CN113644916B - Electric power system steady-state data compression method based on edge calculation - Google Patents
Electric power system steady-state data compression method based on edge calculation Download PDFInfo
- Publication number
- CN113644916B CN113644916B CN202110871555.3A CN202110871555A CN113644916B CN 113644916 B CN113644916 B CN 113644916B CN 202110871555 A CN202110871555 A CN 202110871555A CN 113644916 B CN113644916 B CN 113644916B
- Authority
- CN
- China
- Prior art keywords
- sparse
- matrix
- state data
- algorithm
- steady
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000013144 data compression Methods 0.000 title claims abstract description 35
- 238000004364 calculation method Methods 0.000 title claims abstract description 26
- 239000011159 matrix material Substances 0.000 claims abstract description 58
- 238000007906 compression Methods 0.000 claims abstract description 28
- 230000006835 compression Effects 0.000 claims abstract description 28
- 230000009466 transformation Effects 0.000 claims abstract description 13
- 238000000354 decomposition reaction Methods 0.000 claims description 17
- 238000005259 measurement Methods 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 10
- 239000012895 dilution Substances 0.000 claims description 3
- 238000010790 dilution Methods 0.000 claims description 3
- 239000000243 solution Substances 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims description 2
- 230000004927 fusion Effects 0.000 abstract description 14
- 238000005516 engineering process Methods 0.000 abstract description 10
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 11
- 230000000694 effects Effects 0.000 description 11
- 238000005070 sampling Methods 0.000 description 8
- 230000003068 static effect Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000012669 compression test Methods 0.000 description 2
- 238000007667 floating Methods 0.000 description 2
- 238000011158 quantitative evaluation Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention relates to a steady-state data compression method of an electric power system based on edge calculation, which completes steady-state data fusion of the electric power system under the edge calculation by combining compressed sensing and distributed source coding through the stages of establishing a joint sparse model, establishing a sparse redundant dictionary, definitely measuring a matrix, establishing a joint reconstruction algorithm and the like; and decomposing the obtained data fusion result to each scale level according to the resolution by utilizing a wavelet transformation algorithm to obtain a high-frequency coefficient and a low-frequency coefficient, and outputting a compression result by adopting a lossless coding technology after thresholding the high-frequency coefficient. The invention combines with the edge calculation method to construct a steady-state data compression method, effectively compresses the data of the power system, and reduces the storage space and the data transmission quantity.
Description
Technical Field
The invention relates to an electric power system steady-state data compression method based on edge calculation. Belongs to the technical field of power grids.
Background
The national economy is continuously rising, the power grid scale is increasingly strong, and the power system is gradually developed towards diversification and complexity. And the power management and control requirements of data analysis, fault monitoring, wide area measurement and the like can be met by effectively recording massive power data. Therefore, if the large-scale data generated during the operation of the power system is stored and transmitted, the operation speed and the burden of the storage space are greatly increased, and even the intelligent development of the power grid in a crossing manner is hindered.
The power system is a key component of the grid enterprise. The society and technology develop rapidly, the degree of intellectualization and informatization is increased gradually, and along with the rapid promotion of the power grid communication technology, the data transmission scale is gradually strong, so that the data compression is gradually evolved into one of the hot research subjects in the electric power field.
Disclosure of Invention
The invention aims to overcome the defects and provides an electric power system steady state data compression method based on edge calculation.
