A Data-driven Bidding Strategy of a Wind-Diesel-Electrical Storage Hybrid System in a day-ahead Electricity Market

Variability of renewable energy sources (RES) creates great challenges for their owners and power system operator. In recent years energy storage system is employed to mitigate the fluctuation of RESs. This paper addresses optimal decision-making of the wind-diesel-storage system in the day-ahead electricity market with a data-driven approach. The problem is formulated as MILP to maximize the profit. Uncertain parameters of the day-ahead electricity market and wind speed are forecasted by a deep bidirectional gated recurrent neural network (DBGRUNN). The proposed model is compared with LSTM, GRU, CNN, and Conv models and different evaluation criteria, as well as the mean absolute percentage error (MAPE), the root, mean square error (RMSE), and the mean absolute error (MAE) are used to measure accuracy. The presented model results on a Ferrol 1 Tide Gauge station database at the Spain pool market show that the proposed model performs better than the comparative models.

I. Introduction

Wind power plants (WPPs) have a great potential for energy production. Operation and management of variable wind energy generation pose challenges to keeping a balance between production and consumption in power systems. Erratic behavior of wind power plants increases power system reserve to compensate variable generation [1]. Recently, energy storage technologies are widely used to balance the deviation of wind power plants. Energy storage leads to high reliability and increases efficiency in power systems. Also, they can produce energy and capacity reserve for power system operators [2]. Small/medium scall dispatchable distributed generation (DG) are not big enough to directly participate in wholesale electricity markets. A coordinated operation can facilitate market access to small-scale dispatchable and renewable distributed energy resources (DERs).