CN108872777B - Power transformer winding state evaluation method based on system delay order estimation - Google Patents

Power transformer winding state evaluation method based on system delay order estimation Download PDF

Info

Publication number
CN108872777B
CN108872777B CN201810550863.4A CN201810550863A CN108872777B CN 108872777 B CN108872777 B CN 108872777B CN 201810550863 A CN201810550863 A CN 201810550863A CN 108872777 B CN108872777 B CN 108872777B
Authority
CN
China
Prior art keywords
vibration
winding
power transformer
signal
order
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810550863.4A
Other languages
Chinese (zh)
Other versions
CN108872777A (en
Inventor
郑婧
黄海
潘杰
胡异炜
李灵至
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN201810550863.4A priority Critical patent/CN108872777B/en
Publication of CN108872777A publication Critical patent/CN108872777A/en
Application granted granted Critical
Publication of CN108872777B publication Critical patent/CN108872777B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
    • G01R31/72—Testing of electric windings

Abstract

The invention discloses a power transformer winding state evaluation method based on improved system delay order estimation, which better reflects the mechanical structure state of a winding based on a winding vibration generation mechanism, directly associates winding faults with characteristic quantities and provides a basis for the effectiveness and scientificity of fault diagnosis. Compared with the prior art, the method has the advantages that the extracted characteristic quantity is closer to the actual order of the system, and the system characteristic and the mechanical structure state of the winding can be more accurately reflected; in addition, the invention has no electrical connection with the transformer, does not need to power off the transformer and has little influence on the operation of the whole power system.

