技术领域technical field
本发明涉及配电网开关机械故障诊断技术领域，具体涉及一种基于经验模态分解(empirical mide decomposition，EMD)样本熵和模糊C-均值聚类(fuzzy C-means，FCM)的配电开关机械状态诊断方法。The invention relates to the technical field of mechanical fault diagnosis of distribution network switches, in particular to a distribution switch based on empirical mode decomposition (empirical mide decomposition, EMD) sample entropy and fuzzy C-means clustering (fuzzy C-means, FCM) Machine condition diagnosis method.
背景技术Background technique
配电开关在配电网中具有控制和保护双重功能，它按生产要求实现对电力设备的投切，当设备或线路发生故障时，能够通过合理地控制配电开关将将故障部分从配电网中迅速有效地切除。因此，配电开关故障可能导致严重的电网事故，大部分的配电开关故障都属于机械故障。The power distribution switch has dual functions of control and protection in the distribution network. It realizes switching of power equipment according to production requirements. The mesh is removed quickly and efficiently. Therefore, the failure of distribution switches may lead to serious grid accidents, and most of the failures of distribution switches are mechanical failures.
配电开关分、合闸过程中，操动机构内部各组件依序动作，形成激励源并由此产生振动信号。该振动信号是典型的无规律、非平稳、时变信号，包含有大量的机械状态信息，能够表现出配电开关内部机械结构的正常与否。。振动诊断法就是通过分析配电开关的振动信号，提取有效特征量，从而对其机械状态进行诊断识别的方法。振动诊断法具有可靠性和准确性，有利于实现对开关的非侵入式状态监测。During the opening and closing process of the distribution switch, the components inside the operating mechanism act in sequence to form an excitation source and thus generate a vibration signal. The vibration signal is a typical irregular, non-stationary, time-varying signal, which contains a large amount of mechanical state information, and can show whether the internal mechanical structure of the power distribution switch is normal or not. . The vibration diagnosis method is a method of diagnosing and identifying its mechanical state by analyzing the vibration signal of the distribution switch and extracting effective feature quantities. The reliability and accuracy of vibration diagnostics facilitate non-intrusive condition monitoring of switches.
目前，基于不同原理的配电开关状态监测方法的可靠性、普适性和经济性有待提高；以“事后检修”和“到期必修”为主的配电开关计划检修策略也不能够满足现代电网安全、可靠、经济运行的要求。建立在配电开关状态监测和状态诊断基础上的配电开关状态检修是当前先进的检修策略。在配电开关状态监测的基础上，利用振动诊断法实现配电开关机械状态的诊断识别，能够提前 发现配电开关的潜在故障隐患，从而合理地安排检修策略。At present, the reliability, universality and economy of distribution switch status monitoring methods based on different principles need to be improved; the planned maintenance strategy of distribution switches based on "after-event maintenance" and "due mandatory maintenance" cannot meet the requirements of modern Requirements for safe, reliable and economical operation of power grids. The condition-based maintenance of distribution switches based on the condition monitoring and diagnosis of distribution switches is an advanced maintenance strategy at present. On the basis of the state monitoring of distribution switches, the vibration diagnosis method is used to realize the diagnosis and identification of the mechanical state of distribution switches, which can detect potential failures of distribution switches in advance, so as to arrange maintenance strategies reasonably.
