Anomaly Detection Method of Electrical Power Consumption Based on Deep Autoencoder

doi:10.16182/j.issn1004731x.joss.22-FZ0931

Abstract

Abstract:

Aiming at the nonlinear and non-stationary characteristics of electrical power consumption data, an abnormal electrical power consumption detection model based on deep autoencoder is proposed. Gated recurrent unit (GRU) network of the deep learning is combined with autoencoder structure, and the encoder and decoder parts of traditional autoencoder are realized by gated recurrent unit network, which gives full play to the data feature extraction capability of gated recurrent unit and the data reconstruction function of autoencoder structure. Based on the reconstruction error between original data and reconstructed data, abnormal data points of the electrical power consumption are detected. By applying the proposed method to actual workshop electrical power consumption data set, it is shown that the proposed method can detect the abnormal points of power consumption data, and the detection effect is better.

Key words: anomaly detection of energy consumption, gated recurrent unit, autoencoder, deep autoencoder, reconstruction error

CLC Number:

TP391.9

Ningke Sun, Yan Wang, Zhicheng Ji. Anomaly Detection Method of Electrical Power Consumption Based on Deep Autoencoder[J]. Journal of System Simulation, 2022, 34(12): 2557-2565.

Figures/Tables 13

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Table 1

Fig. 8

Fig. 9

Fig. 10

Table 2

Table 3

References 19

1	江航, 卢暾, 顾寒苏, 等. 一种动态实时高校建筑能耗异常检测方法[J]. 计算机工程, 2017, 43(4): 15-20, 27.
	Jiang Hang, Lu Tun, Gu Hansu, et al. A Dynamic and Real-time Outlier Detection Method for Energy Consumption of Campus Building[J]. Computer Engineering, 2017, 43(4): 15-20, 27.
2	代德宇, 何江涛, 吴中元. 基于PSO-BP算法的铜管生产过程能耗异常检测模型[J]. 机电工程技术, 2016, 45(9): 128-132.
	Dai Deyu, He Jiangtao, Wu Zhongyuan. Copper Tube Production Process Energy Consumption of Anomaly Detection Model Based on PSO-BP Algorithm[J]. Mechanical & Electrical Engineering Technology, 2016, 45(9): 128-132.
3	杨海东, 江海昌, 方华, 等. 基于GA-SVR的挤压机能耗异常检测模型研究[J]. 机床与液压, 2019, 47(5): 163-168.
	Yang Haidong, Jiang Haichang, Fang Hua, et al. Study on Abnormal Energy Consumption Model of Extruder Based on GA-SVR[J]. Machine Tool & Hydraulics, 2019, 47(5): 163-168.
4	段中兴, 梅思雨. 基于数据挖掘的建筑能耗异常检测研究[J]. 计算机测量与控制, 2020, 28(7): 253-259.
	Duan Zhongxing, Mei Siyu. Research on Abnormal Detection of Building Energy Consumption Based on Data Mining[J]. Computer Measurement & Control, 2020, 28(7): 253-259.
5	Zhao Z, Zhang Y, Zhu X, et al. Research on Time Series Anomaly Detection Algorithm and Application[C]//2019 IEEE 4th Advanced Information Technology, Electronic and Automation Control Conference (IAEAC). Chengdu, China: IEEE, 2019: 16-20.
6	EI-Fiqi H, Wang M, Kasmarik K, et al. Weighted Gate Layer Autoencoders[J]. IEEE Transactions on Cybernetics(S2168-2275), 2022, 52(8): 7242-7253.
7	爨莹, 吴越. 基于Seq2Seq深度自编码器的时间序列异常检测方法研究[J]. 现代电子技术, 2022, 45(2): 26-30.
	Ying Cuan, Wu Yue. Method of Time Seriesanomaly Detection Based on Seq2Seq Depth Autoencoder[J]. Modern Electronics Technique, 2022, 45(2): 26-30.
8	Provotar O I, Linder Y M, Veres M M. Unsupervised Anomaly Detection in Time Series Using LSTM-Based Autoencoders[C]//2019 IEEE International Conference on Advanced Trends in Information Theory (ATIT). Kyiv, Ukraine: IEEE, 2019: 513-517.
9	陈磊, 秦凯, 郝矿荣. 基于集成LSTM-AE的时间序列异常检测方法[J]. 华中科技大学学报(自然科学版), 2021, 49(11): 35-40.
	Chen Lei, Qin Kai, Hao Kuangrong. Time Series Anomaly Detection Method Based on Integrated LSTM-AE[J]. Journal of Huazhong University of Science and Technology(Natural Science Edition), 2021, 49(11): 35-40.
10	Li Y L, Jiang J R. Anomaly Detection for Non-Stationary and Non-Periodic Univariate Time Series[C]//2020 IEEE Eurasia Conference on IOT, Communication and Engineering (ECICE). Yunlin, Taiwan: IEEE, 2020: 177-179.
11	Mansouri N, Lachiri Z. Laughter Synthesis: A Comparison between Variational Autoencoder and Autoencoder[C]//2020 5th International Conference on Advanced Technologies for Signal and Image Processing (ATSIP). Sousse, Tunisia: IEEE, 2020: 1-6.
12	Barreto F, Yadav S, Patnaik S, et al. SIFT Features for Deep and Variational Autoencoders: A Performance Comparison[C]//2020 2nd International Conference on Advances in Computing, Communication Control and Networking (ICACCCN). Greater Noida, India: IEEE, 2020: 652-655.
13	Xu X, Zhao H, Liu H, et al. LSTM-GAN-XGBOOST Based Anomaly Detection Algorithm for Time Series Data[C]//2020 11th International Conference on Prognostics and System Health Management (PHM-2020 Jinan). Jinan, China: IEEE, 2020: 334-339.
14	Zhu G, Zhao H, Liu H, et al. A Novel LSTM-GAN Algorithm for Time Series Anomaly Detection[C]//2019 Prognostics and System Health Management Conference (PHM-Qingdao). Qingdao, China: IEEE, 2019: 1-6.
15	Zhai N, Yao P, Zhou X. Multivariate Time Series Forecast in Industrial Process Based on XGBoost and GRU[C]//2020 IEEE 9th Joint International Information Technology and Artificial Intelligence Conference (ITAIC). Chongqing, China: IEEE, 2020: 1397-1400.
16	Yang S, Yu X, Zhou Y. LSTM and GRU Neural Network Performance Comparison Study: Taking Yelp Review Dataset as an Example[C]//2020 International Workshop on Electronic Communication and Artificial Intelligence (IWECAI). Shanghai, China: IEEE, 2020: 98-101.
17	高德亮. 基于AdamW算法的WT-GRU模型在高频股指预测中的应用[D]. 济南: 山东大学, 2021.
	Gao Deliang. Application of WT-GRU Model Based on AdamW Algorithm in High Frequency Stock Index Forecasting[D]. Ji'nan: Shandong University, 2021.
18	Yu K, Long Z. A TCP Friendly Rate Control Algorithm Based on GRU Prediction Model[C]//2021 Asia-Pacific Conference on Communications Technology and Computer Science (ACCTCS). Shenyang, China: IEEE, 2021: 21-24.
19	Xu Q, Wu Z, Yang Y, et al. The Difference Learning of Hidden Layer Between Autoencoder and Variational Autoencoder[C]//2017 29th Chinese Control and Decision Conference (CCDC). Chongqing, China: IEEE, 2017: 4801-4804.