Description

Power transformer winding state evaluation method based on system delay order estimation
Technical Field
The invention belongs to the technical field of power transformer safety fault detection, and particularly relates to a power transformer winding state evaluation method based on improved system delay order estimation.
Background
The large power transformer is used as an important ring in a power system, and the safe operation of the large power transformer is important to ensure the safe and reliable power supply of a power grid. In a power transformer fault, the percentage of the fault in the winding is as high as 46.4%, which is the most important component of the transformer fault, and the serious fault caused by the mechanical deformation of the winding under the action of electrodynamic force accounts for 70% of the total fault of the winding. Even if the winding is slightly deformed, the problems of deterioration of mechanical properties of the winding, reduction of insulation strength and short-circuit resistance and the like are caused, and great potential safety hazards are brought. Therefore, it is necessary and important to monitor and evaluate the condition of the winding of the main fault component in the live monitoring of the transformer.
The vibration analysis method is used for analyzing the collected oil tank wall vibration signals to obtain the state information of the internal parts of the transformer, and when the internal mechanical structure of the transformer is changed, the system response of a vibration system is inevitably changed. Therefore, a relation model of the input electromagnetic force and the vibration response of the output winding can be constructed, and the mechanical structure state of the transformer winding can be reflected by extracting the system characteristics of the model.
Chinese patent No. 201310047702.0 proposes a power transformer fault diagnosis method based on an electric-vibration model, which establishes a regression model between each frequency component of vibration and current, voltage and oil temperature by collecting voltage signals, current signals, oil temperature signals and a plurality of vibration measuring points of a transformer, and compares actual measurement data of vibration with prediction data obtained through the model to diagnose a fault of the transformer; although the method considers the nonlinearity in the transformer vibration to a certain extent, the method still has the problems that partial frequency components introduced by strong nonlinearity such as hysteresis effect cannot be modeled or the model error is large, and the like, so the method has certain limitations on accuracy and effectiveness.
The Chinese patent with the application number of 201710867458.0 proposes and constructs a Hammerstein nonlinear model between current and winding vibration on the basis of the vibration principle of a transformer winding, and reflects the mechanical structure state of the transformer winding by extracting the system characteristic of the model, namely the system delay order; the method considers the physical mechanism of the influence of mechanical faults on the characteristics of a model system, only extracts partial information related to the faults in the model as a diagnosis basis, can effectively avoid misjudgment directly caused by model errors, and has the problems of large estimation errors and even estimation errors when the difference between input orders and output orders of the system is large in the system delay order extraction algorithm, so that the accuracy of the winding state evaluation result is influenced.
Disclosure of Invention
In view of the above, the present invention provides a power transformer winding state evaluation method based on improved system delay order estimation, which can detect the mechanical structure state of the power transformer winding on line, improve the problem of inaccurate order estimation when the input and output order difference is large, and enable the extracted parameters to reflect the actual state of the winding more accurately.
A power transformer winding state evaluation method based on improved system delay order estimation comprises the following steps:
(1) the method comprises the following steps that a plurality of vibration sensors are dispersedly arranged on the surface of an oil tank of the power transformer corresponding to winding positions, vibration signals of the vibration sensors of the power transformer under a low-voltage load operation condition are recorded, and primary side currents of the power transformer are synchronously collected;
(2) preprocessing the vibration signal and the current signal, wherein the preprocessing comprises the demagnetization processing of the current signal;
(3) for any vibration sensor, a relation model of current and winding vibration is constructed, and the delay order N of the model is estimated according to a vibration signal preprocessed by the vibration sensor and a current signal and is used as a characteristic quantity of system characteristics, wherein N is Na+Nb，NaAnd NbActual linear delay order, N, of system output and input, respectivelyaAnd NbAre all natural numbers; further according to N, NaAnd NbJudging the mechanical structure state of the power transformer winding under the condition of the data of the vibration sensor;
(4) traversing all the vibration sensors according to the step (3), and finally judging that the winding of the power transformer is normal when the mechanical structure state of the winding is judged to be normal based on the vibration sensors in a certain proportion, or judging that the winding of the power transformer is abnormal.
