Fast Calculation Method of 3D Nonlinear Magnetic Field Based on Adaptive Model Order Reduction
Liu Yutong, Ren Ziyan, Chi Lianqiang, Zhang Dianhai, Zhang Yanli
Key Laboratory of Special Motors and High-voltage Electrical Apparatus, Ministry of Education （Shenyang University of Technology） Shenyang 110870 China
Abstract Digital twins, as a key technology to achieve real-time monitoring and full life-cycle management of industrial products, are required to reflect the physical characteristics of in-service electrical equipment in real time through simulation. The finite element method (FEM) is commonly used for analyzing electromagnetic properties of low-frequency electrical equipment. For complex three-dimensional problems, the high spatial freedom of geometric model and many time steps result in a large amount of computational time required for finite element analysis. In addition, if the nonlinear behavior of the equipment medium is considered, the stiffness matrix has to contain the parameter information of the previous calculation, and the matrix has to be reassembled for each iteration, which greatly reduces the efficiency of the finite element calculation. The expensive computational cost limits the application of digital twin technology for the optimal design of electrical equipment and for real-time performance analysis. The greedy strategy and improved sparrow search algorithm (ISSA) are combined with the proper orthogonal decomposition (POD) method to propose an adaptive model reduction method applicable for three-dimensional nonlinear magnetic field problems. Firstly, the full-order FEM was used to model and simulate the research object, and the FEM solutions corresponding to different sample points were calculated as the sample library. Secondly, the POD was combined with radial basis function (RBF) interpolation, while the ISSA-POD-RBF reduced-order model considering optimal combination of RBF width parameters was constructed based on the ISSA; Finally, the iterative computation of snapshot matrix and optimal width parameter combination was carried out by the greedy strategy to construct a fast computational model that more closely matches the original full-order model. To verify the effectiveness of the proposed model reduction method, the TEAM24 problem and a traction transformer are taken as the research objects.The speed and accuracy of the reduction model are investigated by comparing computational results and computation time of the reduction model with the FEM full-order model. In the TEAM24 problem, a sample library containing 100 samples corresponding to the FEM solutions is prepared in advance, and the iteration termination condition of the greedy strategy is set to the maximum relative error e2max<0.5%. The final size of the snapshot matrix is determined to be (118 831×3)×13, and the POD basis retains the order of 2nd order.The results show thatthe relative errors of the samples computed by the reduced-order model constructed with the optimal combination of width parameters were overall smaller than the computational errors of the reduced-order model constructed with the same width parameters. The single-point error of flux density comparison at local nodes does not exceed 0.01%. The computational times of the proposed reduced-order model and the FEM full-order model are 205 s and 0.56 s, respectively, and the acceleration ratio is 366.07. In the three-dimensional model, the size of the finalized snapshot matrix is (125 403×3)×40, and the POD basis retains the order of 11th order. The results show that the computational errors of the model are in the same order of magnitude as those introduced during simulation due to uncertainties in material properties, dimensions, and simulation operations. The computational times of the proposed reduced-order model and the FEM full-order model are 386.4 s and 3.9 s, respectively, with an acceleration ratio of 99.08. The proposed fast solution method for the nonlinear magnetic field problem of electrical equipment provides a real-time simulation solution for the construction of digital twin models of electromagnetic performance of electrical equipment.
Liu Yutong,Ren Ziyan,Chi Lianqiang等. Fast Calculation Method of 3D Nonlinear Magnetic Field Based on Adaptive Model Order Reduction[J]. Transactions of China Electrotechnical Society, 2025, 40(1): 1-12.