The purpose of the invention is realized in the following way:
the steady-state data compression method of the electric power system based on edge calculation is characterized by comprising the following steps of: the method comprises the following steps:
s1, establishing a joint sparse model;
assuming that the initial steady state data signal is x j The common part and the innovation part are respectively z c 、z j Then signal x j The sparse representation of (2) is as follows:
x j ＝z c +z j ＝z j ＝ψθ j
wherein, the sparse matrix and the corresponding sparse coefficient are respectively psi and theta j ；
S2, creating a sparse dictionary;
regarding the sparse decomposition stage, an initial sparse dictionary D is known, sparse representation is obtained through solution of an orthogonal matching pursuit algorithm, and then sparse representation coefficients shown in the following formula are obtained
The constraints are as follows:
wherein y represents a reconstructed steady-state data signal, the preset threshold value is epsilon, N represents the number of atoms, and F represents the norm of a sparse coefficient matrix;
in the cyclic calculation during dictionary updating, the dictionary training algorithm only updates one atom at a time; when a new atom d is acquired k When the following equation holds:
in the above, the dilution factors with the line number j and k in the sparse factor matrix X are respectivelyd j The ordinal number j is an atom;
if d is removed k The deviations generated by the other atoms beingThe above formula is rewritten as the following expression:
let atom d k Is indexed by ω k ，N*ω k Matrix of omega k If divided by (omega) k (i) I) matrix elements other than non-zero values are all zero values, the following expression is derived from the above equation,
wherein the index ω avoiding result divergence k The expression is as follows:
in the above, the k-th line sparse coefficientAfter the zero value term in (1) is removed, a row vector is obtained>I.e. < ->Atom d in sparse coding stage k Deviation column is->I.e. < ->
By singular value decompositionPolicy resolution bias columnThe following decomposition expression is obtained:
wherein, two mutually orthogonal matrixes are U, V respectively, the diagonal matrix is delta, the first columns of the two orthogonal matrixes U, V are obtained by decomposition, and the first columns are used for completing the atom d in the initial dictionary k The latter is multiplied by a diagonal matrix delta (1, 1), and x is updated and replaced by the resulting product j Further acquiring a new sparse dictionary;
s3, establishing a measurement matrix;
constructing a Gaussian measurement matrix, reducing the signal dimension represented by the sparse matrix, and simultaneously ensuring that the accuracy of the reconstructed signal and the constraint equidistant condition are established;
s4, establishing a joint reconstruction algorithm;
establishing a joint reconstruction algorithm by fusing a synchronous orthogonal matching pursuit algorithm and a learning training algorithm; reconstructing the acquired steady-state data by using the former algorithm, and updating the sparse dictionary by using the latter algorithm; the algorithm operation flow is specifically described as follows:
s4-1, initializing relevant parameters of a joint reconstruction algorithm; for the initial residual r 0 A residual error r corresponding to the p-th node p The two are equal; index value omega k =0; index set Λ 0 Is an empty set;
s4-2, the initial signal matrix X n*s Initial dictionary ψ n*n Measuring matrix phi m*n Minimum reconstructed SNR def As an input term, where s is the number of nodes, and p= {1, 2..once, s }, data length is n, and the number of measurements is m;
s4-3, establishing a sensing matrix to obtain the following expression:
A m*n ＝ψ m*n *Φ n*n
s4-4, solving each by adopting the following formulaRow residual r p And each column of sensing matrix A q Sum of two norms between:
according to the obtained maximum value of the sum of the two norms, reserving a corresponding sensing matrix array index, and fusing the sensing matrix array index with an index set to obtain a new index set, wherein the new index set is as follows:
Λ τ ＝[Λ τ-1 ξ p ]
s4-5, after the related parameters are obtained through a least square algorithm, updating residual errors by using the following expression:
s4-6, respectively solving a reconstructed intermediate signal and an opposite party root error and a reconstructed signal-to-noise ratio by adopting the following calculation formula:
finally, the lowest reconstructed SNR is compared def When the signal-to-noise ratio SNR is smaller, dictionary atoms are updated by adopting a dictionary training algorithm, and S4-4 is returned; otherwise, an output result, i.e. a reconstruction result x ', is obtained' j And dictionary atoms used;
s5, compressing steady-state data of the power system under wavelet transformation;
known even sequence e j+1 Odd number sequence o j+1 Then adoptThe wavelet decomposition process is described by the following expression:
split(x′ j )＝(e j+1 ,o j+1 )
in the above, even sequence e j+1 ＝a j+1 -U(b j+1 ) Odd number sequence o j+1 ＝b j+1 +P(a j+1 ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein a is j+1 And b j+1 Respectively representing the low frequency coefficient and the high frequency coefficient in the sequence, Y (b) j+1 ) And P (a) j+1 ) Respectively representing an updating result of the high-frequency coefficient and a prediction result of the low-frequency coefficient;
from this, a compression reconstructed signal representation is derived as follows:
x″ j ＝merge(e j+1 ,o j+1 )
where merge represents a merge sort algorithm.