目前，对于开关振动信号的分析处理方法一般基于最大振动事件、振动起始事件及其对应时刻，或基于对振动信号的时域、频域、时频域分析，或利用数据序列直接得到能够表征出开关机械状态的特征参数。EMD根据振动信号序列自身的时间尺度特征将其自适应分解成有限个不同频率、不等带宽的固有模态分量(intrinsic mode function，IMF)和一个残差之和。与小波、小波包等时频分解方法相比，EMD不存在预先选取最优基函数的问题，能够更好地表示出信号的局部特征，实现对非平稳非线性振动信号的时频分解。熵是对系统不确定度的表征，当系统的状态发生变化时，其熵值也会发生相应的变化。样本熵是在近似熵的基础上对其进行修正发展而来的，是一种将非线性数据序列量化为不计自身数据长度比较的统计量。样本熵具有一定的抗噪声干扰能力，算法的计算精度和计算时间较近似熵有了很大的提高。FCM是无监督模糊聚类方法中的一种，算法简单快速，具有比较直观的几何意义。FCM算法在设定n组振动信号向量xj(j＝1,2,…,n)的模糊簇数目c(2≤c≤n)后，通过求取每类簇的聚类中心vi(i＝1,2,…c)，以使得目标函数尽量小为原则，将这n组振动信号的特征数据点按一定的隶属度归于某类簇的聚类中心。At present, the analysis and processing methods for switch vibration signals are generally based on the maximum vibration event, the initial vibration event and its corresponding time, or based on the time domain, frequency domain, and time-frequency domain analysis of the vibration signal, or use the data sequence to directly obtain the characterization The characteristic parameters of the mechanical state of the switch. EMD adaptively decomposes the vibration signal sequence into a limited number of intrinsic mode functions (intrinsic mode function, IMF) with different frequencies and different bandwidths and a sum of residuals according to the time scale characteristics of the vibration signal sequence itself. Compared with time-frequency decomposition methods such as wavelet and wavelet packet, EMD does not have the problem of pre-selecting the optimal basis function, can better represent the local characteristics of the signal, and realize the time-frequency decomposition of non-stationary nonlinear vibration signals. Entropy is a representation of system uncertainty, when the state of the system changes, its entropy value will also change accordingly. Sample entropy is developed on the basis of approximate entropy by modifying it. It is a statistic that quantifies non-linear data sequences as a comparison of their own data length. The sample entropy has a certain ability to resist noise interference, and the calculation accuracy and calculation time of the algorithm have been greatly improved compared with the approximate entropy. FCM is one of the unsupervised fuzzy clustering methods, the algorithm is simple and fast, and has a relatively intuitive geometric meaning. After setting the number of fuzzy clusters c (2≤c≤n) of n groups of vibration signal vectors xj (j=1,2,…,n), the FCM algorithm obtains the cluster center vi ( i=1,2,...c), in order to make the objective function as small as possible, attribute the characteristic data points of these n groups of vibration signals to the cluster center of a certain type of cluster according to a certain degree of membership.
发明内容Contents of the invention
本发明的目的是针对现有技术的不足，提供一种基于EMD样本熵和FCM的配电开关机械状态诊断方法，旨在准确有效地诊断出配电开关运行过程中的机械状态，从而更加合理地制定配电开关的状态检修策略，提高配电开关运行可靠性与稳定性。The purpose of the present invention is to address the deficiencies of the prior art and provide a method for diagnosing the mechanical state of a distribution switch based on EMD sample entropy and FCM, aiming at accurately and effectively diagnosing the mechanical state of the distribution switch during operation, thereby making it more reasonable To formulate the condition-based maintenance strategy for distribution switches to improve the reliability and stability of distribution switches.
本发明的目的通过如下技术方案来实现：The purpose of the present invention is achieved through the following technical solutions:
一种基于EMD样本熵和FCM的配电开关机械状态诊断方法,其特征在于：包括如下步骤：A method for diagnosing the mechanical state of a power distribution switch based on EMD sample entropy and FCM, characterized in that it includes the following steps:
S01：利用配电开关振动信号采集装置，采集配电开关不同机械状态下分、合闸时刻的振动信号；S01: Use the vibration signal acquisition device of the power distribution switch to collect the vibration signal of the power distribution switch at the moment of opening and closing under different mechanical states;
S02：对采集得到的振动信号进行截取处理，得到能够用于分析诊断的有效波形信号并对其进行EMD分解，将振动信号分解成有限个不同频率、不等带宽的IMF和一个残差之和；S02: Intercept and process the collected vibration signal to obtain an effective waveform signal that can be used for analysis and diagnosis, and perform EMD decomposition on it, and decompose the vibration signal into a limited number of IMFs with different frequencies and different bandwidths and a sum of residuals ;
S03：计算不同机械状态下振动信号各阶IMF的样本熵，构成样本熵矩阵，作为配电开关机械状态特征量；S03: Calculate the sample entropy of each order IMF of the vibration signal under different mechanical states to form a sample entropy matrix as the characteristic quantity of the mechanical state of the distribution switch;
S04：以样本熵矩阵作为FCM的输入，通过模糊聚类方法诊断出配电开关的机械状态。S04: Using the sample entropy matrix as the input of FCM, the mechanical state of the distribution switch is diagnosed by the fuzzy clustering method.
进一步的，所述步骤S02中对截取得到的有效振动信号进行EMD自适应分解的具体方法如下：Further, in the step S02, the specific method of performing EMD adaptive decomposition on the intercepted effective vibration signal is as follows:
(1)计算振动信号S(t)的所有局部极大值点和极小值点。(1) Calculate all local maximum and minimum points of the vibration signal S(t).
