【摘要】 电力设备的物理状态感知是建设智能电网的关键点之一。数字孪生技术可实时映射并快速预测设备的物理状态，但现有建模方法难以满足电缆温度场数字孪生模型的实时计算要求。鉴于此，该文提出了一种基于降阶模型的配网电缆温度场数字孪生建模方法。首先，建立电缆多物理场全阶模型，基于奇异值分解和响应面插值方法建立其稳态温度场降阶模型；然后，结合稳态温度场降阶模型及电缆表面温度数据，通过实时求解多回路电缆传热反问题重构当前的电缆内部瞬态温度场，并基于排管敷设电缆温升试验验证了方法的正确性。进一步地，将该方法应用于实际运行的10 kV电缆的内部瞬态温度场重构，将重构结果作为已知初始状态，基于稳态温度场降阶模型和改进叠加法快速预测应急状态下的电缆导体温度。与全阶模型相比，电缆导体温度重构值的最大相对误差为1.76%，应急状态运行的导体温度预测误差为1.01%，单次重构和预测计算时间分别为8.1 s和3.6 s，计算效率分别提高约35 555倍和6 000倍。该方法兼顾了计算速度、计算精度和建模成本，对其他类型电力设备温度场数字孪生建模具有参考意义。更多还原

【Abstract】 State perception of power equipment is one of the key issues in the construction of smart grid. Digital twin(DT) technology can map and quickly predict the physical state of equipment in real-time, but existing modeling methods are difficult to meet the real-time computing requirements. This paper proposes a DT modeling method for distribution cables temperature field based on reduced-order model. The method first establishes a multiphysics full-order model of cables, builds a reduced-order model of the steady-state temperature field based on singular value decomposition and response surface interpolation methods. Combining the steady-state reduced-order model and temperature measurement data, the multi-circuit cables heat transfer inverse problem is solved to reconstruct the transient temperature field inside the cables in real-time, and the correctness is verified based on the temperature rise test results. Furthermore, the proposed method is employed to reconstruct the internal transient temperature field of 10 kV cables in actual operation. Taking the reconstructed result as the known initial state, the cable conductor temperature during emergency operation is quickly predicted based on the reduced-order model and the improved superposition method. As compared with the simulation results of the full-order model, the maximum relative error of the conductor temperature DT reconstruction value is 1. 76%, the conductor temperature prediction error during emergency operation is 1. 01%. The calculation time of single reconstruction and prediction is 8. 1 s and 3. 6 s, and the calculation efficiency is about 35 555 times and 6 000 times that of the full-order model, respectively. The new method takes calculation speed, calculation accuracy and modeling cost into consideration, and has reference significance for the temperature field DT modeling of other types of power equipment.更多还原