Earticle

초록

영어

This paper first presents how to create a data stream of solar power generation from the climate archive open to the public as well as the operation logs accumulated for internal use. Then, a daily prediction model is built to forecast the amount of the future electricity generation according to the weather parameters such as wind speed, temperature, insolation, and sunshine hours in Jeju City. For the regression model built upon the identification of linear dependency of most parameters to the generated solar energy, its fitting accuracy is evaluated in terms of absolute residuals, standardized residuals, Cook’s distance, and error probability distribution. The prediction result shows that the absolute residual stays below 30 kwh and the average at 15.8 kwh, while maximum generation reaching 148 kwh. The prediction model will not only evolve with more record collections, possibly averaging out the effect of some abnormal points, but also make it possible for consumers like electric vehicles to select an energy source out of solar, wind, and legacy grid-providing energies.

목차

Abstract
 1. Introduction
 2. Related Work
 3. Data Management
 4. Linear Regression Model
 5. Prediction
 6. Conclusions
 References

키워드

저자

• Junghoon Lee [ Dept. of Computer Science and Statistics, Jeju National University ]
• Jin-hee Ko [ Dept. of Computer Science and Statistics, Jeju National University ]
• Chan Jung Park [ Dept. of Computer Education, Jeju National University ]
• Gyung-Leen Park [ Dept. of Computer Science and Statistics, Jeju National University ] Corresponding author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Control and Automation

• 간기

  월간

• pISSN

  2005-4297

• 수록기간

  2008~2016

• 십진분류

  KDC 505 DDC 605