Quantification Of End-use Heat Consumer Load Forecasting And Its Flexibility

The heating method in China is mainly based on fossil fuels,but the burning of fossil fuels aggravates the emission of carbon dioxide and causes environmental pollution problems.With the increasing requirements of environmental protection and the expanding demand for energy conservation in China,we need to ensure economic development while transforming existing technologies to save energy,reduce the emission of pollutants and achieve the common development of environment and economy.Centralized heating system with high heating efficiency,low cost and low environmental pollution has become the main form of heat supply in many countries.However,the centralized heating system has problems of complex pipe network design and uneven heating and cooling,which need to be actively faced and solved in the face of these problems.In this paper,the end-user dynamic load prediction is carried out to address the problem of uneven heat supply for end-users,and the predicted heat load is used as a basis to provide a theoretical basis for the heating company to adjust the heat supply data.A heat network model of the heating system is built,and the concept of heat network flexibility is given and flexibility is quantified and predicted from the perspective of flexibility,so as to explore the heat required by end users.Firstly,a comprehensive analysis of the heating district selected in this project was carried out.Relevant studies were conducted from the geographical location of the district,the layout of the building and the installation location of the room temperature monitoring device to determine the object of heat load prediction.The heat supply data were collected through the intelligent heat supply integrated control platform,and the pre-processing of the data was completed.These factors are divided into internal set external factors for correlation analysis,and the strongest correlation is selected as the input variable of the prediction model to prepare for the heat load prediction.Secondly,five algorithms are selected from various machine learning methods for heat load prediction,namely,support vector machine,K-neighborhood classification algorithm,stochastic gradient descent method,adaptive boosting algorithm and dendritic network.The results of single-family load prediction and heat exchange station cell heat load prediction are compared,and the dendritic network is the most suitable method for heat load prediction in this paper and is described in detail in terms of method mechanism and parameter selection,respectively.The results of heat load prediction provide a theoretical basis for heat supply companies to adjust heat.Finally,this paper proposes the concept of flexibility of heat networks,mainly by analogy with the concept of flexibility of electric power systems,and gives a definition of flexibility by combining the characteristics of heat networks themselves and the insights of this paper on flexibility.A model of the heat network is constructed,and flexibility is quantified from the perspective of qualitative regulation of the heat supply system.Based on the actual measured data,the abstract definition of flexibility is transformed into concrete flexibility values,and flexibility is predicted based on the prediction results of heat load.The final result of the quantified flexibility of the heat network is the adjustable range of the upper and lower heat load.The adjustable heat load within this range can meet the heat demand of heat users,ensure the heat experience of heat users and at the same time achieve reasonable heat distribution,avoid unnecessary heat waste.