| 引用本文: | 毕贵红,杨 楠,刘大卫,等.基于时频图像组合和DenseNet-CPSAMs的电能质量复合扰动识别[J].电力系统保护与控制,2025,53(17):156-168.[点击复制] |
| BI Guihong,YANG Nan,LIU Dawei,et al.Composite power quality disturbance identification based on time-frequency image fusion and DenseNet-CPSAMs[J].Power System Protection and Control,2025,53(17):156-168[点击复制] |
|
| 摘要: |
| 针对新一代电力系统的电能质量扰动(power quality disturbances, PQDs)识别难题，提出一种改进的自适应噪声完备集合经验模态分解(improved complete ensemble empirical mode decompositiom with adaptive noise, ICEEMDAN)、两种模态时频图组合和DenseNet-CPSAMs深度学习模型结合的PQDs识别新方法。首先，提出ICEEMDAN分解PQDs信号，并重构分量。其次，通过同步提取变换(synchroextracting transform, SET)和S变换(Stockwell transform, ST)生成对应时频图，组合为6通道输入张量。最后，引入DenseNet-CPSAMs深度学习模型，融合了密集连接卷积神经网络(densely connected convolutional networks, DenseNet)、通道注意力机制(channel attention mechanism, CAM)与并行空间注意力机制(parallel spatial attention mechanisms, PSAMs)，实现融合时频图特征深度提取与强化识别。相比于DenseNet-121模型，DenseNet-CPSAMs模型方法在成功减少模型参数6.5 M的同时，在20 dB高信噪比条件下对31类扰动的平均识别率为99.645%，仿真实验表明该方法特征提取能力强、抗噪性能好，并且对复合扰动识别率高。 |
| 关键词: 电能质量扰动 ICEEMDAN 同步提取变换 S变换 DenseNet 深度学习 |
| DOI：10.19783/j.cnki.pspc.241498 |
| 投稿时间：2024-11-08修订日期：2025-02-25 |
| 基金项目:国家自然科学基金项目资助(51767012) |
|
| Composite power quality disturbance identification based on time-frequency image fusion and DenseNet-CPSAMs |
| BI Guihong,YANG Nan,LIU Dawei,YANG Yi,CHEN Dongjing,CHEN Shilong |
| (School of Electric Power Engineering, Kunming University of Science and Technology, Kunming 650500, China) |
| Abstract: |
| To address the challenge of identifying power quality disturbances (PQDs) in new-generation power systems, a novel PQDs identification method is proposed, which combines improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), dual time-frequency image fusion, and a DenseNet-CPSAMs deep learning model. First, ICEEMDAN is utilized to decompose the PQDs signals and reconstruct their components. Second, corresponding time-frequency images are generated through the synchroextracting transform (SET) and Stockwell transform (ST), which are fused into a 6-channel input tensor. Finally, a DenseNet-CPSAMs deep learning model is introduced, integrating densely connected convolutional networks (DenseNet), channel attention mechanisms (CAM), and parallel spatial attention mechanisms (PSAMs), to achieve multi-scale time-frequency feature extraction and the enhanced disturbance recognition. Compared to the DenseNet-121 model, the DenseNet-CPSAMs method reduces model parameters by 6.5 M, while achieving an average recognition rate of 99.645% for 31 disturbance types under a 20 dB high signal-to-noise ratio. Simulation results demonstrate that the proposed method exhibits strong feature extraction capability, high noise resistance, and superior recognition performance for composite disturbances. |
| Key words: power quality disturbance ICEEMDAN synchronous extraction transformation S transform DenseNet deep learning |
