A multivariate time series forecast model for wind and storage integrated system operation

Abstract

A multivariate time series wind power forecast model, dynamic factor model (DFM), is proposed to tackle a stochastic dynamic optimal power flow (OPF) problem integrated with wind energy and storage devices in this paper. The uncertainty of wind power forecasting is represented by a set of scenarios generated by the DFM. As opposed to other common methods of generating scenarios, such as Monto Carlo simulation, the DFM considers the spatial correlations among wind farms, which manifests a more accurate and real situation. DFM models the wind power as a multivariate stochastic process, which can be decomposed as the multiplication of fixed polynomial matrices and dynamic shocks. Such dynamic shocks are spatially and temporally uncorrelated white noises which in turn become the impulses to generate forecast scenarios. The stochastic optimization is to minimize the total expected generating cost over all scenarios in one day ahead. Storage devices are installed to compensate for the operation cost and power deficiency. The proposed forecast model is verified in both time and frequency domain, and stochastic optimization model is tested over modified IEEE 30-bus system.