| Title |
On-board Switchboard Diagnosis System based on AI Algorithm and High-frequency Current Conversion Sensor |
| Authors |
박준용(Jun-Yong Park) ; 김영일(Il-Young Kim) ; 이동준(Dong-Jun Lee) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.12.1881 |
| Keywords |
Partial discharge; Switchboard; Ai algorithm; Diagnosis system; Pre-processing |
| Abstract |
This paper presents an on-board switchboard diagnosis system based on the Ai algorithm for the purpose of establishing a real-time partial discharge monitoring diagnosis system. For classifying pattern of partial discharge, FCM-based RBFNN is applied to on-board diagnosis system and K-means, NN(Neural Networks) are used for comparative analysis. The data are obtained by HFCT(High Frequency Current Transformer) sensor from designed partial discharge simulation environment and its phase is divided into 128 degree. Commonly used partial discharge data analysis methods such as statistical analysis, PRPS(Phase Resolved Pulse Sequence) and PRPD(Phase Resolved Partial Discharge) are introduced. In this paper PRPDA(Phase Resolved Partial Discharge Analysis) and PCA(Principal Component Analysis) are used as a pre-processing. The Ai algorithm comparison is performed twice in total. Firstly, a comparison analysis of the test data validation of each model trained in the Python environment is conducted through a confusion matrix. Afterwards, an on-board diagnostic device is added to the partial discharge simulation circuit, and the judgment results for the actual operation are compared. Through comparative analysis in virtual and real environments, it is confirmed that the case in which FCM RBFNN is mounted shows excellent performance. |