CN115099280A - A Prony iterative analysis method, computing device, storage medium and power grid system - Google Patents

Abstract

The invention provides a Prony iteration analysis method, a calculation device, a storage medium and a power grid system, which relate to the technical field of oscillation analysis and comprise the following steps: the method comprises the steps of carrying out data sampling on an oscillation signal to be identified according to a set sampling rate, carrying out preprocessing with abnormal data detection and correction on a one-dimensional sample set obtained through sampling, expanding a multi-dimensional sample set, adopting a Prony model of a P order to infer the multi-dimensional sample set according to the multi-dimensional sample set, detecting whether P reaches an expected value or not according to an oscillation parameter set, if not, adopting an iteration method to heighten the Prony model by one order, if so, determining and outputting an oscillation analysis result, overcoming the defect that the accuracy is difficult to promote due to the fact that a traditional Prony oscillation analysis method is limited by the signal-to-noise ratio and the fixed-order accuracy of the sample, avoiding the fixed-order in the Prony oscillation analysis process, simplifying the fixed-order mode, reducing noise interference, promoting the fixed-order accuracy and being beneficial to promoting the accuracy and efficiency of Prony oscillation analysis.

Description

A Prony iterative analysis method, computing device, storage medium and power grid system

technical field

The invention relates to the technical field of oscillation analysis, and in particular, to a Prony iterative analysis method, a computing device, a storage medium and a power grid system.

Background technique

As more and more new energy devices are connected to the power grid system, the damping and inertia of the power grid system are also reduced, which makes it easier to stimulate different oscillation modes, and the oscillation phenomenon becomes more frequent, which seriously threatens the stability and security of the power grid system. The power grid system will generate oscillating signals during the oscillation process. The analysis of the oscillating signals in the power grid system is particularly important for online early warning and power grid transformation.

Compared with the Fourier algorithm and the least squares algorithm, the Prony model fits the signal with a linear combination of a set of exponential functions with arbitrary amplitude, phase, frequency and damping coefficient, with no need for frequency domain response and sample autocorrelation estimation and other advantages, it is more suitable for analyzing oscillating signals.

At present, in the process of Prony oscillation analysis, an extended order matrix is constructed based on multiple data points collected from the oscillation signal to be identified, and the singular value decomposition method is used to determine the effective rank P (that is, the order of the Prony model) for the extended order matrix. , solve the P-dimensional linear equation system based on the extended order matrix, obtain the equation coefficient sequence 1 and the minimum error energy, solve the characteristic equation based on the equation coefficient sequence 1, obtain the root sequence, and combine the root sequence and multiple data points obtained by sampling to solve the van der Monte Equation system, get equation coefficient sequence 2, calculate amplitude sequence and phase sequence based on equation coefficient sequence 2, calculate frequency sequence and damping coefficient sequence based on root sequence, and output amplitude sequence, phase sequence, frequency sequence and damping coefficient sequence as oscillation analysis results .

方程系数序列二可以表示为[b₁，…，b_i，…，b_P]^T，i为从1到P的正整数。Among them, the equation coefficient sequence 1 can be expressed as [a ₁ , ..., a _i , ..., a _P ] ^T , the root sequence can be expressed as [z ₁ , ..., _zi , ..., z _P ] ^T , the approximate point sequence can be expressed as Expressed as

Equation coefficient sequence two can be expressed as [b ₁ , . . . , b _i , . . , b _P ] ^T , where i is a positive integer from 1 to P.

Among them, the amplitude sequence can be expressed as [A ₁ , ..., A _i , ..., A _P ] ^T , the phase sequence can be expressed as [θ ₁ , ..., θ _i , ..., θ _P ] ^T , and the frequency sequence can be expressed as [ f ₁ , ..., f _i , ..., f _P ] ^T , the damping coefficient sequence can be expressed as [α ₁ , ..., α _i , ..., α _P ] ^T .

However, limited by the high signal-to-noise ratio of the sample or/and the low accuracy of the order determination, the existing Prony oscillation analysis method is difficult to improve the accuracy.

