Research On The Method Of Promoting New Energy Consumption Considering Thermal-electric Decoupling In Energy Interne

With the worsening global climate,the supply of traditional energy is gradually decreasing,quietly triggering a new energy revolution.In order to reduce carbon dioxide emissions,renewable energy sources such as wind power have been introduced into power systems.However,the uncertainty of fluctuating wind power will lead to difficulties in dispatching renewable energy and heat energy,which can be even more complicated in power systems with combined heat and power units.Due to the serious conflict between electricity and heat demand,large-scale wind curtailment and environmental pollution may occur without carefully designed thermoelectric decoupling schemes.The existing iterative methods for comprehensive heat and electric model can be used to solve the threats caused by the uncertainty of fluctuating wind power and heat load.However,the ignored details of the model may lead to higher costs.Adopting a distributed non iterative approach can ensure the independent operation of the energy industry,but this approach that does not consider carbon emissions may face higher environmental pollution issues.At the same time,because most energy data is distributed in different departments,their models are managed locally.Considering the security of power data,it is necessary to avoid directly using original data for analysis.Different energy departments want to conduct comprehensive scheduling while maintaining their private data.How to protect the security and privacy of data is also an important issue in decision-making.This article focuses on the construction and application research of integrated power and heat dispatch system,considering the uncertainty of wind power and heat load as well as environmental issues,as well as privacy data protection issues between different departments.The main contents are as follows:Firstly,this paper establishes a two-stage model considering electric power and heat,including day-ahead unit scheduling and resource allocation.It solves the threats posed by the uncertainty of wind power and heat load,as well as the operating costs and environmental issues of the power grid.Energy storage system are used to expand the adjustable space of the power system and enhance the adaptability of wind power generator.The proposed model aims to minimize carbon emissions and system operating costs while meeting electrical and heat loads.Secondly,in order to obtain the optimal scheduling strategy without detailed modeling of complex power plants,this paper uses reinforcement learning algorithms to learn effective strategies.The two stages of the problem are optimized and analyzed experimentally.Experiments are conducted using real electrical and heat load data.Thirdly,in order to solve the problem of privacy protection between different departments,this article establishes a load forecasting model based on federated learning,aiming to jointly build a machine learning model by sharing private data between different departments.This form not only considers the characteristics of different data,but also protects data privacy.Compared to traditional centralized machine learning,federated learning offloads model training to client devices,enabling them to jointly train a more accurate and efficient load forecasting model using their respective load data resources while protecting the data privacy of each load data owner,effectively solving data privacy and security issues.Experimental results on real data demonstrate the effectiveness of the proposed algorithm.