Machine learning enabled reduced-order scenario generation for stochastic analysis of solar power forecasts

Abstract

With increased reliance on solar-based energy generation in modern power systems, the problem of managing uncertainty in power system operation becomes crucial. However, in order to properly capture the uncertainty spread of the power forecast time series along with all its statistical properties, a large number of scenarios are normally required to be simulated at significant computational cost. This work presents a novel and efficient method to generate statistically accurate scenarios from probabilistic forecasts and a method based on unsupervised machine learning to reduce the number of scenarios and speed up the computations, while preserving the statistical properties of the original set. Through a systematic parametric study, an optimum clustering-based machine learning method and its associated parameters are derived. This approach yields statistically equivalent characteristics as a full set with a substantially reduced cardinality (from 7000 to 20). The reduced set of scenarios also preserves the temporal correlation, which is imperative in time-series data and complies with the non-parametric distribution of power obtained from a probabilistic forecast at any particular time. Applying the optimal algorithm to the benchmark RTS-GMLC and the actual California ISO yearly solar production data, it is shown that the uncertainty in the estimation of the statistical moments is reduced to less than 2% and 4.5% of the respective daily peak power values.