SUMMARY OF THE INVENTION

The present invention aims to solve the technical problems in the related art at least to a certain extent. To achieve the above purpose, the present invention provides a Prony iterative analysis method, a computing device, a non-transitory computer-readable storage medium and a power grid system.

A first aspect of the present invention provides a Prony iterative analysis method, which includes:

S1, perform data sampling on the oscillating signal to be identified according to the set sampling rate, and obtain a one-dimensional sample set with a length of N, where N is a positive integer greater than 2;

S2, performing preprocessing with abnormal data detection and correction on the one-dimensional sample set;

S3, expanding a multi-dimensional sample set based on the one-dimensional sample set that has undergone the preprocessing;

S4, according to the multi-dimensional sample set, adopt a P-order Prony model to infer an oscillation parameter set, where P is a positive integer greater than 1;

S5, according to the oscillation parameter set to detect whether P reaches the expected value, if not, then execute S6, if so, execute S7;

S6, after making P=P+1, return to the S4, so that the Prony model is adjusted to one order higher;

S7, determine and output an oscillation analysis result adapted to the oscillation signal to be identified.

A second aspect of the present invention provides a computing device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, when the processor executes the computer program, the implementation is as described in the first aspect The Prony iterative analysis method.

A third aspect of the present invention provides a non-transitory computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, implements the Prony iterative analysis method described in the first aspect.

A fourth aspect of the present invention provides a power grid system, characterized by comprising the computing device described in the second aspect.

The beneficial effects of the above-mentioned Prony iterative analysis method, computing device, non-transitory computer-readable storage medium and power grid system are: considering that data anomalies are likely to occur during the sampling process, the one-dimensional sample set obtained by sampling can be obtained by preprocessing. Correcting abnormal conditions helps to improve the accuracy of samples. The preprocessed one-dimensional sample set expands the sample data, increases the sample capacity, helps to improve the sample signal-to-noise ratio, and expands the obtained multi-dimensional sample set. Based on this, a cyclic mechanism in which Prony oscillation analysis and order iteration promote each other is formed until the order is determined to the desired value, which overcomes the difficulty of improving the accuracy of the traditional Prony oscillation analysis method due to the limitation of the sample signal-to-noise ratio and the accuracy of order determination. There is no need to determine the order in the process of Prony oscillation analysis, which simplifies the order determination method, reduces noise interference, and improves the order determination accuracy, which helps to improve the accuracy and efficiency of Prony oscillation analysis.

Description of drawings

1 is a schematic flowchart of a Prony iterative analysis method according to an embodiment of the present invention;

Fig. 2 is the schematic flow chart of S2 in Fig. 1;

3 is a schematic diagram of a subsequence according to an embodiment of the present invention;

Fig. 4 is the schematic flow chart of S3 in Fig. 1;

5 is a schematic diagram of a sequence of replacing the k+1 th data point according to an embodiment of the present invention;

Fig. 6 is the schematic flow chart of S4 in Fig. 1;

7a is a schematic diagram of waveforms of respectively collecting oscillating signals from multiple bus bars according to an embodiment of the present invention;

FIG. 7b is a schematic diagram of the waveform of the oscillating signal collected from the bus bar 5 according to an embodiment of the present invention;

FIG. 7c is a schematic diagram of the waveform of the oscillating signal collected from the bus bar 13 according to an embodiment of the present invention;

8a is a schematic diagram of an oscillation mode corresponding to P is 3 according to an embodiment of the present invention;

Fig. 8b is a schematic diagram corresponding to the comparison between the oscillation signal to be identified and the fitting oscillation signal corresponding to P is 3 according to an embodiment of the present invention;

FIG. 9a is a schematic diagram corresponding to two oscillation modes where P is 11 according to an embodiment of the present invention;

Fig. 9b is a schematic diagram corresponding to the comparison between the oscillation signal to be identified and the fitting oscillation signal corresponding to P is 11 according to an embodiment of the present invention;

10a is a schematic diagram of three oscillation modes corresponding to P is 15 according to an embodiment of the present invention;

FIG. 10b is a schematic diagram corresponding to the comparison between the to-be-identified oscillation signal and the fitted oscillation signal with P of 15 according to an embodiment of the present invention;

11a is a schematic diagram of three oscillation modes corresponding to P is 17 according to an embodiment of the present invention;

FIG. 11 b is a schematic diagram corresponding to the comparison between the to-be-identified oscillating signal and the fitting oscillating signal when P is 17 according to an embodiment of the present invention.

