Fuzzy System Based Short-term and Very Short-term Solar PV Output Forecasting

  • 周 梓為

Student thesis: Master's Thesis


To enhance the accuracy of the photovoltaic (PV) power output forecasting model for energy management system (EMS) this thesis proposes a one-day ahead short-term and real-time very short-term solar PV output forecasting model based on easier accessible information to solve some problems in practical application The one-day ahead short-term PV output forecast of the proposed forecasting model predicts the PV power output every 15 minutes by using self-organizing map (SOM) and fuzzy inference method The real-time very short-term PV output forecast of the proposed model constantly tunes the short-term predicted values achieved one-day ahead once the actual power output one time-step of 15 minutes before is available Further from the forecast horizontal solar radiation a method is proposed to estimate the actual radiation on the surface of the PV panels to have better forecasting accuracy of PV power generation The proposed method has been tested in a practical 3 kW PV system in Shulin Taiwan and a 499 kW commercial grid-connected PV system in Tainan Taiwan The average short-term forecast error is 4 98% with the average forecast error of 2 67% for real-time very short-term forecast model
Date of Award2016 Jul 26
Original languageEnglish
SupervisorHong-Tzer Yang (Supervisor)

Cite this