期刊菜单

Power Quality Disturbance Classification Method Based on Improved Transfer Learning
DOI: 10.12677/SG.2023.133006, PDF, HTML, XML, 下载: 282  浏览: 704

Abstract: Aiming at the classification of power quality disturbance signals, a power quality disturbance classification method based on gram angle field and multiple transfer learning is proposed in this paper. Firstly, the one-dimensional power quality disturbance signal is transformed into GAF coded image by using gram angle field, and then three RESNET sub model networks are constructed. The disturbance signals with representative signal-to-noise ratios of 0 dB, 20 dB and 40 dB are selected as the input of the sub model to train the three sub models respectively. During this period, the training weights of the sub models are transferred in turn by using the method of multiple transfer learning. The pre training weight of the latter model is inherited from the training weight of the previous model, and the weight processing methods of partial freezing and partial fine-tuning are adopted to ensure the optimal training effect of the model. Finally, the features of the three sub models are used to train the full connection layer classifier, and finally a complete power quality disturbance classification model is obtained. Simulation results show that the method has good classification accuracy and anti noise performance, and the proposed model has good robustness and generalization.

1. 引言

2. 格拉姆角场

$G=\left(\begin{array}{cccc}〈{a}_{1},{b}_{1}〉& 〈{a}_{1},{b}_{2}〉& \cdots & 〈{a}_{1},{b}_{n}〉\\ 〈{a}_{2},{b}_{1}〉& 〈{a}_{2},{b}_{2}〉& \cdots & 〈{a}_{2},{b}_{n}〉\\ ⋮& ⋮& \ddots & ⋮\\ 〈{a}_{n},{b}_{1}〉& 〈{a}_{n},{b}_{2}〉& \cdots & 〈{a}_{n},{b}_{n}〉\end{array}\right)$ (1)

${G}_{S}=\left(\begin{array}{ccc}\mathrm{cos}\left({\theta }_{1}+{\theta }_{1}\right)& \cdots & \mathrm{cos}\left({\theta }_{1}+{\theta }_{n}\right)\\ \mathrm{cos}\left({\theta }_{2}+{\theta }_{1}\right)& \cdots & \mathrm{cos}\left({\theta }_{2}+{\theta }_{n}\right)\\ ⋮& \ddots & ⋮\\ \mathrm{cos}\left({\theta }_{n}+{\theta }_{1}\right)& \cdots & \mathrm{cos}\left({\theta }_{n}+{\theta }_{n}\right)\end{array}\right)={\stackrel{˜}{X}}^{T}\stackrel{˜}{X}-{\sqrt{I-{\stackrel{˜}{X}}^{2}}}^{T}\sqrt{I-{\stackrel{˜}{X}}^{2}}$ (2)

${G}_{D}=\left(\begin{array}{ccc}\mathrm{sin}\left({\theta }_{1}-{\theta }_{1}\right)& \cdots & \mathrm{sin}\left({\theta }_{1}-{\theta }_{n}\right)\\ \mathrm{sin}\left({\theta }_{2}-{\theta }_{1}\right)& \cdots & \mathrm{sin}\left({\theta }_{2}-{\theta }_{n}\right)\\ ⋮& \ddots & ⋮\\ \mathrm{sin}\left({\theta }_{n}-{\theta }_{1}\right)& \cdots & \mathrm{sin}\left({\theta }_{n}-{\theta }_{n}\right)\end{array}\right)={\sqrt{I-{\stackrel{˜}{X}}^{2}}}^{T}\cdot \stackrel{˜}{X}-{\stackrel{˜}{X}}^{T}\cdot \sqrt{I-{\stackrel{˜}{X}}^{2}}$ (3)

${G}_{S}$${G}_{D}$ 分别表示格拉姆角和场(Gramian Summation Angular Field, GASF)与格拉姆角差场(Gramian Difference Angular Field, GADF)。 ${G}_{S}$${G}_{D}$ 的差异主要体现在内积的定义上面。为了利用极坐标系下的角度来反映各点自身以及相互之间连接的关系，无论是 ${G}_{S}$ 还是 ${G}_{D}$ 都对内积进行了重新定义：

${〈{x}_{i},{x}_{j}〉}_{S}=\mathrm{cos}\left({\theta }_{1}+{\theta }_{2}\right)={x}_{i}\cdot x{}_{j}-\sqrt{1-{x}_{i}{}^{2}}\cdot \sqrt{1-{x}_{j}{}^{2}}$ (4)

${〈{x}_{i},{x}_{j}〉}_{D}=\mathrm{sin}\left({\theta }_{1}-{\theta }_{2}\right)=\sqrt{1-{x}_{i}{}^{2}}\cdot {x}_{j}-{x}_{i}\sqrt{1-{x}_{j}{}^{2}}$ (5)

3. 改进的迁移学习

3.1. 迁移学习的概念

Figure 1. Process of transfer learning

3.2. 多重迁移学习

Figure 2. Process of multiple transfer learning

4. 基于改进迁移学习的电能质量扰动分类方法与仿真验证

Figure 3. Flow chart of power quality disturbance classification

Table 1. Classification and recognition results of power quality disturbance signals with different SNRs

