Research On Load Prediction And Demand Response Control Strategy Of Thermal Storage Air-conditioning

With the development of green and sustainable development policy,solar and wind power generation has received more attention.When this part of the power is incorporated into the grid,the instability of the power system increases.The air-conditioning load caused by high temperature in summer,which increases the power difference between peak and valley.As a flexible temperature control load,air-conditioning load is a kind of demand response resource with great potential.It can improve the instability of power system,reduce the peak load and change the status of power supply and demand when it participates in demand response.The most effective and direct way for the air-conditioning system to participate in the demand response is to control the air-conditioning system.Through the control system operation to change the system energy consumption in demand response period.The load prediction in the demand response period is the premise for the development of effective system operation strategy.Machine learning is a data-driven modeling method,while the white-box and gray-box modeling methods need more parameters,workload and complex modeling.Therefore,machine learning is used for predicting air-conditioning load.Taking the full-scale experimental platform of thermal storage air-conditioning in Chang’an university as the research object,the dynamic simulation model of TRNSYS is established.The simulation data are obtained for load prediction,on this basis,the physical experiment platform is run to obtain the data related to the prediction,and the optimal model is used to predict the load.At the same time,Elman neural network is used to predict the energystorage and energy-release time of thermal storage tank.Finally,according to the results of load,energy-storage and release time prediction,a demand response strategy combining active energy storage and global temperature regulation is proposed.The experimental results show that the optimal load prediction model of the thermal storage air-conditioning system is Elman neural network model optimized by improved particle swarm optimization.The outdoor temperature and humidity predicted by Elman neural network meets the demand of load prediction,and the error of energy storage and energy release time predicted by this method is small.A demand response strategy combining active energy storage and global temperature regulation based on load prediction is proposed.Compared with the other four strategies include conventional operation strategy,active energy storage strategy,global temperature regulation strategy,active energy storage strategy and global temperature regulation strategy,on the premise of satisfying the comfort,the first strategy of power consumption in the demand response period is the least and the full-day running cost is the lowest.Therefore,it is feasible to formulate the demand response operation strategy of thermal storage air-conditioning system based on load prediction to reduce the peak load of power grid and save the operation cost for users.