An Analysis for Uncertainty Power Fluctuations of Dynamic Pricing Type FastADR by AI Controlled Large-scale Distributed Air-conditioners

Abstract

AI control for building multi-type air conditioners which adapt to dynamic pricing type FastADR has begun to be considered. In order to adjust supply and demand balance, the grid management system needs to know how AI-controlled air-conditioners respond to unit price fluctuations. In this paper, we clarify the power response distribution of AI-controlled air-conditioners to unit price fluctuations at 30 minute intervals. Large-scale experiments on real buildings are difficult to conduct, and reproducibility cannot be ensured. Therefore, we used an emulator that can simulate the diverse and uncertain minute-by-minute power consumption of building multi-type air-conditioners, and quantify the expected power reduction and uncertainty range based on the power response distribution of 100 diverse virtual buildings.