CN106153183A

CN106153183A - The on-line monitoring method of on-load tap changers of transformers machine performance

Info

Publication number: CN106153183A
Application number: CN201510216661.2A
Authority: CN
Inventors: 陆杨; 张庆富; 刘安宏; 何育; 杨义
Original assignee: NANJING UNITECH ELECTRIC POWER TECHNOLOGY DEVELOPMENT Co Ltd; State Grid Corp of China SGCC; State Grid Jiangsu Electric Power Co Ltd; Yancheng Power Supply Co of State Grid Jiangsu Electric Power Co Ltd
Current assignee: NANJING UNITECH ELECTRIC POWER TECHNOLOGY DEVELOPMENT Co Ltd; State Grid Corp of China SGCC; State Grid Jiangsu Electric Power Co Ltd; Yancheng Power Supply Co of State Grid Jiangsu Electric Power Co Ltd
Priority date: 2015-04-27
Filing date: 2015-04-27
Publication date: 2016-11-23

Abstract

The invention discloses an on-line monitoring method for the mechanical state of a transformer on-load tap changer, comprising: collecting vibration signals on the surface of the transformer on-load tap-changer; performing phase space reconstruction on the vibration signal x(t) to obtain the vibration signal Reconstruct the phase space; calculate the number of cluster centers Kc of the vibration signal in the reconstructed phase space and the position coordinates of the Kc cluster centers; calculate the relative position of the Kc cluster centers in the reconstructed phase space of the vibration signal during the operation of the on-load tap changer Relative error of the I-th cluster center. The invention effectively and highly sensitively monitors the running state of the transformer on-load tap changer on-line, so that the transformer on-load tap-changer can be repaired or replaced in time, and transformer failure and power system failure caused by damage to the on-load tap-changer can be avoided.

Description

On-line Monitoring Method of Mechanical State of Transformer On-Load Tap-Changer

技术领域technical field

本发明涉及一种信号监测方法，尤其涉及一种变压器有载分接开关机械状态的在线监测方法。The invention relates to a signal monitoring method, in particular to an online monitoring method for the mechanical state of a transformer on-load tap changer.

背景技术Background technique

有载调压变压器是电力系统各种设备中非常重要的关键设备之一，主要通过有载调压分接开关(OLTC：On-load Tap Changer)的逐级动作，实现电力变压器的有载调压，因此，有载分接开关是电力变压器的关键核心部件之一。The on-load tap changer is one of the most important key equipment in the power system. It mainly realizes the on-load tap changer of the power transformer through the step-by-step action of the on-load tap changer (OLTC: On-load Tap Changer). Therefore, the on-load tap-changer is one of the key core components of the power transformer.

依靠有载分接开关准确及时的切换动作，不仅可减少和避免电压的大幅度波动，而且可以强制分配负荷潮流，挖掘设备无功和有功出力，增加电网调度的灵活性。Relying on the accurate and timely switching action of the on-load tap-changer, it can not only reduce and avoid large fluctuations in voltage, but also force the distribution of load flow, excavate the reactive power and active power output of equipment, and increase the flexibility of power grid dispatching.

随着对电能质量要求的提高，有载调压电力变压器日均调压次数显著增加，调压次数也随之增加，相应地，导致有载分接开关的故障率呈现增长趋势，影响电力系统的安全稳定运行。With the improvement of power quality requirements, the average daily voltage adjustment times of on-load tap changer power transformers have increased significantly, and the number of voltage adjustments has also increased. Correspondingly, the failure rate of on-load tap-changers has shown an increasing trend, affecting the power system. safe and stable operation.

国外统计资料表明，有载分接开关故障占有载调压变压器故障的41％，几呈上升趋势。国内平均统计数据表明，有载分接开关的故障占变压器故障的20％以上。因此，对运行中的电力变压器的有载分接开关的运行状态进行在线监测和故障诊断，及时发现有载分接开关的潜在故障隐患及损失程度，研究有载分接开关的状态评估技术，实现设备维修的合理化、规范化和科学化，符合智能化变电站中关于实施电气设备状态评估与状态维修的要求，具有较大的研究意义和良好的应用前景。Foreign statistics show that on-load tap-changer failures account for 41% of on-load tap-changing transformer failures, almost showing an upward trend. Domestic average statistics show that on-load tap changer failures account for more than 20% of transformer failures. Therefore, on-line monitoring and fault diagnosis are carried out on the operating status of the on-load tap-changer of the power transformer in operation, and the potential hidden trouble and loss degree of the on-load tap-changer are discovered in time, and the state evaluation technology of the on-load tap-changer is studied. Realizing the rationalization, standardization and scientificization of equipment maintenance meets the requirements of state assessment and condition maintenance of electrical equipment in intelligent substations, and has great research significance and good application prospects.

变压器有载分接开关主要由选择器、切换开关和电动机构组成，包括电气性能和机械性能两个方面。其中电气性能主要指触头的接触电阻，当触头接触电阻增大时，会引起触头过热，甚至烧损。机械性能是指有载分接开关操作过程中选择开关和切换开关等部件的动作顺序和时间配合、以及切换过程中是否存在卡塞和触头切换不到位等。机械故障是电力变压器有载分接开关的主要故障类型，它可能损坏有载分接开关和电力变压器，影响电力设备和系统的正常安全运行并造成严重后果。The transformer on-load tap-changer is mainly composed of a selector, a diverter switch and an electric mechanism, including two aspects of electrical performance and mechanical performance. Among them, the electrical performance mainly refers to the contact resistance of the contact. When the contact resistance of the contact increases, it will cause the contact to overheat or even burn out. Mechanical performance refers to the action sequence and time coordination of the selector switch and diverter switch and other components during the operation of the on-load tap-changer, and whether there is jamming and contact switching is not in place during the switching process. Mechanical failure is the main failure type of the on-load tap-changer of the power transformer. It may damage the on-load tap-changer and the power transformer, affect the normal and safe operation of the power equipment and system, and cause serious consequences.