Further, the specific process of preprocessing the vibration signal and the current signal in the step (2) is as follows: firstly, carrying out normalization processing on a vibration signal and a current signal; the normalized current signal is then non-linearly transformed according to the following equation:
Figure GDA0002482624210000031
wherein: i (t) is the normalized current signal, t is the time, ipAnd (t) is a current signal after nonlinear transformation.
Further, the specific process of estimating the model delay order n in step (3) is as follows:
3.1 calculate the L ipschitz coefficient for all sample point combinations in the relationship model for a delay order of n according to the following formula:
Figure GDA0002482624210000032
wherein: lij (n)The L ipschitz coefficient representing the combination of the ith sampling point and the jth sampling point in the relation model under the condition that the delay order is n, wherein i and j are the serial numbers of the sampling points, and when n is an even number, gamma is1(i)～γn(i) The correspondence is equal to y (i-1), x (i-1), y (i-2), x (i-2)a),x(i-Nb)，γ1(j)～γn(j) Corresponds to y (j-1), x (j-1), y (j-2), x (j-2)a),x(j-Nb) (ii) a When n is an odd number, γ1(i)～γn(i) The correspondence is equal to y (i-1), x (i-1), y (i-2), x (i-2)b),x(i-Nb),y(i-Na)，γ1(j)～γn(j) Corresponds to y (j-1), x (j-1), y (j-2), x (j-2)b),x(j-Nb),y(j-Na) (ii) a y (i) and y (j) are respectively the signal values of the ith sampling point and the jth sampling point in the vibration signal preprocessed by the vibration sensor, y (i-1) and y (j-1) are respectively the signal values of the ith-1 sampling point and the jth-1 sampling point in the vibration signal preprocessed by the vibration sensor, y (i-2) and y (j-2) are respectively the signal values of the ith-2 sampling point and the jth-2 sampling point in the vibration signal preprocessed by the vibration sensor, and y (i-N)a) And y (j-N)a) Respectively corresponding to the ith-N in the vibration signal preprocessed by the vibration sensoraSample points and j-NaSignal values of the sampling points, y (i-N)b) And y (j-N)b) Respectively corresponding to the ith-N in the vibration signal preprocessed by the vibration sensorbSample points and j-NbThe signal values of the sampling points, x (i-1) and x (j-1) are respectively the signal values of the processed current signal corresponding to the i-1 th sampling point and the j-1 th sampling point, x (i-2) and x (j-2) are respectively the signal values of the processed current signal corresponding to the i-2 th sampling point and the j-2 th sampling point, and x (i-N)b) And x (j-N)b) Respectively corresponding to the ith-N in the processed current signalbSample points and j-NbSignal values of the sampling points;
3.2, sorting all L ipsitz coefficients obtained by calculation in the step 3.1 from large to small, cutting off the first m L ipsitz coefficients, and further calculating a L ipsitz average coefficient of the relation model under the condition that the delay order is n according to the following formula:
Figure GDA0002482624210000033
wherein: l(n)L ipsitz average coefficient, l, representing a relational model for a delay order of n(n)(z) is the z-th L ipschitz coefficient after ordering from big to small, and m is usually 0.01Nset，NsetThe total number of sampling points is;
3.3 initialization delay order N2 and Na＝Nb＝1；
3.4 converting NaAdding 1 in total, and then calculating N-N according to the steps 3.1-3.2a+NbAnd N ═ Na-1+NbCorresponding average coefficient l of L ipsitz(Na+Nb)And l(Na-1+Nb)And the following judgments were made:
if l(Na+Nb)-l(Na-1+Nb)≤×l(Na-1+Nb)And N isbIf not, determining NaThe value before the accumulation is carried out at the time and the step 3.5 is carried out;
if l(Na+Nb)-l(Na-1+Nb)≤×l(Na-1+Nb)And N isbIf determined, then determine NaFor the value before the present accumulation and making sure NaAnd NbAdding to obtain a delay order n;
if l(Na+Nb)-l(Na-1+Nb)＞×l(Na-1+Nb)And N isbIf not, executing step 3.5;
if l(Na+Nb)-l(Na-1+Nb)＞×l(Na-1+Nb)And N isbIf it is determined, step 3.4 is repeated until NaDetermining and making sure NaAnd NbAdding to obtain a delay order n;
3.5 converting NbAdding 1 in total, and then calculating N-N according to the steps 3.1-3.2a+NbAnd N ═ Na+Nb-L ipsitz mean coefficient l corresponding to case 1(Na+Nb)And l(Na+Nb-1)And the following judgments were made:
if l(Na+Nb)-l(Na+Nb-1)≤×l(Na+Nb-1)And N isaIf not, determining NbThe value before the accumulation is the value before the accumulation and the step 3.4 is executed again;
if l(Na+Nb)-l(Na+Nb-1)≤×l(Na+Nb-1)And N isaIf determined, then determine NbFor the value before the present accumulation and making sure NaAnd NbAdding to obtain a delay order n;
if l(Na+Nb)-l(Na+Nb-1)＞×l(Na+Nb-1)And N isaIf not, returning to execute the step 3.4;
if l(Na+Nb)-l(Na+Nb-1)＞×l(Na+Nb-1)And N isaIf it is determined, step 3.5 is repeated until NbDetermining and making sure NaAnd NbAdding to obtain a delay order n;
wherein: is the convergence factor.
Further, when N isaAnd NbWhen none of them is determined, the convergence factor is equal to1(ii) a When N is presentaAnd NbWhen one of them is determined, the convergence factor is equal to2(ii) a Wherein the content of the first and second substances,1and2respectively, two different constants are set.
Further, the specific process of determining the mechanical structure state of the power transformer winding in the step (3) is as follows: firstly, calculating the delay orders N and N of the model according to the processes of the steps (1) to (3) under the condition of ensuring that the power transformer winding is normalaAnd Nb(ii) a Then, N, N obtained by actual measurement and calculation based on the current vibration sensor data in the step (3)aAnd NbN and N under normal condition of power transformer windingaAnd NbComparing, and if the two data are matched, judging that the mechanical structure state of the power transformer winding is normal under the condition of the current vibration sensor data; and if the two data are not matched, judging that the mechanical structure state of the power transformer winding is abnormal under the condition of the current vibration sensor data.
