-
小波分解的过程可以近似看作高通滤波器和低通滤波器的组合,示意图如图3(a)所示。其中,S为待处理的信号序列,A表示该信号的低频分量,D则表示该信号的高频分量。信号消噪的简化过程如图3(b)所示,cA和cD分别对应上述两部分信号的小波系数,下标数字则表示当前的分解层次。
具体计算过程如下所示。首先信号f(t)∈L2(R)可以按照以下关系展开:
$$ f(t) = \sum\limits_k {{d_{j,k}} \cdot {\varphi _{j,k}}(t)} + \sum\limits_k {{c_{j,k}} \cdot {\phi _{j,k}}(t)} $$ (1) 式中:
j ——预先设置的分解尺度;
φ(t)、Φ(t)——分解过程中的小波函数和尺度函数;
dj,k、cj,k ——分解尺度j所对应的小波系数和尺度系数。其递推关系式可表示为:
$$ \begin{gathered} {c_{j + 1,k}} = \sum\limits_m {{h_0} \cdot (m - 2k)} \cdot {c_{j,m}} \\ {d_{j + 1,k}} = \sum\limits_m {{h_1} \cdot (m - 2k)} \cdot {c_{j,m}} \\ \end{gathered} $$ (2) 式中:
h0(n)、h1(n)——展开系数,与分解尺度无关。固定计算式为:
$$ \begin{gathered} {h_0}(n) = \left\langle {\phi (t),{\phi _{ - 1,n}}(t)} \right\rangle \\ {h_1}(n) = \left\langle {\varphi (t),{\phi _{ - 1,n}}(t)} \right\rangle \\ \end{gathered} $$ (3) 对小波系数和尺度系数进行重组,成为新的小波分解结构。再运用逆变换对这些系数作重构处理,即可以复现为去除噪声干扰的信号。其递推关系式为:
$$ {c_{j - 1,m}} = \sum\limits_k {{h_0}(m - 2k) \cdot {c_{j,k}}} + \sum\limits_k {{h_1}(m - 2k) \cdot {d_{j,k}}} $$ (4) 为了防止消噪处理后含噪声信号的不连续邻域中可能会出现振荡现象,引入平移不变量小波消噪法。假设待处理信号为s(t)(0tn),其中t为时间,n为信号序列的长度。则经过循环平移的消噪法表达式为:
$$ \bar T(s(t),{S_h}) = \frac{1}{h} \cdot \sum {{S_{ - h}}(T({S_h} \cdot s(t)))} $$ (5) 式中:
h ——平移量,正整数且有0hn;
Sh ——信号序列循环平移h的算子,即Sh·s(t)为经过h循环平移后的信号;
T ——对信号进行阈值消噪处理。
为了检验重构信号中的噪声抑制效果,从而选取最佳重构参数,假设给定的待检验信号序列为{x(n),n=1,2,···,N},则该信号序列的自相关系数可由以下表达式计算得到:
$$ {\rho _k} = \frac{{\sum\limits_{n = 1}^{N - k} {\left( {{x_n} - {{\bar x}_N}} \right) \cdot \left( {{x_{n + k}} - {{\bar x}_N}} \right)} }}{{\sum\limits_{n = 1}^{N - k} {{{\left( {{x_n} - {{\bar x}_N}} \right)}^2}} }} $$ (6) $$ {\bar x_N} = \frac{1}{N} \cdot \sum\limits_{n = 1}^N {{x_n}} $$ (7) 式(6)中,{ρk, k=1,2,···,m}表示给定信号序列的自相关系数估计值;自由度参数m2N,通常取值为5~10。根据时间序列分析理论,若该序列为白噪声信号,则当N足够大时,近似服从于m维标准正态分布。进一步推导知χ2(m)=N·(ρ12+ρ22+···+ρm2)将会近似服从于m维χ2分布。换言之,若序列表现为白噪声的特征,也即含噪声信号的高频分量中仍然存在噪声干扰,则可以得到以下关系式:
$$ \rho _1^2 + \rho _2^2 + \cdots + \rho _m^2 = 0 $$ (8) 假设给定的显著性水平为a,根据自由度m查表得χ2分布的临界值λa,则检验统计量χ2必然符合以下关系式:
$$ {\chi ^{\text{2}}}\left( m \right) = N \cdot \left( {\rho _1^2 + \rho _2^2 + \cdots + \rho _m^2} \right) < {\lambda _\alpha } $$ (9) 反之,当检验统计量χ2>λa时,则表示该序列是相关序列。此时,含噪声信号中的高频分量均为有用信号,噪声干扰已被消除。结合上述原理,设计最优小波分解尺度的判断流程如图4所示。
-
为了验证小波分解尺度选择算法的可行性,首先产生标准测试信号来模拟泄漏电流信号,表达式如下:
$$ {s_{\text{2}}}(t) = 100 \times \sum\limits_k {{A_k} \cdot \sin (2\pi {f_k} \cdot t)} $$ (10) 式中,k=1,3,5;A1=1 mA,A3=A5=0.03 mA。即包含3次、5次谐波分量的正弦波形。然后将高斯白噪声叠加于标准测试信号中,如图5所示。其中,干扰信号由噪声函数随机生成,同样取其均值为0,标准差为20,服从高斯分布。