目前，国内有载分接开关的大都采用离线定期维修方式，它是根据设各的磨损规律，预先确定修理类型、修理间隔及维修工作量对设备进行周期性维修，防止故障的发生。定期维修方式可以使生产和修理均能有计划地进行，可以防止和减少突发故障，适用于已知设备寿命分布规律而且有明显损耗期的设备。缺点是工作量大，效率低和测量精度不高，且不能及时发现维修间隔内的设备故障，已逐渐不能适应形式需求。At present, most of the on-load tap-changers in China adopt the off-line regular maintenance method. According to the wear pattern of each device, the repair type, repair interval and maintenance workload are pre-determined to carry out periodic maintenance on the equipment to prevent the occurrence of failures. The regular maintenance method can make both production and repairs proceed in a planned manner, and can prevent and reduce sudden failures. It is suitable for equipment with known equipment life distribution rules and obvious wear-and-tear periods. The disadvantage is that the workload is heavy, the efficiency is low and the measurement accuracy is not high, and the equipment failure within the maintenance interval cannot be found in time, and it has gradually failed to meet the formal requirements.

有载分接开关在操作过程中，机构零部件之间的碰撞或摩擦会引起机械振动，而机械振动是一个丰富的信息载体，有载分接开关开关机座、外壳上的振动是内部多种现象激励的响应。若称机构零部件的一次碰撞或摩擦为一个振动事件，则不同振动时间产生的机械振动信号会在时域上形成一个振动信号的时间序列。因此，若将振动信号分析法引入到有载分接开关的在线监测与故障诊断中来，通过非介入性地监测电力变压器有载分接开关的机械特性，实时采集、分析和处理有载分接开关表面的振动信号，获取传动机构的状态信息和工作模式，进而对变压器有载分接开关的运行状态进行识别，可及时发现有载分接开关运行过程中的潜伏性故障，提高有载分接开关、变压器及电力系统运行的可靠性和安全性。During the operation of the on-load tap-changer, the collision or friction between the mechanism parts will cause mechanical vibration, and the mechanical vibration is a rich information carrier. response to a stimulus. If a collision or friction of a mechanism component is called a vibration event, the mechanical vibration signals generated at different vibration times will form a time series of vibration signals in the time domain. Therefore, if the vibration signal analysis method is introduced into the on-line monitoring and fault diagnosis of the on-load tap-changer, through non-invasive monitoring of the mechanical characteristics of the power transformer on-load tap-changer, real-time collection, analysis and processing of the on-load tap-changer Connect the vibration signal on the surface of the switch to obtain the state information and working mode of the transmission mechanism, and then identify the operating state of the transformer on-load tap-changer, which can detect latent faults during the operation of the on-load tap-changer in time and improve the on-load tap-changer. Reliability and safety of tap changer, transformer and power system operation.

发明内容Contents of the invention

本发明的目的是提供一种变压器有载分接开关机械状态的在线监测方法，该方法通过对有载分接开关动作过程中的振动信号进行实时监控，能够实现对变压器有载分接开关机械状态的高效、准确地判断。The purpose of the present invention is to provide an on-line monitoring method for the mechanical state of the transformer on-load tap-changer. The method can realize the monitoring of the transformer on-load tap-changer mechanical Efficient and accurate judgment of status.

为了实现上述发明目的，本发明提供了一种变压器有载分接开关机械状态的在线监测方法，其包括下列步骤：In order to achieve the purpose of the above invention, the present invention provides an online monitoring method for the mechanical state of the transformer on-load tap changer, which includes the following steps:

1、一种变压器有载分接开关机械状态的在线监测方法，其特征在于，包括下列步骤：1. A method for on-line monitoring of the mechanical state of a transformer on-load tap changer, characterized in that it comprises the following steps:

(1)将振动传感器设置在变压器有载分接开关的表面上，实时采集变压器有载分接开关表面的振动信号x(t)，t＝1，…No，No为时间序列的长度；(1) The vibration sensor is arranged on the surface of the transformer on-load tap changer, and the vibration signal x(t) on the surface of the transformer on-load tap-changer is collected in real time, t=1, ... No, No is the length of the time series;

(2)使用延迟坐标法对振动信号x(t)进行相空间重构，为(2) Use the delayed coordinate method to reconstruct the phase space of the vibration signal x(t), as

$\{\begin{matrix} X x ((11)) = = [[x x ((11)),, x x ((11 + + τ τ)),, \cdot &Center Dot; \cdot &Center Dot; \cdot \cdot,, x x ((11 + + ((m m - - 11)) τ τ))]] \\ X x ((22)) = = [[x x ((22)),, x x ((22 + + τ τ)),, \cdot \cdot \cdot &Center Dot; \cdot &Center Dot;,, x x ((22 + + ((m m - - 11)) τ τ))]] \\ . . . . . . . . . . . . \\ X x ((k k)) = = [[x x ((k k)),, x x ((k k + + τ τ)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, x x ((k k + + ((m m - - 11)) τ τ))]] \\ . . . . . . . . . . . . \\ X x ((N N)) = = [[x x ((N N)),, x x ((N N + + τ τ)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, x x ((N N + + ((m m - - 11)) τ τ))]] \end{matrix}$

N为相空间重构后的相空间中的时间向量个数，m为嵌入维数，τ为延迟时间，X(k)(k＝1，…，N)为对振动信号x(t)进行重构后的相空间的第k个时间向量，No＝N+(m-1)τ；N is the number of time vectors in the phase space after phase space reconstruction, m is the embedding dimension, τ is the delay time, X(k) (k=1,...,N) is the vibration signal x(t) The kth time vector of the reconstructed phase space, No=N+(m-1)τ;

这N个时间向量形成了一个重构的相空间，这N个延迟时间序列的关联积分函数为The N time vectors form a reconstructed phase space, and the associated integral function of the N delayed time series is

${C C}_{s the s} ((m m,, N N,, r r,, t t)) = = \frac{22}{N N ((N N - - 11))} \underset{11 \leq \leq i i \leq \leq j j \leq \leq N N}{Σ Σ} H h ((r r - - {d d}_{ij ij})),, r r > > 00$

d_ij＝||X(i)-X(j)||d _ij =||X(i)-X(j)||

$H h ((r r)) = = \{\begin{matrix} 11,, & r r &GreaterEqual; &Greater Equal; 00 \\ 00,, & r r < < 00 \end{matrix}$

H(r)为Heaviside阶跃函数，r为半径；H(r) is the Heaviside step function, r is the radius;

延迟坐标法是对一维时间序列进行高维相空间重构的常用数学方法，其是本领域内的普通技术人员均所熟知的，故本文在此不再进行描述。The delayed coordinate method is a commonly used mathematical method for reconstructing high-dimensional phase space of one-dimensional time series, which is well known to those skilled in the art, so it will not be described here.

