A Semi-Decentralized Power Management Strategy in Islanded PV-BESS Based DC Microgrid with Thermostatic Loads

Abstract

In this work, a decentralized coordinated control strategy has been developed for an islanded DC microgrid composed of PV generation units, a battery energy storage system, thermostatically controlled loads, and other loads. Battery unit is used as a main DC bus voltage regulator balancing the total generation and load by supplying/absorbing deficit/surplus power. Due to the limited capacity of the battery and to avoid DC bus overvoltage and overcharging scenarios in case of continuous MPPT operation, PV curtailment control is implemented with seamless transition capability between maximum power and generation curtailment modes. Thermostatically controlled loads are utilized for peak load management to prevent DC link voltage collapse and hence the overall reliability of the microgrid is improved without installing additional generation capacity or using load-shedding techniques. TCLs have been prioritized according to their temperatures when participating in peak load management and therefore, the situation is alleviated for longer periods. In addition, TCLs operational and QoS constraints such as lock-out constraint and temperature higher and lower limits have all been considered in the developed work. A deep learning LSTM based prediction models have been developed to quantify the amount of power from TCLs that can be controlled considering all relevant factors (i.e. forecasted PV maximum power, predicted BESS SOC, predicted loading condition and TCLs operational constraints). If a power deficiency is predicted, the forecasting model sends a signal to the back-up diesel generator in order to synchronize and connect to the microgrid to share the total load by means of droop control. The developed power management strategy ensures the continuity of power supply to all microgrid loads without shedding any part of the load and hence a high reliability is maintained even under adverse operating conditions.

Date of Award	1 May 2024
Original language	American English
Supervisor	Tarek El Fouly (Supervisor)