Power system transient stability assessment method based on multiple STA-GLN ensemble models

doi:10.3969/j.issn.2097-0706.2023.07.006

Abstract

Abstract:

With the continuous access of high-proportion renewable energy to the power grid and advancing of power electronization， power systems are becoming increasingly complex in structure， which threats of systems' stability. To address the poor topology adaptability， difficulty in learning instability samples， and long model training time of the transient stability assessment （TSA） method based on artificial intelligence（AI），an ensemble TSA method based on Spatial-Temporal Attention Mechanism， Graph Convolution and Long Short-Term Memory Network（STA-GLN） is proposed. A power system simulation model is built， in which different line faults are set under three topologies， full connection， N-1 disconnection and N-2 disconnection， and the original sample sets are obtained. The TSA method based on STA-GLN shows stronger adaptability and accuracy to the variation of the system's topologies. Then， Adaptive Boosting （AdaBoost） and transfer learning are integrated into the multi-task TSA model based on STA-GLN， which reduces the false judgment and accelerates the response speed of the model. The effectiveness of the method is verified by the simulation analysis of a New England 10-generator 39-node system.

Key words: power system, renewable energy, transient stability assessment, AI, ensemble learning, transfer learning, multi-task model, power electronization

CLC Number:

TK01

YANG Bo, LI Chengyun, LYU Haoxuan, ZHOU Bowen, LI Guangdi, GU Peng. Power system transient stability assessment method based on multiple STA-GLN ensemble models[J]. Integrated Intelligent Energy, 2023, 45(7): 48-60.

Figures/Tables 29

Fig.1

Table 1

Table 2

Table 3

Fig.2

Fig.3

Fig.4

Fig.5

Fig.6

Table 4

Table 5

Assessment results of different characteristic sets

特征集	评估指标
特征集	$P a c c$ /%	$F 1$ /%
1	94.57	95.33
2	94.15	94.54
3	93.62	94.20

Table 5

Table 6

Table 7

Performances of STA-GLN models with different layers

SA-GCN 层数量	TA-LSTM层数量	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%
1	1	96.72	98.21	88.71	93.22
1	2	88.52	85.42	66.13	74.55
2	1	97.95	98.31	93.55	95.87
2	2	98.77	98.36	96.77	97.56
3	1	95.90	88.24	96.77	92.31
3	2	97.54	95.16	95.16	95.16

Table 7

Table 8

Performances of different models on dataset A

模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%
GCN	89.24	93.55	96.67	95.08
SA-GCN	93.72	95.93	98.33	97.12
LSTM	94.17	95.22	99.58	97.35
TA-LSTM	95.52	98.32	97.50	97.91
GCN-LSTM	96.41	99.57	97.08	98.31
STA-GLN	97.31	99.58	97.92	98.74

Table 8

Table 9

Performances of different models on dataset B

模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%
GCN	88.28	89.50	96.24	92.75
SA-GCN	92.89	96.17	94.62	95.39
LSTM	93.72	93.85	98.39	96.06
TA-LSTM	94.14	97.78	94.62	96.17
GCN-LSTM	94.56	96.76	96.24	96.50
STA-GLN	96.65	97.34	98.39	97.86

Table 9

Table 10

Performances of different models on dataset C

模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%
GCN	78.12	94.74	96.00	95.36
SA-GCN	79.69	94.77	96.67	95.71
LSTM	81.25	95.39	96.67	94.77
TA-LSTM	82.81	96.64	96.00	96.32
GCN-LSTM	84.38	95.45	98.00	96.71
STA-GLN	87.50	96.71	98.80	97.35

Table 10

Table 11

Performances of different models on dataset D

模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%
GCN	90.41	94.08	96.62	95.33
SA-GCN	91.10	94.70	96.62	95.65
LSTM	92.47	95.36	97.30	96.32
TA-LSTM	93.15	95.39	97.97	96.67
GCN-LSTM	94.52	96.67	97.97	97.32
STA-GLN	96.58	98.64	97.97	98.31

Table 11

Table 12

Fig. 7

Table 13

Performances of different models in sample set A

模型	任务1				任务2
模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%	J/%
AdaBoost-STA-GLN	98.77	98.32	97.50	97.91	95.70
AdaBoost-LSTM	97.54	97.28	96.24	96.76	94.55
AdaBoost-CNN	97.31	98.30	96.77	97.53	94.35
RF	96.86	97.83	96.77	97.30	93.67
AdaBoost-DT	96.55	96.28	97.08	96.68	93.38
AdaBoost-GCN	95.82	96.17	94.62	95.39	92.77

