Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to solve the technical problems that the existing high-dimensional random matrix error state evaluation method needs stable output, has low evaluation index reliability and cannot judge the error polarity.
In order to achieve the purpose, the evaluation parameter basically having the stable distribution characteristic is constructed by using a differential method, then, the 'outlier' distribution of the characteristic value by using the single-loop law is avoided by combining the M-P law, the evaluation index sensitivity is higher, and finally, the error polarity of the voltage transformer is judged under the condition of error abnormity based on the correlation of the same-phase voltage.
In a first aspect, the present invention provides an error evaluation method for a non-stationary output voltage transformer, including the following steps:
s1, representing the evaluation parameters of the error state of the voltage transformer by adopting a difference method under the non-steady output condition;
s2, evaluating the evaluation parameters of the error state of the voltage transformer in real time by combining the high-dimensional random matrix and the M-P law, and evaluating the error state of the voltage transformer; the error state comprises an error normal state and an error abnormal state;
and S3, if the voltage transformer is in an error abnormal state, judging the error polarity of the voltage transformer in the error abnormal state based on the correlation between the in-phase voltage signals.
Optionally, the step S1 includes:
and aiming at the unstable condition of the voltage signal on the primary side of the voltage transformer, performing first-order differential processing on the output voltage signal of the voltage transformer, and taking the obtained first-order differential variable quantity as an evaluation parameter of the error state of the voltage transformer.
Optionally, the step S2 includes:
measuring the extracted evaluation parameters of the error states of the N groups of voltage transformers for T times in the intercepted evaluation time window, and constructing an N × T original matrix according to the obtained N groups of T measurement values;
under the condition of sparse state parameters, based on a Kalman filtering algorithm, expanding the original matrix to obtain an expanded matrix;
standardizing the extended matrix to obtain a high-dimensional random matrix, wherein the mean value of elements of the high-dimensional random matrix is 0;
calculating a covariance matrix of the high-dimensional random matrix, solving an eigenvalue of the covariance matrix, and determining a probability density distribution function of the eigenvalue based on statistical analysis;
determining the difference degree of the two functions based on the M-P law probability density distribution function and the probability density distribution function of the characteristic value, and evaluating the error state of the voltage transformer according to the difference degree; and when the difference degree exceeds a threshold value, the voltage transformer is in an error abnormal state, otherwise, the voltage transformer is in an error normal state.
Optionally, the step S3 includes:
l groups of voltage transformers (L is more than or equal to 2) are arranged at the same measuring point, and if the evaluation time period is t1~t2The sliding step length of the sliding time window is TwAnd calculating:
in the formula, sumUijkRepresenting the judgment index, U, of the k phase error polarity of the ith voltage transformer to be evaluated relative to other voltage transformersiktRepresents the k phase voltage amplitude value, U, of the ith voltage transformer at the t momentjktThe voltage amplitude of a k phase at the t moment of a jth voltage transformer is shown, and k represents A, B, C three phases;
according to the physical connection of different voltage transformers at the same measuring point and the probability statistical analysis of the same-phase voltage, if the voltage transformers are normal, the error change forms of the different voltage transformers are basically consistent, and sumUijkAll show positive with the change of t' in time sequenceA trend of negative random fluctuations; if the voltage transformer is at t1~t2When the distribution of the positive errors is abnormal in time, sumU is relative to other voltage transformersijkThe time sequence changes along with t' show an increasing trend; if the voltage transformer is at t1~t2When negative error distribution is abnormal within time, sumU is compared with other voltage transformersijkThe time series change along with t' shows a decreasing trend.
In a second aspect, the present invention provides an error evaluation system for a non-stationary output voltage transformer, comprising:
the evaluation parameter determining unit is used for representing the evaluation parameters of the error state of the voltage transformer by adopting a difference method under the non-steady output condition;
the error state evaluation unit is used for evaluating the evaluation parameters of the error state of the voltage transformer in real time by combining the high-dimensional random matrix and the M-P law and evaluating the error state of the voltage transformer; the error state comprises an error normal state and an error abnormal state;
and the error polarity judging unit is used for judging the error polarity of the voltage transformer in the error abnormal state based on the correlation between the in-phase voltage signals if the voltage transformer is in the error abnormal state.
