Parameter Identification of DC-Link Capacitor in Traction Converter Based on Physical Information Neural Network
Xiang Chaoqun1, Yin Xueyao1, Wu Xun1, Cao Zhonglin2, Liu Yuancai2
1. School of Transportation Engineering Central South University Changsha 410075 China; 2. Tianjin Rail Transit Operation Group Co. Ltd Tianjin 300000 China
Abstract：At present, the fault monitoring dilemma of DC-link capacitors in traction converters mainly focuses on the noise interference of sensor measurement, the aging speed of the capacitor is affected by various environmental factors, and the accuracy and stability requirements are high. Therefore, this paper proposes a DC-link capacitor parameter identification method for traction converter based on physical information neural network (PINN) and capacitor pre-charging model. The sampling frequency of the voltage sensor is very low, and there is no need for the capacitor pre-charging voltage curve to be strictly aligned with the time axis, effectively reducing the influence of the measurement signal-to-noise ratio on the prediction results. Moreover, the amount of capacitance data required is minimal using the cycle consensus generative adversarial network (CycleGAN) algorithm, which can be applied to a wide range of capacitance intervals under the same topology. This method can be applied to rail transit applications. The architecture design of the adaptive physical information neural network model in this method is described, including the construction of partial differential equations and boundary conditions, physical constraint loss function modeling, network structure, and training process. Regarding the adaptive weight PINN execution part, the AdamW algorithm is used to update the adaptive weight and linear network weight of each loss component. After the number of iterations is set, the L-BFGS algorithm performs a new round of iterative optimization on the linear network weight part of the current PINN network. However, the adaptive weight is no longer updated. During the training process using the PINN network, when the number of iterations is greater than the specified number limit and the mean square error of the test set is less than the set number, the CycleGAN model is used to generate data, and the current PINN network is used as the constraint condition label. After adding the generated training set, the training results are monitored by the test set’s accuracy after each iteration. Suppose the accuracy is unchanged or improved in a fixed iteration interval, the generated training set is considered credible and effective, and the generated data set is updated before the next fixed iteration interval is turned on. The laboratory low-power prototype platform is used for example analysis. The results show that under normal conditions, the identification relative error of this method is about 1%. When the signal-to-noise ratio reaches 30 dB, the identification relative error can still be controlled within 5%, and reducing the sampling frequency can alleviate the effect of signal-to-noise ratio reduction. This method provides a new idea for solving the problem of state parameter identification of DC-link capacitors and ensuring the safe operation and long-term stability of DC-link capacitor of traction converters.
向超群, 尹雪瑶, 伍珣, 曹忠林, 刘元才. 基于物理信息神经网络的牵引变流器直流支撑电容参数辨识方法[J]. 电工技术学报, 2024, 39(15): 4654-4667.
Xiang Chaoqun, Yin Xueyao, Wu Xun, Cao Zhonglin, Liu Yuancai. Parameter Identification of DC-Link Capacitor in Traction Converter Based on Physical Information Neural Network. Transactions of China Electrotechnical Society, 2024, 39(15): 4654-4667.