[1] Ghnatios C, Masson F, Huerta A, et al.Proper Generalized Decomposition based dynamic data-driven control of thermal processes[J]. Computer Methods in Applied Mechanics and Engineering, 2012, 213: 29-41. [2] Chakraborty S, Adhikari S, Ganguli R.The role of surrogate models in the development of digital twins of dynamic systems[J]. Applied Mathematical Modelling, 2021, 90: 662-681. [3] 张宁, 刘士利, 郝建, 等. 变压器油中气泡杂质相局部放电特性研究综述[J]. 电工技术学报, 2023, 38(10): 2757-2776. Zhang Ning, Liu Shili, Hao Jian, et al.Review on partial discharge characteristics of bubble impurity phase in transformer oil[J].Transactions of China Electrotechnical Society, 2023, 38(10): 2757-2776. [4] 张宇娇, 赵志涛, 徐斌, 等. 基于U-net卷积神经网络的电磁场快速计算方法[J]. 电工技术学报, 2024, 39(19): 2730-2742. Zhang Yujiao, Zhao Zhitao, Xu Bin et al. Fast calculation method of electromagnetic field based on U-net convolution neural network[J].Transactions of China Electrotechnical Society, 2024, 39(19): 2730-2742. [5] 谭皓元, 兰志勇, 罗元钧, 等. 无铁心圆筒型永磁同步直线电动机磁场分析[J]. 电气技术, 2023, 24(7): 39-46. Tan Haoyuan, Lan Zhiyong, Luo Yuanjun, et al.Design and magnetic field analysis of ironless tubular permanent magnet synchronous linear motor[J]. Electrical Engineering, 2023, 24(7): 39-46. [6] 曹龙飞, 范兴纲, 李大伟, 等. 基于快速有限元的永磁电机绕组涡流损耗半解析高效计算[J]. 电工技术学报, 2023, 38(1): 153-165. Cao Longfei, Fan Xinggang, Li Dawei, et al.Semi analytical and efficient calculation method of eddy current loss in windings of permanent magnet machines based on fast finite element method[J]. Transactions of China Electrotechnical Society, 2023, 38(1): 153-165. [7] 谢飞. 基于POD的电磁场非线性问题快速计算方法研究[D]. 沈阳: 沈阳工业大学, 2022. Xie Fei.Research on fast calculation method of electromagnetic field nonlinear problem based on POD[D]. Shenyang University of Technology, 2022. [8] 王进钊, 严干贵, 刘侃. 基于交替方向隐式平衡截断法的直驱风电场次同步振荡分析的模型降阶研究[J]. 发电技术, 2023, 44(6): 850-858. Wang Jinzhao, Yan Gangui, Liu Kan.Research on model reduction of direct drive wind farm sub-synchronous oscillation analysis based on alternating direction implicit balanced truncation method[J]. Power Generation Technology, 2023, 44(6): 850-858. [9] 田野, 卜凯阳, 李楚杉, 等. 用于IGBT模块温度观测的3-D降阶混合型热模型[J]. 电工技术学报, 2024, 39(16): 5104-5120. Tian Ye, Bu Kaiyang, Li Chushan, et al.A hybrid 3-D reduced-order thermal model for temperature observation of IGBT modules[J]. Transactions of China Electrotechnical Society, 2024, 39(16): 5104-5120. [10] 刘刚, 胡万君, 刘云鹏, 等. 降阶技术与监测点数据融合驱动的油浸式变压器绕组瞬态温升快速计算方法[J]. 电工技术学报, 2024, 39(19): 6162-6174. Liu Gang, Hu Wanjun, Liu Yunpeng, et al.A fast calculation method for transient temperature rise of oil immersed transformer windings driven by fusion of order reduction technology and monitoring point data[J]. Transactions of China Electrotechnical Society, 2024, 39(19): 6162-6174. [11] 邱亚松, 白俊强, 华俊. 基于本征正交分解和代理模型的流场预测方法[J]. 航空学报, 2013, 34(6): 1249-1260. Qiu Yasong, Bai Junqiang, Hua Jun.Flow field estimation method based on proper orthogonal decomposition and surrogate model[J]. Acta Aeronautica et Astronautica Sinica, 2013, 34(6): 1249-1260. [12] 刘钊, 王沐晨, 李金玖, 等. 基于POD的尾流激励叶片气动力降阶模型[J]. 兵器装备工程学报, 2021, 42(2): 82-86, 173. Liu Zhao, Wang Muchen, Li Jinjiu, et al.Reduced order aerodynamic model of wake excited blade based on POD[J]. Journal of Ordnance Equipment Engineering, 2021, 42(2): 82-86, 173. [13] Xie Dan, Xu Min.A comparison of numerical and semi-analytical proper orthogonal decomposition methods for a fluttering plate[J]. Nonlinear Dynamics, 2015, 79(3): 1971-1989. [14] 魏代同, 陈星, 陈玉刚, 等. 基于本征正交分解的叶片碰摩系统降阶方法[J]. 航空动力学报, 2022, 37(4): 711-720. Wei Daitong, Chen Xing, Chen Yugang, et al.Reduced order method of blade rubbing system based on proper orthogonal decomposition[J]. Journal of Aerospace Power, 2022, 37(4): 711-720. [15] 胡金秀, 高效伟. 变系数瞬态热传导问题边界元格式的特征正交分解降阶方法[J]. 物理学报, 2016, 65(1): 273-289. Hu Jinxiu, Gao Xiaowei.Reduced order model analysis method via proper orthogonal decomposition for variable coefficient of transient heat conduction based on boundary element method[J]. Acta Physica Sinica, 2016, 65(1): 273-289. [16] 朱强华, 杨恺, 梁钰, 等. 