Table 13

Table 14

Performances of different models in sample set B

模型	任务1				任务2
模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%	J/%
AdaBoost-STA-GLN	97.51	97.11	97.92	97.51	94.56
AdaBoost-LSTM	96.65	96.64	96.00	96.32	93.27
AdaBoost-CNN	96.58	96.62	95.33	95.97	93.10
RF	95.21	96.03	96.67	96.35	92.57
AdaBoost-DT	94.56	95.39	96.67	94.77	91.35
AdaBoost-GCN	92.47	94.70	96.62	95.65	89.68

Table 14

Table 15

Performances of different models in sample set C

模型	任务1				任务2
模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%	J/%
AdaBoost-STA-GLN	95.52	97.28	96.62	96.95	92.56
AdaBoost-LSTM	92.47	96.17	94.62	95.39	91.40
AdaBoost-CNN	91.78	95.39	96.67	96.03	91.25
RF	91.10	95.57	96.62	96.30	90.77
AdaBoost-DT	89.24	95.33	96.62	95.97	88.78
AdaBoost-GCN	86.55	94.67	95.95	95.30	86.55

Table 15

Table 16

Performances of different models in sample set E

模型	任务1				任务2
模型	$P a c c$ /%	$P r e c$ /%	$R e c$ /%	$F 1$ /%	J/%
AdaBoost-STA-GLN	98.36	98.76	99.58	99.17	95.50
AdaBoost-LSTM	97.31	98.38	97.85	98.11	94.25
AdaBoost-CNN	96.86	98.32	97.50	97.91	94.18
RF	96.41	97.85	97.85	97.85	93.37
AdaBoost-DT	95.40	97.83	96.77	97.30	92.86
AdaBoost-GCN	93.72	96.28	97.08	96.68	91.54

Table 16

Table 17

Fig.8

Fig.9

Fig.10

Table 18

Anti-jamming capability of different ensemble models

评估模型		50 dB噪声样本集		30 dB噪声样本集
评估模型		$P a c c$ /%	J/%	$P a c c$ /%	J/%
1	AdaBoost-STA-GLN	98.20	95.30	97.67	94.75
2	AdaBoost-LSTM	96.85	94.00	95.40	93.10
3	AdaBoost-CNN	96.40	93.75	95.17	92.77
4	RF	95.52	92.46	94.34	91.56
5	AdaBoost-DT	94.17	92.18	93.40	91.49
6	AdaBoost-GCN	92.38	91.34	91.52	90.64