首先,针对叠加高斯白噪声的标准测试信号,运用平移不变量小波变换进行消噪处理,部分结果如图6所示。在本次测试过程中,统一选择sym8作为小波变换的基函数,并依次取分解尺度j=1~9。从图中可以看出,当分解尺度j=5时,经过消噪处理后的测试信号波形与其有用信号部分最为接近。
取自由度m=5,显著性水平a=0.01,查表得2分布的临界值λa=15.086。在不同的分解尺度下,分别计算测试信号高频分量对应的小波系数的检验统计量。结果列于表1。从表中数据可以看出,当分解尺度j=5时,检验统计量大于临界值,应拒绝原假设,即是相关序列。这表示高频分量的小波系数均为有用信号,噪声干扰已被消除。信噪比R同样在该分解尺度下达到最大值,此时的消噪效果达到最优。
表 1 检验统计量及信噪比
Table 1. Test statistics and signal-to-noise ratio
j 1 2 3 4 5 6 7 8 9 χ2 1.98 10.91 13.81 11.26 20.14 41.35 128.11 69.90 171.08 R 14.52 17.65 20.80 23.47 24.89 23.47 21.75 19.03 18.51 然后针对叠加高斯白噪声的泄漏电流原始信号,运用平移不变量小波变换进行消噪处理,部分结果如图7所示。此处同样选择sym8作为小波变换过程的基函数,并依次取分解尺度j=1~9。从图中可以看出,当j=5时,很好地消除了噪声干扰和高次谐波分量,且未过度消噪,较为符合泄漏电流有用信号的波形。
取和上述相同的自由度以及显著水平。在不同分解尺度下,分别计算高频分量对应的小波系数的检验统计量。当j=5时,检验统计量大于临界值。这表示高频分量的小波系数均为有用信号,噪声干扰已被消除,与图6中对波形的判断结果相符合。
Leakage Current Detection for Surge Arrester in Switchgear Based on Time-Frequency Analysis Method
-
摘要:
目的 在电力系统中,开关柜避雷器承担着抑制瞬态过电压和泄放脉冲大电流的重要作用,对于维持其正常稳定运行具有重大意义。 方法 为了有效抑制噪声对泄漏电流信号检测的干扰,提出了一种基于自相关系数与卡方检验优化的时频分析方法。首先通过电流传感器和分流器两种测量结果进行分析,然后利用基于自相关系数与卡方检验优化的小波变换消除信号中的噪声干扰,实现最优分解尺寸的确定,从而更好地适应小信噪比场合。 结果 通过软件平台分析得出在分流器的测量基础上利用优化后的小波算法去噪的抗干扰能力更强,波形质量更好。最后研制了一款泄漏电流在线检测装置,对提出的泄漏电流检测模型加以验证。 结论 实验结果表明该装置能够较好地实现避雷器泄漏电流的实时检测。 Abstract:Introduction In the power system, the surge arrester in switchgear plays an important role in suppressing transient overvoltage and discharging large pulse currents, which is of great significance to maintain its normal and stable operation. Method A time-frequency analysis method based on the autocorrelation coefficient and the Chi-squared test optimization was proposed in the research to effectively suppress the noise interference on leakage current signal detection. Firstly, the measurement results of the current sensor and the shunt were analyzed. Then, the wavelet transform based on the autocorrelation coefficient and the Chi-squared test optimization was used to eliminate the noise interference in the signal in order to determine the optimal decomposition size to better suit the occasion of small signal-noise ratio. Result Based on the analysis in the software platform, it is concluded that the optimized wavelet algorithm denoising has stronger anti-interference capability and better waveform quality on the basis of measurement of the shunt. Finally, an online leakage current detection device is developed to verify the leakage current detection model proposed in the research. Conclusion The experimental results show that the device can realize the real-time detection of leakage current of the surge arresters. -
表 1 检验统计量及信噪比