(2)利用3次样条函数将振动信号的所有极大值点和所有极小值点分别拟合成数据的上包络线a0(t)和下包络线b0(t)，求取上、下包络线的均值m0(t)。(2) Using cubic spline function to fit all maximum points and all minimum points of the vibration signal into upper envelope a0 (t) and lower envelope b0 (t) of the data respectively, Calculate the mean value m0 (t) of the upper and lower envelopes.
m0(t)＝[a0(t)+b0(t)]/2m0 (t)=[a0 (t)+b0 (t)]/2
(3)求出振动信号S(t)与上、下包络线的均值m0(t)的差，得到一个去掉低频成份的振动数据序列，记为h0(t)。(3) Calculate the difference between the vibration signal S(t) and the mean value m0 (t) of the upper and lower envelopes, and obtain a vibration data sequence with low frequency components removed, denoted as h0 (t).
h0(t)＝S(t)-m0(t)h0 (t)=S(t)-m0 (t)
(4)IMF必须满足以下两个条件：a)对于一列振动信号数据，其极值点数目和零点数目必须相等或至多相差一点；b)在振动信号上任意点，由局部极大值点构成的上包络线和局部极小值点构成的下包络线的平均值为零。判断是 否满足条件a)和b)，若满足，则h0(t)为振动信号S(t)的一阶IMF；否则，记h0(t)为S(t)，重复步骤(l)～(3)，直至得到表示振动信号S(t)中高频率分量的第一阶振动信号IMF，记为c1(t)。(4) The IMF must meet the following two conditions: a) For a series of vibration signal data, the number of extreme points and the number of zero points must be equal or at most a little different; b) Any point on the vibration signal must be composed of local maximum points The average value of the upper envelope and the lower envelope formed by the local minimum points is zero. Judging whether the conditions a) and b) are satisfied, if so, then h0 (t) is the first-order IMF of the vibration signal S(t); otherwise, record h0 (t) as S(t), and repeat step (l) ~(3), until the first-order vibration signal IMF representing the high-frequency component in the vibration signal S(t) is obtained, denoted as c1 (t).
(5)记r1(t)＝S(t)-c1(t)为新的待分析信号，重复步骤(1)～(4)，得到第二阶IMF，记为c2(t)，此时余项为r2(t)＝S(t)-c2(t)；继续重复上述步骤，最终可得到n阶IMF，原始振动信号S(t)可表示为(5) Record r1 (t)=S(t)-c1 (t) as the new signal to be analyzed, repeat steps (1) to (4), and obtain the second-order IMF, which is recorded as c2 (t) , the remaining term is r2 (t)=S(t)-c2 (t); continue to repeat the above steps, and finally get n-order IMF, the original vibration signal S(t) can be expressed as
其中，rn(t)为残余函数，表示振动信号的平均趋势。Among them, rn (t) is the residual function, which represents the average trend of the vibration signal.
进一步的，所述步骤S03计算不同机械状态下振动信号各阶IMF的样本熵的具体方法为：Further, the specific method for calculating the sample entropy of each order IMF of the vibration signal in different mechanical states in the step S03 is as follows:
(1)将振动信号数据序列{si}＝{s(1),s(2),…,s(N)}依序构造成m维矢量S(1)，…，S(N-m+1)，其中(1) Construct the vibration signal data sequence {si }={s(1), s(2),...,s(N)} into an m-dimensional vector S(1),..., S(N-m +1), where
S(i)＝{s(i),s(i+1),…s(i+m-1)}S(i)={s(i),s(i+1),...s(i+m-1)}
i＝1,2,…,N-m+1i=1,2,...,N-m+1
(2)将S(i)与S(j)间(i≠j)的距离定义为两者对应元素中差值最大的一个，即(2) Define the distance between S(i) and S(j) (i≠j) as the one with the largest difference among the corresponding elements, that is
d[S(i),S(j)]＝max0-(m-1)|s(i+k)-s(j+k)|d[S(i), S(j)]＝max0-(m-1) |s(i+k)-s(j+k)|
(3)给定数值r(r>0)，统计i、j在不同取值情况下d[S(i),S(j)]<r的数目并计算该数目与总的矢量个数N-m的比值，记为(3) Given a value r (r>0), count the number of d[S(i), S(j)]<r under different values of i and j and calculate the number and the total number of vectors Nm The ratio of is denoted as
(4)计算所有Bim(r)的平均值，记为Bm(r)(4) Calculate the average value of all Bim (r), denoted as Bm (r)
(5)将维数增加至m+1，重复步骤(1)～(4)，得到和Bm+1(r)，其中，Bm(r)是由振动信号构造得到的数据序列中两个数据序列在相似容限r下匹配m个点的概率，Bm+1(r)是振动信号构造得到的数据序列中两个数据序列在相似容限r下匹配m+1个点的概率；(5) Increase the dimension to m+1, repeat steps (1)~(4), and get and Bm+1 (r), where, Bm (r) is the probability that two data sequences in the data sequence constructed by the vibration signal match m points under the similarity tolerance r, Bm+1 (r) is the probability that two data sequences in the data sequence obtained by the vibration signal match m+1 points under the similarity tolerance r;
(6)理论上，此振动信号数据序列的样本熵为(6) Theoretically, the sample entropy of this vibration signal data sequence is
但是，在实际计算过程中，N为有限值，不可能取得无穷大，因此将样本熵记为However, in the actual calculation process, N is a finite value, and it is impossible to obtain infinity, so the sample entropy is recorded as
通常，m＝1或2，r取值为原始振动信号数据序列方差的0.1-0.25倍。Usually, m=1 or 2, and the value of r is 0.1-0.25 times the variance of the original vibration signal data sequence.