Detailed ways

Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Where the description refers to the drawings, the same reference numerals in different drawings designate the same or similar elements unless otherwise indicated. It is to be noted that the implementations described in the following exemplary embodiments do not represent all implementations of the present invention. They are merely examples of apparatus and methods consistent with some aspects of the present disclosure as recited in the claims, and the scope of the present invention is not limited thereto. The features of the various embodiments of the present invention may be combined with each other without inconsistency.

In addition, the terms "sequence one" and "sequence two" are only used for descriptive purposes, and should not be construed as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, features defined as "sequence one" and "sequence two" may explicitly or implicitly include at least one of the features. In the description of the present invention, "multiple" or "multi-dimensional" means at least two dimensions, for example, two or three dimensions, etc., unless otherwise expressly and specifically defined.

Referring to FIG. 1, the Prony iterative analysis method according to an embodiment of the present invention includes S1 to S7.

S1, perform data sampling on the oscillating signal to be identified according to a set sampling rate, and obtain a one-dimensional sample set with a length of N, where N is a positive integer greater than 2.

In the embodiment of the present invention, the one-dimensional sample set may be expressed as X=[ _{x 1} , . . . , _{x n} , . Yes, the expression form of the one-dimensional sample set is not limited to the form of row vectors, but can also be in the form of column vectors, so that the collected N data points can be stored in the form of a sequence, and a one-dimensional sample set can be formed.

S2, perform preprocessing with abnormal data detection and correction on the one-dimensional sample set.

S3, a multi-dimensional sample set is expanded based on the pre-processed one-dimensional sample set.

S4, according to the multi-dimensional sample set, adopt the P-order Prony model to infer the oscillation parameter set, where P is a positive integer greater than 1.

S5, according to the oscillation parameter set to detect whether P reaches the expected value, if not, execute S6, if yes, execute S7.

S6, after setting P=P+1, return to the S4, so as to increase the Prony model by one order.

S7, determine and output an oscillation analysis result that is adapted to the oscillation signal to be identified.

Using the above Prony iterative analysis method, considering that data anomalies are prone to occur during the sampling process, the one-dimensional sample set obtained by sampling can be corrected in the event of data anomalies through preprocessing, which is helpful to improve the accuracy of the samples. The one-dimensional sample set, which expands the sample data, increases the sample capacity, and helps to improve the sample signal-to-noise ratio. Based on the multi-dimensional sample set obtained by the expansion, a circular mechanism in which Prony oscillation analysis and order iteration promote each other is formed, until The order is determined to the expected value, which overcomes the defect that the accuracy of the traditional Prony oscillation analysis method is difficult to improve due to the limitation of the sample signal-to-noise ratio and the accuracy of the order determination. It not only reduces noise interference, but also improves the accuracy of order determination, which helps to improve the accuracy and efficiency of Prony oscillation analysis.

Optionally, referring to FIG. 2 , S2 includes S21 to S26.

S21, select the nth data point as the tested data point in the one-dimensional sample set, and let the multiple data points that are arranged on both sides of the tested data point and are continuous with it and form a subsequence are respectively reference data points, wherein , n starts from 2.

S22, according to a plurality of reference data points, detect whether the detected data point is abnormal, if so, execute S23 first and then execute S24, if not, execute S24 directly.

S23, correcting the detected data points.

S24, check whether n reaches N-1, if not, execute S25, if yes, execute S26;

S25, after setting n=n+1, return to S21, so that the subsequence is moved forward by one synchronously following the detected data point.

