Prediction of Voltage Related Power Quality Values from a Small Renewable Energy Installation

Power from some renewable energy sources can be highly variable and difficult to predict, which in turn can lead to fluctuations in produced power. These fluctuations may lead to a decrease in the power quality, which may be mitigated, or its impact lessened by foreknowledge of its occurrence. In order to relate quality of power produced from small scale renewable energy installations to the meteorological values that drive them, a number of different prediction methods are used. Data from an experimental, small scale, residential installation, which includes photovoltaic and wind power, are used to build the models, which are then used to estimate the values of short and long term flicker severity, as well as the total harmonic distortion of voltage. Forecast horizons of one and ten minutes were considered. Estimates for long term flicker severity and total harmonic distortion of voltage were better predicted than short term flicker severity, which had errors such that the predictions may be useful. The extreme learning machine method consistently performed well, however both it and other methods failed to capture sudden large spikes present in the data.