Optionally, the evaluation parameter determining unit performs first-order difference processing on the output voltage signal of the voltage transformer aiming at the unstable condition of the primary side voltage signal of the voltage transformer, and uses the obtained first-order difference variation as the evaluation parameter of the error state of the voltage transformer.
Optionally, the error state evaluation unit measures evaluation parameters of the extracted N groups of voltage transformer error states for T times in an intercepted evaluation time window, constructs an N × T original matrix according to the obtained N groups of T measurement values, expands the original matrix to obtain an expanded matrix based on a Kalman filtering algorithm under a sparse state parameter condition, standardizes the expanded matrix to obtain a high-dimensional random matrix, the mean value of elements of the high-dimensional random matrix is 0, calculates a covariance matrix of the high-dimensional random matrix, obtains a characteristic value of the covariance matrix, determines a probability density distribution function of the characteristic value based on statistical analysis, determines the difference degree of two functions based on an M-P law probability density distribution function and the probability density distribution function of the characteristic value, evaluates the error state of the voltage transformer according to the difference degree, and when the difference degree exceeds a threshold value, the voltage transformer is in an error abnormal state, otherwise, the voltage transformer is in an error normal state.
Optionally, the error polarity determination unit is arranged at the same measurement point, and has L groups of voltage transformers (L is more than or equal to 2), if the evaluation time period is t1~t2The sliding step length of the sliding time window is TwAnd calculating:
in the formula, sumUijkRepresenting the judgment index, U, of the k phase error polarity of the ith voltage transformer to be evaluated relative to other voltage transformersiktRepresents the k phase voltage amplitude value, U, of the ith voltage transformer at the t momentjktThe voltage amplitude of a k phase at the t moment of a jth voltage transformer is shown, and k represents A, B, C three phases; according to the physical connection of different voltage transformers at the same measuring point and the probability statistical analysis of the same-phase voltage, if the voltage transformers are normal, the error change forms of the different voltage transformers are basically consistent, and sumUijkThe change of t' in time sequence shows the trend of positive and negative random fluctuation; if the voltage transformer is at t1~t2When the distribution of the positive errors is abnormal in time, sumU is relative to other voltage transformersijkThe time sequence changes along with t' show an increasing trend; if the voltage transformer is at t1~t2When negative error distribution is abnormal within time, sumU is compared with other voltage transformersijkThe time series change along with t' shows a decreasing trend.
Generally, compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:
1. the invention provides the error evaluation method and the error evaluation system of the voltage transformer with non-stable output, which can realize the real-time evaluation of the error state and the error polarity only according to the non-stable output data of the voltage transformer, greatly reduce the evaluation cost and be beneficial to improving the operation and maintenance level of the voltage transformer.
2. The invention provides an error evaluation method and system for a non-stable output voltage transformer.
3. The invention provides an error evaluation method and system for a non-stable output voltage transformer, which can roughly obtain the error polarity of the voltage transformer under the condition of abnormal error according to the correlation of in-phase voltage, and are beneficial to the development of subsequent operation and maintenance.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The invention discloses an error evaluation method and system of a voltage transformer with non-steady output, aiming at evaluating the error state and error polarity of the voltage transformer by adopting a difference and high-dimensional random matrix and considering the correlation of the same-phase voltage only according to the output of the voltage transformer under the condition of non-steady output. The specific process comprises the following steps: according to the analysis of the voltage signal characteristics, under the non-stable output condition, an evaluation parameter capable of representing the error state of the voltage transformer is constructed by using a difference method; constructing a random matrix in real time based on the sliding time window; expanding a random matrix based on a Kalman filter; standardizing the expansion matrix; calculating the characteristic value of the covariance matrix; solving the M-P law probability density difference degree, and evaluating the error state of the voltage transformer; and judging the positive and negative error conditions. The method and the device realize effective evaluation on the error state and the error polarity of the voltage transformer on the premise of not depending on standard appliances, and the evaluation method has universality, effectiveness and instantaneity.
