Research On Dynamic Control Of Chilled Water Based On Chilled Model And Load Forecasting In Air-conditioning System

Building energy consumption has accounted for a large proportion of the totalsocial energy consumption, and central air-conditioning energy consumption accountsfor about50%of the total building energy consumption, so the energy-saving ofcentral air-condition system has significance for energy-saving.The energyconsumption in water system takes main part of total energy consumption inair-conditioning system. However most of the time the system is running at part load.In order to achieve energy-saving variable flow air-conditioning system emerges. Inactual process, due to the constantly change of the running environment and thesystem load changes, usually system is running in partial load. Accurate prediction ofair conditioning load can guide the implementation of the control strategy andimprove the operation economy of air conditioning system. The control based on loadforecast in air conditioning system is one of the research hotspots and trend at present.This paper analyzed the VAV system chiller water system and the load of the room,changes the pump frequency to achieve variable flow area to meet the needs of airconditioning load on the lower part of the heat, Minimize energy consumption underthe premise to meet the load requirements.In this paper variable air volume air-conditioning system of intelligent buildingand building automation laboratory in Xi’an University of Architecture&Technologyis taken as the research object. First, support vector machine method was used toestablish a load forecasting model, and the model was validated. A kind of dynamiccontrol of chilled water flow based on load forecasting was proposed.The chilledwater pump model was identified and the design of flow controller was completed. Second, recursive least squares method was used to establish a chiller multivariatepolynomial regression model, based on the model chiller group control strategies wereproposed and the feasibility of the policy was analyzed. For cooler model in practicalengineering prone to traffic parameters can not be obtained or the measurementaccuracy is low, a solution of flow measurement that bases on RBF neural network isproposed.The prediction load and the actual load were calculated and compared byusing Labview software. According to the results to control the Chilled water pump,thus b changing the flow rate of chilled water supply to match the load demand tosome extent can reduce energy consumption, avoid the waste of energy that becauseof the mismatch between supply and demand.