Statistical Analysis of PV Power Forecast Error

PV power forecast error usually has been assumed to have a near Gaussian distribution. With a simple statistical analysis, it can be shown that this is neither valid nor accurate. To obtain a more appropriate probability density function (pdf) of the PV power forecast error, an approach based on statistical analysis is proposed. An artificial neutral network (ANN) power forecasting model based on digital weather forecast is used to generate the forecast data. Three different forecast scenarios are simulated and a statistical analysis model of PV power forecast error is proposed. A principle is proposed to verify its validity. A case is studied and it is found that the forecast error pdf for each bin has different shapes, and the model is valid to describe the characteristic of PV power forecast error. Result of this case shows the usefulness of the proposed approach in spinning reserve scheduling.