Backward neural network (BNN) based multilevel control for enhancing the quality of an islanded RES DC microgrid under variable communication network

Abstract

Microgrids (MGs) and energy communities have been widely implemented, leading to the participation of multiple stakeholders in distribution networks. Insufficient information infrastructure, particularly in rural distribution networks, is leading to a growing number of operational blind areas in distribution networks. An optimization challenge is addressed in multi-feeder microgrid systems to handle load sharing and voltage management by implementing a backward neural network (BNN) as a robust control approach. The control technique consists of a neural network that optimizes the control strategy to calculate the operating directions for each distributed generating point. Neural networks improve control during communication connectivity issues to ensure the computation of operational directions. Traditional control of DC microgrids is susceptible to communication link delays. The proposed BNN technique can be expanded to encompass the entire multi-feeder network for precise load distribution and voltage management. The BNN results are achieved through mathematical analysis of different load conditions and uncertain line characteristics in a radial network of a multi-feeder microgrid, demonstrating the effectiveness of the proposed approach. The proposed BNN technique is more effective than conventional control in accurately distributing the load and regulating the feeder voltage, especially during communication failure.

Keywords

• Renewable energy sources
• Backward NN
• NN microgrid control
• Communication latencies
• Multi-level control
• Distributed control