Further, in S3, there is a constant δ in the value range from 0 to 1, and the measurement matrix Φ may be established for all the sparse coefficient matrices X as follows:
a compression effect evaluation method of an electric power system steady-state data compression method based on edge calculation is characterized by comprising the following steps: an electric energy acquisition device is connected to the electric interface to acquire steady state data of a study object; quantitatively evaluating three indexes of the data compression space occupation ratio, the Fan Junfang error and the data compression ratio; wherein,
compared with the prior art, the invention has the beneficial effects that:
(1) The invention relates to an electric power system steady-state data compression method based on edge calculation, which completes electric power system steady-state data fusion under the edge calculation by combining compressed sensing and distributed source coding through the stages of establishing a joint sparse model, establishing a sparse redundant dictionary, definitely measuring a matrix, establishing a joint reconstruction algorithm and the like; and decomposing the obtained data fusion result to each scale level according to the resolution by utilizing a wavelet transformation algorithm to obtain a high-frequency coefficient and a low-frequency coefficient, and outputting a compression result by adopting a lossless coding technology after thresholding the high-frequency coefficient.
(2) According to the steady-state data compression method of the electric power system based on edge calculation, according to quantitative evaluation results of three indexes of the data compression space ratio, the Fan Junfang error and the data compression ratio, the method has obvious compression advantages, and the waveform fitting degree of a compressed signal and an actual sampling signal is high.
Drawings
Fig. 1 is a schematic diagram of the principle of reconstruction compression.
FIG. 2 is a flow chart for fusing and compressing steady-state data signals of a power system.
FIG. 3 is a schematic diagram of a steady state data sampling signal.
Fig. 4 is a waveform diagram of a reconstructed signal based on a fusion of multiple compression modes.
Fig. 5 is a waveform diagram of an error signal based on a fusion of multiple compression modes.
Fig. 6 is a waveform diagram of a reconstructed signal based on tensor Tucker decomposition.
Fig. 7 is a waveform diagram of an error signal based on tensor Tucker decomposition.
Fig. 8 is a waveform diagram of a reconstructed signal according to the present invention.
Fig. 9 is a waveform diagram of an error signal according to the present invention.
Fig. 10 is a diagram showing the results of the error and data compression ratio evaluation index of Fan Junfang.
FIG. 11 is a schematic diagram of steady state data compression space duty cycle index for different methods.
Detailed Description
The following describes the embodiments of the present invention further with reference to the drawings and examples. The following examples are only for more clearly illustrating the technical aspects of the present invention and are not to be construed as limiting the scope of the invention.
The invention is based on a distributed compressed sensing technology, and the steady-state data fusion of the power system under the edge calculation is completed through four steps of establishing a joint sparse model, establishing a sparse redundant dictionary, definitely measuring a matrix and establishing a joint reconstruction algorithm. And decomposing the obtained data fusion result to each scale level according to the resolution by utilizing a wavelet transformation algorithm to obtain a high-frequency coefficient and a low-frequency coefficient, and outputting a compression result by adopting a lossless coding technology after thresholding the high-frequency coefficient.
The invention comprises the following steps:
s1, establishing a joint sparse model;
the steady-state data of the power system in the time domain has no sparsity, and the distributed compressed sensing technology can be used for collection and fusion only after the steady-state data are sparsely decomposed by using a sparse basis. Because the signal frequency of the steady-state data contains subharmonic and fundamental wave, the initial sparse basis is represented by a Fourier positive transformation matrix, and a joint sparse model is established to acquire the steady-state data. The signals do not have a common part, so the sparse representation of the innovation part (i.e. the difference between the coefficient vector and the common part) can be done by one sparse basis. Assuming that the initial steady state data signal is x j The common part and the innovation part are respectively z c 、z j Then signal x j The sparse representation of (2) is as follows:
x j ＝z c +z j ＝z j ＝ψθ j (1)
wherein, the sparse matrix and the corresponding sparse coefficient are respectively psi and theta j 。
S2, creating a sparse dictionary;
the sparseness is inversely related to the number of atoms and the amount of data uploaded. In order to ensure that atoms and initial signals realize self-adaptive matching, a learning type sparse dictionary in a dictionary training algorithm is introduced, and the deviation between the initial signals and the reconstructed signals is reduced through repeated dictionary updating, so that the signal-to-noise ratio of the reconstructed signals accords with a preset threshold value.