参数名称	参数设置
optimizer	Adam
dropout rate	0.2
learning rate	0.001
epoch	200
batch size	24

检测模型	P	R	F₁
传统RNN自编码器	0.948	0.891	0.918
LSTM深度自编码器	0.970	0.951	0.961
GRU深度自编码器	0.973	0.953	0.963

检测模型	平均运行时间
传统RNN自编码器	157.63
LSTM深度自编码器	474.66
GRU深度自编码器	204.95

[1]	Nie Kai, Zeng Kejun, Meng Qinghai. Uncertainty Simulation Method Based on Deep Bayesian Networks Learning [J]. Journal of System Simulation, 2022, 34(1): 79-85.
[2]	Lin Shuo, An Lei, Gao Zhijun, Shan Dan, Shang Wenli. Research on Intrusion Detection Based on Stacked Autoencoder and Long-short Memory [J]. Journal of System Simulation, 2021, 33(6): 1288-1296.
[3]	Wu Dinghui, Zhu Zhichao, Han Xinhong. Fault Detection of Wind Turbine Bearing Based on BO-SDAE Multi-source Signal [J]. Journal of System Simulation, 2021, 33(5): 1148-1156.
[4]	Dong Shuqin, Zhang Bin. Network Traffic Anomaly Detection Method for Imbalanced Data [J]. Journal of System Simulation, 2021, 33(3): 679-689.
[5]	Guo Yecai, Zhang Haoran. Modulation Recognition Algorithm Based on Improved LDA and Autoencoders [J]. Journal of System Simulation, 2021, 33(2): 494-500.
[6]	Liu Fang, Lu Lixia, Wang Hongjuan, Wang Xin. Image Classification Based on Sparse Autoencoder and Support Vector Machine [J]. Journal of System Simulation, 2018, 30(8): 3007-3014.
[7]	Gu Congcong, Wang Yan, Yan Dahu, Ji Zhicheng. Research on Classification Based on Autoencoder Combination Features Extraction Method [J]. Journal of System Simulation, 2018, 30(11): 4132-4140.
[8]	Jiang Mingxin, Pan Zhigeng, WangLanfang, Hu Taoxin. Visual Object Tracking Algorithm Based on Deep Denoising Autoencoder over RGB-D Data [J]. Journal of System Simulation, 2018, 30(11): 4276-4283.