Day-Ahead Solar Generation Prediction from Weather Forecasts in the UK Using LSR-Fuzzy-Markov Chain

Day-ahead forecast is a necessity for optimal management of peer-to-peer energy trading in an “energy community”, which is proposed as the foundation block of the future energy market. This paper proposes a hybrid technique that consists of (1) a linear day-ahead PV generation forecast using weather predictions and historical measurements, and (2) a combined model of fuzzy classifier and Markov Chain to improve the forecast error. The final LSR-Fuzzy-Markov technique shows superiority when tested on two-years data from the Active Office Building in Swansea University, UK.