S26, output the preprocessed one-dimensional sample set for use in S3.

In an example, the second data point is the tested data point, the first data point and the third data point can each be the reference data point; the third data point is the tested data point, the second data point and the fourth data point can each be a reference data point; and so on, the N-1th data point is a tested data point, and the N-2th data point and the Nth data point can each be a reference data point.

In another example, the 2nd data point is the test data point, the 1st data point, the 3rd data point, the 4th data point and the 5th data point may each be the reference data point; the 3rd data point The data point is the tested data point, the first data point, the second data point, the fourth data point and the fifth data point can each be the reference data point, and so on to the N-2th data point; The N-1 data points are the inspected data points, and the N-4th data point, the N-3th data point, the N-2th data point, and the Nth data point may each be a reference data point.

It should be understood that the reference data points located before the tested data point and the reference data points located after the tested data point may be the same in number and remain unchanged, or may be adaptive depending on the location of the tested data point. Change; the subsequence can be regarded as a data window to limit the data range, and the length of the subsequence can depend on the actual situation of the one-dimensional sample set, for example, if N equals 100, the length of the subsequence can be 5, if N equals 150, the subsequence The length of the sequence can be 8.

For a one-dimensional sample set, during the process of traversing the N-1th data point from the second data point, abnormality detection is performed on each data point, and only each data point with abnormal conditions is corrected to maintain the normal condition. Each data point remains unchanged to prevent errors and omissions, taking into account simplicity and accuracy.

Optionally, the multiple reference data points and the abnormal detected data points satisfy the following abnormal data detection model:

where x' _n represents a new data point suitable for replacing the tested data point x _n , from x _ni to x _n-1 each represents a reference data point located before the tested data point, from x _n+1 to x _{L +ni-1} each represents a reference data point located after the tested data point, C ₀ represents a preset threshold greater than zero, L represents the length of the subsequence, L is a positive integer in [3, N-1], and i is A positive integer in [1, L-2].

Exemplarily, referring to FIG. 3, the nth data point in which an abnormality occurs is the inspected data point, and the respective normal n-2th data points, n-1th data points, n+1th data points, and n+1th data points are normal. The n+1 data points are all reference data points, and the four reference data points are monotonically increasing. For example, the five data points can be 1.2, 3.6, 10, 6.8, and 9 in sequence, C ₀ can be 3, and the new The data point x′ _n of , is 5.2, and the examples of the present invention are intended to help understand the processing process of the sample set, and should not constitute a limitation on the actual situation of the sample set.

In the abnormal data detection model, it not only restricts the monotonically increasing or monotonically decreasing multiple reference data points, but also restricts the abnormal detected data points from losing monotonicity, and the multiple reference data points do not satisfy the normal detected data points. Anomaly data detection model, taking into account simplicity and accuracy.

Optionally, S23 includes: between the two reference data points x _n-1 and x _n+1 , after deleting the tested data point x _n , interpolating a new data point x' _n to prevent disorder, Simplicity and reliability.

Optionally, referring to FIG. 4 , S3 includes S31 to S35.

S31 , constructing and initializing a two-dimensional array, and setting the preprocessed one-dimensional sample set as a first-dimensional vector in the two-dimensional array.

S32, the kth 1st dimension vector is copied into the k+1th dimension vector, when the k+1th dimension vector is obtained by performing linear interpolation on the kth data point and the k+2th data point. The new data point replaces the k+1th data point, where k starts at 1.

Exemplarily, referring to FIG. 5 , the two solid circles represent the kth data point and the k+2th data point respectively, and the arithmetic mean value of the two data points is calculated as a new data point, and the one represented by the cross is deleted. After the k+1th data point, a new data point indicated by a dashed circle is interpolated.

S33, check whether k reaches N-2, if not, execute S34, if yes, execute S35.

S34, after setting k=k+1, return to S32 to reconstruct the next-dimensional vector in the two-dimensional array according to the first-dimensional vector.

S35, output a two-dimensional array with N-1-dimensional vectors as a multi-dimensional sample set for use in S4.