Table 18

Fig.11

References 30

[1]	薛巍, 舒继武, 严剑峰, 等. 基于集群机的大规模电力系统暂态过程并行仿真[J]. 中国电机工程学报, 2003, 23(8):38-43.
	XUE Wei, SHU Jiwu, YAN Jianfeng, et al. Cluster-based parallel simulation for power system transient stability analysis[J]. Proceedings of the CSEE, 2003, 23(8):38-43.
[2]	李亚楼, 李芳, 刘赫川, 等. 基于PSASP的综合能源仿真分析系统[J]. 电网技术, 2019, 43(7):2498-2505.
	LI Yalou, LI Fang, LIU Hechuan, et al. Framework design of simulation system for integrated energy systems based on PSASP[J]. Power System Technology, 2019, 43(7):2498-2505.
[3]	CHANG H D, WU F F, VARIAYA P P. A BCU method for direct analysis of power system transient stability[J]. IEEE Transactions on Power Systems, 1994, 19(2):11-17.
[4]	汪小明, 刘涤尘, 吴军, 等. 基于能量函数法的电网暂态稳定性分析[J]. 电网技术, 2011, 35(8):13-17.
	WANG Xiaoming, LIU Dichen, WU Jun, et al. Energy function-based power system transient stability analysis[J]. Power System Technology, 2011, 35(8):13-17.
[5]	张若愚, 吴俊勇, 李宝琴, 等. 基于迁移学习的电力系统暂态稳定自适应预测[J]. 电网技术, 2020, 44(6):2196-2203.
	ZAHNG Ruoyu, WU Junyong, LI Baoqin, et al. Self-adaptive power system transient stability prediction based on transfer learning[J]. Power System Technology, 2020, 44(6):2196-2203.
[6]	钟智, 管霖, 苏寅生, 等. 基于图注意力深度网络的电力系统暂态稳定评估[J]. 电网技术, 2020, 45(6):2122-2130.
	ZHONG Zhi, GUAN Lin, SU Yinsheng, et al. Power system transient stability assessment based on graph attention deep network[J]. Power System Technology, 2020, 45(6):2122-2130.
[7]	HUANG Jiyu, GUAN Lin, SU Yinsheng, et al. Recurrent graph convolutional network-based multi-task transient stability assessment framework in power system[J]. IEEE Access, 2020, 8: 93283-93296. doi: 10.1109/Access.6287639
[8]	HE H B, GARCIA E A. Learning from imbalanced data[J]. IEEE Transactions on Knowledge and Data Engineering, 2009, 21(9):1263-1284. doi: 10.1109/TKDE.2008.239
[9]	谭本东, 杨军, 赖秋频, 等. 基于改进CGAN的电力系统暂态稳定评估样本增强方法[J]. 电力系统自动化, 2019, 43(1):149-157.
	TAN Bendong, YANG Jun, LAI Qiupin, et al. Data augment method for power system transient stability assessment based on improved conditional generative adversarial network[J]. Automation of Electric Power Systems, 2019, 43(1):149-157.
[10]	田芳, 周孝信, 史东宇, 等. 基于卷积神经网络综合模型和稳态特征量的电力系统暂态稳定评估[J]. 中国电机工程学报, 2019, 39(14):4025-4031.
	TIAN Fang, ZHOU Xiaoxin, SHI Dongyu, et al. Power system transient stability assessment based on comprehensive convolutional neural network model and steady-state features[J]. Proceedings of the CSEE, 2019, 39(14):4025-4031.
[11]	李宝琴, 吴俊勇, 邵美阳, 等. 基于集成深度置信网络的精细化电力系统暂态稳定评估[J]. 电力系统自动化, 2020, 44(6):17-26.
	LI Baoqin, WU Junyong, SHAO Meiyang, et al. Refined transient stability evaluation for power system based on ensemble deep belief network[J]. Automation of Electric Power Systems, 2020, 44(6):17-26.
[12]	邵美阳, 吴俊勇, 李宝琴, 等. 基于两阶段集成深度置信网络的电力系统暂态稳定评估[J]. 电网技术, 2020, 44(5):1776-1787.
	SHAO Meiyang, WU Junyong, LI Baoqin, et al. Transient stability assessment of power system based on two-stage ensemble deep belief network[J]. Power System Technology, 2020, 44(5):1776-1787.
[13]	ZHOU Y Z, GUO Q L, SUN H B, et al. A novel data-driven approach for transient stability prediction of power systems considering the operational variability[J]. International Journal of Electrical Power and Energy Systems, 2019, 107:379-394. doi: 10.1016/j.ijepes.2018.11.031
[14]	李楠, 李保罗, 朱建华, 等. 计及样本不平衡与重叠的暂态稳定评估方法[J]. 电力系统自动化, 2020, 44(21):64-71.
	LI Nan, LI Baoluo, ZHU Jianhua, et al. Transient stability assessment method considering sample imbalance and overlap[J]. Automation of Electric Power Systems, 2020, 44(21):64-71.
[15]	陈振, 肖先勇, 李长松, 等. 基于代价敏感极端学习机的电力系统暂态稳定评估方法[J]. 电力自动化设备, 2016, 36(2):118-123.
	CHEN Zhen, XIAO Xianyong, LI Changsong, et al. Power system transient stability assessment based on cost-sensitive extreme learning machine[J]. Electric Power Automation Equipment, 2016, 36(2):118-123.
[16]	WANG H Y, CHEN Q F, Zhang B H. Transient stability assessment combined model framework based on cost-sensitive method[J]. IET Generation, Transmission & Distribution, 2020, 14(12): 2256-2262. doi: 10.1049/gtd2.v14.12
