Bengang Wei

Mr. Bengang Wei

orcid

Research Focus & Metrics

Academic Career Length

10Years (2017-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
6 May 2020 in Journal of Superconductivity and Novel Magnetism
Verified
Journal Article
Closed access
PDF file
WXWei Xie
Wei Xie
BWBengang Wei
Bengang Wei
ZYZhoufei Yao
Zhoufei Yao

13 June 2022 in Sensors
Verified
Journal Article
Open access
PDF file
WMWeiyun Mao
Weiyun Mao
BWBengang Wei
Bengang Wei
XXXiangyi Xu
Xiangyi Xu
LCLu Chen
Lu Chen
TWTianyi Wu
Tianyi Wu
CRChen Ren
Chen Ren
The fault diagnosis of power transformers is a challenging problem. The massive multisource fault is heterogeneous, the type of fault is undetermined sometimes, and one device has only met a few kinds of faults in the past. We propose a fault diagnosis method based on deep neural networks and a semi-supervised transfer learning framework called adaptive reinforcement (AR) to solve the above limitations. The innovation of this framework consists of its enhancement of the consistency regularization algorithm. The experiments were conducted on real-world 110 kV power transformers’ three-phase fault grounding currents of the iron cores from various devices with four types of faults: Phases A, B, C and ABC to ground. We trained the model on the source domain and then transferred the model to the target domain, which included the unbalanced and undefined fault datasets. The results show that our proposed model reaches over 95% accuracy in classifying the type of fault and outperforms other popular networks. Our AR framework fits target devices’ fault data with fewer dozen epochs than other novel semi-supervised techniques. Combining the deep neural network and the AR framework helps diagnose the power transformers, which lack diagnosis knowledge, with much less training time and reliable accuracy.

29 October 2021 in Electronics
Verified
Journal Article
Open access
PDF file
ZPZhengrui Peng
Zhengrui Peng
XGXinyi Gong
Xinyi Gong
BWBengang Wei
Bengang Wei
XXXiangyi Xu
Xiangyi Xu
SMShixiong Meng
Shixiong Meng
Due to the huge demand for textile production in China, fabric defect detection is particularly attractive. At present, an increasing number of supervised deep-learning methods are being applied in surface defect detection. However, the annotation of datasets in industrial settings often depends on professional inspectors. Moreover, the methods based on supervised learning require a lot of annotation, which consumes a great deal of time and costs. In this paper, an approach based on self-feature comparison (SFC) was employed that accurately located and segmented fabric texture images to find anomalies with unsupervised learning. The SFC architecture contained the self-feature reconstruction module and the self-feature distillation. Accurate fiber anomaly location and segmentation were generated based on these two modules. Compared with the traditional methods that operate in image space, the comparison of feature space can better locate the anomalies of fiber texture surfaces. Evaluations were performed on the three publicly available databases. The results indicated that our method performed well compared with other methods, and had excellent defect detection ability in the collected textile images. In addition, the visual results showed that our results can be used as a pixel-level candidate label.

9 November 2017 in IOP Conference Series: Earth and Environmental Science
Verified
Conference Paper
Open access
PDF file
BWBengang Wei
Bengang Wei
XWX Y Wu
X Y Wu
ZYZ F Yao
Z F Yao
HHH Huang
H Huang
Transformers are essential devices of the power system. The accurate computation of the highest temperature (HST) of a transformer's windings is very significant, as for the HST is a fundamental parameter in controlling the load operation mode and influencing the life time of the insulation. Based on the analysis of the heat transfer processes and the thermal characteristics inside transformers, there is taken into consideration the influence of factors like the sunshine, external wind speed etc. on the oil-immersed transformers. Experimental data and the neural network are used for modeling and protesting of the HST, and furthermore, investigations are conducted on the optimization of the structure and algorithms of neutral network are conducted. Comparison is made between the measured values and calculated values by using the recommended algorithm of IEC60076 and by using the neural network algorithm proposed by the authors; comparison that shows that the value computed with the neural network algorithm approximates better the measured value than the value computed with the algorithm proposed by IEC60076.

1 February 2021 in Energies
Verified
Journal Article
Open access
PDF file
XBXiaoyan Bian
Xiaoyan Bian
YZyao zhang
yao zhang
QZQibin Zhou
Qibin Zhou
TCTing Cao
Ting Cao
BWBengang Wei
Bengang Wei
Building Integrated Photovoltaic (BIPV) modules are a new type of photovoltaic (PV) modules that are widely used in distributed PV stations on the roof of buildings for power generation. Due to the high installation location, BIPV modules suffer from lightning hazard greatly. In order to evaluate the risk of lightning stroke and consequent damage to BIPV modules, the studies on the lightning attachment characteristics and the lightning energy withstand capability are conducted, respectively, based on numerical and experimental methods in this paper. In the study of lightning attachment characteristics, the numerical simulation results show that it is easier for the charges to concentrate on the upper edge of the BIPV metal frame. Therefore, the electric field strength at the upper edge is enhanced to emit upward leaders and attract the lightning downward leaders. The conclusion is verified through the long-gap discharge experiment in a high voltage lab. From the experimental study of multi-discharge in the lab, it is found that the lightning interception efficiency of the BIPV module is improved by 114% compared with the traditional PV modules. In the study of lightning energy withstand capability, a thermoelectric coupling model is established. With this model, the potential, current and temperature can be calculated in the multi-physical field numerical simulation. The results show that the maximum temperature of the metal frame increases by 16.07 °C when 100 kA lightning current flows through it and does not bring any damage to the PV modules. The numerical results have a good consistency with the experimental study results obtained from the 100 kA impulse current experiment in the lab.

25 December 2020
Verified
Conference Paper
Closed access
WLWenhe Li
Wenhe Li
Yubo FanYubo Fan
Yubo Fan
AJAnfeng Jiang
Anfeng Jiang
ZLZhe Liu
Zhe Liu
APAiqiang Pan
Aiqiang Pan
BWBengang Wei
Bengang Wei
The existing single machine working mode of live working robots cannot meet the computational power requirements of artificial intelligence high-precision algorithm, and it is difficult to adapt to the complex working scene, and cannot be applied to the scene where a large number of robots work at the same time to generate massive data management and control. In view of the existing shortcomings, this paper analyzes and compares the working modes of existing power grid inspection robots and live working robots, and proposes an intelligent management and control method for live working robots based on cloud and edge computing terminals. By deploying different parts of deep learning algorithm in cloud and edge computing terminal, the problems existing in the previous working mode are solved. A typical example of live working robot scene in substation is analyzed. The results show that the intelligent control method proposed in this paper can meet the computational power requirements of artificial intelligence deep learning, neural network coordinated control, recognition and control model self-learning, self-renewal and other high-precision algorithm, and can be applied to the scene of complex environment or a large number of robots working at the same time to produce massive data control.