Research On Energy Management Of Hybrid Energy Commercial Buildings And Dispatch Of Electirc Vehicle Charging Stations

As society and the economy rapidly develop,the demand for energy continues to grow.Smart buildings that can automatically adjust their operational state according to changes in indoor and outdoor environments,improving energy efficiency and user comfort,have become one of the directions for the future transformation of the construction industry.However,the energy efficiency of smart buildings depends on energy management methods.In existing literature,traditional rule-based methods are simple but cannot utilize uncertain factors,while model-based methods rely heavily on model accuracy and cannot effectively utilize historical data.In addition,with the development of electric vehicles and related industries,the coupling of a large number of electric vehicles with buildings has become inevitable.Effective energy scheduling for electric vehicles is indispensable for the stable operation of building microgrids.However,in commercial building environments,direct scheduling methods for electric vehicles are often ineffective due to limitations such as communication and privacy.Therefore,this paper uses deep reinforcement learning methods to address these shortcomings and studies energy scheduling for various loads in multi-energy commercial buildings.The coupling relationship between electric vehicle charging stations and commercial buildings is analyzed and dynamic pricing strategies are used to indirectly schedule electric vehicles.The specific research content of this paper is as follows.Firstly,this paper proposes a sma building energy management system model for mixedenergy commercial buildings,consisting of photovoltaic power generation systems,HVAC systems,combined heat and power systems,thermal storage systems and energy storage systems.This system is connected to the main power grid and natural gas stations.The goal of this paper is to minimize the cumulative energy cost of the system while ensuring the comfort of residents within the building.To achieve this goal,an optimization problem is established considering various system constraints.The optimization problem is then transformed into a Markov decision process and the corresponding state and action spaces and reward functions are designed.A datadriven real-time energy management method based on deep deterministic policy gradient algorithm is proposed.This method has obvious advantages in saving energy costs and maintaining resident comfort.Compared with other benchmark methods,the method proposed in this paper saves at least 5.4% of energy costs.Secondly,this paper addresses the coupling scheduling problem between electric vehicles and commercial buildings.Based on the characteristics of electric vehicles and users,an electric vehicle charging station service model is proposed.The electric vehicle charging station is powered by the commercial building.In addition,to promote building decarbonization,a carbon emission model is established for energy supply from different sources.The goal is to minimize cumulative operating costs and carbon emissions while ensuring resident comfort.An optimization problem is established and expressed as a Markov decision process.To address the shortcomings of deep deterministic policy gradient algorithm being overly sensitive to hyperparameters and having poor exploratory ability,a real-time energy management method based on soft actor-critic algorithm is proposed.Dynamic pricing is used to indirectly schedule electric vehicles.Simulation results show that the method proposed in this paper performs well in increasing the revenue of electric vehicle charging stations and reducing energy costs and carbon emissions.Compared with other benchmark methods,the method proposed in this paper achieves the best balance and reduces total operating costs by at least 5.2% compared to other methods.