The invention provides an error evaluation method and system of a non-stable output voltage transformer, which comprises the following steps:
step 1: analyzing the characteristics of the voltage signals, and constructing an evaluation parameter capable of representing the error state of the voltage transformer;
for the output signal U of the voltage transformer, the abnormal fluctuation thereof can be considered to be caused by two aspects, the primary side voltage fluctuation and the transformer self error, as shown in the following formula:
the voltage of the primary side is represented by k, and the fluctuation coefficient of the voltage is represented by k, and the transformer output abnormality caused by the transformer error is represented by the load change, the adjustment of a transformer tap switch and the like.
In the actual operation process of a power grid, the random fluctuation of primary voltage is far greater than the requirement (generally 0.2 level) of the measurement accuracy of a voltage transformer, when a primary side voltage signal is a stable sequence, a primary side voltage fluctuation coefficient k is close to the transformation ratio of the transformer, and abnormity caused by transformer errors is easy to identify from an output sequence.
For the case that the primary side voltage signal is not stable, when the first order difference processing is performed on the actual voltage signal, the following formula is shown:
in the formula, t represents a previous time, t +1 represents a next time, the variation of the actual voltage signal after the first-order difference processing substantially meets the characteristic of stable distribution, and the influence of the measurement error occupies a large proportion, and at this time, the first-order difference variation of the voltage signal can be used as the state evaluation parameter.
Firstly, the stability of the voltage signal output by the voltage transformer is checked.
In view of wide application and simple calculation of the run length inspection, firstly, the stability of the run length inspection parameters is inspected by using the following steps:
suppose the output voltage signal sequence is { x }tThe sequence length is M, M1And M2Representing the number of numbers in the sequence that are greater than the mean and the number of numbers in the sequence that are less than the mean. M1And M2All do not exceed 15 (small samples), the run total obeys r distribution:
when M is1,M2Greater than 15 (large samples), statisticsQuantity:
the statistic Z progressively obeys to an N (0,1) distribution.
Thus, statistics r and Z can be calculated for the sequence tested, given a significance level of α if rL<r<rUOr | Z | < Zα(ZαCritical), the sequence is considered stationary, otherwise the sequence is non-stationary.
When the output voltage signal sequence is a non-stationary sequence, first-order differential processing needs to be carried out on the output voltage signal, an evaluation parameter capable of representing the error state of the voltage transformer is constructed, and when the output voltage signal is a stationary sequence, first-order differential processing does not need to be carried out.
Step 2: constructing an original matrix in real time by utilizing the evaluation parameters based on the sliding time window;
within the intercepted evaluation time window, measuring the extracted N groups of state evaluation parameters for T times, and constructing an original matrix with the matrix size of N × T by using the T measured values of the N groups of evaluation parameters1:
Wherein x isijWhich represents the value of the i-th evaluation parameter at time j.
And step 3: under the condition of sparse state parameters, based on a Kalman filtering algorithm, matrix expansion is carried out;
when the evaluation parameter types are few, the purpose of engineering application cannot be achieved, and in order to solve the problem, the original matrix is expanded on the basis of a Kalman filtering algorithm under the condition of sparse state parameters. The method specifically comprises the following steps:
evaluating parameter measurement value estimation:
wherein Z (k) is the state of the system at time k; x (k) is the measured value of the evaluation parameter at time k; w (k), process noise following a gaussian distribution with covariance q (k); v (k) is the observed noise, subject to a Gaussian distribution, with covariance R (k).
Obtaining a state estimator Z (k) by a state estimation parameter measured value X (k) through a Kalman filtering algorithm, superposing Gaussian white noise equivalent to the measured noise level as a matrix expansion line, namely:
Xi'=Z+ρR1(i=N+1,N+2,…,N') (8)
in the formula, R1In order to construct the extended matrix D, the white Gaussian noise obeying the standard normal distribution is used, p is used for adjusting the signal-to-noise ratio of the extended matrix, N 'is the size of the extended matrix, preferably, p can be selected as the size of the standard deviation of the measured value of the evaluation parameter, and N' can be selected as 2002:
In the formula, N' is an expansion matrix D2The number of rows of (c).