所述的嵌入维数m和延迟时间τ的计算过程为The calculation process of the embedding dimension m and delay time τ is

(I)计算振动信号x(t)的标准差σ；(1) calculate the standard deviation σ of vibration signal x (t);

(1I)将振动信号x(t)分成t₀个不相交的子序列，为(1I) Divide the vibration signal x(t) into t ₀ disjoint subsequences, as

${{{x x}_{11},, {x x}_{{t t}_{00} + + 11},, {x x}_{22 {t t}_{00} + + 11},, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;}}$

${{{x x}_{22},, {x x}_{{t t}_{00} + + 22},, {x x}_{22 {t t}_{00} + + 22},, \cdot \cdot \cdot \cdot \cdot \cdot}}$

……...

${{{x x}_{{t t}_{00}},, {x x}_{{22 t t}_{00}},, {x x}_{{33 t t}_{00}},, \cdot \cdot \cdot \cdot \cdot \cdot}}$

此处，每个子序列的长度为1＝N₀/t₀；Here, the length of each subsequence is 1=N ₀ /t ₀ ;

(III)定义这t₀个子序列的检验统计量为嵌入维数为m时t₀个子序列的关联积分函数与嵌入维数为I时t₀子序列的关联积分函数差值之和的平均值，其计算公式为(III) Define the test statistic of the t ₀ subsequences as the average value of the sum of the difference between the associated integral function of the t ₀ subsequence when the embedding dimension is m and the difference of the associated integral function of the t ₀ subsequence when the embedding dimension is I , its calculation formula is

$S S ((m m,, {N N}_{00},, r r,, {t t}_{00})) = = \frac{11}{{t t}_{00}} {Σ Σ}_{s the s = = 11}^{{t t}_{00}} [[{C C}_{s the s} ((m m,, {N N}_{00} / / {t t}_{00},, r r,, {t t}_{00})) - - {C C}_{s the s}^{m m} ((11,, {N N}_{00} / / {t t}_{00},, r r,, {t t}_{00}))]]$

则当N₀→∞时，有Then when N ₀ →∞, we have

$S S ((m m,, r r,, {t t}_{00})) = = \frac{11}{{t t}_{00}} {Σ Σ}_{s the s = = 11}^{{t t}_{00}} [[{C C}_{s the s} ((m m,, r r,, {t t}_{00})) - - {C C}_{s the s}^{m m} ((11,, r r,, {t t}_{00}))]]$

(IV)对t₀个子序列，分别计算：检验统计量的均值检验统计量的差值和检验统计量的方差Scor(t₀)，其表达式分别为(IV) For t ₀ subsequences, calculate respectively: the mean value of the test statistic difference of test statistic and the variance of test statistics Scor(t ₀ ), the expressions are respectively

$\overset{&OverBar; &OverBar;}{S S} (({t t}_{00})) = = \frac{11}{1616} {Σ Σ}_{m m = = 22}^{55} {Σ Σ}_{j j = = 11}^{44} S S ((m m,, {r r}_{j j},, {t t}_{00}))$

$Δ Δ \overset{&OverBar; &OverBar;}{S S} (({t t}_{00})) = = \frac{11}{44} {Σ Σ}_{m m = = 22}^{55} ΔS ΔS ((m m,, {t t}_{00}))$

${S S}_{cor cor} (({t t}_{00})) = = Δ Δ \overset{&OverBar; &OverBar;}{S S} (({t t}_{00})) + + | | \overset{&OverBar; &OverBar;}{S S} (({t t}_{00})) | |$

ΔS(m，t₀)＝max{S(m，r_j，t₀)}-min{S(m，r_j，t₀)}ΔS(m,t ₀ )=max{S(m,r _j ,t ₀ )}-min{S(m,r _j ,t ₀ )}

式中，r_j＝jσ/2；In the formula, r _j =jσ/2;

(IV)寻找的第一个零点或的第一个极小值点，第一个零点或第一个极小值点对应的时间即为延迟时间τ；(IV) looking for the first zero of or The first minimum value point of , the time corresponding to the first zero point or the first minimum value point is the delay time τ;

(V)寻找Scor(t₀)的最小值点，对应的时间为振动信号时间序列x(t)的第一个整体最大值时间窗口τ_w，.计算嵌入维数m，公式如下：(V) Find the minimum value point of Scor(t ₀ ), the corresponding time is the first overall maximum time window τ _w of the vibration signal time series x(t). Calculate the embedding dimension m, the formula is as follows:

m＝int(τ_w/τ+1)m=int(τ _w /τ+1)

式中，int为取整函数。In the formula, int is the rounding function.

(3)根据K-means聚类分析法计算振动信号在重构相空间的簇中心个数Kc及Kc个簇中心的位置坐标。所述的簇中心个数Kc及Kc个簇中心的位置坐标的计算过程为(3) Calculate the number of cluster centers Kc and the position coordinates of Kc cluster centers of the vibration signal in the reconstructed phase space according to the K-means cluster analysis method. The calculation process of the number of cluster centers Kc and the position coordinates of Kc cluster centers is

(I)在重构相空间的N个向量中任选Kc₁个相点作为簇中心，此处取Kc₁＝2，Kc₁个簇中心在重构相空间的位置分别为X(i)(1＜i Kc₁)。计算相空间中其余相点与这Kc₁个相点的欧氏距离，按照距离最近的原则将其余相点归入到这Kc₁个簇中。绝对距离是求得两点距离的常用值数学方法，其是本领域内的普通技术人员均所熟知的，故本文在此不再进行列式表达。(1) In the N vectors of the reconstructed phase space, Kc ₁ phase points are selected as cluster centers, where Kc ₁ = 2, and the positions of Kc ₁ cluster centers in the reconstructed phase space are respectively X(i) (1<i Kc ₁ ). Calculate the Euclidean distance between the remaining phase points and the Kc ₁ phase points in the phase space, and classify the remaining phase points into the Kc ₁ clusters according to the principle of the closest distance. Absolute distance is a commonly used mathematical method to obtain the distance between two points, which is well known to those skilled in the art, so this article will not express it in a column form here.

