Short-term Load Forecasting Model of WGAN Based on GRU Neural Network

doi:10.11985/2022.02.019

Abstract

Abstract:

To improve the accuracy of short-term load forecasting, a short-term load forecasting model of WGAN based on GRU neural network is proposed. The Wasserstein distance is taken as the loss function of GAN, compared with the traditional GAN, it can solve the problems of gradient disappearance and mode collapse in the training process. At the same time, its generator and discriminator model uses GRU neural network to solve the gradient problem in recurrent neural network. Compared with GRU neural network model, traditional GAN model, and GAN model with KL divergence as loss function and GRU as generator and discriminator, it is proved that the new model has better prediction accuracy and stability.

Key words: Short-term load forecasting, GAN model, Wasserstein distance, GRU neural network, power system

CLC Number:

TM715

GAO Ao, WANG Shuai, HAN Xingchen, ZHANG Zhisheng. Short-term Load Forecasting Model of WGAN Based on GRU Neural Network[J]. Journal of Electrical Engineering, 2022, 17(2): 168-175.

Figures/Tables 10

References 28

[1]	郑睿程, 顾洁, 金之俭, 等. 数据驱动与预测误差驱动融合的短期负荷预测输入变量选择方法研究[J]. 中国电机工程学报, 2020, 40(2):487-500.
	ZHENG Ruicheng, GU Jie, JIN Zhijian, et al. Research on short-term load forecasting variable selection based on fusion of data driven method and forecast error driven method[J]. Proceedings of the CSEE, 2020, 40(2):487-500.
[2]	张淑清, 段晓宁, 张立国, 等. TSNE降维可视化分析及飞蛾火焰优化ELM算法在电力负荷预测中应用[J]. 中国电机工程学报, 2021, 41(9):3120-3130.
	ZHANG Shuqing, DUAN Xiaoning, ZHANG Liguo, et al. TSNE dimension reduction visualization analysis and MFOELM algorithm applied in power load forecasting[J]. Proceedings of the CSEE, 2021, 41(9):3120-3130.
[3]	康重庆, 夏清, 张伯明. 电力系统负荷预测研究综述与发展方向的探讨[J]. 电力系统自动化, 2004(17):1-11.
	KANG Chongqing, XIA Qing, ZHANG Boming. Review of power system load forecasting and its development[J]. Automation of Electric Power Systems, 2004(17):1-11.
[4]	廖旎焕, 胡智宏, 马莹莹, 等. 电力系统短期负荷预测方法综述[J]. 电力系统保护与控制, 2011, 39(1):147-152.
	LIAO Nihuan, HU Zhihong, MA Yingying, et al. Review of the short-term load forecasting methods of electric power system[J]. Power System Protection and Control, 2011, 39(1):147-152.
[5]	董骁翀, 孙英云, 蒲天骄. 基于条件生成对抗网络的可再生能源日前场景生成方法[J]. 中国电机工程学报, 2020, 40(17):5527-5536.
	DONG Xiaochong, SUN Yingyun, PU Tianjiao. Day-ahead scenario generation of renewable energy based on conditional GAN[J]. Proceedings of the CSEE, 2020, 40(17):5527-5536.
[6]	吴云, 雷建文, 鲍丽山, 等. 基于改进灰色关联分析与蝙蝠优化神经网络的短期负荷预测[J]. 电力系统自动化, 2018, 42(20):67-72.
	WU Yun, LEI Jianwen, BAO Lishan, et al. Short-term load forecasting based on improved grey relational analysis and neural network optimized by bat algorithm[J]. Automation of Electric Power Systems, 2018, 42(20):67-72.
[7]	彭湃, 刘敏. 基于Prophet-LSTM组合模型的短期负荷预测方法[J]. 电力系统及其自动化学报, 2021, 33(11):15-20.
	PENG Pai, LIU Min. Short-term load forecasting method based on Prophet-LSTM combined model[J]. Proceedings of the CSU-EPSA, 2021, 33(11):15-20.
[8]	王增平, 赵兵, 纪维佳, 等. 基于GRU-NN模型的短期负荷预测方法[J]. 电力系统自动化, 2019, 43(5):53-58.
	WANG Zengping, ZHAO Bing, JI Weijia, et al. Short-term load forecasting method based on GRU-NN model[J]. Automation of Electric Power Systems, 2019, 43(5):53-58.
[9]	赵兵, 王增平, 纪维佳, 等. 基于注意力机制的CNN-GRU短期电力负荷预测方法[J]. 电网技术, 2019, 43(12):4370-4376.
	ZHAO Bing, WANG Zengping, JI Weijia, et al. A short-term power load forecasting method based on attention mechanism of CNN-GRU[J]. Power System Technology, 2019, 43(12):4370-4376.
[10]	ZHANG Jianguang, ZHANG Xuyang, YANG Jianfeng, et al. Deep LSTM and GAN based short-term load forecasting method at the zone level[C]// 2020 International Conference on Artificial Intelligence in Information and Communication (ICAIIC), 2020:613-618.
[11]	ARJOVSKY M, CHINTALA S, BOTTOU L. Wasserstein generative adversarial networks[C]// International Conference on Machine Learning. PMLR, 2017:214-223.
[12]	肖白, 黄钰茹, 姜卓, 等. 数据匮乏场景下采用生成对抗网络的空间负荷预测方法[J]. 中国电机工程学报, 2020, 40(24):7990-8001.
	XIAO Bai, HUANG Yuru, JIANG Zhuo, et al. The method of spatial load forecasting based on the generative adversarial network for data scarcity scenarios[J]. Proceedings of the CSEE, 2020, 40(24):7990-8001.