And 4, step 4: standardizing the extended matrix to make the average value of elements of the high-dimensional random matrix 0;
for matrix D2The following normalization operation is performed to become a matrix Dstd:
In the formula, yijIs a matrix DstdElement of (2), xijIs a matrix D2Of (2) is used.
And 5: calculating a covariance matrix as a state evaluation matrix, and solving the characteristic value probability density distribution of the matrix;
first, a normalized matrix D is calculatedstdCovariance matrix of (2):
subsequently, the eigenvalue λ of the covariance matrix S is determinedi(i=1,2,…,N')。
And finally, based on statistical analysis, solving the probability density distribution characteristics of the characteristic values.
Step 6: and extracting the evaluation index of the M-P law probability density difference degree and evaluating the error state of the voltage transformer.
Assuming a probability density distribution function of the feature values as f1(x),x∈[c1,d1]The standard M-P law probability density distribution function is f2(x),x∈[c2,d2]Defining the difference degree v of the two functions as:
wherein c is min (c)1,c2),d=min(d1,d2) And n is the order.
Preferably, the degree of difference when n is 1 may be selected as the state evaluation index. According to the M-P law, when the error state of the voltage transformer is in a normal range, the probability density distribution of the characteristic values approaches to a standard M-P law probability density distribution function, and the difference degree is smaller; when the error state of the voltage transformer exceeds the allowable range, the more obvious the difference degree between the probability density distribution of the characteristic values and the standard M-P law probability density distribution function is, the larger the difference degree is. Therefore, the abnormal state of the voltage transformer can be identified according to the evaluation index.
And 7: judging error polarity, namely judging the error polarity of the voltage transformer under the condition of error abnormity based on the correlation between the in-phase voltage signals;
more than two groups of transformers are generally arranged at the same measuring point to measure the same voltage signal, and if a certain measuring point is assumed to have L groups of voltage transformers (L is more than or equal to 2), if the evaluation time period is t1~t2The sliding step length of the sliding time window is TwAnd calculating:
in the formula of UiktRepresenting a certain phase voltage amplitude, U, of the voltage transformer to be evaluatedjktRepresenting other voltage transformer voltage magnitudes, k representing A, B, C three phases.
According to the physical connection of different voltage transformers at the same measuring point and the probability statistical analysis of the same-phase voltage, if the transformers are normal, the error change forms of the different transformers are basically consistent, and sumUijkThe change of t' in time sequence shows the trend of positive and negative random fluctuation; if the mutual inductor is at t1~t2When the positive error distribution is abnormal in the time, sumU is relative to other transformersijkThe time sequence changes along with t' show an increasing trend; if the mutual inductor is at t1~t2When negative error distribution is abnormal within time, sumU is compared with other transformersijkThe time series change along with t' shows a decreasing trend. According to t1~t2Within a time period, sumUijkAnd evaluating the error polarity of the voltage transformer under the condition of error abnormity in the time in the variation condition of the time sequence.
For further understanding of the present invention, the following brief description of the related principles of the present invention is provided:
1) basic principle of error state estimation
The voltage transformer collects primary voltage information of a power grid, characteristic parameters such as phases, amplitudes, frequencies and three-phase unbalance degrees can be obtained through calculation, the parameters possibly have stable distribution characteristics within a certain time in the normal operation process of the voltage transformer, and whether the parameters accord with known rules can be judged by utilizing a high-dimensional random matrix theory, so that whether the error state of the voltage transformer has abnormal change or not is judged, and accordingly, the error state is evaluated. When the parameters do not have stable distribution characteristics, evaluation parameters capable of representing the error state can be constructed by analyzing the characteristics of parameter signals and utilizing a difference method, the parameters basically have stable distribution characteristics, and then the error state evaluation is carried out by combining a high-dimensional random matrix theory.
2) M-P law
Let matrix H ═ xijIs a random matrix of m × n, where xijIs a mean of 0 and a variance of σ2When c is m/n ∈ (0,1)]And when M → ∞, n → ∞ time, the empirical spectrum of the sample covariance matrix converges with a probability 1, the convergence trend conforms to the M-P law, and the probability density function is:
wherein the content of the first and second substances,
when sigma is
2When 1, the rule of convergence of the empirical spectral distribution of the covariance matrix is the standard M-P law.