Illustratively, a multidimensional sample set can be represented as follows:

where x _(k,n) represents the nth data point in the kth dimension vector, and k is from 1 to N-1.

Using the preprocessed one-dimensional sample set, with the help of two-dimensional arrays, in addition to the first-dimensional vector, N-2-dimensional vectors are extended, taking into account simplicity and accuracy.

Optionally, referring to FIG. 6 , S4 includes S41 to S47.

S41 , solving the P-dimensional linear equation system based on the k-th dimension vector in the multi-dimensional sample set to obtain the k-th equation coefficient sequence 1, where k is re-iterated from 1, and 2p≤N.

Substituting each data point in the k-th dimensional vector into the P-dimensional system of linear equations can be expressed as follows:

Wherein, the k-th equation coefficient sequence 1 can be expressed as [a _{(k, 1)} , . . . , a _{(k, j)} , . . , a _{(k, P)} ] ^T , where j is a positive integer from 1 to P.

S42, based on the coefficient sequence 1 of the k-th equation, solve the characteristic equation to obtain the k-th root sequence.

Substituting the kth equation coefficient sequence into the characteristic equation, it can be expressed as follows:

Among them, z _{(k, j)} represents the root adapted to the j th coefficient in the k th equation coefficient sequence 1, and the k th root sequence can be expressed as [z _{(k, 1)} , . . . , z _{(k, j)} , ..., z _{(k, P)} ] ^T .

S43, based on the kth root sequence and the kth dimension vector, solve the Vandermonte equation system to obtain the second kth equation coefficient sequence.

Substituting the kth root sequence and the first P data points in the kth dimension vector into the Vandermont equations can be expressed as follows:

Wherein, the k-th equation coefficient sequence two can be expressed as [b _{(k, 1)} , . . . , b _{(k, j)} , . . , b _{(k, P)} ] ^T .

S44, respectively calculating the kth amplitude sequence and the kth phase sequence based on the kth equation coefficient sequence 2; and, respectively calculating the kth frequency sequence and the kth damping coefficient sequence based on the kth root sequence.

Measure the coefficient sequence 2 of the k-th equation by formula (1) to obtain the k-th amplitude sequence, which can be expressed as [A _{(k, 1)} , ..., A _{(k, 2)} , ..., A _(k,P) ] ^T .

Formula (1): A _{(k, j)} = |b _{(k, j)} |.

The k-th equation coefficient sequence 2 is calculated by formula (2), and the k-th phase sequence is obtained. The k-th phase sequence can be expressed as [θ _{(k, 1)} , ..., θ _{(k, j)} , ..., θ _(k,P) ] ^T .

Formula (2): θ _{(k, j)} = arctan[Im(b _{(k, j)} )/Re(b _{(k, j)} )], where arctan represents the arc tangent function, and Im represents the complex imaginary part function , Re means to take the complex real part function.

The kth root sequence is measured by formula (3), and the kth frequency sequence is obtained, and the kth frequency sequence can be expressed as [f _(k,1) ,...,f _(k,j) ,...,f _{( k, P)} ] ^T .

Formula (3): f _{(k, j)} = arctan[Im(z _{(k, j)} )/Re(z _{(k, j)} )].

The k-th root sequence is measured by formula (4), and the k-th damping coefficient sequence is obtained. The k-th damping coefficient sequence can be expressed as [α ( _{k , 1)} , . . . α _{(k, P)} ] ^T .

Formula (4): α _{(k, j)} =In|z _{(k, j)} |/Δt, where In represents the natural logarithmic function, and Δt represents the sampling time interval.

S45, check whether k reaches N-1, if not, execute S46, if yes, execute S47.

S46, after setting k=k+1, the process returns to S41.

S47, combine N-1 amplitude sequences, N-1 phase sequences, N-1 frequency sequences and N-1 damping coefficient sequences to obtain an oscillation parameter set.