Regarding the sparse decomposition stage, an initial sparse dictionary D is known, sparse representation is obtained through solution of an orthogonal matching pursuit algorithm, and then sparse representation coefficients shown in the following formula are obtained
The constraints are as follows:
wherein y represents a reconstructed steady-state data signal, the preset threshold value is epsilon, N represents the number of atoms, and F represents the norm of the sparse coefficient matrix.
In loop computation at dictionary update time, the dictionary training algorithm performs update processing for only one atom at a time. When a new atom d is acquired k When the following equation holds:
in the above, the dilution factors with the line number j and k in the sparse factor matrix X are respectivelyd j The ordinal number j is an atom.
If d is removed k The deviations generated by the other atoms beingThe above formula is rewritten as the following expression:
let atom d k Is indexed by ω k ，N*ω k Matrix of omega k If divided by (omega) k (i) I) matrix elements other than non-zero values are zero values, the following expression is derived from the above equation, wherein the index ω of the result divergence is avoided k The expression is shown as formula (7):
in the above, the k-th line sparse coefficientAfter the zero value term in (1) is removed, a row vector is obtained>I.e. < ->Atom d in sparse coding stage k Deviation column is->I.e. < ->
Decomposing bias columns by singular value decomposition strategyThe following decomposition expression is obtained:
wherein, two mutually orthogonal matrixes are U, V respectively, the diagonal matrix is delta, the first columns of the two orthogonal matrixes U, V are obtained by decomposition, and the first columns are used for completing the atom d in the initial dictionary k The latter is multiplied by a diagonal matrix delta (1, 1), and x is updated and replaced by the resulting product j And further obtaining a new sparse dictionary.
S3, establishing a measurement matrix;
and (3) constructing a Gaussian measurement matrix, reducing the signal dimension represented by the sparse matrix, and simultaneously ensuring that the accuracy of the reconstructed signal and the constraint equidistant condition are established. That is, there is a constant δ in the value range of 0 to 1, and the measurement matrix Φ may be established for all the sparse coefficient matrix X as follows:
s4, establishing a joint reconstruction algorithm;
and establishing a joint reconstruction algorithm by fusing the synchronous orthogonal matching pursuit algorithm and the learning training algorithm. The former algorithm is used for reconstructing the acquired steady-state data, and the latter algorithm is used for updating the sparse dictionary. The algorithm operation flow is specifically described as follows:
s4-1, initializing relevant parameters of a joint reconstruction algorithm. For the initial residual r 0 A residual error r corresponding to the p-th node p The two are equal; index value omega k =0; index set Λ 0 Is an empty set;
s4-2, the initial signal matrix X n*s Initial dictionary ψ n*n Measuring matrix phi m*n Minimum reconstructed SNR def As an input term, where s is the number of nodes, and p= {1, 2..once, s }, data length is n, and the number of measurements is m;
s4-3, establishing a sensing matrix to obtain the following expression:
A m*n ＝ψ m*n *Φ n*n (10)
s4-4, solving residual errors r of each row by adopting the following formula p And each column of sensing matrix A q Sum of two norms between:
according to the obtained maximum value of the sum of the two norms, reserving a corresponding sensing matrix array index, and fusing the sensing matrix array index with an index set to obtain a new index set, wherein the new index set is as follows:
Λ τ ＝[Λ τ-1 ξ p ] (12)
s4-5, after the related parameters are obtained through a least square algorithm, updating residual errors by using the following expression:
s4-6, respectively solving a reconstructed intermediate signal and an opposite party root error and a reconstructed signal-to-noise ratio by adopting the following calculation formula:
finally, the lowest reconstructed SNR is compared def When the signal-to-noise ratio SNR is smaller, dictionary atoms are updated by adopting a dictionary training algorithm, and S4-4 is returned; otherwise, an output result, i.e. a reconstruction result x ', is obtained' j And dictionary atoms used.