(II)计算Kc₁个簇中所有相点的平均中心位置，作为Kc₁个簇的在相空间的新的簇中心位置，记为X’(i)(1＜i≤Kc₁，口，即X’(i)在相空间的mτ维坐标为Kc₁簇内所有相点的mτ维坐标的平均值。(II) Calculate the average central position of all phase points in the Kc ₁ cluster, as the new cluster central position in the phase space of the Kc ₁ cluster, denoted as X'(i) (1<i≤Kc ₁ , That is, the mτ-dimensional coordinate of X'(i) in the phase space is the average value of the mτ-dimensional coordinates of all phase points in the Kc ₁ cluster.

(III)计算新簇的簇中心X’(i)相对于原有簇的簇中心X(i)的偏移比例(BiasProportion，BP)，当偏移比例小于I％时，可认为新簇的位置中心X’(i)是稳定的；当偏移比例大于1％时，可认为新簇的位置中心X’(i)是不稳定的，重复步骤(I)、步骤(II)和步骤(1II)，直至得到Kc₁个稳定的新簇中心位置。偏移比例的计算公式为(III) Calculate the offset ratio (BiasProportion, BP) of the cluster center X'(i) of the new cluster relative to the cluster center X(i) of the original cluster. When the offset ratio is less than 1%, the new cluster can be considered The position center X'(i) is stable; when the deviation ratio is greater than 1%, it can be considered that the position center X'(i) of the new cluster is unstable, and repeat steps (I), step (II) and step ( 3II) until Kc ₁ stable new cluster center positions are obtained. The formula for calculating the offset ratio is

(IV)根据确定的Kc₁个新的簇中心X’(i)，重新计算相空间中其余相点相对于Kc₁个簇中心X’(i)的欧氏距离，按照距离最近的原则将其余相点归入到这Kc₁个簇中。(IV) According to the determined Kc ₁ new cluster centers X'(i), recalculate the Euclidean distances of the remaining phase points in the phase space relative to the Kc ₁ cluster centers X'(i), according to the principle of the closest distance to The remaining phase points are classified into this Kc ₁ cluster.

(V)分别计算Kc₁个簇中簇中心与属于该簇的各个相点的距离，其计算公式为(V) Calculate the distance between the cluster center and each phase point belonging to the cluster in Kc ₁ cluster respectively, the calculation formula is

$J J (({C C}_{{K K}_{C C 11}})) = = \underset{{X x}_{i i} &Element; &Element; {C C}_{{K K}_{C C}}}{Σ Σ} {| | | | {X x}^{,,} ((i i)) - - X x ((j j)) | | | |}^{22},, 11 < < i i \leq \leq {K K}_{C C 11},, 11 \leq \leq j j \leq \leq {P P}_{t t}$

X’(i)为第i个簇的簇中心的相空间位置，1＜i≤Kc₁；X(j)为属于第i个相点的相空间坐标；Pt为属于第i个簇的相点数目。X'(i) is the phase space position of the cluster center of the i-th cluster, 1<i≤Kc ₁ ; X(j) is the phase space coordinate of the i-th phase point; Pt is the phase space belonging to the i-th cluster number of points.

将每个簇中心与属于该簇的相点的距离进行累加，得到该簇的总体距离，其计算公式为The distance between each cluster center and the phase points belonging to the cluster is accumulated to obtain the overall distance of the cluster, and its calculation formula is

$J J ((C C)) = = {Σ Σ}_{K K = = 11}^{{K K}_{C C 11}} J J (({K K}_{C C 11}))$

(VI)令Kc＝Kc₁+1，重复步骤(I)～步骤(V)，得到Kc个簇的总体距离，记为J’(C)。计算总体距离的减小速率(Decreasing Rate，DR)，当减小速率小于5％时，可认为总体距离的减小速率基本稳定，重构信号已经被充分表示，此时K’。即为所求的簇中心的数目，X’(i)(1＜i≤Kc₁)即为Kc个簇中心的位置坐标。否则重复步骤(I)～步骤(VI)，直至总体距离的减小速率基本稳定，此时对应的Kc即为所求的簇中心的数目。(VI) Set Kc=Kc ₁ +1, repeat steps (I) to (V), and obtain the overall distance of Kc clusters, denoted as J'(C). Calculate the Decreasing Rate (DR) of the overall distance. When the decreasing rate is less than 5%, it can be considered that the decreasing rate of the overall distance is basically stable, and the reconstructed signal has been fully represented. At this time, K'. That is, the number of cluster centers to be sought, and X'(i) (1<i≤Kc ₁ ) is the position coordinates of Kc cluster centers. Otherwise, repeat step (I) to step (VI) until the decrease rate of the overall distance is basically stable, at this time the corresponding Kc is the number of cluster centers sought.

所述的减小速率的计算公式为The formula for calculating the reduction rate is

$DR DR = = \frac{| | {J J}^{' '} ((C C)) - - J J ((C C)) | |}{| | J J ((C C)) | |} \times \times 100100 % %$

(4)分别计算Kc个簇中心与第I个簇中心的绝对距离。绝对距离是求得两点距离的常用值数学方法，其是本领域内的普通技术人员均所熟知的，故本文在此不再进行列式表达；(4) Calculate the absolute distances between Kc cluster centers and the I-th cluster center respectively. Absolute distance is a commonly used mathematical method to obtain the distance between two points, which is well known to those of ordinary skill in the art, so this article will not express it in column form here;

(5)计算有载分接开关操作过程中振动信号重构相空间中K_C个簇中心相对于第1个簇中心的相对误差。若第2个簇中心与第1个簇中心的绝对距离减小为原来的20％且第2个簇中心与第1个簇中心的绝对距离大于其余Kc-2个簇中心与第1个簇中心的绝对距离，则判断有载分接开关操作过程中运行状态异常；否则，认为有载分接开关操作过程中运行状态正常。(5) Calculate the relative error of K _C cluster centers relative to the first cluster center in the reconstructed phase space of the vibration signal during the operation of the on-load tap-changer. If the absolute distance between the second cluster center and the first cluster center is reduced to 20% of the original and the absolute distance between the second cluster center and the first cluster center is greater than the remaining Kc-2 cluster centers and the first cluster If the absolute distance of the center is determined, it is judged that the operating state of the on-load tap-changer is abnormal during operation; otherwise, it is considered that the operating state of the on-load tap-changer is normal during operation.

