Day-ahead TCLs dispatch optimization: An integer genetic algorithm approach based on microgrids composition

Abstract

Day-ahead microgrid optimization has been extensively studied in recent technical literature, which predominantly focuses on microgrids comprised of loads, renewable energy systems (RES), and energy storage systems (ESS). However, many microgrids are only composed of loads (such as homes in buildings). This work studies microgrid optimization through a specific focus on thermostatically controllable loads (TCLs), prevalent components in such microgrids. The optimization objectives are tailored to account for the unique characteristics of each microgrid’s composition. Additionally, the study considers the TCLs’ ability to participate in demand response programs within the power system and addresses challenges stemming from discrepancies between dayahead dispatch and real-time operation. Importantly, the optimization process employs a genetic algorithm (GA) to derive optimal on/off sequences and corresponding temperature profiles for each TCL, instead of adjusting variable temperature setpoints. Furthermore, the GA initial population is generated using a novel method called stratified random sampling, proposed in this work. The study presents a procedure for TCL optimization aimed at maximizing each microgrid’s performance relative to its composition. Results demonstrate a reduction in the targeted metric ranging from 2.4% to 18%.