基于数字孪生的变压器多粒度建模与仿真分析

Multi-granularity Modeling and Simulation Analysis of Transformer Based on Digital Twin

为了实现电力系统的数字孪生,针对常规数字孪生建模方法中单一粒度的模型不能满足复杂系统仿真运行需求的问题,将多粒度建模思想融入到数字孪生建模理论中,提出了基于多粒度的数字孪生建模方法并基于COMSOL有限元仿真实现。为了解决系统变阶方法删除关键参数无法实现降阶的问题,提出了系统属性变阶方法,并在基于聚合解聚对变压器模型进行层次化粒度划分的基础上,利用该方法对变压器模型作进一步的粒度划分。以400 kVA变压器为例,利用能量守恒原理分析变压器热传递特性。应用磁-热-流耦合温度场进行仿真计算,分析变压器的温升特性及不同粒度下仿真耗时,由系统属性变阶得到的粗粒度模型在较小的误差下,仿真耗时更短。

To achieve digital twin of power systems,the multi-granularity modeling idea is integrated into the digital twin modeling theory,since the single granularity model used in the conventional digital twin modeling method cannot meet the requirements of complex systems.Based on this,a digital twinning modeling method is proposed that utilizes multi-granularity and finite element simulation based on COMSOL.Additionally,to address the limitation of the system variable order method in achieving order reduction by deleting key parameters,a system attribute variable order method is introduced.The hierarchical granularity division using the aggregated and disaggregated method is applied to further divide the transformer model into smaller granules.As an example,a 400 kVA transformer is analyzed using the principle of energy conservation to examine its heat transfer characteristics.The magnetic-thermal-current coupling temperature field is then used for simulation calculation to analyze the temperature rise characteristics of the transformer and the simulation time under different granularity.The results show that the coarse-grained model obtained from the system attributes with varying order has shorter simulation time with smaller errors.