A Very Short-Term Energy Forecasting Technique for Small Scale Photovoltaic Systems using k-Nearest Neighbour Algorithm

The field of photovoltaic (PV) forecasting has beenevolving rapidly in the recent years. This paper provides a veryshort-term forecasting technique on energy harvested from smallscale PV systems. It makes use of a supervised machine learningtechnique, k-nearest neighbours (kNN), to provide PV owners 5easily-comprehensible output levels of Very Low, Low, Medium,High and Very High. This proposed technique uses readilyavailable data from inverters, namely time of the day andinstantaneous power, and data from commonly used additionalweather measurement equipment namely outdoor temperature,panel temperature and on-site irradiance. The proposed techniquetargets very short-term forecasting period of 15 minutes ahead,which is sufficient for building owners to activate alternatives suchas powering up backup generator or switching off non-criticalloads to reduce load demand. The short-term forecasting resultsare useful in small localized areas where the weather changes veryquickly. Its results can be passed to smart energy managementsystem to aid in their decision makings. Historical data of anexisting 30kWp PV system located in Singapore is used to evaluatethe accuracy of the kNN short term forecasting technique. Despitethe lack of expensive and complicated resources such as numericalweather prediction models and satellite and sky imageryobservations of clouds, the proposed technique achieved anacceptable accuracy of over 68%. The paper compares anddiscusses the parameters and the number of neighbours to be usedin the technique