Optimal Scheduling Of Building Clusters Integrated Energy System Considering Renewable Energy Uncertainty

In today’s world,the accelerated urbanization and the improvement of economic level lead to the rapid increase of energy consumption in buildings and the increasingly serious environmental pollution,while the rapid development of new energy technology provides important support for cities to reduce energy consumption and environmental pollution.But at the same time,with the increasing penetration of new energy,the randomness and volatility of new energy itself brings uncertainty to the safe and stable operation of power systems.How to effectively cope with the uncertainty brought by new energy,and further improve the level of new energy consumption and building energy utilization is of great significance.Therefore,this paper focuses on the integrated energy system of building clusters under new energy access,and conducts research on the uncertainty of new energy and the optimal operation of building cluster integrated energy system,with the following details:In this paper,firstly,a new energy generation power prediction method based on conditional generation adversarial network is proposed for the problem of low accuracy of new energy generation power prediction.First,the analysis of historical data and meteorological information of new energy sites is launched,and the similarity between meteorological information in the past multi-day period and meteorological information at the predicted moment is analyzed based on gray correlation analysis,and the similar moment power is filtered.Further,we construct conditional values containing key meteorological factors and similar moment power,which are used to guide the operation of the generative and discriminative models.Secondly,for the meteorological information of new energy field stations,the historical data are classified into various weather types based on K-means clustering method,which facilitates the establishment of more accurate prediction models.Then,convolutional neural networks are used to build the structure of the generative and discriminative models to improve the generating ability of the generative model and the discriminative ability of the discriminative model,and the prediction model is updated by the Adam optimizer.Finally,the feasibility of this method is further verified in the case study based on three prediction error evaluation indexes.Secondly,since the new energy power day-ahead prediction still has a certain prediction error,a two-layer optimization model of the building integrated energy system day-ahead considering the new energy prediction error is proposed for the uncertainty caused by the new energy.Firstly,based on the master-slave game theory,the leader role of building cluster integrated energy system operator and the follower role of building cluster users are determined,and both parties are considered to pursue their own interest maximization.For the upper-level operator model,the electric system and natural gas system models are constructed with the objective functions of revenue maximization and adjustment cost minimization.For the lowerlevel user model,the electric and gas load models are built with cost minimization as the obj ective function,and the RC thermal dynamic model of the heating area is further considered for the building thermal storage characteristics and envelope structure.Secondly,based on the historical prediction error of the new energy prediction method and the improved K-means method,the generation and reduction of new energy output scenarios are completed.Finally,the Karush-Kuhn-Tucker condition and strong dual theory are used to convert the two-layer optimization problem into a single-layer optimization problem,and the single-layer optimization problem is solved based on the multi-scene stochastic programming theory.The algorithm shows that the proposed model can effectively balance the interests of both operators and users.Finally,based on the above research,an intra-day optimal scheduling strategy for building integrated energy systems considering multiple heating methods is proposed.Firstly,we analyze the multiple energy demands of building users and construct a network model including electric system,natural gas system and thermal system as well as a mathematical model of CHP units.Secondly,according to the users’ heating demand,the mathematical models of air conditioning heating and hot water heating are established,and the usage of different types of heating equipment is determined based on the electricity and heat prices at each moment.Finally,an intra-day optimization model based on Model Prediction Control(MPC)rolling optimization is proposed for the integrated energy system of building clusters to reduce the influence of uncertainties and reduce the system operation cost through intra-day rolling optimization.The example shows that the proposed method can effectively reduce the operating cost of building integrated energy system,and the MPC algorithm can better cope with the impact of intra-day prediction errors of new energy and conventional load.