5. 结论

 [1] 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. [2] 尹柏强, 何怡刚, 朱彦卿. 一种广义S变换及模糊SOM网络的电能质量多扰动检测和识别方法[J]. 中国电机工程学报, 2015, 35(4): 866-872. [3] Chandrasekar, P. and Kamaraj, V. (2010) Detection and Classification of Power Quality Disturbance Waveform Using MRA Based Modified Wavelet Transform and Neural Networks. Journal of Electrical Engineering, 61, 235-240. https://doi.org/10.2478/v10187-010-0033-4 [4] 于志勇, 张卫辉, 王新库, 黄南天, 黄喜旺. 基于GA和ELM的电能质量扰动识别特征选择方法[J]. 电测与仪表, 2016, 53(23): 62-66. [5] 邢建平. 基于多特征量的电能质量复合扰动识别方法研究[J]. 山东电力技术, 2017, 44(4): 16-21. [6] 仇新艳, 李付亮. 基于ITD和K均值聚类的电能质量扰动分析与识别[J]. 电力系统及其自动化学报, 2015, 27(8): 54-59. [7] LeCun, Y., Bottou, L., Bengio, Y. and Haffner, P. (1998) Gradient-Based Learning Applied to Document Recognition. Proceedings of the IEEE, 86, 2278-2324. https://doi.org/10.1109/5.726791 [8] Keogh, E.J. and Pazzani, M.J. (2000) Scaling up Dynamic Time Warping for Datamining Applications. Proceedings of the 6th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Boston, 20-23 August 2000, 285-289. https://doi.org/10.1145/347090.347153 [9] Simonyan, K. and Zisserman, A. (2015) Very Deep Convolutional Networks for Large-Scale Image Recognition. 3rd International Conference on Learning Representations, San Diego, 7-9 May 2015, 1-14. [10] Szegedy, C., Liu, W., Jia, Y., et al. (2015) Going Deeper with Convolutions. 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, 7-12 June 2015, 1-9. https://doi.org/10.1109/CVPR.2015.7298594 [11] 彭宇文, 李瑞, 李沁雪, 杨国荣, 陈晓华, 许海文. 基于小波变换和GA-LSSVM的电能质量扰动识别与分类[J]. 黑龙江电力, 2023, 45(1): 1-9. [12] 杨华勋. 基于麻雀搜索算法优化支持向量机的电能质量扰动分类研究[J]. 红水河, 2023, 42(2): 93-97. [13] 杨金东, 吴万军, 唐立军, 杨子龙. 基于改进型匹配滤波的非稳态电能质量信号分析方法[J]. 电工电能新技术, 2023, 42(2): 30-38. [14] Hinton, G.E., Osindero, S. and The, Y.-W. (2006) A Fast Learning Algorithm for Deep Belief Nets. Neural Computation, 18, 1527-1554. https://doi.org/10.1162/neco.2006.18.7.1527 [15] Krizhevsky, A., Sutskever, I. and Hinton, G. (2012) ImageNet Classification with Deep Convolutional Neural Networks. Advances in Neural Information Processing Systems, 25, 1097-1105. [16] He, K., Zhang, X., Ren, S., et al. (2016) Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, 27-30 June 2016, 770-778. https://doi.org/10.1109/CVPR.2016.90 [17] Schmidhuber, J. (2015) Deep Learning in Neural Networks: An Overview. Neural Networks, 61, 85-117. https://doi.org/10.1016/j.neunet.2014.09.003 [18] 赵隆乾, 路长宝, 张强, 赵健勃, 朱文超. 基于深度学习与SVM的配电网工程数据预测分析模型设计[J]. 电子设计工程, 2023, 31(8):90-94. [19] 王述, 许海棋, 陈中豪, 任佳. 一种基于改进深度迁移学习的睡姿识别算法[J]. 工业控制计算机, 2023, 36(5): 78-80. [20] 吴忠强, 卢雪琴. 基于深度迁移学习和LSTM网络的微电网故障诊断[J]. 计量学报, 2023, 44(4): 582-590. [21] Shuang, F., Jianing, C., Yujian, Y., et al. (2022) A Two-Stage Deep Transfer Learning for Localisation of Forced Oscillations Disturbance Source. International Journal of Electrical Power & Energy Systems, 135, Artıcle ID: 107577. https://doi.org/10.1016/j.ijepes.2021.107577 [22] 马鹏, 樊艳芳. 基于深度迁移学习的小样本智能变电站电力设备部件检测[J]. 电网技术, 2020, 44(3): 1148-1159. [23] Wang, Z. and Oates, T. (2015) Encoding Time Series as Images for Visual Inspection and Classification Using Tiled Convolutional Neural Networks. Workshops at the 29th AAAI Conference on Artificial Intelligence, Austin, 25-30 January 2015, 176-186. [24] Wang, Z. and Oates, T. (2015) Imaging Time-Series to Improve Classification and Imputation. 24th International Joint Conference on Artificial Intelligence, Buenos Aires, 25-31 July 2015, 3939-3946. [25] 梁涛, 陈春宇, 谭建鑫, 井延伟. 基于多方面特征提取和迁移学习的风速预测[J]. 太阳能学报, 2023, 44(4): 132-139. [26] 欧阳福莲, 王俊, 周杭霞. 基于改进迁移学习和多尺度CNN-BiLSTM-Attention的短期电力负荷预测方法[J]. 电力系统保护与控制, 2023, 51(2): 132-140. [27] 陈伟, 何家欢, 裴喜平. 基于相空间重构和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, 2018, 46(14): 87-93.