Exemplarily, N-1 amplitude sequences can be combined into a two-dimensional array. Similarly, N-1 phase sequences, N-1 frequency sequences, and N-1 damping coefficient sequences can be combined with N-1 amplitude sequences. The combination method is the same and will not be repeated here.

As shown above, combine N-1 amplitude sequences in a two-dimensional array.

Optionally, S5 includes:

Perform averaging and calculation processing on N-1 amplitude sequences, N-1 phase sequences, N-1 frequency sequences, and N-1 damping coefficient sequences, respectively, to obtain amplitude mean sequence, phase mean sequence, frequency mean sequence and damping. mean series;

Based on the amplitude mean sequence, the phase mean sequence, the frequency mean sequence and the damped mean sequence, construct a fitted oscillatory signal;

It is detected whether the fitting degree of the fitted oscillating signal is greater than the preset convergence accuracy, and if so, it is determined that P has not reached the expected value, and if not, it is determined that P has reached the expected value.

表示振幅均值序列，以

表示相位均值序列，以

表示频率均值序列，以

表示阻尼均值序列。exemplarily, with

represents the amplitude mean series, with

represents the phase-averaged sequence, with

represents the frequency mean series, with

represents a damped mean series.

其中，

表示拟合振荡信号，

表示振幅均值序列中的第j个平均振幅，

表示阻尼均值序列中的第j个平均阻尼系数,

表示频率均值序列中的第j个平均频率,

表示相位均值序列中的第j个平均相位，

t表示时间。Illustratively,

in,

represents the fitted oscillatory signal,

represents the j-th mean amplitude in the amplitude mean sequence,

represents the jth average damping coefficient in the damped mean sequence,

represents the jth average frequency in the frequency mean sequence,

represents the j-th averaged phase in the phase-averaged sequence,

t represents time.

Exemplarily, the fitting degree mentioned above is measured and calculated by formula (5), and formula (5) is as follows:

表示拟合振荡信号中的第m个数据点，x_m表示待辨识振荡信号中的对应于第m个数据点

的数据点。Among them, s ² represents the degree of fit,

represents the mth data point in the fitted oscillatory signal, x _m represents the mth data point in the oscillatory signal to be identified corresponding to the mth

data points.

Exemplarily, the preset convergence accuracy can be 0.0001. If the fitting degree is greater than 0.0001, it reflects that P has not yet reached the expected value. If the fitting degree is less than or equal to 0.0001, it reflects that P has reached the expected value. Under the condition of , the amplitude mean sequence, phase mean sequence, frequency mean sequence and damping mean sequence can be combined as oscillation analysis results.

Fitting the signal by averaging the processed four oscillation parameter sequences helps to improve the accuracy of fitting the oscillation signal, and thus helps to improve the accuracy of the identification order.

A computing device according to another embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program program, the above-mentioned Prony iteration is implemented. Analytical method. It can be understood that the aforementioned computing device may be a server or a terminal device such as a desktop computer or a portable computer, wherein the processor may be connected to the memory through a universal serial control bus.

A power grid system according to another embodiment of the present invention includes the computing device mentioned above.

In the embodiment of the present invention, a period of time after a fault occurs in the power grid system, multiple busbars oscillate, and the oscillation modes are mainly divided into 0.08Hz, 0.15Hz and 0.31Hz, wherein the two busbars are busbar 5 and busbar 13 respectively. , FIG. 7a shows a schematic diagram of waveforms of oscillating signals collected from multiple buses, FIG. 7b shows a waveform of oscillating signals collected from bus 5, and FIG. 7c shows a waveform of oscillating signals collected from bus 13.

In the embodiment of the present invention, referring to FIG. 8a and FIG. 8b, when the order P of the Prony model is 3, the 0.31 Hz oscillation mode is analyzed, and the two oscillation modes of 0.15 Hz and 0.08 Hz are not analyzed. There is a big difference between the combined oscillation signal and the signal to be identified.

