Renewable generation in power systems has proved to be challenging for system operators owing to the increasing levels of penetration. The operation of power systems currently requires additional flexibility and reserves due to the intermittency and unpredictability of renewable generators. However, it is difficult to precisely predict and control the stochastic nature of renewable sources; nevertheless, its capacity continues to increase. To monitor and control renewable generators efficiently, the entire system needs to be established in a hierarchical order. This study proposed the concept of a substation that is uniquely designed for renewable interconnection. The purpose of this substation is simple: to make the renewable generators dispatchable to operators such that each group of renewable generators is sufficiently stable to be considered as conventional generators. For this purpose, methods for sizing and controlling energy storage system are proposed based on forecasts and error distributions.

In recent years, the number of wind farms consisting of type 3 and type 4 wind turbines located within the distribution system has been growing rapidly. Wind turbines can be utilized as a continuous reactive power source to support the system voltage by taking advantage of their reactive power control capability. This paper aims to further develop the reactive power assignment strategy in order to minimize losses in wind farms described in the published paper. We introduce the method of reconfiguration and numbering to apply the algorithm to the wind farm structure and develop the previously-defined allocation ratio into two types of allocation ratios. The goal is to apply the loss minimization algorithm to a wind farm configuration with up to two wind turbines connected to one ring main unit (RMU). The proposed strategy reduces power loss and increases the real power flow in the wind farm by allocating reactive power to connected wind turbines taking into account the resistance value. The proposed allocation technique is validated in a Real Time Digital Simulator (RTDS)-based Hardware-in-the-loop Simulation (HILS) environment considering the Dongbok wind farm configuration in Jeju, South Korea. In the simulation, a Raspberry Pi acts as a wind farm controller sending a reactive power dispatch signal to each wind turbine via Modbus TCP/IP protocol. The simulation results mean that, applying the proposed algorithm, we can expect loss reduction effects in the wind farm.

Power system operators have recently introduced some AI-based techniques in load prediction, fault diagnosis, scheduling, and maintenance. Operators require a grid analysis that includes wind turbines to mitigate impacts by environmental factors. Among them, a method for modeling wind turbines that reflects the dynamic characteristics and their output characteristics is receiving attention. Recently, a data-based power curve modeling method has been adopted for a simplified model that characteristics as much as possible. However, the simplified EMT simulation is difficult to reflect the output characteristics according to nonlinear wind conditions accurately. This paper proposes a wind turbine model based on artificial neural network techniques using real supervisory control and data acquisition (SCADA) data from a wind farm. The proposed strategy derive the similar to real output value through the trained wind turbine model in various wind scenarios. For the verification of the proposed strategy, the case study was conducted using a real-time digital simulator (RTDS).