[13]	姚程文, 杨苹, 刘泽健. 基于CNN-GRU混合神经网络的负荷预测方法[J]. 电网技术, 2020, 44(9):3416-3424.
	YAO Chengwen, YANG Ping, LIU Zejian. Load forecasting method based on CNN-GRU hybrid neural network[J]. Power System Technology, 2020, 44(9):3416-3424.
[14]	梁俊杰, 韦舰晶, 蒋正锋. 生成对抗网络GAN综述[J]. 计算机科学与探索, 2020, 14(1):1-17.
	LIANG Junjie, WEI Jianjing, JIANG Zhengfeng. Generative adversarial networks GAN overview[J]. Journal of Frontiers of Computer Science and Technology, 2020, 14(1):1-17.
[15]	袁培, 王舶仲, 毛文奇, 等. 基于多重生成对抗网络的智能开关设备状态感知与诊断研究[J]. 电力系统保护与控制, 2021, 49(6):67-75.
	YUAN Pei, WANG Bozhong, MAO Wenqi, et al. Research on state perception and diagnosis of intelligent switches based on triple generative adversarial networks[J]. Power System Protection and Control, 2021, 49(6):67-75.
[16]	张雄, 杨琳琳, 上官宏, 等. 基于生成对抗网络和噪声水平估计的低剂量CT图像降噪方法[J]. 电子与信息学报, 2021, 43(8):2404-2413.
	ZHANG Xiong, YANG Linlin, SHANGGUAN Hong, et al. A low-dose CT image denoising method based on generative adversarial network and noise level estimation[J]. Journal of Electronics & Information Technology, 2021, 43(8):2404-2413.
[17]	王德文, 杨凯华. 基于生成式对抗网络的窃电检测数据生成方法[J]. 电网技术, 2020, 44(2):775-782.
	WANG Dewen, YANG Kaihua. A data generation method for electricity theft detection using generative adversarial network[J]. Power System Technology, 2020, 44(2):775-782.
[18]	滕伟, 丁显, 史秉帅, 等. 基于WGAN-GP的风电机组传动链故障诊断[J]. 电力系统自动化, 2021, 45(22):167-173.
	TENG Wei, DING Xian, SHI Bingshuai, et al. Fault diagnosis of wind turbine drivetrain based on Wasserstein generative adversarial network-gradient penalty[J]. Automation of Electric Power Systems, 2021, 45(22):167-173.
[19]	何佳美. 基于生成对抗网络的电力设备图像扩充模型及算法研究[D]. 成都: 电子科技大学, 2020.
	HE Jiamei. Research on image expansion model and algorithm of power equipment based on generative adversarial network[D]. Chengdu: University of Electronic Science and Technology of China, 2020.
[20]	王婷婷, 朱江. 基于差分WGAN的网络安全态势预测[J]. 计算机科学, 2019, 46(S2):433-437.
	WANG Tingting, ZHU Jiang. Network security situation forecast based on differential WGAN[J]. Computer Science, 2019, 46(S2):433-437.
[21]	于雅娜, 李红娇, 李晋国. 差分隐私保护WGAN-GP算法研究[J]. 计算机应用研究, 2021, 38(9):2837-2841.
	YU Yana, LI Hongjiao, LI Jinguo. Research on differential privacy protection for WGAN-GP algorithm[J]. Application Research of Computers, 2021, 38(9):2837-2841.
[22]	廖一帆, 武志刚. 基于迁移学习与Wasserstein生成对抗网络的静态电压稳定临界样本生成方法[J]. 电网技术, 2021, 45(9):3722-3728.
	LIAO Yifan, WU Zhigang. The method to generate static voltage stability critical sample based on transfer learning and wasserstein generative adversarial network[J]. Power System Technology, 2021, 45(9):3722-3728.
[23]	党存禄, 武文成, 李超锋, 等. 基于CatBoost算法的电力短期负荷预测研究[J]. 电气工程学报, 2020, 15(1):76-82.
	DANG Cunlu, WU Wencheng, LI Chaofeng, et al. Short-term load forecasting based on CatBoost algorithm[J]. Journal of Electrical Engineering, 2020, 15(1):76-82.
[24]	王康, 张智晟, 撖奥洋, 等. 基于Bagging的双向GRU集成神经网络短期负荷预测[J]. 电力系统及其自动化学报, 2021, 33(10):24-30.
	WANG Kang, ZHANG Zhisheng, HAN Aoyang, et al. Short-term load forecasting based on bidirectional GRU neural network integrated by Bagging[J]. Proceedings of the CSU-EPSA, 2021, 33(10):24-30.
[25]	谢谦, 董立红, 厍向阳. 基于Attention-GRU的短期电价预测[J]. 电力系统保护与控制, 2020, 48(23):154-160.
	XIE Qian, DONG Lihong, SHE Xiangyang. Short-term electricity price forecasting based on Attention-GRU[J]. Power System Protection and Control, 2020, 48(23):154-160.
[26]	王俊, 李霞, 周昔东, 等. 基于VMD和LSTM的超短期风速预测[J]. 电力系统保护与控制, 2020, 48(11): 45-52.
	WANG Jun, LI Xia, ZHOU Xidong, et al. Ultra-short-term wind speed prediction based on VMD-LSTM[J]. Power System Protection and Control, 2020, 48(11):45-52.
[27]	张智晟, 于道林. 考虑需求响应综合影响因素的RBF-NN短期负荷预测模型[J]. 中国电机工程学报, 2018, 38(6):1631-1638,1899.
	ZHANG Zhisheng, YU Daolin. RBF-NN based short-term load forecasting model considering comprehensive factors affecting demand response[J]. Proceedings of the CSEE, 2018, 38(6):1631-1638,1899.
[28]	孔祥玉, 李闯, 郑锋, 等. 基于经验模态分解与特征相关分析的短期负荷预测方法[J]. 电力系统自动化, 2019, 43(5):46-52.
	KONG Xiangyu, LI Chuang, ZHENG Feng, et al. Short-term load forecasting method based on empirical mode decomposition and feature correlation analysis[J]. Automation of Electric Power Systems, 2019, 43(5):46-52.