In the embodiment of the present invention, referring to FIGS. 9a and 9b, when the order P of the Prony model is 11, two oscillation modes of 0.31 Hz and 0.15 Hz are analyzed, and the 0.08 Hz oscillation mode is not analyzed. When P is 3, the fitted oscillating signal is closer to the signal to be identified. However, for the fitted oscillating signal, the 0.15Hz oscillating mode belongs to the decay process, which is inconsistent with the actual situation of the oscillating signal to be identified, so it is necessary to continue to increase the order.

In the embodiment of the present invention, referring to FIGS. 10a and 10b, when the order P of the Prony model is 15, three oscillation modes of 0.31 Hz, 0.15 Hz and 0.08 Hz are analyzed, and the fitted oscillation signal approximates the oscillation to be identified. The degree of the signal is the highest, and 15 is the expected value, representing the optimal order.

In the embodiment of the present invention, referring to FIGS. 11a and 11b , when the order P of the Prony model is 17, compared with P being 15, although the order is increased by two, the fitting effect is not improved. The oscillation modes that do not exist in the signal to be identified are analyzed to generate interference.

It should be noted that any one of Fig. 8b, Fig. 9b, Fig. 10b and Fig. 11b is presented in the upper and lower sub-graphs, wherein the solid line in the upper sub-graph represents the oscillating signal to be identified corresponding to the bus 5, and the lower sub-graph The solid line of represents the oscillating signal to be identified corresponding to the bus bar 13 , and the dashed line represents the corresponding fitting oscillating signal.

Another embodiment of the present invention is a non-transitory computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, implements the Prony iterative analysis method mentioned above.

For the computing device, the non-transitory computer-readable storage medium and the power grid system mentioned above, reference may be made to the specific description of the Prony iterative analysis method and its beneficial effects mentioned above, which will not be repeated here.

In general, computer instructions for implementing the methods of the present invention may be carried in any combination of one or more computer-readable storage media. A non-transitory computer-readable storage medium may include any computer-readable medium except for the temporarily propagated signal itself.

The computer readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples (non-exhaustive list) of computer readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAk), read only memory (ROk), Erasable Programmable Read Only Memory (EPROk or Flash), fiber optics, Portable Compact Disk Read Only Memory (CD-ROk), optical storage devices, magnetic storage devices, or any suitable combination of the above. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

Computer program code for carrying out operations of the present invention may be written in one or more programming languages, including object-oriented programming languages—such as Java, Skalltalk, C++, and conventional Procedural programming language - such as "C" language or similar programming language, especially Python language suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. Where a remote computer is involved, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or wide area network (WAN), or, to an external computer (eg, using an Internet service provider to connect through the Internet) ).

Although the embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and those of ordinary skill in the art can Embodiments are subject to variations, modifications, substitutions and variations.

Claims (10)

1. A Prony iterative analysis method, comprising:

s1, performing data sampling on the oscillation signal to be identified according to a set sampling rate to obtain a one-dimensional sample set with the length of N, wherein N is a positive integer greater than 2;

s2, preprocessing the one-dimensional sample set with abnormal data detection and correction;

s3, expanding a multi-dimensional sample set based on the preprocessed one-dimensional sample set;

s4, according to the multi-dimensional sample set, adopting a Prony model of order P to deduce an oscillation parameter set, wherein P is a positive integer larger than 1;

s5, detecting whether P reaches the expected value according to the oscillation parameter set, if not, executing S6, if yes, executing S7;

s6, returning to S4 after P +1 to raise the Prony model by one step;

and S7, determining and outputting an oscillation analysis result matched with the oscillation signal to be identified.

2. The Prony iterative analysis method of claim 1, wherein said S2 comprises:

s21, selecting the nth data point as the detected data point in the one-dimensional sample set, and making each of the plurality of data points listed on both sides of the detected data point and consecutive thereto in a subsequence as a reference data point, wherein n starts from 2;

s22, detecting whether the detected data point is abnormal or not according to the reference data points, if so, executing S23 and then executing S24, and if not, directly executing S24;

s23, correcting the detected data point;

s24, detecting whether N reaches N-1, if not, executing S25, if yes, executing S26;

s25, after n is equal to n +1, returning to S21 to make the subsequence advance by one bit synchronously with the detected data point;

s26, outputting the preprocessed one-dimensional sample set for use by the S3.