Tab. 1. Test statistics and signal-to-noise ratio
j 1 2 3 4 5 6 7 8 9 χ2 1.98 10.91 13.81 11.26 20.14 41.35 128.11 69.90 171.08 R 14.52 17.65 20.80 23.47 24.89 23.47 21.75 19.03 18.51 -
[1] 巫聪云, 刘斌, 沈梓正, 等. 二次设备集中安防运维及主动感知和诊断预警关键技术的研究 [J]. 南方能源建设, 2021, 8(4): 85-94. DOI: 10.16516/j.gedi.issn2095-8676.2021.04.012. WU C Y, LIU B, SHEN Z Z, et al. Research on secondary equipment centralized security operation and maintenance and key technologies of active perception, diagnosis and early warning [J]. Southern Energy Construction, 2021, 8(4): 85-94. DOI: 10.16516/j.gedi.issn2095-8676.2021.04.012. [2] 姜绍艳, 李京平, 蔡志平. 智能分布式加电压-时间型双策略就地故障判别方法 [J]. 南方能源建设, 2020, 7(增刊1): 48-52. DOI: 10.16516/j.gedi.issn2095-8676.2020.S1.009. JIANG S Y, LI J P, CAI Z P. Intelligent distributed voltage-time dual strategy local fault identification method [J]. Southern Energy Construction, 2020, 7(Supp. 1): 48-52. DOI: 10.16516/j.gedi.issn2095-8676.2020.S1.009. [3] 郑幸, 陈逸鹏, 齐冬辉. 输电线路三相不平衡电流及高抗过补偿仿真分析 [J]. 南方能源建设, 2021, 8(2): 78-84. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.012. ZHENG X, CHEN Y P, QI D H. Simulation analysis of three phase unbalanced current and over compensation of high voltage reactor in transmission line [J]. Southern Energy Construction, 2021, 8(2): 78-84. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.012. [4] 陈红武. 高压开关柜局部放电特性的检测研究 [J]. 电工材料, 2020(5): 40-42+47. DOI: 10.16786/j.cnki.1671-8887.eem.2020.05.011. CHEN H W. Research on detection of partial discharge characteristics of high voltage switchgear [J]. Electrical Engineering Materials, 2020(5): 40-42+47. DOI: 10.16786/j.cnki.1671-8887.eem.2020.05.011. [5] 张斌. 10kV开关柜局部放电检测技术及测试平台的研究 [D]. 北京: 华北电力大学(北京), 2017. ZHANG B. Research on partial discharge detection technology and testing platform of 10kV switch cabinet [D]. Beijing: North China Electric Power University(Beijing), 2017. [6] 熊文, 危国恩, 王莉, 等. 智能配电网广域同步相量测量体系设计方案研究 [J]. 南方能源建设, 2021, 8(2): 85-90. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.013. XIONG W, WEI G E, WANG L, et al. Research on the design scheme of wide area synchronous phasor measurement system for smart distribution grid [J]. Southern Energy Construction, 2021, 8(2): 85-90. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.013. [7] 刘生, 巴砚. 油田群项目海上降压变电站布置方案研究 [J]. 南方能源建设, 2021, 8(2): 99-104. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.015. LIU S, BA Y. Research on layout plan of offshore step-down substation for oilfield group project [J]. Southern Energy Construction, 2021, 8(2): 99-104. DOI: 10.16516/j.gedi.issn2095-8676.2021.02.015. [8] 张明瀚, 简翔浩, 陆子凯. 