The concept of a predictive analytics system for operating modes of the electric power system in the Central Energy System of Mongolia has been developed and applied at Novosibirsk State Technical University (NETI). The proposed software modules improve the quality and accuracy of planning the operating modes of the power system for the day ahead.
The relevance of the topic is due to the lack of automated systems in the Central Energy System of Mongolia for planning normal modes for the day ahead. This planning is performed manually, which requires a lot of time, reduces the accuracy and validity of decisions made, and increases the risk of errors.
According to Osgonbaatar Tuvshin, a graduate student at NSTU-NETI, who defended his thesis on this topic in July, the main task was to create a single software package that provides the possibility of conducting a series of simulation calculations and is used for monitoring, controlling and planning normal modes of an electric power system with a high proportion of renewable energy sources.
The analysis of existing methods of managing the modes of electric power systems, in particular, methods of short-term forecasting and optimization, was carried out. Based on this analysis, mathematical models have been developed for predicting processes in the energy system based on a combination of classical approaches and ensemble machine learning methods. The results of comparative experiments have justified the choice of specific algorithms for predicting the daily load schedules of the energy system and its nodes, as well as generation based on renewable energy sources (RES), such as wind and solar power plants.
The use of the obtained forecasts of daily load schedules and renewable energy generation in combination with actual data on network equipment and topology made it possible to simulate the functioning of Mongolia's Central Energy System and calculate its steady-state operating modes. The optimization of the distribution of generation between sources, for example, a thermal power plant, based on technical and economic criteria, was also carried out.
"The software implementation of all functions is performed in the Python programming language, which made it possible to develop an effective system for analyzing the operating modes of the power system, which represents the state of the electric power system in the form of diagrams, graphs and other visual solutions," Osgonbaatar Tuvshin noted.
According to him, the scientific novelty of the project lies in the development of multifactorial mathematical models for predicting time series, including graphs of the load of the electric power system and its nodes, graphs of the generation of renewable energy sources, as well as in the development of algorithms for optimizing the normal operation of the electric power system, taking into account multifactor models. The concept of a predictive analytics system for operating modes of the electric power system in the Central Energy System of Mongolia has been developed and applied at Novosibirsk State Technical University (NETI). The proposed concept includes three program blocks. Certificates of state registration have been obtained for these programs.
The developed predictive analytics system will allow the system operator of the Central Energy System of Mongolia to solve key tasks of daily mode planning: optimize dispatching schedules for CHP generation, reducing fuel costs by up to 4%; prevent overloading of network elements by calculating power flows taking into account the forecast of renewable energy sources; improve the accuracy of load balancing and generation thanks to forecasts based on machine learning with an error of up to 2%.
Based on the results of the dissertation, a scientific article was published in the journal "Electric Stations". Currently, the predictive analytics system is being implemented and tested in real-world operating conditions of the Central Power System of Mongolia, in particular, the introduction of modules for predicting power consumption and optimizing normal modes for the day ahead.