[1] 卓超然, 张笑天, 张雄, 等. 支撑电容可分离的直流变压器短路故障电流限流控制方法[J]. 电工技术学报, 2022, 37(2): 424-432. Zhuo Chaoran, Zhang Xiaotian, Zhang Xiong, et al.Short-circuit fault current limiting control method of DC transformer with separable supporting capaci- tor[J]. Transactions of China Electrotechnical Society, 2022, 37(2): 424-432. [2] 伍珣, 于天剑, 李凯迪, 等. 基于预充电模型与RSNA的直流支撑电容器电容量辨识方法[J]. 铁道科学与工程学报, 2023, 20(7): 2664-2675. Wu Xun, Yu Tianjian, Li Kaidi, et al.A capacitance estimation method for DC-link capacitors based on pre-charging model and noise evaluation[J]. IEEE Transactions on Industrial Electronics, 2023, 20(7): 2664-2675. [3] Rachev E, Petrov V.DC link capacitor selection for DC-DC converters[C]//2020 12th Electrical Engin- eering Faculty Conference (BulEF), Varna, Bulgaria, 2020: 1-5. [4] Zhao Nannan, Gao Runfeng, Wang Gaolin, et al.Online estimation method of DC-link capacitors for reduced DC-link capacitance IPMSM drives[J]. IEEE Transactions on Power Electronics, 2021, 36(11): 12196-12201. [5] 罗丹, 陈民铀, 赖伟, 等. 基于Haar小波变换重构开关序列的MMC子模块电容值在线监测方法[J]. 电工技术学报, 2022, 37(20): 5278-5289. Luo Dan, Chen Minyou, Lai Wei, et al.Online monitoring method for sub-module capacitance in modular multilevel converter based on Haar wavelet transform reconstruction switch sequence[J]. Transa- ctions of China Electrotechnical Society, 2022, 37(20): 5278-5289. [6] Zhao Hu, Shen Jianwu, Zhang Yizhou, et al.An intelligent tension sensor for high speed railway catenary monitoring[C]//2022 IEEE 2nd International Conference on Power, Electronics and Computer Applications (ICPECA), Shenyang, China, 2022: 500-503. [7] 周志恒, 贺德强, 陈彦君, 等. 基于GA-BP神经网络的列车关键部件预防性维修优化模型及应用[J]. 铁道科学与工程学报, 2021, 18(6): 1382-1391. Zhou Zhiheng, He Deqiang, Chen Yanjun, et al.Optimization model and application of preventive maintenance of key components of train based on GA-BP neural network[J]. Journal of Railway Science and Engineering, 2021, 18(6): 1382-1391. [8] Zhao Zhaoyang, Davari P, Lu Weiguo, et al.An overview of condition monitoring techniques for capacitors in DC-link applications[J]. IEEE Transa- ctions on Power Electronics, 2021, 36(4): 3692-3716. [9] Soualhi A, Makdessi M, German R, et al.Heath monitoring of capacitors and supercapacitors using the neo-fuzzy neural approach[J]. IEEE Transactions on Industrial Informatics, 2018, 14(1): 24-34. [10] McGrew T, Sysoeva V, Cheng Chihao, et al. Con- dition monitoring of DC-link capacitors using time- frequency analysis and machine learning classifi- cation of conducted EMI[J]. IEEE Transactions on Power Electronics, 2022, 37(10): 12606-12618. [11] Soliman H, Wang Huai, Gadalla B, et al.Condition monitoring for DC-link capacitors based on artificial neural network algorithm[C]//2015 IEEE 5th Inter- national Conference on Power Engineering, Energy and Electrical Drives (POWERENG), Riga, Latvia, 2015: 587-591. [12] Hu Jiayu, Qiu Guanqun, Wang Weiyao, et al.An on-line capacitor condition monitoring method based on switching frequencies for modular multilevel converters[C]//2021 4th International Conference on Energy, Electrical and Power Engineering (CEEPE), Chongqing, China, 2021: 183-187. [13] 伍珣, 田睿, 李凯迪, 等. 基于预充电模型与低频采样的直流支撑电容器电容辨识方法[J]. 中南大学学报(自然科学版), 2023, 54(8): 3325-3334. Wu Xun, Tian Rui, Li Kaidi, et al.Pre-charging model based capacitance estimation method for DC-link capacitors with low sampling frequency[J]. Journal of Central South University (Science and Technology), 2023, 54(8): 3325-3334. [14] Wang Zhongxu, Zhang Yi, Wang Huai, et al.Capacitor condition monitoring based on the DC-side start-up of modular multilevel converters[J]. IEEE Transactions on Power Electronics, 2020, 35(6): 5589-5593. [15] Wu Yu, Du Xiong.A VEN condition monitoring method of DC-link capacitors for power converters[J]. IEEE Transactions on Industrial Electronics, 2019, 66(2): 1296-1306. [16] Li Hao, Xiang Dawei, Han Xu, et al.High-accuracy