珠海“互联网+”柔性直流配电网换流站设计方案 [J]. 南方能源建设, 2020, 7(1): 95-100. DOI: 10.16516/j.gedi.issn2095-8676.2020.01.015. ZHANG M H, JIAN X H, LU Z K. Design of converter stations of Zhuhai "Internet +" flexible DC distribution network [J]. Southern Energy Construction, 2020, 7(1): 95-100. DOI: 10.16516/j.gedi.issn2095-8676.2020.01.015. [9] CHEN J, XU L Q, WAN Z L, et al. Research on monitor technology of MOA state based on third harmonic principle [C]// IEEE. 2016 IEEE International Conference on High Voltage Engineering and Application, Chengdu, 19-22, September, 2016. Chengdu: IEEE, 2016: 1-4. DOI: 10.1109/ICHVE.2016.7800737. [10] HAN Y S, ZHENG H, LI Z H. Measurement of the resistive current of MOA based on an improved equivalent model [C]// IEEE. 2015 IEEE 11th International Conference on the Properties and Applications of Dielectric Materials, Sydney, 19-22, July, 2015. Sydney: IEEE, 2015: 1003-1006. DOI: 10.1109/ICPADM.2015.7295444. [11] HAN Y S, LI Z H, ZHENG H, et al. A decomposition method for the total leakage current of MOA based on multiple linear regression [J]. IEEE Transactions on Power Delivery, 2016, 31(4): 1422-1428. DOI: 10.1109/TPWRD.2015.2462071. [12] 魏绍东, 邓维, 雷红才, 等. 500kV避雷器故障模拟及缺陷检测试验研究 [J]. 电瓷避雷器, 2019(2): 109-114. DOI: 10.16188/j.isa.1003-8337.2019.02.018. WEI S D, DENG W, LEI H C, et al. Study on fault simulation and defect detection of 500 kV MOA [J]. Insulators and Surge Arresters, 2019(2): 109-114. DOI: 10.16188/j.isa.1003-8337.2019.02.018. [13] 胡锡幸, 鲍巧敏, 华盛继. 基于温度补偿的金属氧化物避雷器带电检测技术研究 [J]. 电瓷避雷器, 2017(1): 96-99. DOI: 10.16188/j.isa.1003-8337.2017.01.019. HU X X, BAO Q M, HUA S J. Research on MOA's on-line detection technology based on temperature compensation [J]. Insulators and Surge Arresters, 2017(1): 96-99. DOI: 10.16188/j.isa.1003-8337.2017.01.019. [14] LIU Q Y, GAO X G, MENG J S, et al. Design and realization of online monitoring device for arrester in substation [C]// IEEE. 2017 Chinese Automation Congress, Ji'nan, 20-22, October, 2017. Ji'nan: IEEE, 2017: 3164-3169. DOI: 10.1109/CAC.2017.8243320. [15] DE C CUNHA C F F, CARVALHO A T D, PETRAGLIA M R, et al. An improved scale dependent wavelet selection for data denoising of partial discharge measurement [C]// IEEE. 2013 IEEE International Conference on Solid Dielectrics, Bologna, 30 June-4 July, 2013. Bologna: IEEE, 2013: 100-104. DOI: 10.1109/ICSD.2013.6619894.