S5, compressing steady-state data of the power system under wavelet transformation;
based on a steady-state data fusion result obtained by edge calculation, decomposing the steady-state data fusion result to each scale level according to resolution by utilizing a wavelet transformation algorithm to obtain a high-frequency coefficient and a low-frequency coefficient, zeroing the relatively small high-frequency coefficient by threshold processing, and only preserving the low-frequency coefficient and the high-frequency coefficient with signal characteristic presenting capability. The conversion from integer to integer is fundamentally realized, the floating point calculation step is reduced, and the method is more suitable for the practical application of an electric power system. The reconstruction compression based on the wavelet transformation algorithm is divided into several stages of splitting, predicting, updating and the like, and the principle of the reconstruction compression is shown in figure 1.
Known even sequence e j+1 Odd number sequence o j+1 The wavelet decomposition process is described using the following expression:
split(x′ j )＝(e j+1 ,o j+1 ) (17)
in the above, even sequence e j+1 ＝a j+1 -U(b j+1 ) Odd number sequence o j+1 ＝b j+1 +P(a j+1 ). Wherein a is j+1 And b j+1 Respectively representing the low frequency coefficient and the high frequency coefficient in the sequence, Y (b) j+1 ) And P (a) j+1 ) The update result of the high frequency coefficient and the prediction result of the low frequency coefficient are respectively shown.
From this, a compression reconstructed signal representation is derived as follows:
x″ j ＝merge(e j+1 ,o j+1 ) (18)
where merge represents a merge sort algorithm.
The implementation flow of the fusion and compression method of the steady-state data signals of the power system is shown in figure 2. The steady-state data signals are fused by utilizing an edge algorithm, then the multi-scale transformation processing is carried out by utilizing a wavelet algorithm, the high-frequency coefficient is processed by a threshold value, and the compression ratio is improved by adopting a lossless coding technology.
Embodiment one, power System steady-state data compression Experimental analysis
S1, a preparation stage;
and carrying out a compression test of static data aiming at a test running power system of a certain power grid company, and verifying the feasibility and applicability of the method. And an EAC5000D type electric energy acquisition device is connected to the electric interface to acquire steady state data of a study object, and a sampling signal of the steady state data is shown in figure 3. In the acquisition process, the sampling rate is 50kHz, and the number of sampling points of an initial voltage signal is 36000.
In order to measure the data compression effect, a method integrating multiple compression modes and tensor Tucker decomposition and a method thereof are adopted, collected static data are compressed one by one, and a reconstruction signal and an error signal waveform of the static data are observed; and quantitatively evaluating three indexes including the data compression space ratio, the Fan Junfang error and the data compression ratio. The calculation formulas of the indexes are respectively as follows:
of the three indexes, the other two compression evaluation indexes are in negative correlation with the compression effect except that the data compression space ratio index value and the compression effect are in positive correlation, and the smaller the index value is, the more ideal the compression effect is.
S2, analyzing steady-state data compression effect based on sampling signals;
the static data compression reconstruction signal and the error signal of the different methods are shown in fig. 4-6, respectively. As can be seen from the signal waveform diagram, the method fuses the acquired steady state data by adopting the edge algorithm, and decomposes the fused signal to each scale level by utilizing the wavelet transformation algorithm, so that the final compressed signal has higher fitting degree with the actual steady state data sampling signal waveform (see fig. 3).
By combining the signal waveforms of different methods and the error of the assigned Fan Junfang with the data compression ratio evaluation index result (shown in fig. 7), it can be seen that the invention fundamentally realizes the conversion from integer to integer by referring to the wavelet transformation algorithm, greatly reduces the floating point calculation step, and improves the compression ratio by the lossless coding technology, so that the two index values of the data compression ratio and the assigned Fan Junfang error are far smaller than the index values of the other two methods. This illustrates that the method herein can remove more redundant data, and the initial signal characteristics are better preserved, with significant compression advantages.