3. The Prony iterative analysis method of claim 2, wherein a plurality of said reference data points and said subject data points where an anomaly occurs satisfy the following anomaly data detection model:

wherein, x' _n Representing data points x suitable for replacing said examined data point _n New data point of from x _n-i To x _n-1 Each representing the reference data point preceding the examined data point, from x _n+1 To x _L+n-i-1 Each representing the reference data point, C, after the examined data point ₀ Represents a preset threshold, L represents the length of the subsequence, and L is [3, N-1 ]]Wherein i is [1, L-2 ]]Is a positive integer of (1).

4. The Prony iterative analysis method of claim 3, wherein the S23 comprises: at two of said reference data points x _n-1 And x _n+1 For the examined data point x _n After deletion, the new data point x 'is interpolated' _n 。

5. The Prony iterative analysis method of claim 1, wherein said S3 comprises:

s31, constructing and initializing a two-dimensional array, wherein the preprocessed one-dimensional sample set is set as a 1 st-dimensional vector in the two-dimensional array;

s32, copying the 1 st-dimensional vector into a (k + 1) th-dimensional vector for the kth time, and replacing the (k + 1) th data point with a new data point obtained by performing linear interpolation measurement on the kth data point and the (k + 2) th data point in the (k + 1) th-dimensional vector, wherein k starts from 1;

s33, detecting whether k reaches N-2, if not, executing S34, if yes, executing S35;

s34, returning to S32 after k is k +1, so as to reconstruct a next one-dimensional vector in the two-dimensional array according to the 1 st-dimensional vector;

s35, outputting the two-dimensional array with N-1 dimensional vectors for use by the S4 as the multi-dimensional sample set.

6. The Prony iterative analysis method of claim 5, wherein the S4 comprises:

s41, solving a P-dimensional linear equation set based on the kth-dimensional vector in the multi-dimensional sample set to obtain a kth equation coefficient sequence I, wherein k is iterated again from 1;

s42, solving a characteristic equation based on the kth equation coefficient sequence I to obtain a kth root sequence;

s43, solving a Van der Monte equation set based on the kth root sequence and the kth-dimension vector to obtain a kth equation coefficient sequence II;

s44, respectively measuring a kth amplitude sequence and a kth phase sequence based on the kth equation coefficient sequence II, and respectively measuring a kth frequency sequence and a kth damping coefficient sequence based on the kth root sequence;

s45, detecting whether k reaches N-1, if not, executing S46, and if yes, executing S47;

s46, when k is k +1, returns to S41;

and S47, combining the N-1 amplitude sequences, the N-1 phase sequences, the N-1 frequency sequences and the N-1 damping coefficient sequences to obtain the oscillation parameter set.

7. The Prony iterative analysis method of any of claims 1-6, wherein said set of oscillation parameters comprises N-1 sequences of amplitude, N-1 sequences of phase, N-1 sequences of frequency, and N-1 sequences of damping coefficients;

the S5 includes:

respectively carrying out averaging measurement and calculation processing on the N-1 amplitude sequences, the N-1 phase sequence, the N-1 frequency sequences and the N-1 damping coefficient sequences to obtain an amplitude mean value sequence, a phase mean value sequence, a frequency mean value sequence and a damping mean value sequence;

constructing a fitting oscillation signal based on the amplitude mean sequence, the phase mean sequence, the frequency mean sequence and the damping mean sequence;

and detecting whether the fitting degree of the fitting oscillation signal is greater than preset convergence precision, if so, determining that P does not reach the expected value, and if not, determining that P has reached the expected value.

8. A computing device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor, when executing the computer program, implements the Prony iterative analysis method of any of claims 1-7.

9. A non-transitory computer-readable storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, implements the Prony iterative analysis method of any of claims 1-7.

10. A power grid system comprising the computing device of claim 8.

Images