Based on the technical scheme, the invention has the following beneficial technical effects:
1. the invention has no electrical connection with the transformer, does not need to power off the transformer and has little influence on the operation of the whole power system.
2. The method better reflects the mechanical structure state of the winding based on the winding vibration generation mechanism, directly associates the winding fault with the characteristic quantity, and provides a basis for the effectiveness and the scientificity of fault diagnosis.
3. Compared with the method before improvement, the extracted characteristic quantity is closer to the actual order of the system, and the system characteristic and the mechanical structure state of the winding can be more accurately reflected.
Drawings
FIG. 1 is a schematic flow chart of the steps of the method of the present invention.
FIG. 2 is a layout diagram of measuring points on the surface of a fuel tank of a power transformer.
FIG. 3 shows the input current i (t) and the vibration component
Figure GDA0002482624210000051
Graph of the relationship of (c).
FIG. 4 shows the input current i after the hysteresis removalp(t) and vibration component
Figure GDA0002482624210000052
Graph of the relationship of (c).
Fig. 5 is a schematic flow chart of the improved delay order estimation algorithm of the present invention.
Fig. 6(a) is a diagram illustrating the order estimation process and the result of the original algorithm.
FIG. 6(b) is a diagram illustrating the order estimation process and the result of the improved algorithm according to the present invention.
Detailed Description
In order to more specifically describe the present invention, the following detailed description is provided for the technical solution of the present invention with reference to the accompanying drawings and the specific embodiments.
The specific experimental object of the embodiment is a 110kV three-phase oil-immersed power transformer, and in order to verify the effectiveness of the method, the vibration characteristics of the same winding under a normal condition and a fault condition are specially compared. The fault winding is realized by artificial damage of short circuit impact on a normal winding, the transformer short circuit experiment is a test method specially aiming at the winding, and the current of the transformer winding can reach a rated value by short circuit at a low-voltage end and applying voltage at a high-voltage end.
During diagnosis, in order to obtain vibration signals with different amplitudes without distortion, a vibration sensor with higher sensitivity needs to be selected; in order to ensure the vibration response of the sensor within the sampling filtering frequency band, the vibration sensor is fixed on the side wall of the oil tank by adopting a magnetic seat adsorption or glue bonding mode. The vibration sampling device comprises main modules such as pre-amplification, anti-aliasing filtering, AD sampling and the like, wherein the number of AD sampling bits is at least 12, and the cut-off frequency of an anti-aliasing filter is 2000 Hz; when the vibration signal is sampled, the sampling frequency is at least 4000 Hz. In this embodiment, the sampling frequency for collecting the vibration signal is set to 10000Hz, the sampling number of the AD module is 16 bits, and the whole experiment process is recorded in a continuous sampling mode.
As shown in fig. 1, the method for estimating the winding state of a power transformer based on the improved system delay order estimation of the present invention includes the following steps:
(1) and arranging vibration measuring points.
The method is characterized in that 5 measuring points are arranged on the surface of the oil tank of the power transformer, and as shown in FIG. 2, the A-phase winding can be artificially damaged in an experiment, so that the 5 measuring points are arranged on the wall of the oil tank corresponding to the A-phase winding.
(2) And collecting vibration signals and current signals under normal conditions and abnormal conditions.
Firstly, carrying out a short-circuit experiment on a new transformer, gradually increasing the voltage on a high-voltage side to increase the current on a winding, increasing the current according to the proportion of 10% every time, and gradually reaching a rated value; after the current is increased by 10%, the current is kept stable for 30 seconds, and the vibration of all the measuring points at all times is recorded by using a continuous sampling mode.
And then, artificially damaging the A-phase winding by using the short-circuit impact current, confirming that the A-phase winding is deformed to a greater extent at the middle position, performing the same short-circuit experiment on the transformer, and recording the vibration of all measuring points in the whole process.
(3) The signal is pre-processed.
When the input current i (t) is used as the independent variable, the winding vibrates at the measuring point
Figure GDA0002482624210000061
The vibration component contributed
Figure GDA0002482624210000062
When the dependent variable is plotted as a relation curve between the dependent variable and the dependent variable (measured point)
Figure GDA0002482624210000063
For example, as shown in fig. 3), it can be seen that the output-input relationship curve exhibits obvious hysteresis characteristics due to the strong nonlinear characteristics introduced by the magnetic field of the winding, after the following nonlinear transformation is performed on the input signal:
Figure GDA0002482624210000064
wherein: i' (t) is a differentiated version of the input signal i (t), ipAnd (t) is the transformed input signal. The non-linear variation maps the original input signal from the low latitude space to the high dimension space, removes the hysteresis characteristic of the original non-linear module in the system, and fig. 4 shows the input current i after the conversionp(t) and vibration component
Figure GDA0002482624210000065
The relationship of (1).
(4) And extracting the delay order of the relation model by using an improved order estimation algorithm, and judging the mechanical structure state of the winding.
The unit electric distribution force of the winding is mainly generated by current flowing through a winding coil and is related to a leakage magnetic field near the winding, the size of the leakage magnetic field is directly related to the current, but is related to the structure of the winding and the height of the winding coil; thereby electrically distributing force to unit
Figure GDA0002482624210000071
Non-linear function F regarded as being related only to current i (t)1(i (t)) and a non-linear function related only to coil position
Figure GDA0002482624210000072
Combinations of (a) and (b). Thus, the equation for winding vibration versus input current can be written as:
Figure GDA0002482624210000073
wherein:
Figure GDA0002482624210000074
indicating the point of response of the winding vibration exciting force to the tank wall
Figure GDA0002482624210000075
The overall equivalent unit impulse response of (a).
According to the above formula, the current-winding vibration relation model is composed of a static nonlinear module F1() And dynamic linear module
Figure GDA0002482624210000076
The nonlinear system formed is in accordance with the structure of a classical nonlinear Hammerstein model, namely, the nonlinear system is formed by connecting an input terminal nonlinear static module and a linear dynamic module in series.
From the above model, it can be seen that the mechanical structural characteristics of the winding will directly affect the position-dependent nonlinear function
Figure GDA0002482624210000077
Characteristics and transfer characteristics, so that the linear module characteristics in the model can be directly influenced, and therefore, the linear module system characteristics can be extracted as the winding mechanical structure state characteristic quantity.
For a linear system, the change of the delay order directly affects the system response, so if the delay order of the system can be directly estimated according to the input and output signal characteristics, the change of the delay order can be used for evaluating the mechanical structure characteristics of the transformer winding.
For a nonlinear system, the output y (t) can be described as a function of the delay input, output, i.e.:
y(t)＝g(y(t-1),…,y(t-Na),x(t-1),…,x(t-Nb))
＝g(γ1,γ2,…,γn)
wherein: x (t) and y (t) are the input and output, respectively, of the nonlinear system, and NaAnd NbThe actual linear delay order of the system output and input, respectively, g () is a non-linear function and assuming that the function has a continuity characteristic, the L ipschitz continuity condition is satisfied, where γ is usediI-1, …, N as a marker for the argument of the nonlinear function, and let N-Na+NbIs the number of arguments.
Define L ipsitz coefficient lij (n)To characterize the non-linear function g (gamma)1,γ2,…,γn) Continuity of (c):
Figure GDA0002482624210000081
when essential variable gamma is absent from function argumenthIn the absence of this variable, L ipschitz factor lij (h-1)The value will be much greater than if the necessary variable γ was presenthCoefficient of time value lij (h)At this time iij (h-1)lij (h)(ii) a In contrast thereto, if the variable γh+1A redundant or unnecessary argument, which corresponds to the L ipsitz coefficient lij (h+1)And coefficient l in the absence of this variableij (h)The values being approximately equal, i.e. lij (h+1)≈lij (h)。
To avoid the effect of the noise introduced during the measurement on the above results, an average coefficient l of L ipschitz was used(n)Substitution coefficient lij (n):
Figure GDA0002482624210000082
Wherein: l(n)(z) is lij (n)The z-th coefficient value after reordering according to decreasing rule, m is usually 0.01Nset,NsetIs the number of input-output pairs used for order estimation.
The main idea of improving the pre-order extraction algorithm is to find the minimum independent variable number N (N equals to N)a+Nb) And let its corresponding L ipschitz mean coefficient satisfy the following condition:
Figure GDA0002482624210000083
wherein: an empirical threshold is typically set to 0.1. However, the order obtained by the above method, the order N, is readily apparentaAnd NbThe steps are increased in an alternating manner in the process of searching for the optimal order, so that when the optimal order n is finally determined, when the order n is an even number,
Figure GDA0002482624210000084
when the order N is an even number, Na＝Nb+1. For a practical system, the practical order is NaAnd NbAre often not equal. With Na《NbFor example, in the order estimation process, the order NaAnd NbAlternatively increasing in a stepwise manner until NaWhen approaching to the actual value, the order estimation process satisfies the stopping condition and stops the process in advance, and the order N is at the momentbMuch smaller than the actual value and inaccurate.
Therefore, in order to accurately estimate the delay order and thus more accurately evaluate the system state, the method of the present invention adopts an improved order estimation algorithm, and the steps thereof are shown in fig. 5:
① the delay order N is first initialized to 2, i.e. N is seta＝1，NbThe current L ipsitz coefficient, l, is recorded and calculated at 1(2)。
② equivalent to the order NaWhen not determined, increasing undetermined order number to order Na＝Na+1, record and calculate the current L ipschitz mean coefficient
Figure GDA0002482624210000091
If the optimum order is NaIf so, step ④ is performed.
③ comparison
Figure GDA0002482624210000092
And
Figure GDA0002482624210000093
if the difference between the two satisfies the condition
Figure GDA0002482624210000094
Figure GDA0002482624210000095
The optimum order NaThe estimation is finished, let it be Na＝Na-1; if not, executing the next step.
④ equivalent to the order NbWhen not determined, increasing undetermined order number to order Nb＝Nb+1, record and calculate the current L ipschitz mean coefficient
Figure GDA0002482624210000096
If the optimum order is NbIt has been determined that the algorithm is terminated and the order estimation is finished.
⑤ comparison
Figure GDA0002482624210000097
And
Figure GDA0002482624210000098
if the difference between the two satisfies the condition
Figure GDA0002482624210000099
Figure GDA00024826242100000910
The optimum order NbThe estimation is finished, let it be Nb＝Nb-1, if the condition is not met, re-executing step ②.
In the above algorithm, the convergence threshold is set, and
Figure GDA00024826242100000911
wherein1＝0.1，2＝0.04。
In order to verify the accuracy and effectiveness of the above algorithm and to illustrate the advancement of the algorithm compared with the original algorithm before improvement, the delay order of a nonlinear system is first estimated by a mathematical simulation method, and the actual order is compared with the estimated order. The nonlinear system is formed by connecting a nonlinear module and a linear module in series, wherein the relation expression of the nonlinear module is characterized as follows:
v(t)＝sin(0.6πx(t))+0.1cos(1.5πx(t))
and the difference equation characterizing the linear module is:
y(t)＝-0.28y(t-1)+0.47y(t-6)+0.11v(t-1)+0.08v(t-3)
as can be seen from the above formula, the actual delay order of the system is Na＝6，NbAnd the system input x (t) is defined as a random sequence distributed over (-1, 1).
Order estimation is performed using 500 data pairs and is performed as gamma1＝y(t-1)，γ2＝x(t-1)，γ3＝y(t-2)，γ4… define potential input variables and add variable units in sequence according to the steps in the algorithm and calculate the corresponding L ipschitz coefficients fig. 6(a) and 6(b) show the L ipschitz coefficient difference ratio obtained using the original order estimation algorithm and the improved algorithm respectively
Figure GDA00024826242100000912
From the graph, Δ l in the original algorithm can be easily found(3+3)≤0.1l(3+3)That is, the algorithm stopping condition is satisfied at the order (3, 3), the early stopping algorithm jumps out, and the final order estimation value of (3, 3) is obviously inconsistent with the actual order. Analysis ofThe order estimation process knows that at order (3, 3) due to the actual order NbEqual to the actual value, 3, thus satisfying the abort condition, and then another order N a3 is still much smaller than the actual order 6. When the order estimation is performed by using the improved algorithm, the process firstly meets the stopping condition for the first time at the order (3, 3) and selects NaAs the final estimate, 3, after which the algorithm proceeds and increments N stepwisebOrder value until the stopping condition is satisfied again at order (6, 3), ending the algorithm, and finally selecting N a3 and N b3 is taken as the optimal estimated value of the order and is consistent with the actual order value, and the algorithm result is correct. The simulation results show that the improved algorithm is in NaAnd NbThe delay order of the nonlinear system can be accurately and effectively estimated under the condition of larger difference with the actual order. Therefore, the order estimation algorithm can be used for evaluating the system characteristic change of the linear module of the winding vibration system, so that the order estimation algorithm can be further applied to the evaluation and diagnosis of the mechanical structure state of the transformer winding.
In a winding fault detection experiment, input and output data with the time length of 0.5s are selected for delay order estimation, and the delay orders of a relation model of vibration and current of all measuring points under the states of a normal winding and an abnormal winding are estimated to obtain the results shown in the table 1:
TABLE 1 estimation of the order of all the measurement points in the different states of the winding (100% load)
Figure GDA0002482624210000101
As can be seen from table 1, when the winding mechanical structure is abnormal (deformed), the delay orders of the current-vibration relationship model corresponding to most of the measuring points will be significantly changed, so that the state of the winding mechanical structure can be monitored by monitoring the change of the delay orders of the model.
The embodiments described above are presented to enable a person having ordinary skill in the art to make and use the invention. It will be readily apparent to those skilled in the art that various modifications to the above-described embodiments may be made, and the generic principles defined herein may be applied to other embodiments without the use of inventive faculty. Therefore, the present invention is not limited to the above embodiments, and those skilled in the art should make improvements and modifications to the present invention based on the disclosure of the present invention within the protection scope of the present invention.