也就是说，本技术方案是将变压器有载分接开关动作过程中的一段振动信号进行相空间重构，在高维空间内计算空间相点的簇中心个数与在相空间的对应位置，然后计算有载分接开关动作前后计算Kc个簇中心与第1个簇中心的绝对距离与绝对距离的相对误差，根据绝对误差与绝对距离的相对误差变化就可以判断出变压器有载分接开关操作过程中的运行状态。That is to say, this technical solution is to reconstruct the phase space of a section of the vibration signal during the operation of the transformer on-load tap changer, and calculate the number of cluster centers of the space phase points in the high-dimensional space and the corresponding position in the phase space. Then calculate the relative error of the absolute distance and the absolute distance between Kc cluster centers and the first cluster center before and after the operation of the on-load tap-changer, and the transformer on-load tap-changer can be judged according to the relative error change of the absolute error and the absolute distance The running state during operation.

本发明所述的变压器有载分接开关运行状态的在线监测方法由于采用了上述技术方案，使得其可以通过对变压器有载分接开关表面振动信号的实时监控，来直接判断变压器有载分接开关的工作状态，该判断方法高效、准确，且易于实施，便于操作人员及时发现变压器有载分接开关的旱期机械故障隐患，从而根据异常情况对变压器有载分接开关进行及时地维护与检修，大大降低了变压器有载分接开关的故障损坏率。The online monitoring method of transformer on-load tap-changer operating state according to the present invention adopts the above-mentioned technical scheme, so that it can directly judge the transformer on-load tap-changing by monitoring the surface vibration signal of the transformer on-load tap-changer in real time. The working state of the switch, this judgment method is efficient, accurate, and easy to implement. It is convenient for the operator to find out the hidden dangers of mechanical failure of the transformer on-load tap-changer in a timely manner, so as to timely maintain and monitor the transformer on-load tap-changer according to the abnormal situation. Overhaul greatly reduces the fault damage rate of the transformer on-load tap-changer.

附图说明Description of drawings

图1为本发明变压器有载分接开关机械状态的在线监测方法在实施例中采集到的有载分接开关动作时的振动信号。Fig. 1 is the vibration signal when the on-load tap-changer is in action collected in the embodiment of the online monitoring method for the mechanical state of the transformer on-load tap-changer in the present invention.

图2为采用本发明变压器有载分接开关机械状态的在线监测方法在实施例中得到的Kc个簇中心与第1个簇中心的绝对距离曲线。Fig. 2 is the absolute distance curve between Kc cluster centers and the first cluster center obtained in the embodiment by using the online monitoring method of the mechanical state of the transformer on-load tap changer of the present invention.

具体实施方式detailed description

以下结合附图和具体实施例来对本发明所述的变压器有载分接开关机械状态的在线监测方法做进一步的详细说明。The method for on-line monitoring of the mechanical state of the transformer on-load tap changer according to the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

以某电力公司的某变电站的35kV变压器的M型有载分接开关为试验对象进行在线监测，按照下列步骤判断该变压器有载分接开关操作过程中的机械状态：Taking the M-type on-load tap-changer of a 35kV transformer in a certain substation of a power company as the test object to conduct online monitoring, follow the steps below to judge the mechanical state of the transformer’s on-load tap-changer during operation:

(1)将振动传感器设置在该变压器有载分接开关的表面上，实时采集变压器有载分接开关表面的振动信号x(t)，t＝1，---No，No为时间序列的长度：(1) Set the vibration sensor on the surface of the transformer on-load tap-changer, collect the vibration signal x(t) on the surface of the transformer on-load tap-changer in real time, t=1,---No, No is the time series length:

(2)将采集到的振动信号进行抗混迭数字滤波和高速缓存，然后通过高速总线传输至数据分析模块；(2) Anti-aliasing digital filtering and high-speed buffering are performed on the collected vibration signals, and then transmitted to the data analysis module through a high-speed bus;

(3)数据分析模块通过延迟坐标法对振动信号x(t)进行相空间重构，为(3) The data analysis module reconstructs the phase space of the vibration signal x(t) through the delayed coordinate method, which is

$\{\begin{matrix} X x ((11)) = = [[x x ((11)),, x x ((11 + + τ τ)),, \cdot &Center Dot; \cdot \cdot \cdot \cdot,, x x ((11 + + ((m m - - 11)) τ τ))]] \\ X x ((22)) = = [[x x ((22)),, x x ((22 + + τ τ)),, \cdot \cdot \cdot &Center Dot; \cdot &Center Dot;,, x x ((22 + + ((m m - - 11)) τ τ))]] \\ . . . . . . . . . . . . \\ X x ((k k)) = = [[x x ((k k)),, x x ((k k + + τ τ)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, x x ((k k + + ((m m - - 11)) τ τ))]] \\ . . . . . . . . . . . . \\ X x ((N N)) = = [[x x ((N N)),, x x ((N N + + τ τ)),, \cdot &Center Dot; \cdot &Center Dot; \cdot &Center Dot;,, x x ((N N + + ((m m - - 11)) τ τ))]] \end{matrix}$

d_ij＝||X(i)-X(j)||d _ij =||X(i)-X(j)||

(I)计算振动信号时间序列x(t)的标准差σ；(1) calculate the standard deviation σ of vibration signal time series x (t);

(II)将振动信号x(t)分成t₀个不相交的子序列，为(II) Divide the vibration signal x(t) into t ₀ disjoint subsequences, as

${{{x x}_{11},, {x x}_{{t t}_{00} + + 11},, {x x}_{22 {t t}_{00} + + 11},, \cdot \cdot \cdot \cdot \cdot \cdot}}$

${{{x x}_{22},, {x x}_{{t t}_{00} + + 22},, {x x}_{22 {t t}_{00} + + 22},, \cdot &Center Dot; \cdot &Center Dot; \cdot \cdot}}$

${{{x x}_{{t t}_{00}},, {x x}_{{22 t t}_{00}},, {x x}_{{33 t t}_{00}},, \cdot &Center Dot; \cdot \cdot \cdot \cdot}}$