In one specific embodiment, as shown in fig. 1, the present invention evaluates the error state of a voltage transformer according to the following steps:
1) a110 KV transformer substation in Jiangsu is adopted to acquire A-phase voltage amplitude data from two groups of electronic voltage transformers (0.2 level), the sampling frequency is 4K, the sampling period is 1s, and the sampling time is 7 days. Fig. 2 and fig. 3 respectively show the ratio difference time sequence distribution of two groups of electronic voltage transformers, where a negative error out-of-tolerance abnormality exists in the first group of transformers, and the error of the second group of transformers is within an allowable range.
2) Analyzing the characteristics of the voltage signals, and constructing an evaluation parameter capable of representing the error state of the voltage transformer;
and carrying out stationarity test on the obtained fundamental wave voltage amplitudes of the first group of electronic voltage transformers, finding that the fundamental wave voltage amplitudes do not have stable distribution characteristics, carrying out differential calculation, and taking the first-order differential variation as a state evaluation parameter.
3) Based on the sliding time window, the evaluation time window is taken as 10 minutes, the sliding step length is taken as 5 minutes, and the evaluation parameters are utilized to construct the original matrix D in real time1;
4) Based on the Kalman filtering algorithm, the method comprises the following steps of,will matrix D1After expansion, a high-dimensional random matrix D is formed2. The original matrix and extended matrix sizes are shown in table 1.
TABLE 1 high-dimensional random matrix parameters
5) Using equation (10) to matrix D2Carrying out standardization operation to obtain a matrix Dstd。
6) And (3) solving a covariance matrix S by using the formulas (11) and (12), and extracting an M-P law probability density difference degree evaluation index v.
7) Fig. 4 is a probability density distribution diagram of the eigenvalues of the evaluation matrix under normal conditions, and fig. 5 is a probability density distribution diagram of the eigenvalues at the time of error anomaly. The curve in the graph is a standard M-P law probability density distribution graph, and the histogram is the characteristic value probability density distribution of actual data. Comparing fig. 4 and fig. 5, it can be known that when the voltage transformer is in a normal state, the eigenvalue probability density distribution of the covariance matrix is close to the standard M-P law probability density distribution diagram; when the voltage transformer is in an error abnormal state, the probability density distribution of the characteristic values is deviated from the standard M-P law probability density distribution.
The evaluation index of the degree of difference was calculated according to the formula (12), and the calculation results are shown in table 2. As can be seen from the table, when the electronic voltage transformer is in an error abnormal state, the evaluation index v is remarkably increased, and the error state of the voltage transformer can be effectively evaluated according to the evaluation index.
TABLE 2 comparison of evaluation indexes
8) Calculating sumU with respect to the second group of electronic voltage transformers according to equation (13)ijk. And (5) repeating the steps (2) to (8), so that the real-time evaluation of the error state and the error polarity of the voltage transformer can be realized. Shown in FIG. 6, sumUijkThe value on the time sequence is a negative value and is in a descending trend, and the mutual inductor is judged to have negative error and out-of-tolerance in the evaluation timeAnd (4) abnormity, which accords with the actual error change condition of the first group of electronic voltage transformers.
Fig. 7 is an architecture diagram of an error evaluation system of a voltage transformer with non-stationary output according to the present invention, as shown in fig. 7, including:
the evaluation parameter determining unit 710 is configured to represent an evaluation parameter of an error state of the voltage transformer by using a difference method under a non-stationary output condition;
the error state evaluation unit 720 is used for evaluating the evaluation parameters of the error state of the voltage transformer in real time by combining the high-dimensional random matrix and the M-P law, and evaluating the error state of the voltage transformer; the error state comprises an error normal state and an error abnormal state;
and an error polarity determination unit 730, configured to determine, if the voltage transformer is in an error abnormal state, an error polarity in the error abnormal state of the voltage transformer based on a correlation between the in-phase voltage signals.
The functions of each unit can be referred to the description in the foregoing method embodiments, and are not described herein again.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.