Statistical Machine Learning Model for Distributed Energy Planning in Industrial Park
Chen Zhang,Xiurong Zhang,Xianping Wu,Saddam Aziz,,,,
DOI: https://doi.org/10.47852/bonviewaia42021969
2024-01-01
Artificial Intelligence and Applications
Abstract:With the advancement of industrial modernization, industrial parks have become the main body of new energy production and consumption. However, due to the large demand for energy in industrial agglomeration, the way of energy utilization is changing to sustainable. The direct connection of distributed energy resources in industrial parks, including photovoltaic power generation systems, has an important impact on its planning and operation. Furthermore, weather scenarios can have an impact on distributed photovoltaic generation, and the uncertainty in photovoltaic power output will, in turn, affect the uncertainty in industrial park planning. Therefore, this paper aims to address the issues of inaccurate prediction of distributed electricity generation during the planning period and the non-uniform distribution of energy resources such as electricity, heating, and cooling. This is achieved through the application of statistical machine learning. The paper intends to incorporate the ideas of statistical machine learning into the model for industrial park distributed energy random opportunity-constrained planning, aiming to resolve the problems of non-uniform distribution of distributed energy sources within the park, along with uncertainty in their outputs and high overall investment costs. The model takes the planning, construction and operating costs of the industrial park as the objective function, uses the lost load cost to ensure the safety of the industrial park, and uses the Chebyshev inequality probability to limit the output characteristics of distributed energy equipment. In terms of operation, the planning period is subdivided into heating period, cooling period and transition period, and the balance of electricity, heat and cold is considered. Finally, an actual example of an industrial park is used to verify the effectiveness of this method. Experimental validation shows that this approach can ensure safety requirements in industrial parks during the heating season, cooling season, and transitional periods by flexibly adjusting the confidence threshold. Simultaneously, it delivers significant economic benefits.