[17]	王怀远, 陈启凡. 基于代价敏感堆叠变分自动编码器的暂态稳定评估方法[J]. 中国电机工程学报, 2020, 40(7):2213-2220.
	WANG Haiyuan, CHEN Qifan. A transient stability assessment method based on cost-sensitive stacked variational auto-encoder[J]. Proceedings of the CSEE, 2020, 40(7):2213-2220.
[18]	WANG H Y, WANG Q Y, CHEN Q F, el at. Transient stability assessment model with improved cost-sensitive method based on the fault severity[J]. IET Generation, Transmission & Distribution, 2020, 14(20): 4605-4611. doi: 10.1049/gtd2.v14.20
[19]	LI N, LI B L, HAN Y Q, et al. Dual cost-sensitivity factors-based power system transient stability assessment[J]. IET Generation, Transmission & Distribution, 2020, 14(24): 5858-5869. doi: 10.1049/gtd2.v14.24
[20]	ZHANG R, XU Y, DONG Z Y, et al. Post-disturbance transient stability assessment of power systems by a self-adaptive intelligent system[J]. IET Generation, Transmission & Distribution, 2015, 9(3):296-305. doi: 10.1049/gtd2.v9.3
[21]	TAN B D, YANG J, ZHOU T, et al. Spatial-temporal adaptive transient stability assessment for power system under missing data[J]. International Journal of Electrical Power and Energy Systems, 2020, 123, 106237. doi: 10.1016/j.ijepes.2020.106237
[22]	SAUER P W, PAI M A. Power system dynamics and stability[M]. USA: Prentice Hall, 1998.
[23]	王莉丽, 刘崇茹. DIgSILENT在电力系统稳定计算中的应用[J]. 中国科技论文, 2016(11):1283-1287.
	WANG Lili, LIU Chongru. Application of DIgSILENT in power system stability calculation[J]. China Science Paper, 2016(11):1283-1287.
[24]	LAI G K, CHANG W C, YANG Y M, el at. Modeling long- and short-term temporal patterns with deep neural networks[C]// The 41st International ACM SIGIR Conference on Research & Development in Information Retrieval. SIGIR, 2018:95-104.
[25]	刘文慧, 严博文, 吴江, 等. 基于平行控制理论的循环流化床锅炉床温智能预测模型[J]. 综合智慧能源, 2022, 44(3): 50-57. doi: 10.3969/j.issn.2097-0706.2022.03.008
	LIU Wenhui, YAN Bowen, WU Jiang, et al. Intelligent prediction model of CFB boiler bed temperature based on parallel control theory[J]. Integrated Intelligent Energy, 2022, 44(3): 50-57. doi: 10.3969/j.issn.2097-0706.2022.03.008
[26]	周挺, 杨军, 詹祥澎, 等. 一种数据驱动的暂态电压稳定评估方法及其可解释性研究[J]. 电网技术, 2021, 45(11):4416-4425.
	ZHOU Ting, YANG Jun, ZHAN Xiangpeng, et al. Data-driven method and interpretability analysis for transient voltage stability assessment[J]. Power System Technology, 2021, 45(11):4416-4425.
[27]	梁修锐, 刘道伟, 杨红英, 等. 数据驱动的电力系统静态电压稳定态势评估[J]. 电力建设, 2020, 41(1):126-132. doi: 10.3969/j.issn.1000-7229.2020.01.015
	LIANG Xiurui, LIU Daowei, YANG Hongying, et al. Data-driven situation assessment of power system static voltage stability[J]. Electric Power Construction, 2020, 41(1):126-132. doi: 10.3969/j.issn.1000-7229.2020.01.015
[28]	孙黎霞, 彭嘉杰, 田屹昀, 等. 基于混合门控循环单元子层的多任务暂态稳定评估[J]. 电力建设, 2022, 43(2): 63-69. doi: 10.12204/j.issn.1000-7229.2022.02.008
	SUN Lixia, PENG Jiajie, TIAN Yiyun, et al. Multi-task transient stability assessment based on sub-layer of hybrid gated recurrent unit[J]. Electric Power Construction, 2022, 43(2):63-69. doi: 10.12204/j.issn.1000-7229.2022.02.008
[29]	郑真, 朱峰, 马小丽, 等. 基于TL-LSTM的新能源功率短期预测[J]. 综合智慧能源, 2023, 45(1): 41-48. doi: 10.3969/j.issn.2097-0706.2023.01.005
	ZHENG Zhen, ZHU Feng, MA Xiaoli, et al. Short-term new energy power prediction based on TL-LSTM[J]. Integrated Intelligent Energy, 2023, 45(1): 41-48. doi: 10.3969/j.issn.2097-0706.2023.01.005
[30]	PAN S J, YANG Q. A survey on transfer learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2010, 22(10):1345-1359. doi: 10.1109/TKDE.2009.191

样本集	稳定	失稳	总计
A（全接线）	2 436	1 235	3 671
B（N-1断线）	1 517	958	2 475
C（N-2断线）	845	755	1 600

特征类型	维度	描述	输入维度
I	时间	系统各母线的电压幅值与相角	39×2
II	空间	系统拓扑连接关系的拓扑矩阵	39×39

采样序号	采样时间
1	故障前的稳态母线电压幅值与相角
2	故障发生时刻的母线电压幅值与相角
3	故障发生过程中间时刻的母线电压幅值与相角
4	故障切除时刻的母线电压幅值与相角
5	故障切除后1 s末时的母线电压幅值与相角

特征集	描述	输入维度	采样点数	总维度	样本数
1	39条母线的电压幅值与相角	78	5	390	5 000
2	34条传输线的有功、无功功率	68	5	340	5 000
3	发电机母线电压幅值	10	10	100	5 000

数据集	稳定	失稳	总计
A（全接线）	2 436	1 235	3 671
B（N-1断线）	1 517	958	2 475
C（N-2断线）	845	755	1 600
D（分层抽样混合）	1 249	757	2 006