Impact of Solar Power and Load Demand Forecast Uncertainty on the Optimal Operation of Microgrid

Operation optimization of a microgrid is formulated based on the forecast of renewable energy resources and electricity demand. The forecast error will introduce uncertainties thereby affecting the accuracy and optimality of the solution. In this paper, the impact of this uncertainty is investigated as it receives little attention in the literature. First, an accurate forecasting model for solar irradiance and electricity demand using long short-term memory recurrent neural network and feedforward neural network is developed. To improve the forecasting accuracy, the k-means clustering algorithm is used to partition the datasets into a number of clusters. Second, MDSTool is used to simulate a one-year operation optimization of the microgrid using both the actual and forecasted data. MDSTool is a decision support tool that we developed in our previous work. We find that the forecast errors have a significant impact on the microgrid's annual energy savings.