此处，每个子序列的长度为1＝No/t₀，此处有t₀＝6；Here, the length of each subsequence is 1=No/t ₀ , where t ₀ =6;

(III)定义这t₀个子序列的检验统计量为嵌入维数为m时t₀个子序列的关联积分函数与嵌入维数为1时t₀子序列的关联积分函数差值之和的平均值，其计算公式为(III) Define the test statistic of the t ₀ subsequences as the average of the sum of the difference between the associated integral function of the t ₀ subsequence when the embedding dimension is m and the difference of the associated integral function of the t ₀ subsequence when the embedding dimension is 1 , its calculation formula is

则当No→·∞|有Then when No→·∞|has

(IV)对t₀个子序列，分别计算：检验统计量的均值、检验统计量的差值和检验统计量的方差S_c0r(t₀)，其表达式分别为(IV) For t ₀ subsequences, calculate respectively: the mean value of the test statistic , the difference of the test statistic and the variance of the test statistic S _c0r (t ₀ ), the expressions are respectively

R_j＝jσ/2； _Rj = jσ/2;

(V)寻找S_cor(t₀)的最小值点，对应的时间为振动信号时间序列X(t)的第一个整体最大值时间窗口τ_w，计算嵌入维数m，公式如下：(V) Find the minimum value point of S _cor (t ₀ ), the corresponding time is the first overall maximum time window τ _w of the vibration signal time series X(t), and calculate the embedding dimension m, the formula is as follows:

m＝int(τ_w/T+1)m=int(τ _w /T+1)

int为取整函数。int is the rounding function.

此处有m＝3，τ＝11Here m=3, τ=11

(4)根据K-means聚类分析法计算振动信号在重构相空间的簇中心个数Kc及Kc个簇中心的位置坐标。所述的簇中心个数Kc及Kc个簇中心的位置坐标的计算过程为(4) Calculate the number of cluster centers Kc and the position coordinates of Kc cluster centers of the vibration signal in the reconstructed phase space according to the K-means cluster analysis method. The calculation process of the number of cluster centers Kc and the position coordinates of Kc cluster centers is

(I)在重构相空间的N个向量中任选K_c1个相点作为簇中心，此处取K_c1＝2，K_c1个簇中心在重构相空间的位置为X(i)(1＜i≤K_c1，)。分别计算相空间中其余相点与这K_c1个相点的欧氏距离，按照距离最近的原则将其余相点归入到这K_c1个簇中。绝对距离是求得两点距离的常用值数学方法，其是本领域内的普通技术人员均所熟知的，故本文在此不再进行列式表达。(1) Select K _c1 phase points as cluster centers in the N vectors of the reconstructed phase space, where K _c1 =2, and the positions of K _c1 cluster centers in the reconstructed phase space are X(i)( 1<i≤K _c1 ,). Calculate the Euclidean distances between the remaining phase points in the phase space and the K _c1 phase points, and classify the remaining phase points into the K _c1 clusters according to the principle of the closest distance. Absolute distance is a commonly used mathematical method to obtain the distance between two points, which is well known to those skilled in the art, so this article will not express it in a column form here.

(II)计算Kc个簇中所有相点的平均中心位置，作为K_c1个簇的在相空间的新的簇中心位置，记为X’(i)(1＜i≤K_c1)，即X’(i)在相空间的m维坐标K_c1簇内所有相点的m维坐标的平均值。(II) Calculate the average center position of all phase points in Kc clusters, as the new cluster center position of K _c1 clusters in phase space, denoted as X'(i) (1<i≤K _c1 ), that is, X '(i) The average value of the m-dimensional coordinates of all phase points in the m-dimensional coordinate K _c1 cluster of the phase space.

(III)计算新簇的簇中心X’(i)相对于原有簇的簇中心X(i)的偏移比例(Bias Proportion，BP)，当偏移比例小于1％时，可认为新簇的位置中心X’(i)是稳定的；当偏移比例大于1％时，可认为新簇的位置中心X’(i)是不稳定的，重复步骤(I)、步骤(II)和步骤(1II)，直至得到K_c1个稳定的新簇中心位置。偏移比例的计算公式为(III) Calculate the offset ratio (Bias Proportion, BP) of the cluster center X'(i) of the new cluster relative to the cluster center X(i) of the original cluster. When the offset ratio is less than 1%, the new cluster can be considered The position center X'(i) of the new cluster is stable; when the deviation ratio is greater than 1%, it can be considered that the position center X'(i) of the new cluster is unstable, repeat steps (I), step (II) and step (III), until K _c1 stable new cluster center positions are obtained. The formula for calculating the offset ratio is

(IV)根据确定的K_c1个新的簇中心X’(i)，重新计算相空间中其余相点相对于这K_c1个簇中心X’(i)的欧氏距离，按照距离最近的原则将其余相点归入到这K_c1个簇中。(IV) According to the determined K _c1 new cluster centers X'(i), recalculate the Euclidean distances of the remaining phase points in the phase space relative to the K _c1 cluster centers X'(i), according to the principle of the closest distance The remaining phase points are classified into these K _c1 clusters.

(V)分别计算这K_c1个簇中簇中心与属于该簇的各个相点的距离，其计算公式为(V) Calculate the distance between the center of the cluster in these K _c1 clusters and each phase point belonging to the cluster, the calculation formula is

X’(i)为第i个簇的簇中心的相空间位置，1＜i≤K_c1，X(j)为属于第j个相点的相空间坐标：Pt为属于第i个簇的相点数目。X'(i) is the phase space position of the cluster center of the i-th cluster, 1<i≤K _c1 , X(j) is the phase space coordinate of the j-th phase point: Pt is the phase space belonging to the i-th cluster number of points.