S3, analyzing steady-state data compression effect based on data specification;
in order to detect the influence of the data size on the compression effect, aiming at steady-state data with specifications of 64kB, 128kB, 256kB, 512kB and 1MB, the steady-state data compression effect of different methods is evaluated by adopting the data compression space ratio. The index data results are shown in fig. 8.
It can be seen that the method of the present invention zeroes relatively small high frequency coefficients by thresholding, and only retains low frequency coefficients and high frequency coefficients with signal feature rendering capabilities, so that the compression space occupation ratio is larger and the compression effect is more ideal than the other two methods. As can be seen from the trend of the space duty ratio curve of the method, the index value is reduced along with the increase of the data specification, which indicates that a certain correlation exists between the compression effect and the data size of the method, and the method can be used as the research key point of the next stage so as to cope with the mass data scale of the electric power system in the information age.
The method completes steady-state data fusion of the power system under edge calculation by combining compressed sensing and distributed source coding through the stages of establishing a joint sparse model, establishing a sparse redundant dictionary, definitely measuring a matrix, establishing a joint reconstruction algorithm and the like. And decomposing the obtained data fusion result to each scale level according to the resolution by utilizing a wavelet transformation algorithm to obtain a high-frequency coefficient and a low-frequency coefficient, and outputting a compression result by adopting a lossless coding technology after thresholding the high-frequency coefficient. In the test, aiming at the static data expansion and compression test of the test running power system of a certain power grid company, according to quantitative evaluation results of three indexes of the data compression space ratio, the Fan Junfang error and the data compression ratio, the method has obvious compression advantages, and the waveform fitting degree of the compressed signal and the actual sampling signal is higher.
In the above embodiments, the present invention is described only by way of example, but various modifications of the invention can be made by those skilled in the art after reading the present patent application without departing from the spirit and scope of the invention.
Claims (3)
1. An electric power system steady state data compression method based on edge calculation comprises the following steps:
s1, establishing a joint sparse model;
assuming that the initial steady state data signal is x j The common part and the innovation part are respectively z c 、z j Then signal x j The sparse representation of (2) is as follows:
x j ＝z c +z j ＝z j ＝ψθ j
wherein, the sparse matrix and the corresponding sparse coefficient are respectively psi and theta j ；
S2, creating a sparse dictionary;
regarding the sparse decomposition stage, an initial sparse dictionary D is known, sparse representation is obtained through solution of an orthogonal matching pursuit algorithm, and then sparse representation coefficients shown in the following formula are obtained
The constraints are as follows:
wherein y represents a reconstructed steady-state data signal, the preset threshold value is epsilon, N represents the number of atoms, and F represents the norm of a sparse coefficient matrix;
in the cyclic calculation during dictionary updating, the dictionary training algorithm only updates one atom at a time; when a new atom d is acquired k When the following equation holds:
in the above, the dilution factors with the line number j and k in the sparse factor matrix X are respectivelyd j The ordinal number j is an atom;
if d is removed k The deviations generated by the other atoms beingThe above formula is rewritten as the following expression:
let atom d k Is indexed by ω k ，N*ω k Matrix of omega k If divided by (omega) k (i) I) matrix elements other than non-zero values are all zero values, the following expression is derived from the above equation,
wherein the index ω avoiding result divergence k The expression is as follows:
in the above, the kth lineSparse coefficientAfter the zero value term in (1) is removed, a row vector is obtained>I.e. < ->Atom d in sparse coding stage k Deviation column is->I.e. < ->
Decomposing bias columns by singular value decomposition strategyThe following decomposition expression is obtained:
wherein, two mutually orthogonal matrixes are U, V respectively, the diagonal matrix is delta, the first columns of the two orthogonal matrixes U, V are obtained by decomposition, and the first columns are used for completing the atom d in the initial dictionary k The latter is multiplied by a diagonal matrix delta (1, 1), and x is updated and replaced by the resulting product j Further acquiring a new sparse dictionary;
s3, establishing a measurement matrix;
constructing a Gaussian measurement matrix, reducing the signal dimension represented by the sparse matrix, and simultaneously ensuring that the accuracy of the reconstructed signal and the constraint equidistant condition are established;
s4, establishing a joint reconstruction algorithm;
establishing a joint reconstruction algorithm by fusing a synchronous orthogonal matching pursuit algorithm and a learning training algorithm; reconstructing the acquired steady-state data by using the former algorithm, and updating the sparse dictionary by using the latter algorithm; the algorithm operation flow is specifically described as follows:
s4-1, initializing relevant parameters of a joint reconstruction algorithm; for the initial residual r 0 A residual error r corresponding to the p-th node p The two are equal; index value omega k =0; index set Λ 0 Is an empty set;
s4-2, the initial signal matrix X n*s Initial dictionary ψ n*n Measuring matrix phi m*n Minimum reconstructed SNR def As an input term, where s is the number of nodes, and p= {1, 2..once, s }, data length is n, and the number of measurements is m;
s4-3, establishing a sensing matrix to obtain the following expression:
A m*n ＝ψ m*n *Φ n*n
s4-4, solving residual errors r of each row by adopting the following formula p And each column of sensing matrix A q Sum of two norms between:
according to the obtained maximum value of the sum of the two norms, reserving a corresponding sensing matrix array index, and fusing the sensing matrix array index with an index set to obtain a new index set, wherein the new index set is as follows:
Λ τ ＝[Λ τ-1 ξ p ]
s4-5, after the related parameters are obtained through a least square algorithm, updating residual errors by using the following expression:
s4-6, respectively solving a reconstructed intermediate signal and an opposite party root error and a reconstructed signal-to-noise ratio by adopting the following calculation formula:
finally, the lowest reconstructed SNR is compared def When the signal-to-noise ratio SNR is smaller, dictionary atoms are updated by adopting a dictionary training algorithm, and S4-4 is returned; otherwise, an output result, i.e. a reconstruction result x ', is obtained' j And dictionary atoms used;
s5, compressing steady-state data of the power system under wavelet transformation;
known even sequence e j+1 Odd number sequence o j+1 The wavelet decomposition process is described using the following expression:
split(x′ j )＝(e j+1 ,o j+1 )
in the above, even sequence e j+1 ＝a j+1 -U(b j+1 ) Odd number sequence o j+1 ＝b j+1 +P(a j+1 ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein a is j+1 And b j+1 Respectively representing the low frequency coefficient and the high frequency coefficient in the sequence, Y (b) j+1 ) And P (a) j+1 ) Respectively representing an updating result of the high-frequency coefficient and a prediction result of the low-frequency coefficient;
from this, a compression reconstructed signal representation is derived as follows:
x″ j ＝merge(e j+1 ,o j+1 )
where merge represents a merge sort algorithm.
