Volume 58

A Dynamic Modeling Method for Integrated Energy System Equipment Based on Mechanism and Residual Correction Xinyu Yin, Junli Zhang, Jiong Shen

Abstract

To address the modeling of integrated energy system equipment oriented towards dispatch optimization, a hybrid mechanism-data-driven dynamic modeling approach is proposed. This method employs a transfer function model to characterize the dynamic behavior of the equipment and a Multilayer Perceptron (MLP) to model the nonlinear residuals of the transfer function model. To evaluate the model performance, a water-source heat pump was modeled and comparisons were made among the first-principles model, the transfer function model, the NARX model, and the hybrid model. The results show that the simulation time of the hybrid model is comparable to that of the transfer function model, with reductions of 85.72% and 95.56% compared to the first-principles model and the NARX model, respectively. In terms of simulation accuracy of heat power and electrical power, the hybrid model is comparable to the NARX model, with improvements of 50.76% and 89.30% compared to the transfer function model. The proposed hybrid modeling approach achieves model light-weighting while maintaining accuracy, thereby better supporting the dispatch optimization of integrated energy systems.

Keywords Integrated energy systems, hybrid driven modeling, dynamic modeling