A multivariate statistical method for risk parameter scenario generation and renewable energy bidding in electricity markets

To maximize the expected profits and manage the risks of renewable energy system under electricity market environment, scenario-based- stochastic optimization model can be established to generate energy bidding strategies, in which the probabilistic scenarios of risk parameters are usually obtained by using statistical or machine learning methods. This paper proposes a practical multivariate statistical method for risk parameter scenario generation, which is used by a wind energy system faced with uncertain electricity prices and wind power productions, and it considers the correlation between dependent risk parameters by using historical data directly. The probabilities of scenarios containing correlated risk parameters are calculated by using multivariate histograms, in which the asymmetric correlation between different parameters existing in the historical data are preserved. Additionally, in order to make the stochastic optimization problem with large numbers of scenarios tractable, a multivariate scenario reduction method is used to trim down the scenario number. By solving the stochastic optimization problem, optimal day-ahead bidding curves for the wind energy system are generated, and Douglas-Peucker algorithm is used to fit the bidding curves according to market requirements. Case studies based on real world data in electricity markets are performed to prove the effectiveness of the proposed risk parameter scenario generation method and energy bidding strategies. Finally, conclusions and practical suggestions on future research works are provided.