Compressed sensing method and system for source-load characteristic signals in typical power usage scenarios

Abstract

The present invention discloses a method and system for compressing the sensing of source-load characteristic signals of a typical power usage scenario, the method comprising the steps of: 1) acquiring the source-load characteristic signals of the typical power usage scenario; 2) performing sparse transformation on the source-load characteristic signals to obtain corresponding frequency domain signals; 3) sampling the frequency domain signals through a measurement matrix; 4) performing adaptive matching pursuit reconstruction on the sampled frequency domain signals to obtain reconstructed signals. The present invention has the advantages of simple and fast implementation process and accurate signal reconstruction.

Description

Technical Field

[0001] The present invention mainly relates to the field of signal processing technology, and specifically to a method and system for compressing a source-load characteristic signal in a typical power usage scenario. Background Art

[0002] At present, in typical electricity usage scenarios, there is a large amount of signal data that needs to be collected. These signal data themselves have a relatively comprehensive coverage, so there are many types of signal data, and the signal collection rate is relatively high, which will generate a large amount of data. When these data need to be processed or stored, it requires large hardware costs or time costs. Summary of the invention

[0003] The technical problem to be solved by the present invention is: in view of the problems existing in the prior art, the present invention provides a method and system for compressing source-load characteristic signals of typical power usage scenarios, which has a simple and fast process and can realize accurate signal reconstruction.

[0004] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0005] A method for compressing a source-load characteristic signal of a typical power usage scenario comprises the following steps:

[0006] 1) Obtain source-load characteristic signals of typical power consumption scenarios;

[0007] 2) Perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal;

[0008] 3) Sampling the frequency domain signal through the measurement matrix;

[0009] 4) Perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal.

[0010] Preferably, in step 4), an adaptive matching pursuit algorithm is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal.

[0011] Preferably, the specific process of the adaptive matching pursuit algorithm is:

[0012] 4.1) Initial margin r0 = u, initial step size step_size, stage stage = 1, number of iterations t = 1, index value set

[0013] 4.2) If ||r||2≤ε1, stop the iteration and use the obtained atoms to reconstruct the signal; otherwise, go to step 4.3);

[0014] 4.3) Calculate the correlation coefficient V, and find the index value j corresponding to the maximum value of step_size from V and store it in J;

[0015]

[0016] 4.4) Regularize the correlation coefficients of atoms corresponding to index values ​​in J, and save the regularized results in set J0. The correlation coefficients of atoms in this set satisfy the following formula:

[0017] |V(i)|≤2|V(j)|,i,j∈J

[0018] 4.5) Update support set Where Λ=Λ∪J0;

[0019] 4.6) Apply the following formula to obtain the reconstructed value of the sparse signal And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new

[0020]

[0021]

[0022] 4.7) If ||r new -r||2≤ε2, then set stage=stage+1, step_size=step_size+1, and go to step 4.3), otherwise set r=r new , iteration number t=t+1, go to step 4.2).

[0023] Preferably, in step 4), the ROMP algorithm is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal, and the specific process is:

[0024] S41, calculate the initial margin r0 = u, the sparsity is K, the number of iterations t = 1, the index value set

[0025] S42, calculate the correlation coefficient V, and find the index value j corresponding to the K maximum values ​​from V and store it in J;

[0026]

[0027] in is the support set;

[0028] S43, regularize the correlation coefficients of atoms corresponding to the index values ​​in J, and save the regularized results in a set J0, where the correlation coefficients of atoms in the set satisfy the following formula:

[0029] |V(i)|≤2|V(j)|,i,j∈J

[0030] S44, Update support set Where Λ=Λ∪J0;

[0031] S45. Apply the following formula to obtain the reconstructed value of the sparse signal: And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new

[0032]

[0033]

[0034] S46. If |Λ|≥2K, stop the iteration, otherwise let r=r new , iteration number t=t+1, go to step S42.

[0035] Preferably, in step 1), the source-load characteristic signal includes one or more of voltage sag, voltage swell, voltage interruption, transient oscillation, transient pulse, voltage spike, voltage notch, voltage flicker and harmonic wave.

[0036] Preferably, in step 2), the source-charge characteristic signal is sparsely transformed by wavelet transform.

[0037] Preferably, in step 3), signal sampling is performed by means of a random Gaussian measurement matrix.

[0038] The present invention also discloses a source-load characteristic signal compression sensing system for a typical power usage scenario, including:

[0039] The first program module is used to obtain source-load characteristic signals of typical power usage scenarios;

[0040] The second program module is used to perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal;

[0041] A third program module is used for sampling the frequency domain signal through the measurement matrix;

[0042] The fourth program module is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal.

[0043] The present invention further discloses a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are executed.

[0044] The present invention also discloses a computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is run by the processor, the steps of the method described above are executed.

[0045] Compared with the prior art, the advantages of the present invention are:

[0046] The present invention can realize accurate reconstruction of sparse or compressible signals by collecting a small amount of signal values ​​through a compressed sensing method, and the implementation process is simple and fast.

