基于预充电模型与RSNA的直流支撑电容器电容量辨识方法  

A Capacitance Estimation Method for DC-link Capacitors Based on Pre-charging Model and RSNA Algorithm

作　　者：伍珣 于天剑[1] 李凯迪 田睿 WU Xun;YU Tianjian;LI Kaidi;TIAN Rui(School of Traffic and Transportation Engineering,Central South University,Changsha 410075,China;Shenzhen Metro Group Co.,Ltd.,Shenzhen 518040,China;State Grid Hunan Extra High Voltage Substation Company,Changsha 410004,China)

机构地区：[1]中南大学交通运输工程学院,湖南长沙410075 [2]深圳地铁集团,广东深圳518040 [3]国网湖南超高压变电公司,湖南长沙410004

出　　处：《铁道科学与工程学报》2023年第7期2664-2675,共12页Journal of Railway Science and Engineering

基　　金：国家自然科学基金资助项目(52072414)。

摘　　要：直流支撑电容器是牵引变流系统的关键器件,也是极易出现故障的电气部件。由于电压、充放电频率、温度等因素的影响,电容器的老化速度加快,其实际寿命与制造商手册中的数据相差很大。因此,有必要对直流支撑电容器状态进行辨识,从而保障列车牵引变流系统安全运行。目前,已有较多学者对电容器状态辨识开展研究。但是,铁路应用中的直流支撑电容器状态辨识噪声问题尚未得到很好的解决。在某些情况下,电压传感器测量噪声与电压纹波分量相近会导致直流支撑电容器状态辨识结果出现较大偏差。针对上述问题,提出一种基于预充电模型与递推随机牛顿法(RSNA)的直流支撑电容器电容量辨识方法。在利用现有电压传感器信号以及不对原有系统进行改动的前提下建立电容器预充电模型,采用RSNA算法对电容量进行参数辨识,可以有效抵御噪声干扰并实现电容量的准确估计。该方法原理简单、计算量小,不需要加装额外传感器,以较低的采样频率即可达到较高的计算精度。同时,该方法不易受信噪比(SNR)与信号偏移等的影响,具有较好的鲁棒性。验证结果表明,该方法在正常情况下的辨识误差可以控制在3%以内,在信噪比为35 dB时仍然可以保持在5%左右,当信号偏移达到±10 V时可以保持在3%左右。DC-link capacitors are important parts of traction converters as well as the vulnerable electronic parts.Due to the influence of voltage,charge-discharge frequency,temperature and other factors,the aging of the capacitor is accelerated,and its actual life is very different from the data in the manufacturer’s manual.Therefore,the state estimation of capacitors is necessary and of great significance to the safe operation of trains.Currently,various studies are presented for the capacitor parameter estimations,but the noise problem existing in DC-link capacitor estimation has not been well solved.In some cases,the noise is basically equal to the ripple component,which will lead to a large deviation in the state identification results of DC-link capacitors.Thus,a capacitance estimation method based on pre-charging model and RSNA was proposed for DC-link capacitors in this paper.Existing voltage sensor signals were used for estimation and the modification of the system was not required.The pre-charging model was built and RSNA was utilized for the capacitance calculation.The impact of noise was well decreased and good estimation results could be obtained.Moreover,the proposed method is simple,extra sensor is not required,and low sampling frequency is used.The proposed method is also robust to SNR and signal bias.Experimental results show that the estimation error of the proposed method is within 3%under normal operations,about 5%when SNR is 35 dB,and 3%when signal bias is over±10 V.

关 键 词：电容器 状态辨识 预充电模型 RSNA 

分 类 号：TM464[电气工程—电器]