进一步的，所述步骤S03将不同机械状态下振动信号各阶IMF的样本熵组合成一个样本熵矩阵，作为配电开关机械状态特征量，其具体方法为：假设共采集l组振动信号，每组振动信号EMD分解为m阶IMF，各组振动信号的m阶IMF可得到m个样本熵值，构成一维样本熵向量，l组振动信号的样本熵向量组合成一个样本熵矩阵X，Further, the step S03 combines the sample entropy of each order IMF of the vibration signal under different mechanical states into a sample entropy matrix, which is used as the characteristic quantity of the mechanical state of the distribution switch. A group of vibration signals EMD is decomposed into m-order IMFs, m-order IMFs of each group of vibration signals can obtain m sample entropy values, which constitute a one-dimensional sample entropy vector, and the sample entropy vectors of l group vibration signals are combined into a sample entropy matrix X,
则but
其中，各行自上而下分别表示定义好的配电开关机械状态，各列自左而右分别表示该机械状态下配电开关振动信号第m阶IMF的样本熵。Among them, each row represents the defined mechanical state of the distribution switch from top to bottom, and each column represents the sample entropy of the mth order IMF of the vibration signal of the distribution switch in the mechanical state from left to right.
进一步的，所述步骤S04中FCM具体步骤为：Further, the specific steps of FCM in the step S04 are:
首先，将振动信号的各阶IMF样本熵值构成的矩阵X＝{xj}，作为FCM的输入，预先给定分类数c和加权指数m，初始化隶属度矩阵First, the matrix X={xj } formed by the entropy values of the IMF samples of each order of the vibration signal is used as the input of the FCM, the classification number c and the weighting index m are given in advance, and the membership degree matrix is initialized
Uc×n＝{uij}，其中Uc×n = {uij }, where
接着，计算聚类中心Next, calculate the cluster centers
构造新的隶属度矩阵Construct a new membership matrix
Uc×n＝{uij}Uc×n ＝{uij }
其中表示所有xj中任意第j个样本属于第i类模糊簇的隶属度，||xj-vi||表示xj到聚类中心vi的欧氏距离。in Indicates that any jth sample in all xj belongs to the membership degree of the i-th fuzzy cluster, and ||xj -vi || represents the Euclidean distance from xj to the cluster center vi .
uij满足以下3个条件：uij satisfies the following three conditions:
1)uij∈[0,1]；1) uij ∈ [0,1];
FCM的目标函数为The objective function of FCM is
其中，1<m<+∞为模糊加权指数，一般取m＝2，d(xj,vi)＝||xj-vi||2，若目标函数Jm(U,V)小于迭代终止因子ε，则聚类过程结束；若大于迭代终止因子ε，则重新计算聚类中心，继续上述计算步骤直到聚类结束。Among them, 1<m<+∞ is the fuzzy weighted index, generally m=2, d(xj ,vi )=||xj -vi ||2 , if the objective function Jm (U,V) is less than If the iteration termination factor ε is greater than the iteration termination factor ε, the clustering center is recalculated, and the above calculation steps are continued until the clustering ends.
