•  
  •  
 

Keywords

central air conditioning, demand response, random forest, on/off duration evaluation, indoor comfort

Abstract

Central air conditioning in commercial buildings is a high-quality controllable resource，and the rational utilization of its regulation capacity is of significance to the supply-demand balance of power systems.In particular，during summer peak electricity consumption periods，effective regulation of massive central air conditioning loads can greatly relieve the pressure of source-load balance in power grids.To this end，a demand response capability evaluation of central air conditioning based on random forest （RF） is proposed.Assuming that the participation of central air conditioning in demand response is limited to controlling the on/off status of the central refrigeration system，then the demand response capability can be characterized by the maximum on/off duration starting from any given moment.Firstly，the chilled water circulation process，indoor thermal dynamics，and on/off control logic of central refrigeration systems for central air conditioning are sorted out，and the key state variables affecting the maximum on/off duration are clarified.Secondly，a dataset containing indoor and outdoor temperatures，historical supply and return temperatures of chilled water，and maximum on/off duration is established based on physical model simulation or field measurement data.Then，an RF model is trained with the above state variables as inputs and the maximum on/off duration as outputs.Finally，case comparison and comfort verification are conducted to validate the evaluation accuracy and applicability of the proposed method.The proposed RF model avoids reliance on the complex thermal dynamics parameters of buildings and features low implementation barriers.Simulation comparison and analysis demonstrate that the proposed method exhibits high accuracy in evaluating the maximum on/off duration of central air conditioning，and it will not significantly reduce indoor occupants ’ comfort when assessing demand response capacity.

DOI

10.19781/j.issn.1673-9140.2026.04.010

First Page

114

Last Page

124

Download

Share

COinS