$J J ((C C)) = = {Σ Σ}_{K K = = 11}^{{K K}_{C C 11}} J J (({K K}_{C C 11}))$

(VI)令Kc＝K_c1+1，重复步骤(I)～步骤(V)，得到Kc个簇的总体距离，记为J’(C)。计算总体距离的减小速率(Decreasing Rate，DR)，当减小速率小于5％时，可认为总体距离的减小速率基本稳定，重构信号已经被充分表示，此时K’c即为所求的簇中心的数目，X’(i)(1＜i≤K_c1)即为Kc个簇中心的位置坐标。否则重复步骤(I)一步骤(VI)，直至总体距离的减小速率基本稳定，此时对应的Kc即为所求的簇中心的数目。(VI) Make Kc= _Kc1 +1, repeat steps (I) to (V), and obtain the overall distance of Kc clusters, denoted as J'(C). Calculate the decreasing rate of the overall distance (Decreasing Rate, DR). When the decreasing rate is less than 5%, it can be considered that the decreasing rate of the overall distance is basically stable, and the reconstructed signal has been fully represented. At this time, K'c is the The number of cluster centers to be obtained, X'(i) (1<i≤K _c1 ) is the position coordinates of Kc cluster centers. Otherwise, repeat step (I) to step (VI) until the decreasing rate of the overall distance is basically stable, at this time the corresponding Kc is the number of cluster centers sought.

此处，Kc＝4，Here, Kc=4,

(5)分别计算Kc个簇中心个簇中心与第1个簇中心的绝对距离。绝对距离是求得两点距离的常用值数学方法，其是本领域内的普通技术人员均所熟知的，故本文在此不再进行列式表达；(5) Calculate the absolute distances between the Kc cluster centers and the first cluster center respectively. Absolute distance is a commonly used mathematical method to obtain the distance between two points, which is well known to those of ordinary skill in the art, so this article will not express it in column form here;

(6)计算有载分接开关操作过程中振动信号重构相空间中Kc个簇中心相对于第1个簇中心的相对误差。若第2个簇中心与第1个簇中心的绝对距离减小为原来的20％且第2个簇中心与第1个簇中心的绝对距离大于其余Kc-2个簇中心与第1个簇中心的绝对距离，则判断有载分接开关操作过程中运行状态异常；否则，认为有载分接开关操作过程中运行状态正常。(6) Calculate the relative error of Kc cluster centers relative to the first cluster center in the phase space of vibration signal reconstruction during the operation of the on-load tap-changer. If the absolute distance between the second cluster center and the first cluster center is reduced to 20% of the original and the absolute distance between the second cluster center and the first cluster center is greater than the remaining Kc-2 cluster centers and the first cluster If the absolute distance of the center is determined, it is judged that the operating state of the on-load tap-changer is abnormal during operation; otherwise, it is considered that the operating state of the on-load tap-changer is normal during operation.

图2显示了本实施例中根据上述方法得到的有载分接开关切换过程中的Kc个簇中心与第1个簇中心的绝对距离曲线，这两条曲线中第2个簇中心与第1个簇中心的绝对距离大于其余性-2个簇中心与第I个簇中心的绝对距离。表1显示了本实施例中根据上述方法得到的有载分接开关切换过程中的Kc个簇中心与第1个簇中心的绝对距离的相对误差，第2个簇中心与第1个簇中心的绝对距离减小为原来的23.38％，说明有载分接开关操作过程中运行状态异常，有可能出现弹簧松动等机械故障，需要及时进行维修，避免发生重大故障。Fig. 2 shows the absolute distance curves of the Kc cluster centers and the first cluster center in the on-load tap changer switching process obtained according to the above method in this embodiment, the second cluster center and the first cluster center in these two curves The absolute distance of a cluster center is greater than the absolute distance of the rest-2 cluster centers from the I-th cluster center. Table 1 shows the relative error of the absolute distance between the Kc cluster centers and the first cluster center in the on-load tap-changer switching process obtained according to the above method in this embodiment, the second cluster center and the first cluster center The absolute distance of OLTC is reduced to 23.38% of the original value, indicating that the operation state of the on-load tap-changer is abnormal during operation, and mechanical failures such as loose springs may occur, and timely maintenance is required to avoid major failures.

表1为采用本发明所述的变压器有载分接开关机械状态的在线监测方法在本实施例中得到的Kc个簇中心与第1个簇中心的绝对距离的相对误差。Table 1 shows the relative error of the absolute distance between the Kc cluster centers and the first cluster center obtained in this embodiment by using the on-line monitoring method for the mechanical state of the transformer on-load tap changer according to the present invention.

表1Table 1

要注意的是，以上列举的仅为本发明的具体实施例，显然本发明不限于以上实施例，随之有着许多的类似变化。本领域的技术人员如果从本发明公开的内容直接导出或联想到的所有变形，均应属于本发明的保护范围。It should be noted that the above examples are only specific embodiments of the present invention, and obviously the present invention is not limited to the above embodiments, and there are many similar changes accordingly. All modifications directly derived or associated by those skilled in the art from the content disclosed in the present invention shall belong to the protection scope of the present invention.

Claims

1. the on-line monitoring method of an on-load tap changers of transformers machine performance, it is characterised in that comprise the following steps:

(1) being arranged on the surface of on-load tap changers of transformers by vibrating sensor, Real-time Collection transformator loaded tap-off is opened Vibration signal x (t) on surface, pass, t=1 ... .No, No are seasonal effect in time series length；

(2) use delay coordinate method that vibration signal x (t) is carried out phase space reconfiguration, for

\{\begin{matrix} X (1) = [x (1), x (1 + τ), . . . x (1 + (m - 1) τ)] \\ X (2) = [x (2), x (2 + τ), . . ., x (2 + (m - 1) τ)] \\ . . . . . . \\ X (k) = [x (k), x (k + τ), . . ., x (k + (m - 1) τ)] \\ . . . . . . \\ X (N) = [x (N), x (N + τ), . . ., x (N + (m - 1) τ)] \end{matrix}

In formula, N is the time arrow number in the phase space after phase space reconfiguration, and m is Embedded dimensions, when T is for postponing Between, X (k) (k=1 ..., N) for the kth time arrow of the phase space after vibration signal x (t) is reconstructed, No =N+ (m-1) τ；

This N number of time arrow defines the phase space of a reconstruct, and this, correlation integral function of sequence was N number of time delay

C_{s} (m, N, r, t) = \frac{2}{N (N - 1)} \underset{1 \leq i \leq j \leq N}{Σ} H (r - d_{ij}), r > 0

d_ij=| | X (i)-X (j) | |

H (r) = \{\begin{matrix} 1, & r &GreaterEqual; 0 \\ 0, & r < 0 \end{matrix}

In formula, H (r) is Heaviside jump function, and r is radius；

Described Embedded dimensions m and the calculating process of delay time T be:

