Research On Optimal Scheduling Strategy Of Ice Storage Air Conditioning System Based On Load Forecasting

Ice storage air conditioning system can store cold at night when there is a high demand for air conditioning in summer,and then combine with ice tanks and refrigeration units to supply cold during the peak of daytime electricity consumption,so as to alleviate the contradiction between power supply and demand and realize the "peak-shifting and valley-filling" of power load,thus reducing carbon emissions from the power grid.However,ice storage air conditioning system has many equipment and complex structure,and improper cooling strategy may lead to high system operation cost,low equipment operation efficiency,energy waste and other problems.Currently,the two most common cooling strategies for ice storage air conditioning systems are chiller priority and ice melt priority.The former can reduce the overall energy consumption by reducing the ice storage at night,but there is the problem of increased operating costs,which does not give full play to its advantage of "peak shifting and valley filling",resulting in an insignificant reduction in carbon emissions from the power grid;the latter effectively reduces the overall operating costs of the system by increasing the ice storage at night with low valley tariffs,although the overuse of the ice tank will Although the overuse of the ice tank will give full play to its advantage of "peak-shifting and valley-filling" and significantly reduce the carbon emissions of the grid,the low efficiency of the chiller operation in the ice storage process will also increase the overall energy consumption of the system.Therefore,it is of great significance to study the optimal scheduling strategy of ice storage air conditioning system from the perspective of optimal overall efficiency,in order to balance the contradictory relationship between economy,energy consumption and carbon emission of the power grid.Load forecasting of ice storage and cooling air-conditioning system can provide data basis for system optimization and scheduling scheme,so establishing accurate load forecasting model is the first step to realize optimization and scheduling of ice storage and cooling air-conditioning system.However,the cold load of ice storage air conditioning system is susceptible to the influence of weather,personnel and other factors,and the load is highly stochastic and shows complex characteristics of multidimensional nonlinearity,which leads to the problems of difficult model construction and difficult to determine the number of model dimensions.Based on this,a load forecasting method based on e Xtreme Gradient Boosting(XGBoost)algorithm and Long-term and Short-term Time-series Network(LSTNet)is proposed in this paper,and A two-layer load forecasting model for ice storage air conditioning system is designed.The data processing layer of the model uses the XGBoost model to filter out the feature sets with more significant correlation with the cold load forecasting and determine the input dimensions of the prediction model;then this thesis proposes a load forecasting layer based on LSTNet,and finally adopts an autoregressive model to extract the linear components of the load and integrate the prediction results of multiple modules to achieve high accuracy building cold load forecasting.The two-layer load forecasting model combines the advantages of Convolutional Neural Network(CNN)and Recurrent Neural Network(RNN),using the powerful feature extraction ability of CNN to extract the fine-grained features in the load data on the one hand,and using the property of RNN to be sensitive to On the other hand,we take advantage of the sensitivity of RNN to the periodic information of time series to capture the long-term and shortterm information in the cold load series.The experimental results show that the accuracy of the proposed prediction model is better than that of the traditional BP neural network and singlelayer load forecasting model,and can meet the needs of the subsequent research on optimal scheduling strategies.Based on the data of load forecasting results,this thesis establishes a multi-objective optimization mathematical model of the ice storage and cooling air conditioning system within one operation cycle(one day).In this thesis,the energy consumption model of refrigeration units,cooling tower energy consumption model,water pump energy consumption model and ice storage tank storage and release model are established,and the equipment operating constraints and optimization decision variables are determined.From the perspective of economy,energy saving and environmental protection,the optimization objectives of reducing operation cost,reducing energy consumption and improving the load change rate of the grid are established under the premise of meeting the cold load demand,so as to balance the contradictory relationship between economy,energy consumption and carbon emission of the grid,and to improve the comprehensive operation efficiency of the ice storage air conditioning system.Based on the above research results,this thesis proposes a multi-objective optimal scheduling method based on NSGA-II-DE hybrid optimization algorithm for ice storage and cooling air conditioning system to solve the real-time cooling capacity of ice tanks and dualstage refrigeration units.The NSGA-II-DE hybrid optimization algorithm is proposed to address the disadvantages of slow convergence and poor convergence of the traditional Nondominated Sorting Genetic Algorithm-II(NSGA-II)in solving complex ice storage and cooling air conditioning system problems,using the Differential Evolution Algorithm(Differential Evolution Algorithm,DE)of variational crossover operator instead of the simulated binary hybrid operator in NSGA-II,which reduces the sensitivity of the hybrid operator to parameters.Experiments show that compared with the original NSGA-II algorithm,the optimized algorithm in this thesis expands the search range and reduces the average computation time by 29.58%and improves the average HV value by 8.5%.