| 【Technical Introduction】
As more and more renewable energy sources are integrated into the power grid, their intermittency and uncertainty can lead to power flow fluctuation. This would increase electric stress on the equipment in transmission and distribution substation, so as to increase the risk of unpredictable outages. Therefore, NARI has developed a real-time intelligent monitoring and diagnostic system for power transformers using AI and big data technology. The items analyzed by the system include partial discharge, cooling system, core/winding vibration, ground current, oil/winding temperature, on-load tap changer, dissolved gas analysis, and so on.
-
The oil/winding temperature baseline model is established on the base of improved k-nearest neighbor (k-NN) algorithm, which uses the inverse distance weighting in multiple dimensions to improve the prediction accuracy.
-
In the partial discharge part, a large number of historical Phase-Resolved Partial Discharge (PRPD) graphs are used to establish a diagnostic model by Generative Adversarial Network (GAN) and Convolutional Neural Network (CNN).
-
The On-Load Tap-Changer (OLTC) is monitored online for a long term to acquire its instantaneous vibration and electrical signals during switching. Furthermore, numerous analyses, such as waveform envelope, energy density statistics, spectrograph, and cepstrum, are applied to develop its pattern recognition model.
-
In dissolved gas analysis (DGA), changes in the carbon oxides ratio are adopted as an auxiliary index for the traditional Duval analyses in specifically identifying the severity of winding carbonization.
Currently, the system has been installed in an operating substation of Taipower Company to perform online monitoring and diagnosis for the 200MVA power transformer.
-
User interface of the intelligent monitoring and diagnostic system for power transformers.
-
The actual execution status of the system baseline model
The oil/winding temperature baseline model
PD diagnostic model
OLTC diagnostic model
DGA auxiliary index
【Project Planning/Technical Applications】
NARI has been working on the R&D project of “development of intelligent management and efficiency enhancement for distribution system with renewable energy,” since 2021. One of the project goals is realistically running this system in an operating substation of Taipower Company to help improve the stability of the power grid. In addition, NARI also plans to apply the invented techniques to the diagnosis and prognosis of critical equipment in power plants to enhance their operational safety, reliability, and availability.
【Future Development】
NARI continues to develop substation expert decision-making modules in assessing health indices, failure risk, and maintenance priority; and help well manage the remaining life of all critical equipment. Through this comprehensive system, the predictive maintenance and asset management could be realized in substations, so their unpredictable outages would be diminished and the policy goal of stabilizing national power supply achieved.
【Contact Information】
Name：Chen, Chang-Kuo
Tel：886-3-4711400 ext.6255
E-mail： changkuochen@nari.org.tw |
