Abstract
Management services for solar photovoltaic power plants (PVs) have become an important issue in today's era of increasing PV installations. The existence of error-free data is necessary for data-driven PV management; however, real-world data contain an unignorable portion of errors, thereby reducing the reliability of corresponding analyses. A report on the plausible errors within the data is necessary to help PV administrators configure proper data cleaning algorithms. This study provides an error case report derived from an analysis of nearly 2000 PVs distributed throughout Korea. Real error cases are categorized according to their causes and symptoms. The effect of typical errors on the data-driven PV analysis is particularly evaluated for the day-ahead hourly PV power forecast. Errors in the system specification data significantly decrease the forecast accuracy, thereby addressing the impact of human errors. Regression imputation in the time domain demonstrates acceptable results as a simple ad-hoc method in most error situations.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
