Influence of spatial clustering on the result of heat supply system optimization

Kurzfassung

Nowadays, energy forecasting, using AI and machine learning, is common for predicting demand patterns and optimizing energy production. The energy system modeling sector consists of many available tools and software, which tend to be costly and frequently demand advanced expertise. So the aim of this research work is to use open source algorithms and walk through maximum heat energy production. The research focuses on the physical world: Energy planning typically considers settlement structures, e.g. in the form of building blocks. While grouping the buildings is a necessity to reduce complexity, in particular because of measures like redensification or reconstruction, building blocks can be diverse when it comes to their energetic properties. PVLib library is used for generating electric time-series. The work of PVLib is to make simulations of photovoltaic energy systems. Demandlib is used for generating hourly load profiles, especially for heat and electricity demand in various regions. The work of demandlib is to generate various load profiles of energy models with the help of annual energy consumption data. For optimizing purposes, the OEMOF algorithm is used because OEMOF is an open source tool. This thesis compares energy system optimization by following these steps as described. Coming on the topic name, KMeans clustering is carried out first on n number of houses, followed by PV data and heat demand data for n number of houses. The analysis is carried out in two different parts: first geographically and second incl. energy. Then, the clustering algorithm is used to find hidden patterns and some shapes in the data set. Subsequently, the photovoltaic data set is analyzed with various parameters such as solar irradiance, locations, azimuth angle, tilt angle, etc. and ac-power of the module is generated in time-series data. The demandlib is used to generate different heat load profiles and electrical standard load profiles for energy models both in time-series data. The energy modelling setup OEMOF is used to analyze and optimize energy usage. Finally, OEMOF algorithm is used for producing cost-effective investment optimize data.