ShuffleNet‐based comprehensive diagnosis for insulation and mechanical faults of power equipment

Abstract

Despite the complicated fault mechanism of power equipment, with the increasing promotion and development of the Ubiquitous Power Internet of Things (UPIoT), the fault information of power equipment can be instantly saved, which makes possible intelligent diagnosis via fault samples. This study proposes a new method to implement comprehensive intelligent diagnosis by adopting the ShuffleNet lightweight convolution neural network (SLCNN). Considering the requirements of the UPIoT intelligent terminal, this study constructs six models, which are measured in terms of recognition accuracy, model storage, and calculation cost when applied to insulation and mechanical datasets. Compared with the existing models, the SLCNN outperforms them significantly in terms of recognition accuracy, with an accuracy of 95.77% and 99.9% in insulation and mechanical fault diagnoses, respectively. The SLCNN also demonstrates obvious advantages in other performance indicators, all of which contribute to its use in the accurate and reliable fault diagnosis of power equipment in the UPIoT context. Furthermore, through comparing feature maps, it is discovered that the aliasing degree and boundary of insulation defects are not as obvious as those of mechanical faults, which means that insulation fault diagnosis is much more difficult than mechanical fault diagnosis.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

A novel meta‐learning network for partial discharge source localization in gas‐insulated switchgear via digital twin

“…Through virtual model, representative and comprehensive samples representing GIS PD localization can be obtained to build a simulation database, so as to guide the implementation of the GIS PD localization method. To simulate the propagation process of UHF signal in the GIS cavity and build a simulation database, this paper adopts xfdtd [20] to build a 3D virtual model of GIS with a ratio of 1:1. A large number of studies have verified the effectiveness of using fdtd simulation to assist in the study of GIS UHF signal propagation, localization and diagnosis [11,15,[21][22][23][24].…”

SR-POSE: A Novel Non-Contact Real-Time Rehabilitation Evaluation Method Using Lightweight Technology

“…After inputting the image, it undergoes a convolution operation and maximum pooling operation, and then passes through three stages (Stage2, Stage3, and Stage4), which are stacked by varying numbers of shuffle units [33]. In each stage, the first module employs a shuffle unit with a stride of 2 (as displayed in Figure 1(b) Shuffle unit2 [34][35][36]) to realize the down-sampling function. Stage2 and Stage4 are composed of a shuffle unit with a stride=2 and three shuffle units with a stride=1(as displayed in Figure 1(b) Shuffle unit1).…”

A novel adversarial transfer learning in deep convolutional neural network for intelligent diagnosis of gas‐insulated switchgear insulation defect

“…We select four types of typical defects: free metal particles (0 type defect), metal tips (1 type defect), floating electrodes (2 type defect) and insulator air gaps (3 type defect) for GIS insulation defect diagnosis. In [27,28], massive GIS PD samples were obtained through the finite-difference time-domain (FDTD), and high-precision diagnosis of GIS PD under simulation data was realised through CNN. Therefore, we use the FDTD method to obtain massive source domain samples.…”