进一步的，所述配电开关振动信号采集装置由压电式加速度传感器、信号调理模块、数据采集卡和波形显示存储平台组成本发明是在对配电开关分、合闸振动信号进行EMD的基础上，计算分解得到的振动信号各阶IMF的样本熵值，作为机械诊断的量化依据，最后利用FCM实现机械状态的识别，具有如下有益效果：Further, the vibration signal acquisition device of the power distribution switch is composed of a piezoelectric acceleration sensor, a signal conditioning module, a data acquisition card and a waveform display storage platform. In the above, the sample entropy value of IMF of each order of the decomposed vibration signal is calculated and used as the quantitative basis for mechanical diagnosis, and finally the identification of mechanical state is realized by using FCM, which has the following beneficial effects:
(1)利用EMD对非平稳、非线性的配电开关分、合闸振动信号进行自适 应分解，得到能够表示信号内在特征振动形式的IMF分量，更好地表征出了信号的局部特征；(1) Use EMD to adaptively decompose the non-stationary and nonlinear distribution switch opening and closing vibration signals to obtain the IMF component that can represent the inherent characteristic vibration form of the signal, and better characterize the local characteristics of the signal;
(2)以配电开关振动信号各阶IMF的样本熵作为表征配电开关机械状态的有效特征量，算法的计算精度高、计算时间短，且具有一定的抗干扰能力；(2) The sample entropy of each order IMF of the distribution switch vibration signal is used as an effective feature quantity to characterize the mechanical state of the distribution switch. The calculation accuracy of the algorithm is high, the calculation time is short, and it has a certain anti-interference ability;
(3)基于划分的模糊聚类算法FCM对特征量数据的比例变化具有鲁棒性，以其作为诊断识别方法，实现对配电开关机械状态的准确诊断。(3) The partition-based fuzzy clustering algorithm FCM is robust to the proportional change of feature quantity data, and it is used as a diagnostic identification method to realize accurate diagnosis of the mechanical state of the distribution switch.
附图说明Description of drawings
下面结合附图对本发明作进一步详细说明。The present invention will be described in further detail below in conjunction with the accompanying drawings.
图1本发明中的EMD算法流程图。Fig. 1 is the flow chart of the EMD algorithm in the present invention.
图2本发明的某种机械状态下配电开关振动信号原始波形图。Fig. 2 is the original waveform diagram of the vibration signal of the power distribution switch in a certain mechanical state of the present invention.
图3本发明中图2振动信号经EMD后的一系列IMF分量图。Fig. 3 is a series of IMF component diagrams of the vibration signal in Fig. 2 in the present invention after EMD.
图4本发明流程图。Fig. 4 is the flow chart of the present invention.
具体实施方式detailed description
参照图1和图4所示，一种基于EMD样本熵和FCM的配电开关机械状态诊断方法,包括如下步骤：Referring to Figures 1 and 4, a method for diagnosing the mechanical state of a distribution switch based on EMD sample entropy and FCM includes the following steps:
S01：利用配电开关振动信号采集装置，采集配电开关不同机械状态下分、合闸时刻的振动信号；S01: Use the vibration signal acquisition device of the power distribution switch to collect the vibration signal of the power distribution switch at the moment of opening and closing under different mechanical states;
S02：对采集得到的振动信号进行截取处理，得到能够用于分析诊断的有效波形信号并对其进行EMD分解，将振动信号分解成有限个不同频率、不等带宽的IMF和一个残差之和；S02: Intercept and process the collected vibration signal to obtain an effective waveform signal that can be used for analysis and diagnosis, and perform EMD decomposition on it, and decompose the vibration signal into a limited number of IMFs with different frequencies and different bandwidths and a sum of residuals ;
S03：计算不同机械状态下振动信号各阶IMF的样本熵，构成样本熵矩阵， 作为配电开关机械状态特征量；S03: Calculate the sample entropy of each order IMF of the vibration signal under different mechanical states, form a sample entropy matrix, and use it as the characteristic quantity of the mechanical state of the distribution switch;
S04：以样本熵矩阵作为FCM的输入，通过模糊聚类方法诊断出配电开关的机械状态。S04: Using the sample entropy matrix as the input of FCM, the mechanical state of the distribution switch is diagnosed by the fuzzy clustering method.
