Research On Renewable Energy Power Forecasting,Load Forecasting And Energy Management Strategy Of CCHP System With Compressed Air Energy Storage

Multi-energy supplement Combined Cooling Heating and Power(CCHP)system can effectively absorb intermittent renewable energy and realize"grade-to-grade and cascade utilization"of energy,which is an important direction of world energy technology development under the background of current energy and environmental crisis.Compressed Air Energy Storage(CAES)has attracted much attention from the international community because of its advantages of low environmental pollution,high energy conversion efficiency and no phase change loss.This thesis makes full use of the natural cooling,heating,power and other multi-energy interfaces of compressed air energy storage system,and connects CAES to CCHP system by multi-energy coupling method,Besides,the study of the source-load prediction and energy management strategy of CCHP system with compressed air energy storage proves that this method not only suppresses the intermittence and fluctuation of renewable energy,but also improves the flexibility of multi-energy joint dispatch and the economy and cleanliness of system operation.Firstly,the structure of CCHP system with CAES is designed.Based on the analysis of the operation principles of gas turbine,absorption chiller,gas boiler,compressed air energy storage,wind turbine,photovoltaic power generation system and other core equipment,a mathematical model is establishedthat which is suitable for system energy scheduling.Aiming at the randomness and volatility of renewable energy,a combination forecasting method based on Empirical Mode Decomposition(EMD)is proposed,and the improved Induced Ordered Weighted Averaging(IOWA)operator is used to adjust the combination weight coefficients in real time.The study results of wind/solar power prediction show that the proposed algorithm has high prediction accuracy,which can effectively reduce the interference caused by randomness and avoid the limitation of a single prediction algorithm.Accurate load forecasting is an important basis for operation scheduling and energy management of CCHP system.In this thesis,a multi-variable load forecasting method based on deep learning algorithm Long Short-Term Memory(LSTM)is proposed,which utilizes the periodicity of load and considers the coupling relationship between current load data and historical load data,as well as the coupling relationship among heating load,cooling load and electrical load.The simulation results show that the multi-variable LSTM forecasting model can improve the accuracy of load forecasting effectively compared with the single-variable forecasting model and the traditional load forecasting model.Finally,a mixed integer non-linear programming model is established to minimize the total operation cost of the system,with the constraints of electricity,cooling,heating balance,capacity of output equipment and upper and lower limits of charging and discharging power.A new energy management strategy for CCHP system with CAES is proposed,in which generation side,storage side and user side resources cooperate with each other.The advantages of adding energy storage system and considering demand side response to system peak-shaving and valley-filling,smoothing load curve and economy of system operation are verified.