Failure Diagnosis Method of Photovoltaic Generator Using Support Vector Machine
- PDF / 4,798,597 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 76 Downloads / 259 Views
ORIGINAL ARTICLE
Failure Diagnosis Method of Photovoltaic Generator Using Support Vector Machine Kyeong‑Hee Cho1 · Hyung‑Chul Jo1 · Eung‑sang Kim1 · Hyang‑A. Park1 · June Ho Park2 Received: 24 September 2019 / Revised: 18 February 2020 / Accepted: 13 April 2020 © The Korean Institute of Electrical Engineers 2020
Abstract The capacity of photovoltaic (PV) generators can increase owing to the 4030 policy of the Government of South Korea.. In addition, there has been significant interest in developing a technology for the maintenance of PV generators owing to an increase in the number of outdated PV generators. This paper describes a failure diagnosis method that uses operational data for power generation and solar radiation of PV generators. The measured data stored since four years in an operational 50-kW PV generator that was installed in 2014, were analyzed. The proposed failure diagnosis logic uses support vector machine classification as a failure diagnosis method that can classify normal and failure data. The failure data were processed to be used as the fault diagnosis logic for solar power generators. A new 50-kW PV generator, which contained no fault data, was used for a case study in this paper. Fault data were generated and the operation data of the PV generators were diagnosed by applying the proposed method. In addition, the accuracy was calculated and the results were analyzed. Keywords Photovoltaic (PV) generator · Failure diagnosis · Fault data · Support vector machine (SVM)
1 Introduction Photovoltaic (PV) generators in the Republic of Korea have been in operation since 1992. PV generators have continuously been installed and operated for 19 years. Various failures, such as the degradation of power generation, hot spots, fire, and communication errors, have occurred and have resulted in a lower efficiency. As a result, it is very
* Kyeong‑Hee Cho [email protected] Hyung‑Chul Jo [email protected] Eung‑sang Kim [email protected] Hyang‑A. Park [email protected] June Ho Park [email protected] 1
Digital Energy System Research Center, Korea Electrotechnology Research Institute, Changwon, South Korea
Department of Electrical and Computer Engineering, Pusan National University, Busan, South Korea
2
important to analyze the causes of these failures to improve the utilization of PV generators [1]. According to the power statistics information system from South Korea, the solar power system capacity was 7.2%, or 8.6 GW out of 119 GW, for domestic power generation in 2019. According to the third basic energy plan of the Government of South Korea, renewable energy will increase by 30–35% by 2040 [2]. The government’s policy and interest in the development of a technology for maintenance of PV generators has been increasing owing to the increase in the number of outdated PV generators [3, 4]. Currently, many diagnostic techniques are being developed to detect failures in PV systems. There was a recent study that used stochastic and regression data collected from PV panels that used probabilist
Data Loading...