具体实施方式如下：将压电式加速度传感器安装在开关操动机构主轴附近，以80kHz的采样频率采集正常状态、机械结构卡涩状态和底座螺丝松动等三类状态下的开关分、合闸振动信号。考虑到实验中的环境等随机因素可能对振动信号产生影响，故对上述三类机械状态下的振动信号分别进行3-5次的采集，比较同类状态下的波形信号，若波形形状相似、呈现出相同的特征信息，则可作为进一步分析的有效信号。对上述三类机械状态下的配电开关合闸振动信号进行分析，对各类机械状态各取3组作为样本数据，再任意取2组作为待检测状态数据。对这11组振动信号波形数据进行编号，编号1、2、3为正常状态，编号4、5、6为机械结构卡涩状态，编号7、8、9为底座螺丝松动状态，编号10、11为待检测状态。对这11组数据进行EMD分解，具体步骤如下：The specific implementation method is as follows: install the piezoelectric acceleration sensor near the main shaft of the switch operating mechanism, and collect the switch opening and closing vibrations under three types of states: normal state, mechanical structure jamming state and base screw loosening with a sampling frequency of 80kHz Signal. Considering that random factors such as the environment in the experiment may have an impact on the vibration signal, the vibration signals in the above three types of mechanical states were collected 3-5 times respectively, and the waveform signals in the same state were compared. If the same characteristic information is obtained, it can be used as an effective signal for further analysis. The closing vibration signals of the distribution switch under the above three types of mechanical states are analyzed, and 3 groups are taken as sample data for each type of mechanical state, and 2 groups are randomly selected as the state data to be detected. Number the 11 sets of vibration signal waveform data, numbers 1, 2, and 3 are normal states, numbers 4, 5, and 6 are mechanical structure jamming states, numbers 7, 8, and 9 are loose base screws, and numbers 10, 11 to be detected. Perform EMD decomposition on these 11 sets of data, the specific steps are as follows:
(1)计算振动信号S(t)的所有局部极大值点和极小值点。(1) Calculate all local maximum and minimum points of the vibration signal S(t).
(2)利用3次样条函数将振动信号的所有极大值点和所有极小值点分别拟合成数据的上包络线a0(t)和下包络线b0(t)。求出上、下包络线的均值m0(t)(2) Fit all the maximum points and all the minimum points of the vibration signal to the upper envelope a0 (t) and the lower envelope b0 (t) of the data respectively by using the cubic spline function. Calculate the mean value m0 (t) of the upper and lower envelopes
m0(t)＝[a0(t)+b0(t)]/2m0 (t)=[a0 (t)+b0 (t)]/2
(3)求出振动信号S(t)与上、下包络线的均值m0(t)的差，得到一个去掉低频的振动数据序列，记为h0(t)。(3) Calculate the difference between the vibration signal S(t) and the mean value m0 (t) of the upper and lower envelopes, and obtain a vibration data sequence with low frequencies removed, denoted as h0 (t).
h0(t)＝S(t)-m0(t)h0 (t)=S(t)-m0 (t)
(4)IMF必须满足以下两个条件：a)对于一列振动信号数据，极值点数目和零点数目必须相等或至多相差一点；b)在振动信号上任意点，由局部极大值点构成的上包络线和局部极小值点构成的下包络线的平均值为零。判断是否 满足条件a)和b)，若满足。若满足，则h0(t)为振动信号S(t)的一阶IMF；否则，记h0(t)为S(t)，重复步骤(l)～(3)，直至得到表示振动信号S(t)中高频率分量的第一阶振动信号IMF，记为c1(t)。(4) The IMF must meet the following two conditions: a) For a series of vibration signal data, the number of extreme points and the number of zero points must be equal or at most a little different; b) At any point on the vibration signal, the number of local maximum points The average value of the lower envelope formed by the upper envelope and the local minimum points is zero. Determine whether conditions a) and b) are met, and if so. If it is satisfied, then h0 (t) is the first-order IMF of the vibration signal S(t); otherwise, record h0 (t) as S(t), and repeat steps (1) to (3) until the vibration signal The first-order vibration signal IMF of the high-frequency component in S(t), denoted as c1 (t).
(5)记r1(t)＝S(t)-c1(t)为新的待分析信号，重复步骤(1)～(4)，得到第二阶IMF，记为c2(t)，此时余项为r2(t)＝S(t)-c2(t)。继续重复上述步骤，最终可得到n阶IMF，原始振动信号S(t)可表示为(5) Record r1 (t)=S(t)-c1 (t) as the new signal to be analyzed, repeat steps (1) to (4), and obtain the second-order IMF, which is recorded as c2 (t) , and the remaining term is r2 (t)=S(t)-c2 (t). Continue to repeat the above steps, and finally the n-order IMF can be obtained, and the original vibration signal S(t) can be expressed as
其中，rn(t)为残余函数，表示振动信号的平均趋势。Among them, rn (t) is the residual function, which represents the average trend of the vibration signal.
EMD算法流程如图2所示。The EMD algorithm process is shown in Figure 2.
某种状态下的配电开关振动信号的原始波形如图3所示，经EMD后得到的一系列IMF分量如图4所示。The original waveform of the vibration signal of the distribution switch in a certain state is shown in Figure 3, and a series of IMF components obtained after EMD are shown in Figure 4.
取维数m＝2，计算11组振动信号EMD分解后各阶IMF的样本熵值，得到一组11×10样本熵矩阵X。Take the dimension m=2, calculate the sample entropy values of each order IMF after 11 groups of vibration signal EMD decomposition, and obtain a set of 11×10 sample entropy matrix X.