2. The method for compressing steady-state data of an electric power system based on edge calculation according to claim 1, wherein the method comprises the following steps: in S3, there is a constant δ in the value range from 0 to 1, and the measurement matrix Φ holds the following inequality for all the sparse coefficient matrix X:
3. the method for evaluating the steady-state data compression method of the electric power system based on edge calculation according to claim 1, wherein the method comprises the following steps: an electric energy acquisition device is connected to the electric interface to acquire steady state data of a study object; quantitatively evaluating three indexes of the data compression space occupation ratio, the Fan Junfang error and the data compression ratio; wherein,
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110871555.3A CN113644916B (en) | 2021-07-30 | 2021-07-30 | Electric power system steady-state data compression method based on edge calculation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110871555.3A CN113644916B (en) | 2021-07-30 | 2021-07-30 | Electric power system steady-state data compression method based on edge calculation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113644916A CN113644916A (en) | 2021-11-12 |
| CN113644916B true CN113644916B (en) | 2023-12-26 |
Family
ID=78419220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110871555.3A Active CN113644916B (en) | 2021-07-30 | 2021-07-30 | Electric power system steady-state data compression method based on edge calculation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113644916B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114553538A (en) * | 2022-02-22 | 2022-05-27 | 国网山东省电力公司电力科学研究院 | Active protection method and system for power grid information safety |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103778919A (en) * | 2014-01-21 | 2014-05-07 | 南京邮电大学 | Speech coding method based on compressed sensing and sparse representation |
| CN104159112A (en) * | 2014-08-08 | 2014-11-19 | 哈尔滨工业大学深圳研究生院 | Compressed sensing video transmission method and system based on dual sparse model decoding |
| CN104331913A (en) * | 2014-11-19 | 2015-02-04 | 西安电子科技大学 | Polarized SAR polarization method based on sparse K-SVD (Singular Value Decomposition) |
| CN110135488A (en) * | 2019-05-10 | 2019-08-16 | 南京邮电大学 | Merge the high-quality compression method of data of dictionary training and observing matrix optimization |
-
2021
- 2021-07-30 CN CN202110871555.3A patent/CN113644916B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103778919A (en) * | 2014-01-21 | 2014-05-07 | 南京邮电大学 | Speech coding method based on compressed sensing and sparse representation |
| CN104159112A (en) * | 2014-08-08 | 2014-11-19 | 哈尔滨工业大学深圳研究生院 | Compressed sensing video transmission method and system based on dual sparse model decoding |
| CN104331913A (en) * | 2014-11-19 | 2015-02-04 | 西安电子科技大学 | Polarized SAR polarization method based on sparse K-SVD (Singular Value Decomposition) |
| CN110135488A (en) * | 2019-05-10 | 2019-08-16 | 南京邮电大学 | Merge the high-quality compression method of data of dictionary training and observing matrix optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113644916A (en) | 2021-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Foucart et al. | An invitation to compressive sensing | |
| Otten et al. | Recovering noise-free quantum observables | |
| AU2020214409A1 (en) | Structural damage identification method based on ensemble empirical mode decomposition and convolution neural network | |
| Maji et al. | Theory and application of cellular automata for pattern classification | |
| CN106649658B (en) | Recommendation system and method for user role non-difference treatment and data sparsity | |
| WO2007075757A2 (en) | Parallel multi-rate circuit simulation | |
| He et al. | Improved FOCUSS method with conjugate gradient iterations | |
| CN111324861B (en) | Deep learning magnetic resonance spectrum reconstruction method based on matrix decomposition | |
| CN113644916B (en) | Electric power system steady-state data compression method based on edge calculation | |
| Guillemard et al. | On Groupoid C*-Algebras, Persistent Homology and Time-Frequency Analysis | |
| Billings et al. | Discrete wavelet models for identification and qualitative analysis of chaotic systems | |
| Krishnamurthy et al. | Sample complexity of learning mixture of sparse linear regressions | |
| CN111783631A (en) | Deep learning magnetic resonance spectrum reconstruction method based on sparse representation | |
| CN111915449A (en) | VMD and OMP-based power load data dimension reduction reconstruction processing method | |
| CN115238759A (en) | Electric power data compression and electric energy quality disturbance identification method based on compressed sensing | |
| Lami et al. | Attainability and Lower Semi-continuity of the Relative Entropy of Entanglement and Variations on the Theme | |
| CN115758249A (en) | Non-intrusive load identification method and device | |
| CN114880948A (en) | Harmonic prediction modeling method and system based on random forest optimization algorithm | |
| Parada-Mayorga et al. | Graphon pooling for reducing dimensionality of signals and convolutional operators on graphs | |
| CN103036576A (en) | Two-value sparse signal reconstruction algorithm based on compressive sensing theory | |
| Xie et al. | Probabilistic imaginary-time evolution algorithm based on nonunitary quantum circuits | |
| Seki et al. | Energy-filtered random-phase states as microcanonical thermal pure quantum states | |
| Markovich et al. | More accurate and efficient bath spectral densities from super-resolution | |
| Porat et al. | Signal representation in the combined phase-spatial space: Reconstruction and criteria for uniqueness | |
| CN115051363A (en) | Distribution network area user change relation identification method and device and computer storage medium |
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 |