输入	学习率α=0.1;权重裁剪参数c=0.01; 批量值m=64;n=5;w为判别器参数; θ为生成器参数
1	while θ do
2	for t=0,1,…,n do
3	样本$\left\{x^{(i)}\right\}_{i=1}^{m} \sim P_{r}$为真实数据
4	样本$\left\{g\left(z^{(i)}\right)\right\}_{i=1}^{m} \sim P_{g}$为生成数据
5	$\mathrm{d} w \leftarrow \nabla w\left[\frac{1}{m} \sum_{i=1}^{m} f_{w}\left(x^{(i)}\right)-\frac{1}{m} \sum_{i=1}^{m} f_{w}\left(g\left(z^{(i)}\right)\right)\right]$ 计算判别器梯度
6	w←w+a·dw判别器参数梯度更新
7	w←clip(w,-c,c) 将判别器参数梯度更新的数值控制在(-c, c)之间
8	end for
9	$\mathrm{d} \theta \leftarrow-\nabla \theta\left[\frac{1}{m} \sum_{i=1}^{m} f_{w}\left(g\left(z^{(i)}\right)\right)\right]$计算生成器梯度
10	θ←θ-dθ生成器参数梯度更新
11	end while

预测模型	E_MAPE(%)	E_max(%)	ME/MW	训练耗时/s
模型1	1.2	3.7	32.67	766
模型2	2.2	4.96	63.27	492
模型3	1.89	3.87	54.69	970
模型4	1.62	3.85	47.76	892

星期	模型1
星期	E_MAPE(%)	E_max(%)	ME/MW
周一	1.87	4..03	53.46
周二	1.32	3.75	39.05
周三	1.28	3.76	37.04
周四	1.25	3.63	36.84
周五	1.23	3.74	35.73
周六	2.05	4.2	60.03
周天	2.0	4.17	58.05
平均值	1.57	3.90	45.74

星期	模型1
星期	E_MAPE(%)	E_max(%)	ME/MW
周一	1.94	3.89	56.81
周二	1.43	3.29	41.29
周三	1.37	3.66	42.53
周四	1.12	3.79	30.16
周五	1.45	3.87	42.64
周六	2.0	4.35	58.67
周日	2.39	4.2	68.15
平均值	1.67	3.86	48.61