其中各行从上至下依次对应编号1-11所定义的机械状态，各列从左至右依次对应该编号机械状态下合闸振动信号各阶IMF的样本熵。Among them, each row corresponds to the mechanical state defined by numbers 1-11 from top to bottom, and each column corresponds to the sample entropy of each order IMF of the closing vibration signal in the numbered mechanical state from left to right.
以X作为FCM聚类的输入，设置FCM聚类数目c为3，加权指数m为2， 迭代终止因子为ε＝10-5，最大迭代数为100。聚类结果显示编号1、2、3振动信号为同类信号，编号4、5、6、10振动信号为同类信号，7、8、9、11振动信号为同类信号，与真实情况相符。Taking X as the input of FCM clustering, set the number of FCM clusters c to 3, the weighting index m to 2, the iteration termination factor to ε=10-5 , and the maximum iteration number to 100. The clustering results show that the vibration signals of numbers 1, 2, and 3 are of the same type, the vibration signals of numbers 4, 5, 6, and 10 are of the same type, and the vibration signals of 7, 8, 9, and 11 are of the same type, which is consistent with the real situation.
通过以上实验步骤与实验结果的详细描述，可以看出本发明以EMD作为配电开关振动信号的时频分解方法，并在此基础上计算IMF样本熵矩阵作为特征量，最后通过FCM聚类进行模式识别。本发明能够准确有效地诊断出配电开关的机械状态，具有一定的工程应用价值。Through the detailed description of the above experimental steps and experimental results, it can be seen that the present invention uses EMD as the time-frequency decomposition method of the distribution switch vibration signal, and on this basis, calculates the IMF sample entropy matrix as the feature quantity, and finally performs through FCM clustering pattern recognition. The invention can accurately and effectively diagnose the mechanical state of the power distribution switch, and has certain engineering application value.
以上所述，仅为本发明的较佳实施例而已，故不能以此限定本发明实施的范围，即依本发明申请专利范围及说明书内容所作的等效变化与修饰，皆应仍属本发明专利涵盖的范围内。The above is only a preferred embodiment of the present invention, so it cannot limit the scope of the present invention, that is, equivalent changes and modifications made according to the patent scope of the present invention and the content of the specification should still belong to the present invention covered by the patent.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410196703.6ACN103968937B (en) | 2014-05-09 | 2014-05-09 | A kind of distribution switch mechanical state diagnostic method based on EMD Sample Entropies and FCM |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410196703.6ACN103968937B (en) | 2014-05-09 | 2014-05-09 | A kind of distribution switch mechanical state diagnostic method based on EMD Sample Entropies and FCM |
| Publication Number | Publication Date |
|---|---|
| CN103968937A CN103968937A (en) | 2014-08-06 |
| CN103968937Btrue CN103968937B (en) | 2017-12-26 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410196703.6AActiveCN103968937B (en) | 2014-05-09 | 2014-05-09 | A kind of distribution switch mechanical state diagnostic method based on EMD Sample Entropies and FCM |
| Country | Link |
|---|---|
| CN (1) | CN103968937B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105352726B (en)* | 2015-10-20 | 2018-11-30 | 宁波大学 | A kind of method for diagnosing faults of gear |
| CN105510687B (en)* | 2015-12-24 | 2019-02-19 | 合肥工业大学 | A kind of electric voltage exception characteristic recognition method based on empirical mode decomposition |
| CN105891707A (en)* | 2016-05-05 | 2016-08-24 | 河北工业大学 | Opening-closing fault diagnosis method for air circuit breaker based on vibration signals |
| CN106510751A (en)* | 2016-09-28 | 2017-03-22 | 华南师范大学 | Multifunctional electronic stethoscope and auscultation system |
| CN106990303A (en)* | 2017-03-15 | 2017-07-28 | 国家电网公司 | A kind of Diagnosis Method of Transformer Faults |
| CN108535638B (en)* | 2018-01-25 | 2020-05-12 | 国网浙江省电力有限公司电力科学研究院 | On-load tap-changer mechanical state monitoring method and system based on multilayer filter |
| CN109357751B (en)* | 2018-09-26 | 2020-12-29 | 国网上海市电力公司 | A power transformer winding loose defect detection system |
| CN109784310B (en)* | 2019-02-02 | 2020-12-04 | 福州大学 | Feature extraction method of distribution switch mechanical fault based on CEEMDAN and weighted time-frequency entropy |
| CN110907155A (en)* | 2019-12-02 | 2020-03-24 | 吉林松江河水力发电有限责任公司 | A kind of fault monitoring method of rotating shaft of hydraulic turbine |
| CN111122162B (en)* | 2019-12-25 | 2020-12-01 | 杭州电子科技大学 | Industrial system fault detection method based on Euclidean distance multi-scale fuzzy sample entropy |