基于特征正交分解的一类瞬态非线性热传导问题的新型快速分析方法[J]. 力学学报, 2020, 52(1): 124-138. Zhu Qianghua, Yang Kai, Liang Yu, et al.A novel fast algorithm based on model order reduction for one class of transient nonlinear heat conduction problem[J]. Chinese Journal of Theoretical and Applied Mechanics, 2020, 52(1): 124-138. [17] Pierquin A, Henneron T, Clénet S, et al.Model-order reduction of magnetoquasi-static problems based on POD and arnoldi-based Krylov methods[J]. IEEE Transactions on Magnetics, 2015, 51(3): 7206204. [18] Farzam Far M, Martin F, Belahcen A, et al.Orthogonal interpolation method for order reduction of a synchronous machine model[J]. IEEE Transactions on Magnetics, 2018, 54(2): 8100506. [19] 蒋耀林. 模型降阶方法[M]. 北京: 科学出版社, 2010. Jiang Yaolin.Model reduction methods [M]. Beijing: Science Press, 2010. [20] Sato Y, Clemens M, Igarashi H.Adaptive subdomain model order reduction with discrete empirical interpolation method for nonlinear magneto-quasi-static problems[J]. IEEE Transactions on Magnetics, 2016, 52(3): 1100204. [21] Henneron T, Montier L, Pierquin A, et al.Comparison of DEIM and BPIM to speed up a POD-based nonlinear magnetostatic model[J]. IEEE Transactions on Magnetics, 2017, 53(6): 7204204. [22] Müller F, Siokos A, Kolb J, et al.Efficient estimation of electrical machine behavior by model order reduction[J]. IEEE Transactions on Magnetics, 2021, 57(6): 8105804. [23] 谢裕清, 李琳. 基于本征正交分解的换流变压器极性反转电场降阶模型[J]. 中国电机工程学报, 2016, 36(23): 6544-6551, 6622. Xie Yuqing, Li Lin.A reduced order model via proper orthogonal decomposition for polarity reverse electric fields in converter transformers[J]. Proceedings of the CSEE, 2016, 36(23): 6544-6551, 6622. [24] 胡万君, 刘刚, 朱章宸, 等. 油浸式电力变压器绕组稳态温升降阶计算方法研究[J]. 中国电机工程学报, 2023, 43(16): 6505-6516. Hu Wanjun, Liu Gang, Zhu Zhangchen, et al.Study on calculation method of steady-state temperature rise and fall order of oil-immersed power transformer winding[J]. Proceedings of the CSEE, 2023, 43(16): 6505-6516. [25] Farzam Far M, Martin F, Belahcen A, et al.Orthogonal interpolation method for order reduction of a synchronous machine model[J]. IEEE Transactions on Magnetics, 2018, 54(2): 8100506. [26] Pierquin A, Henneron T.Nonlinear data-driven model order reduction applied to circuit-field magnetic problem[J]. IEEE Transactions on Magnetics, 2021, 57(11): 7402509. [27] 刘刚, 高成龙, 胡万君, 等. 基于鲸鱼优化算法超参数优化的径向基函数响应面模型的油浸式变压器绕组挡板结构优化[J]. 电工技术学报, 2024, 39(17): 5331-5343. Liu Gang, Gao Chenglong, Hu Wanjun, et al.optimization of winding block washer structure for oil immersed transformers based on radial basis function response surface model with whale optimization algorithm hyper-parameters optimization[J]. Transactions of China Electrotechnical Society, 2024, 39(17): 5331-5343. [28] 刘刚, 高成龙, 胡万君, 等. 基于改进的连续局部枚举采样和径向基函数响应面法的变压器静电环结构优化设计[J]. 电工技术学报, 2023, 38(23): 6266-6278. Liu Gang, Gao Chenglong, Hu Wanjun, et al.optimized design of transformer electrostatic ring based on radial basis function response surface method with enhanced successful local enumeration sampling[J]. Transactions of China Electrotechnical Society, 2023, 38(23): 6266-6278. [29] 余波, 陶盈盈. 基于POD-RBF方法的管道内壁几何识别[J]. 应用数学和力学, 2023, 44(4): 406-418. Yu Bo, Tao Yingying.Identification of pipeline inner wall geometry based on the POD-RBF method[J]. Applied Mathematics and Mechanics, 2023, 44(4): 406-418. [30] Wu Zeping, Wang Donghui, Patrick Okolo N, et al.Unified estimate of Gaussian kernel width for surrogate models[J]. Neurocomputing, 2016, 203: 41-51. [31] 李昕燃, 靳伍银. 基于改进麻雀算法优化支持向量机的滚动轴承故障诊断研究[J]. 振动与冲击, 2023, 42(6): 106-114. Li Xinran, Jin Wuyin.Fault diagnosis of rolling bearings based on ISSA-SVM[J]. Journal of Vibration and Shock, 2023, 42(6): 106-114. [32] Beccar-Varela M P, Gonzalez-Huizar H, Mariani M C, et al. Lévy flights and wavelets analysis of volcano-seismic data[J]. Pure and Applied Geophysics, 2020, 177(2): 723-736.
2025, Vol. 40 
