W. Schellong and F. Hentges (Germany)
Energy forecast, energy efficiency, modeling, regression analysis, and neural networks
The paper describes different mathematical modeling methods for the heat demand forecast of a district heating system. Mainly the regression analysis and the design of neural networks are tested on the basis of real consumption data of the heating system. The forecast tools are necessary to control and optimize the operating schedule of a cogeneration plant in combination with the district heating system. The heat demand forecast implemented in an energy management system helps to increase the energy efficiency and supports the sustainable energy development. An analysis of the consumption data and of the main influence factors on the heat demand is necessary in order to obtain suitable forecast models. The paper describes the data management as well as the process of the mathematical modeling. The design of clusters depending on seasonal impacts and the influence of climate factors are investigated. Linear multiple regression models are compared with individually designed neural networks. The experiences of the application of both methods to real data sets are presented.
Important Links:
Go Back