capacitance monitoring of DC-link capacitor in VSI systems by LC resonance[J]. IEEE Transactions on Power Electronics, 2019, 34(12): 12200-12211. [17] Zhao Shuai, Peng Yingzhou, Zhang Yi, et al.Parameter estimation of power electronic converters with physics-informed machine learning[J]. IEEE Transactions on Power Electronics, 2022, 37(10): 11567-11578. [18] Gao Yihang, Cheung K C, Ng M K.SVD-PINNs: transfer learning of physics-informed neural networks via singular value decomposition[C]//2022 IEEE Symposium Series on Computational Intelligence (SSCI), Singapore, 2022: 1443-1450. [19] Liu Jie, Zhang Rongwei, Luan Mengxiao, et al.GL-PINN algorithm for inferring velocity and pressure fields from sparse concentration field[C]// 2022 IEEE 17th International Conference on Control & Automation (ICCA), Naples, Italy, 2022: 326-331. [20] Wang Yuzhang, Alkhadhr S, Almekkawy M.PINN simulation of the temperature rise due to ultrasound wave propagation[C]//2021 IEEE International Ultrasonics Symposium (IUS), Xi’an, China, 2021: 1-4. [21] Zhao Zhenyu, Fan Fei, Sun Quqin, et al.Physics informed neural network-based high-frequency modeling of induction motors[J]. Chinese Journal of Electrical Engineering, 2022, 8(4): 30-38. [22] Stiasny J, Misyris G S, Chatzivasileiadis S.Physics- informed neural networks for non-linear system identification for power system dynamics[C]//2021 IEEE Madrid PowerTech, Madrid, Spain, 2021: 1-6. [23] Lu Lu, Meng Xuhui, Mao Zhiping, et al.DeepXDE: a deep learning library for solving differential equations[J]. SIAM Review, 2021, 63(1): 208-228. [24] Misyris G S, Venzke A, Chatzivasileiadis S.Physics- informed neural networks for power systems[C]//2020 IEEE Power & Energy Society General Meeting (PESGM), Montreal, QC, Canada, 2020: 1-5. [25] Bin Waheed U.Kronecker neural networks overcome spectral bias for PINN-based wavefield com- putation[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 8029805. [26] 臧国强, 刘晓莉, 徐颖菲, 等. 深度学习在电力设备缺陷识别中的应用进展[J]. 电气技术, 2022, 23(6): 1-7. Zang Guoqiang, Liu Xiaoli, Xu Yingfei, et al.Application progress of deep learning in power equipment defect identification[J]. Electrical Engin- eering, 2022, 23(6): 1-7. [27] 徐宝昌, 张学智, 王雅欣, 等. 用于两相流环空压力预测的自适应物理信息神经网络模型[J]. 石油学报, 2023, 44(3): 545-555. Xu Baochang, Zhang Xuezhi, Wang Yaxin, et al.self-adaptive physical information neural network model for prediction of two-phase flow annulus pressure[J]. Journal of Petroleum, 2023, 44(3): 545-555. [28] 仲林林, 胡霞, 刘柯妤. 基于改进生成对抗网络的无人机电力杆塔巡检图像异常检测[J]. 电工技术学报, 2022, 37(9): 2230-2240, 2262. Zhong Linlin, Hu Xia, Liu Keyu.Power tower anomaly detection from unmanned aerial vehicles inspection images based on improved generative adversarial network[J]. Transactions of China Elec- trotechnical Society, 2022, 37(9): 2230-2240, 2262. [29] 郭亮, 王祥业, 姜文聪. 基于生成对抗网络的注入电流式热声成像逆问题研究[J]. 电工技术学报, 2021, 36(增刊1): 22-30. Guo Liang, Wang Xiangye, Jiang Wenchong.The study on the inverse problem of applied current thermo-acoustic imaging based on generative adversarial network[J]. Journal of Electrical Engin- eering Technology, 2021, 36(S1): 22-30. [30] Bi Xiaoyang, He Enhao, Wang Qipeng.Research on style transfer model based on improved CycleGAN algorithm[C]//2023 IEEE 5th International Con- ference on Civil Aviation Safety and Information Technology (ICCASIT), Dali, China, 2023: 129-135. [31] Zhang Xingwei, Zhang Ke, Xie Lingwang, et al.An enhancement and detection method for a glue dispensing image based on the CycleGAN model[J]. IEEE Access, 2022, 10: 92036-92047. [32] 李广全, 刘志明, 张子璠, 等. 基于EMD和ISA方法的轨道车辆动态应变信号去工频干扰研究[J]. 北京交通大学学报, 2018, 42(1): 106-112. Li Guangquan, Liu Zhiming, Zhang Zifan, et al.Research on removing power frequency interference on dynamic strain signal of rolling stock based on EMD and ISA algorithm[J]. Journal of Beijing Jiaotong University, 2018, 42(1): 106-112.
2024, Vol. 39 