Claims (4)

1. A power transformer winding state evaluation method based on improved system delay order estimation comprises the following steps:
(1) the method comprises the following steps that a plurality of vibration sensors are dispersedly arranged on the surface of an oil tank of the power transformer corresponding to winding positions, vibration signals of the vibration sensors of the power transformer under a low-voltage load operation condition are recorded, and primary side currents of the power transformer are synchronously collected;
(2) preprocessing the vibration signal and the current signal, wherein the preprocessing comprises the demagnetization processing of the current signal;
(3) for any vibration sensor, a relation model of current and winding vibration is constructed, and the delay order N of the model is estimated according to a vibration signal preprocessed by the vibration sensor and a current signal and is used as a characteristic quantity of system characteristics, wherein N is Na+Nb，NaAnd NbActual linear delay order, N, of system output and input, respectivelyaAnd NbAre all natural numbers; further according to N, NaAnd NbJudging the mechanical structure state of the power transformer winding under the condition of the data of the vibration sensor;
the specific process for estimating the model delay order n is as follows:
3.1 calculate the L ipschitz coefficient for all sample point combinations in the relationship model for a delay order of n according to the following formula:
Figure FDA0002387071860000011
wherein: lij (n)L representing combination of ith and jth sampling points in relation model with delay order nThe ipschitz coefficient, i and j are sampling point serial numbers; when n is an even number, γ1(i)～γn(i) The correspondence is equal to y (i-1), x (i-1), y (i-2), x (i-2)a),x(i-Nb)，γ1(j)～γn(j) Corresponds to y (j-1), x (j-1), y (j-2), x (j-2)a),x(j-Nb) (ii) a When n is an odd number, γ1(i)～γn(i) The correspondence is equal to y (i-1), x (i-1), y (i-2), x (i-2)b),x(i-Nb),y(i-Na)，γ1(j)～γn(j) Corresponds to y (j-1), x (j-1), y (j-2), x (j-2)b),x(j-Nb),y(j-Na) (ii) a y (i) and y (j) are respectively the signal values of the ith sampling point and the jth sampling point in the vibration signal preprocessed by the vibration sensor, y (i-1) and y (j-1) are respectively the signal values of the ith-1 sampling point and the jth-1 sampling point in the vibration signal preprocessed by the vibration sensor, y (i-2) and y (j-2) are respectively the signal values of the ith-2 sampling point and the jth-2 sampling point in the vibration signal preprocessed by the vibration sensor, and y (i-N)a) And y (j-N)a) Respectively corresponding to the ith-N in the vibration signal preprocessed by the vibration sensoraSample points and j-NaSignal values of the sampling points, y (i-N)b) And y (j-N)b) Respectively corresponding to the ith-N in the vibration signal preprocessed by the vibration sensorbSample points and j-NbThe signal values of the sampling points, x (i-1) and x (j-1) are respectively the signal values of the processed current signal corresponding to the i-1 th sampling point and the j-1 th sampling point, x (i-2) and x (j-2) are respectively the signal values of the processed current signal corresponding to the i-2 th sampling point and the j-2 th sampling point, and x (i-N)b) And x (j-N)b) Respectively corresponding to the ith-N in the processed current signalbSample points and j-NbSignal values of the sampling points;
3.2, sorting all L ipsitz coefficients obtained by calculation in the step 3.1 from large to small, cutting off the first m L ipsitz coefficients, and further calculating a L ipsitz average coefficient of the relation model under the condition that the delay order is n according to the following formula:
Figure FDA0002387071860000021
wherein: l(n)L ipsitz average coefficient, l, representing a relational model for a delay order of n(n)(z) is the z-th L ipschitz coefficient after ordering from big to small, and m is usually 0.01Nset，NsetThe total number of sampling points is;
3.3 initialization delay order N2 and Na＝Nb＝1；
3.4 converting NaAdding 1 in total, and then calculating N-N according to the steps 3.1-3.2a+NbAnd N ═ Na-1+NbCorresponding average coefficient l of L ipsitz(Na+Nb)And l(Na-1+Nb)And the following judgments were made:
if l(Na+Nb)-l(Na-1+Nb)≤×l(Na-1+Nb)And N isbIf not, determining NaThe value before the accumulation is carried out at the time and the step 3.5 is carried out;
if l(Na+Nb)-l(Na-1+Nb)≤×l(Na-1+Nb)And N isbIf determined, then determine NaFor the value before the present accumulation and making sure NaAnd NbAdding to obtain a delay order n;
if l(Na+Nb)-l(Na-1+Nb)＞×l(Na-1+Nb)And N isbIf not, executing step 3.5;
if l(Na+Nb)-l(Na-1+Nb)＞×l(Na-1+Nb)And N isbIf it is determined, step 3.4 is repeated until NaDetermining and making sure NaAnd NbAdding to obtain a delay order n;
3.5 converting NbAdding 1 in total, and then calculating N-N according to the steps 3.1-3.2a+NbAnd N ═ Na+Nb-L ipsitz mean coefficient l corresponding to case 1(Na+Nb)And l(Na+Nb-1)And the following judgments were made:
if l(Na+Nb)-l(Na+Nb-1)≤×l(Na+Nb-1)And N isaIf not, determining NbThe value before the accumulation is the value before the accumulation and the step 3.4 is executed again;
if l(Na+Nb)-l(Na+Nb-1)≤×l(Na+Nb-1)And N isaIf determined, then determine NbFor the value before the present accumulation and making sure NaAnd NbAdding to obtain a delay order n;
if l(Na+Nb)-l(Na+Nb-1)＞×l(Na+Nb-1)And N isaIf not, returning to execute the step 3.4;
if l(Na+Nb)-l(Na+Nb-1)＞×l(Na+Nb-1)And N isaIf it is determined, step 3.5 is repeated until NbDetermining and making sure NaAnd NbAdding to obtain a delay order n;
wherein: is a convergence factor;
(4) traversing all the vibration sensors according to the step (3), and finally judging that the winding of the power transformer is normal when the mechanical structure state of the winding is judged to be normal based on the vibration sensors in a certain proportion, or judging that the winding of the power transformer is abnormal.
2. A power transformer winding state evaluation method according to claim 1, characterized by: the specific process of preprocessing the vibration signal and the current signal in the step (2) is as follows: firstly, carrying out normalization processing on a vibration signal and a current signal; the normalized current signal is then non-linearly transformed according to the following equation:
Figure FDA0002387071860000031
wherein: i (t) is the normalized current signal, t is the time, ipAnd (t) is a current signal after nonlinear transformation.
3. A power transformer winding state evaluation method according to claim 1, characterized by: when N is presentaAnd NbWhen none of them is determined, the convergence factor is equal to1(ii) a When N is presentaAnd NbWhen one of them is determined, the convergence factor is equal to2(ii) a Wherein the content of the first and second substances,1and2respectively, two different constants are set.
4. A power transformer winding state evaluation method according to claim 1, characterized by: the specific process of judging the mechanical structure state of the power transformer winding in the step (3) is as follows: firstly, calculating the delay orders N and N of the model according to the processes of the steps (1) to (3) under the condition of ensuring that the power transformer winding is normalaAnd Nb(ii) a Then, N, N obtained by actual measurement and calculation based on the current vibration sensor data in the step (3)aAnd NbN and N under normal condition of power transformer windingaAnd NbComparing, and if the two data are matched, judging that the mechanical structure state of the power transformer winding is normal under the condition of the current vibration sensor data; and if the two data are not matched, judging that the mechanical structure state of the power transformer winding is abnormal under the condition of the current vibration sensor data.
CN201810550863.4A 2018-05-31 2018-05-31 Power transformer winding state evaluation method based on system delay order estimation Active CN108872777B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810550863.4A CN108872777B (en) 2018-05-31 2018-05-31 Power transformer winding state evaluation method based on system delay order estimation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810550863.4A CN108872777B (en) 2018-05-31 2018-05-31 Power transformer winding state evaluation method based on system delay order estimation