(I) standard deviation sigma of calculating vibration signal x (t):

(II) vibration signal x (t) is divided into t0 disjoint subsequence, for

{x_{1}, x_{t_{0} + 1}, x_{{2 t}_{0} + 1}, . . .}

{x_{2}, x_{t_{0} + 2}, x_{{2 t}_{0} + 2}, . . .}

……

{x_{t_{0}}, x_{{2 t}_{0}}, x_{{3 t}_{0}}, . . .}

Herein, a length of 1=No/t of each subsequence_o；

(III) defining the statistic of test of this t subsequence is Embedded dimensions t when being m₀The correlation integral of individual subsequence T when function and Embedded dimensions are 1_oThe meansigma methods of the correlation integral function difference sum of subsequence, its computing formula is

S (m, N_{0}, r, t_{0}) = \frac{1}{t_{0}} Σ_{s = 1}^{t_{0}} [C_{s} (m, N_{0} / t_{0}, r {, t}_{0}) - C_{s}^{m} (1, N_{0} / t_{0}, r {, t}_{0})]

Then work as N₀During → ∞, have

S (m, r, t_{0}) = \frac{1}{t_{0}} Σ_{s = 1}^{t_{0}} [C_{s} (m, r, t_{0}) - C_{s}^{m} (1, r, t_{0})]

(IV) to t_oIndividual subsequence, calculates: the average of statistic of test respectivelyThe difference of statistic of test ValueVariance S with statistic of test_c0r(t_o) its expression formula is respectively

\overset{&OverBar;}{S} (t_{0}) = \frac{1}{16} Σ_{m = 2}^{5} Σ_{j = 1}^{4} S (m, r_{j}, t_{0})

Δ \overset{&OverBar;}{S} (t_{0}) = \frac{1}{4} Σ_{m = 2}^{5} ΔS (m, t_{0})

S_{cor} (t_{0}) = Δ \overset{&OverBar;}{S} (t_{0}) + | \overset{&OverBar;}{S} (t_{0}) |

Δ S (m, t₀)=max{S (m, r_j, t₀)-min{S (m, r_j, t₀)}

In formula, r_j=j σ/2；

(IV) findFirst zero point orFirst minimum point, first zero point or first The time that individual minimum point is corresponding is delay time T

(V) S is found_c0r(t_o) minimum point, time is Vibration Signal Time Series X (t) first corresponding entirety is It is worth greatly time windowCalculating Embedded dimensions m, formula is as follows:

In formula, int is bracket function.

(3) vibration signal is calculated at bunch Center Number Pang Jigecu center of phase space reconstruction according to K-means clustering methodology Position coordinates；

(4) absolute distance at each bunch of center and the 1st Ge Cu center is calculated respectively:

(5) each bunch of center relative error relative to i-th bunch center is calculated: if the 2nd Ge Cu center and the 1st Ge Cu center Absolute distance be reduced to original 20% and the 2nd Ge Cu center absolute distance with the 1st Ge Cu center more than remaining Kc-2 Ge Cu center and the absolute distance at i-th bunch center, then judge that in load ratio bridging switch operating process, running status is abnormal；No Then, it is believed that in load ratio bridging switch operating process, running status is normal.

The on-line monitoring method of on-load tap changers of transformers machine performance the most according to claim 1, its feature exists In, described bunch Center Number Kc and the circular of the position coordinates at each bunch of center are as follows:

(I) optional Kc in N number of vector of phase space reconstruction₁Individual phase point as a bunch center, this Kc₁Ge Cu center is in reconstruct phase The position in space is X (i) (1 ＜ i ＜ Kc1), calculates remaining phase point and this Kc in phase space respectively₁The Euclidean of individual phase point away from From, according to closest principle, remaining phase point is included into this Kc₁In individual bunch；

(II) this Kc is calculated₁The average central of all phase points in bunch, as this Kc₁Individual bunch new bunch of phase space Center, is designated as X ' (i) (1 ＜ i ＜ Kc₁), i.e. X ' (i) is respectively at the m dimension coordinate of phase space with Kc₁Institute in bunch There is the meansigma methods of the m dimension coordinate of phase point；

(III) calculating Xin Cucu center X ' (i) shift ratio BP relative to original Cu Cu center X (i), formula is such as Under

When shift ratio is less than 1%, it is believed that place-centric X ' (i) of new bunch is stable, enter step (IV)；

When shift ratio is more than 1%, it is believed that place-centric X ' (i) of new bunch is unstable, return step (I)；

(IV) according to the Kc determined₁Ge Xincu center X ' (i), recalculates in phase space remaining phase point relative to this Kc₁Individual The Euclidean distance at bunch center X ' (i), is included into this Kc according to closest principle by remaining phase point₁In individual bunch.

(V) this Kc is calculated respectively₁The distance of each bunch of center and each phase point belonging to this bunch in individual bunch, its computing formula is

J (C_{K_{C 1}}) = \underset{X_{i} &Element; C_{K_{C}}}{Σ} {| | X^{'} (i) - X (j) | |}^{2} 1 < i \leq K_{C 1} 1 \leq j \leq P_{t}

In formula, X ' (i) is the phase space position at i-th Cu Cu center, 1 ＜ i≤Kc₁；X (j) is for belong to i-th bunch The Phase Coordinates of jth phase point；P_tFor belonging to counting out mutually of i-th bunch.

The distance at each bunch of center with the phase point belonging to this bunch being added up, obtain the overall distance of this bunch, it calculates public affairs Formula is

J (C) = Σ_{K = 1}^{K_{C 1}} J (K_{C 1})

(VI) Kc=Kc is made₁+ 1, repeat step (I)～step (V), obtain the overall distance of Kc bunch, be designated as J ' (C)；

Calculating reduction speed DR of overall distance, formula is

DR = \frac{| J^{'} (C) - J (C) |}{| J (C) |} \times 100 %

When reducing speed less than 5%, it is believed that the reduction speed of overall distance is basicly stable, and reconstruction signal is by abundant table Showing, now K ' c is the number at required bunch center, X ' (i) (1 ＜ i≤Kc₁) it is the position coordinates at Kc Ge Cu center； Otherwise repeating step (I)～step (VI), until the reduction speed of overall distance is basicly stable, now corresponding Kc is institute The number at Qiu Cu center.