Artificial-Intelligence-Based Reduced Sensor Voltage Control Strategy for DC Microgrid Applications

The expeditious advancement in renewable energy technologies enables the concept of microgrids to boost the incorporation of renewable energy into power systems. In this context, distributed generation (DG)-based DC microgrids (MGs) are favoured because of their higher efficiency, greater reliability, and simpler development and control compared to their AC counterparts. This paper presents an artificial neural network (ANN) voltage control for a DC-DC step-up converter to reduce the number of sensors in the DC microgrids. The proposed approach offered cost-effective and better voltage regulation in multi-bus DC MG. The proposed methodology employs quasi-stationary line (QSL) modeling to account for DC MG uncertainties and disturbances, while simultaneously developing and implementing a model predictive voltage control (MPVC) strategy to generate the comprehensive dataset. The converter's voltage error and switching signals, extracted from the generated dataset, serve as input features for offline training of an artificial neural network (ANN). Once trained, the ANN is deployed online to regulate distributed generators (DGs) within a multi-bus DC MG. Real-time hardware-in-the-loop simulations using OPAL-RT 4510 demonstrate that the proposed controller effectively regulates voltage with reduced sensors, ensuring improved reliability and efficiency.