We propose an ensemble short-term wind power forecasting model that is based on our novel approaches and advanced forecasting techniques in the modern literature. The performance of our model has been verified by forecasting wind power up to 48 hours ahead at seven wind farms for one and a half years. Our model ranked fourth in the Power and Energy Society (PES) wind power forecasting competition. The forecasting model uses 52 Neural Network (NN) sub-models and five Gaussian Process (GP) sub-models in parallel. For 48 hours, the NN sub-models forecast the future wind power based on historical wind power data and forecasted wind information. In parallel, for the first five hours, five GP sub-models are used to forecast wind power using only historical wind power in order to provide accurate wind power forecasts to NN sub-models. These models provide various forecasts for the same hour, so the optimal forecast should be decided from overlapped forecasts by the decision process.

With increasing demand for energy, the penetration of alternative sources such as renewable energy in power grids has increased. Solar energy is one of the most common and well-known sources of energy in existing networks. But because of its non-stationary and non-linear characteristics, it needs to predict solar irradiance to provide more reliable Photovoltaic (PV) plants and manage the power of supply and demand. Although there are various methods to predict the solar irradiance. This paper gives the overview of recent studies with focus on solar irradiance forecasting with ensemble methods which are divided into two main categories: competitive and cooperative ensemble forecasting. In addition, parameter diversity and data diversity are considered as competitive ensemble forecasting and also preprocessing and post-processing are as cooperative ensemble forecasting. All these ensemble forecasting methods are investigated in this study. In the end, the conclusion has been drawn and the recommendations for future studies have been discussed.

Load and wind power scenarios are synthesized through the generalized dynamic factor model (GDFM), which represents the load and wind power as the sum of a periodic component, idiosyncratic noise component, and common component, where the GDFM preserves the correlation structure between load and wind. The common component consists of the dynamic shock, which is white noise, and the matrix polynomial, which represents the temporal and geographical correlation between load and wind power. Since the dimension of dynamic shocks is less than that of actual load and wind power, the GDFM requires fewer dimensions and variables than multivariate time series models. Scenarios are verified through statistical, spectral density, and correlation analysis. The usefulness of scenarios is also verified by calculating the total generation and transmission upgrade costs on the IEEE 300-bus benchmark. Using correlated scenarios results in higher generation and upgrade costs than using uncorrelated or weakly correlated scenarios. Therefore, correlated scenarios should be used in order to more accurately estimate power system planning costs.

We build an advanced offer curve in the day-ahead electricity market for wind power producers based on the multivariate (bivariate or two-dimensional) distribution of the real-time (RT) price and wind power forecasting errors. The bivariate distribution results in more profitable offers based on the conditional probability of wind power forecasting errors with respect to the RT price forecasting errors. The standard deviation of the bivariate distribution is reduced through a novel probabilistic wind power forecasting algorithm based on a parametric approach to further increase the profitability of the offer curve. In conclusion, the test of the advanced offer curve on the data sampled from the Iberian peninsula shows that the offer curve of the bivariate distribution has a higher profitability than offer curves of the marginal distribution.

We propose an optimal operation strategy for an electric vehicle (EV) aggregator (AGG), which performs energy arbitrage in the energy market and provides ancillary services from aggregated EVs, while providing charging services to EVs to maximize the profit in a future energy market. We group EV batteries as several virtual batteries (VB) with respect to their departure time and stages to implement multiple schedules simultaneously in the multi-stage stochastic optimization (MSSO). They are continuously grouped and regrouped as EVs arrive and depart. We predict VB scenario trees of stepwise EV driving routes and optimize the decisions in responding to uncertain EV movements. The VB states change as EVs enter and exit through the stages, so we should indirectly track time-varying VB characteristics at each stage. Then, we distribute the regulation bids for the VBs to the individual EVs. To reduce mismatches between the bidding amounts for VBs and actual transacted amounts for EVs, we suggest a novel binary progressive hedging algorithm to quickly determine the VB operations. Based on data from historical vehicle statistics, we verify that our strategy with the MSSO model provides a higher profit to the AGG than two-stage stochastic models.

We propose a two-stage probabilistic solar power (SP) forecasting algorithm to utilize the solar irradiation (SI) observations measured from different locations. In the first stage, we predict the SI based on the numerical weather prediction (NWP) after interpolating SI observations. Since the SI on the target location is not measured, we interpolate it using the spatio-temporal Kriging technique based on the SI observed from nearby weather stations. In the second stage, we forecast the SP based on the SI predictions after training the SI and SP observations. The model is trained by observations, but it forecasts based on predictions. Furthermore, in the two-stage model, forecasting errors can propagate across stages. We overcome these problems by using probabilistic forecasting. We design distributions of SI predictions through the probabilistic graphical model. Then, we extract SI scenarios from the distributions and predict SP scenarios based on these SI scenarios. We also group the NWP with respect to its prediction time, and we subdivide these groups as subgroups with respect to weather conditions. Furthermore, we propose a changeable ensemble model, where we have different weights for each weather condition. We verify our algorithm based on the data from the Korea power exchange renewable energy forecasting competition 2019. We finished the competition in 2nd place among a few hundred participants.