Publications (2)

Publication Number Publication Date
CN108872777A CN108872777A (en) 2018-11-23
CN108872777B true CN108872777B (en) 2020-07-17

Family

ID=64335888

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810550863.4A Active CN108872777B (en) 2018-05-31 2018-05-31 Power transformer winding state evaluation method based on system delay order estimation

Country Status (1)

Country Link
CN (1) CN108872777B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109991508B (en) * 2019-04-15 2021-01-12 中国计量大学 Transformer winding state diagnosis method based on dynamic nonlinear characteristic sequence
CN110162743A (en) * 2019-05-08 2019-08-23 孙力勇 A kind of data administering method based on k neighborhood nonlinear state Eq algorithm

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102778633A (en) * 2012-08-07 2012-11-14 绍兴电力局 Method for selecting vibration sensitive area of power transformer tank surface winding
CN104034289A (en) * 2014-07-04 2014-09-10 国家电网公司 Condition monitoring method and device for windings of power transformer
CN105205285A (en) * 2015-10-20 2015-12-30 河海大学 BSRM stator mode finite element modeling method considering influence of windings
CN107239596A (en) * 2017-04-27 2017-10-10 株洲中车时代电气股份有限公司 A kind of rail traffic vehicles transformer limit element artificial module construction method
CN107741271A (en) * 2017-09-22 2018-02-27 浙江大学 A kind of Winding in Power Transformer state evaluating method based on system delay Order- reduction

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6620157B1 (en) * 2000-12-28 2003-09-16 Senorx, Inc. High frequency power source
JP5866680B2 (en) * 2011-08-23 2016-02-17 国立大学法人九州工業大学 Partial discharge position locating method and partial discharge position locating apparatus
US8266085B1 (en) * 2012-02-06 2012-09-11 The United States Of America As Represented By The Secretary Of The Army Apparatus and method for using analog circuits to embody non-lipschitz mathematics and properties using attractor and repulsion modes
CN106128046A (en) * 2014-07-16 2016-11-16 曹新民 Monitoring Power Transformer state, the method for work of the gradable warning system do not reported to the police
CN106443229A (en) * 2016-08-04 2017-02-22 内蒙古电力(集团)有限责任公司内蒙古电力科学研究院分公司 Transformer winding deformation diagnosis method
CN106780126B (en) * 2016-12-09 2020-06-16 西安交通大学 Simplified solving method for transient response of direct current system of direct current feed-in power grid

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102778633A (en) * 2012-08-07 2012-11-14 绍兴电力局 Method for selecting vibration sensitive area of power transformer tank surface winding
CN104034289A (en) * 2014-07-04 2014-09-10 国家电网公司 Condition monitoring method and device for windings of power transformer
CN105205285A (en) * 2015-10-20 2015-12-30 河海大学 BSRM stator mode finite element modeling method considering influence of windings
CN107239596A (en) * 2017-04-27 2017-10-10 株洲中车时代电气股份有限公司 A kind of rail traffic vehicles transformer limit element artificial module construction method
CN107741271A (en) * 2017-09-22 2018-02-27 浙江大学 A kind of Winding in Power Transformer state evaluating method based on system delay Order- reduction

Also Published As

Publication number Publication date
CN108872777A (en) 2018-11-23

Similar Documents

Publication Publication Date Title
CN111487046B (en) Fault diagnosis method for circuit breaker voiceprint and vibration entropy feature fusion
CN104090214B (en) A kind of Cable fault examination and aging analysis method
US7127364B2 (en) Method of compensating for distorted secondary current of current transformer
CN109001594B (en) Fault traveling wave positioning method
CN110794271A (en) Power cable intermediate joint damp positioning diagnosis method based on input impedance spectrum
WO2013185405A1 (en) Power transformer iron core looseness fault diagnosis and preliminary fault location system and method
CN108872777B (en) Power transformer winding state evaluation method based on system delay order estimation
US20170030958A1 (en) Transformer parameter estimation using terminal measurements
CN102998545A (en) Online transformer winding work state monitoring method
US11243267B2 (en) Ground-wall insulation aging monitoring and locating method for converter transformer
CN110728257B (en) Transformer winding fault monitoring method based on vibration gray level image
CN110174255B (en) Transformer vibration signal separation method based on radial basis function neural network
CN103439631A (en) Method and system for detecting corrosion state of grounding grid
CN114325361A (en) Transformer tap switch state online monitoring method based on vibration principle
Zhang et al. Research on estimating method for the smart electric energy meter’s error based on parameter degradation model
CN106526359A (en) Prony algorithm and ill-conditioned data analysis-based power grid low-frequency oscillation on-line detection algorithm
CN111273100B (en) Power transformer winding state evaluation method based on vibration phase
Akçay et al. Identification of power transformer models from frequency response data: A case study
CN107741271A (en) A kind of Winding in Power Transformer state evaluating method based on system delay Order- reduction
CN110208657B (en) Novel intelligent T-connection line fault identification method
CN116540016A (en) Cable defect diagnosis method based on reflection coefficient discrete sequence
Hashim et al. Single-ended fault location for transmission lines using traveling wave and multilayer perceptron network
CN112881839B (en) Transformer diagnosis method based on mutual information of frequency concentration and vibration stability
CN107576259A (en) Deformation of transformer winding online test method based on very fast transient overvoltage characteristic
CN112114215A (en) Transformer aging evaluation method and system based on error back propagation algorithm

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant