电能质量扰动识别方法能够对暂态扰动进行高效、准确的识别分类，平均测试准确率为97.89%，优于传统方法10%左右；同时所提方法的架构清晰，且具有较好的收敛性和泛化能力，适用于电力系统电能质量暂态扰动的快速、精确识别。
Transient power quality disturbance identification method based on AGA-SSAE and SDP domain transformation
ZHU Yakui1, ZHAO Shasha1, LI Zheng2
1. Marketing Service Center, State Grid Hebei Electric Power Co., Ltd., Shijiazhuang 050000, China;
2. Hebei University of Science and Technology, Shijiazhuang 050000, China
Abstract: The non-stationary and non-linear characteristics of complex power quality disturbance signal make it difficult to extract and identify the signal features directly. In this paper, a transient power quality disturbance identification method based on stack sparse autoencoder (SSAE) and symmetric dot pattern (SDP) domain conversion of adaptive genetic algorithm (AGA) is proposed, Six kinds of single disturbance signals and nine kinds of compound disturbance signals are randomly generated by Matlab simulation. The original disturbance signals in time domain are transformed into polar coordinate domain by SDP method to realize the visualization of disturbance signals and generate corresponding disturbance maps, and the parameters of disturbance maps are optimized. Then, the SSAE recognition model is built based on tensorflow open-source framework, and the AGA algorithm is used to optimize the model structure and its parameters to realize the depth feature extraction and mining of disturbance spectrum. Finally, unsupervised learning classification is carried out by the end classifier, and the advantages and disadvantages of common disturbance recognition methods are compared. The results show that the proposed transient power quality disturbance identification method based on AGA-SSAE and SDP domain transformation can identify and classify the transient disturbances efficiently and accurately, and the average test accuracy is 97.89%, which is about 10% better than the traditional method. At the same time, the proposed method has clear architecture, good convergence and generalization ability, and is suitable for fast and accurate identification of power quality transient disturbances.
Keywords: power quality;symmetrical dot pattern;stack sparse autoencoder;transient system
2023, 49(8):28-35,66 收稿日期: 2021-07-12;收到修改稿日期: 2021-09-23
基金项目: 国网河北营销中心市场开发室2020年“网上国网”线上线下全渠道融合建设费用（B704YF200028）
作者简介: 朱雅魁(1988-),男,河北石家庄市人,工程师,研究方向为电力营销信息技术研究与应用
参考文献
[1] 杨新法, 苏剑, 吕志鹏, 等. 微电网技术综述[J]. 中国电机工程学报, 2014, 34(1): 57-70
[2] 于会群, 张晓丹, 李增峰, 等. 基于电能质量等级的孤岛微电网分层控制策略[J]. 电力科学与技术学报, 2020, 35(6): 68-75
[3] 聂晓华. 一种基于卡尔曼滤波的电能质量扰动检测新方法[J]. 中国电机工程学报, 2017, 37(22): 6649-6658, 6777
[4] 吴志宇, 朱云芳, 侯怡爽, 等. 电能质量扰动识别的小波压缩感知方法[J]. 电力系统及其自动化学报, 2019, 31(5): 1-7
[5] 黄建明, 瞿合祚, 李晓明. 基于短时傅里叶变换及其谱峭度的电能质量混合扰动分类[J]. 电网技术, 2016, 40(10): 3184-3191
[6] 徐永海, 赵燕. 基于短时傅里叶变换的电能质量扰动识别与采用奇异值分解的扰动时间定位[J]. 电网技术, 2011, 35(8): 174-180
[7] HE S, TIAN W, ZHANG J, et al. A high efficient approach for power disturbance waveform compression in the view of heisenberg uncertainty[J]. IEEE Transactions on Industrial Informatics, 2018, 15(5): 2580-2591
[8] 刘军, 黄纯, 江亚群, 等. 电能质量扰动分析中广义S变换算法的改进与实现[J]. 电力系统及其自动化学报, 2017, 29(3): 35-41
[9] 武昭旭, 杨岸, 祝龙记. 基于循环神经网络的电能质量扰动识别[J]. 电力系统保护与控制, 2020, 48(18): 88-94
[10] 张然, 郭俊宏, 蓝新斌, 等. 基于深度置信网络的电能质量扰动检测与分类[J]. 广东电力, 2020, 33(6): 92-98
[11] 王维博, 张斌, 曾文入, 等. 基于特征融合一维卷积神经网络的电能质量扰动分类[J]. 电力系统保护与控制, 2020, 48(6): 53-60
[12] 郝惠敏, 梁永国, 武海彬, 等. 对称点模式-深度卷积神经网络的红外光谱识别方法[J]. 光谱学与光谱分析, 2021, 41(3): 782-788
[13] WANG H, XU J, YAN R. Yan, et al. A new intelligent bearing fault diagnosis method using SDP representation and SE-CNN[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 69: 2377-2389
[14] 徐舒玮, 邱才明, 张东霞, 等. 基于深度学习的输电线路故障类型辨识[J]. 中国电机工程学报, 2019, 39(1): 65-74, 321
[15] 黄南天, 王文婷, 蔡国伟, 等. 计及复杂气象耦合特性的模块化去噪变分自编码器多源–荷联合场景生成[J]. 中国电机工程学报, 2019, 39(10): 2924-2934
[16] 张聪, 许晓慧, 孙海顺, 等. 基于自适应遗传算法的规模化电动汽车智能充电策略研究[J]. 电力系统保护与控制, 2014, 42(14): 19-24
[17] 徐岩, 郅静. 基于改进自适应遗传算法的PMU优化配置[J]. 电力系统保护与控制, 2015, 43(2): 55-62
[18] AHILA R, SADASIVAM V, MANIMALA K. An integrated PSO for parameter determination and feature selection of ELM and its application in classification of power system disturbances[J]. Applied Soft Computing, 2015, 32: 23-37
[19] KHOKHAR S, ZIN A A B M, MOKHTAR A S B, et al. A comprehensive overview on signal processing and artificial intelligence techniques applications in classification of power quality disturbances[J]. Renewable and Sustainable Energy Reviews, 2015, 51: 1650-1663
[20] 陈伟, 何家欢, 裴喜平. 深度信念网络在电能质量复合扰动识别中的应用[J]. 电力系统及其自动化学报, 2018, 30(9): 75-82
[21] 王燕, 李群湛, 周福林. 一种暂态电能质量扰动检测的新方法[J]. 中国电机工程学报, 2017, 37(24): 7121-7132