| CN111474410B (en)* | 2020-04-07 | 2022-09-09 | 宁波市重强电器有限公司 | Detection method of electric connector |
| CN112636642B (en)* | 2020-12-17 | 2023-03-24 | 广东工业大学 | Method and device for evaluating performance state of numerical control cutting tool bit made of flexible material |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2465811C2 (en)* | 2007-12-10 | 2012-11-10 | БАЙЕР ХЕЛТКЭА ЭлЭлСи | Control of quick charging and supply to metre of analytes in fluid medium arranged with battery supply |
| CN101187593A (en)* | 2007-12-18 | 2008-05-28 | 华南理工大学 | Kolmogorov Entropy Diagnosis Method for Rotor Vibration Faults of Steam Turbine in Power Station |
| CN101763034B (en)* | 2009-12-01 | 2013-04-17 | 深圳市科陆电子科技股份有限公司 | Method for realizing self-adapting network reconfiguration at distribution network terminal |
| CN101782426B (en)* | 2010-01-29 | 2012-04-18 | 浙江大学 | Detection method of looseness fault vibration of power transformer winding |
| CN101887547A (en)* | 2010-06-30 | 2010-11-17 | 广西电网公司电力科学研究院 | Assistant decision system for condition-based maintenance and risk evaluation of power transmission and transformation equipment |
| CN103217613B (en)* | 2013-03-23 | 2015-07-08 | 李景禄 | Power transmission and distribution line fault property judgment and fault state tracing and detecting method |
| Title |
|---|
| 基于EMD和样本熵的滚动轴承故障SVM识别;来凌红 等;《煤矿机械》;20110131;第32卷(第01期);249-252* |
| Publication number | Publication date |
|---|---|
| CN103968937A (en) | 2014-08-06 |
| Publication | Publication Date | Title |
|---|---|---|
| CN103968937B (en) | A kind of distribution switch mechanical state diagnostic method based on EMD Sample Entropies and FCM | |
| CN113673346B (en) | A method for motor vibration data processing and state recognition based on multi-scale SE-Resnet | |
| CN116625685B (en) | Rolling bearing combination fault diagnosis method based on multi-domain feature construction | |
| CN113420691A (en) | Mixed domain characteristic bearing fault diagnosis method based on Pearson correlation coefficient | |
| CN112014047B (en) | Mechanical fault diagnosis method for on-load tap-changer | |
| CN119959781B (en) | A method and system for battery pack health prediction | |
| CN112577743B (en) | A Fault Diagnosis Method for Rolling Bearings Based on Maximum Local Boundary Criterion | |
| CN115221930A (en) | Fault diagnosis method for rolling bearing | |
| CN110399854B (en) | Rolling bearing fault classification method based on hybrid feature extraction | |
| CN113542241A (en) | An intrusion detection method and device based on CNN-BiGRU hybrid model | |
| CN117556344B (en) | Fault diagnosis method and system for ball mill transmission system based on multi-source information fusion | |
| CN118734052A (en) | A multi-dimensional fusion GIS combined electrical appliance fault diagnosis method | |
| CN117010442A (en) | Equipment residual life prediction model training method, residual life prediction method and system | |
| CN116665483A (en) | A Novel Method for Predicting Remaining Parking Spaces | |
| CN114167237A (en) | A GIS partial discharge fault identification method, system, computer equipment and storage medium | |
| Tan et al. | Fault diagnosis method for rolling bearing based on VMD and improved SVM optimized by METLBO | |
| CN115453336B (en) | GIS disconnecting link running state judging method, GIS disconnecting link running state judging system and computer readable storage medium | |
| CN119375811A (en) | An intelligent fault judgment system for electric energy meter verification | |
| CN110186557A (en) | A kind of Reactor Fault diagnostic method | |
| CN118606783A (en) | Insulator surface contamination degree assessment method and system | |
| CN108375729A (en) | Degenerate state appraisal procedure is clustered based on the motor SOM that Fisher differentiates | |
| CN120296387A (en) | UHV converter transformer OLTC fault diagnosis method integrating multi-feature information and RIME-CNN-BiLSTM-SAM | |
| CN115563480A (en) | Gear fault identification method based on screening symplectic geometric mode decomposition based on kurtosis ratio coefficient | |
| CN115293208A (en) | A kind of isolation switch fault diagnosis method and system | |
| CN118730283A (en) | A device fault diagnosis method, device, computer device and storage medium |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |