Data-Driven Decentralized Online System Identification-Based Integral Model-Predictive Voltage and Frequency Control in Microgrids

This article proposes a decentralized online system identification-based integral model-predictive control strategy for the voltage and frequency regulation in the microgrids (MGs). The main advantage of the proposed architecture is that the approach is inherently adaptive to the changing system conditions and does not require the knowledge of system parameters. In addition, as the decentralized system parameters are identified online utilizing local/global measurements, the controller is robust against unknown system disturbances. The decentralized nature of the proposed architecture is evolved from an extended Kalman filter that is utilized to synchronize the states of the identified model, thereby enabling a plug-and-play capability. The proposed controller is constructed utilizing the identified augmented incremental model that incorporates optimal integral action required to mitigate the steady-state errors in the MG voltage and frequency. Furthermore, controller formulation as a quadratic optimization problem including the constraints limits the control inputs within the bounds during electrical faults, thus assisting the downstream primary controller in accomplishing ride-through capability. The proposed framework is validated utilizing a section of the IEEE 123-bus distribution network for the different grid events including the effect of communication latency. Model-in-the-loop real-time simulation results demonstrate that the proposed framework offers significant transient performance improvement compared to the optimal proportional–integral strategy, especially around 50% faster regulation, when the communication latency is considered.