Affiliations: State Grid Shanxi Electric Power Research Institute, Taiyuan, Shanxi, China
Correspondence:
[*]
Corresponding author: Huiqiang Zhi, State Grid Shanxi Electric Power Research Institute, Taiyuan, Shanxi 030000, China. E-mail: [email protected].
Abstract: Power quality analysis and governance need the identification of power quality issues. With the use of smart meters and various smart collection devices, more and more power quality data are collected, and the massive data collection brings pressure on communication, storage and computation to the conventional algorithm for identifying and classifying power quality disturbances based on cloud computing. In the paper, a classification algorithm for power quality disturbance identification based on edge computing and fusion model is proposed. The algorithm’s key concept is to compress and sense the power quality signals at the edge side, and then transmit the compressed power quality data to the cloud, which uses an improved Dense-Net and LSTM fusion model to identify and classify the compressed power quality data. Through experiments, it is proved that the method can compress the power quality signal to 70% of the original signal size while satisfying the recognition and data on power quality disturbance categorization accuracy, reducing the communication cost of data transmission, lowering the computational pressure and caching pressure on the cloud, and having certain robustness.
Keywords: Edge computing, multi-scale parallel dense network, power quality