[0047] The adaptive matching pursuit reconstruction algorithm of the present invention sets two thresholds ε1 and ε2 to adaptively adjust the step size, and does not need to use the sparsity K as a priori knowledge for signal reconstruction, thus avoiding the problem of multiple iterations that still cannot match or overmatch due to the deviation of the estimated value of the sparsity. The regularization process ensures that the support set for accurate signal reconstruction can be obtained after a maximum of K iterations, which greatly shortens the running time while ensuring the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a flowchart of an embodiment of the method.

[0049] Figure 2 The graphs are time domain and frequency domain signal curves of the normal voltage in the present invention; (a) is the time domain graph of the normal voltage; and (b) is the amplitude-frequency spectrum graph.

[0050] Figure 3 The graphs are time domain and frequency domain signal curves of the voltage swell signal in the present invention; wherein (a) is the time domain graph of the voltage swell signal; and (b) is the amplitude-frequency spectrum graph.

[0051] Figure 4 The time domain and frequency domain signal curve diagrams of the voltage sag signal in the present invention; wherein (a) is the time domain diagram of the voltage sag signal; and (b) is the amplitude-frequency spectrum diagram.

[0052] Figure 5 The time domain and frequency domain signal curve diagrams of the voltage interruption signal in the present invention; wherein (a) is the time domain diagram of the voltage interruption signal; and (b) is the amplitude-frequency spectrum diagram.

[0053] Figure 6 The graphs are time domain and frequency domain signal curves of the voltage flicker signal in the present invention; wherein (a) is the time domain graph of the voltage flicker signal; and (b) is the amplitude-frequency spectrum graph.

[0054] Figure 7 The time domain and frequency domain signal curve diagrams of the voltage oscillation signal in the present invention; wherein (a) is the time domain diagram of the voltage oscillation signal; and (b) is the amplitude-frequency spectrum diagram.

[0055] Figure 8 The time domain and frequency domain signal curve diagrams of the voltage pulse signal in the present invention; wherein (a) is the time domain diagram of the voltage pulse signal; and (b) is the amplitude-frequency spectrum diagram.

[0056] Fig. 9The graphs are time domain and frequency domain signal curves of the voltage spike signal in the present invention; wherein (a) is the time domain graph of the voltage spike signal; and (b) is the amplitude-frequency spectrum graph.

[0057] Fig.10 The time domain and frequency domain signal curve diagrams of the voltage notch signal in the present invention; wherein (a) is the time domain diagram of the voltage notch signal; and (b) is the amplitude-frequency spectrum diagram.

[0058] Fig.11 The time domain and frequency domain signal curve diagrams of the harmonic signal in the present invention; wherein (a) is the time domain diagram of the harmonic signal; and (b) is the amplitude-frequency spectrum diagram.

[0059] Fig.12 The comparison diagram of the normal voltage signal before and after reconstruction of the present invention; wherein (a) is the normal voltage signal; (b) is the compressed signal; (c) is the reconstructed signal; and (d) is the reconstruction error.

[0060] Fig.13 The diagram is a comparison of the voltage harmonic signal before and after reconstruction of the present invention; wherein (a) is the voltage harmonic signal; (b) is the compressed signal; (c) is the reconstructed signal; and (d) is the reconstruction error.

[0061] Fig.14 The figures are comparison diagrams of the voltage swell signal before and after reconstruction of the present invention; wherein (a) is the voltage swell signal; (b) is the compressed signal; (c) is the reconstructed signal; and (d) is the reconstruction error.

[0062] Fig.15 The diagram is a comparison of the voltage sag signal before and after reconstruction of the present invention; wherein (a) is the voltage sag signal; (b) is the compressed signal; (c) is the reconstructed signal; and (d) is the reconstruction error.

[0063] Fig.16 The diagram is a comparison of the voltage interruption signal before and after reconstruction according to the present invention; wherein (a) is the voltage interruption signal; (b) is the compressed signal; (c) is the reconstructed signal; and (d) is the reconstruction error. DETAILED DESCRIPTION

[0064] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0065] like Figure 1 As shown, the method for compressing the source-load characteristic signal of a typical power usage scenario in an embodiment of the present invention includes the following steps:

[0066] 1) Obtain source-load characteristic signals of typical power consumption scenarios;

[0067] 2) Perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal;

[0068] 3) Sampling the frequency domain signal through the measurement matrix;

[0069] 4) Perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal.

[0070] The present invention can realize accurate reconstruction of sparse or compressible signals by collecting a small amount of signal values ​​through a compressed sensing method, and the implementation process is simple and fast.

[0071] In the theory of compressed sensing, compressed sampling is achieved directly through the compressed sensing observation matrix φ. The compressed sampling value of the power quality signal f can be expressed as:

[0072]

[0073] Where u is the M×1 dimensional observation vector, f is the N×1 dimensional original harmonic signal, φ is the M×N (M<<N) dimensional compressed sensing measurement matrix, is the N×N dimensional sparse transformation basis matrix, x is the N×1 dimensional sparse transformation signal, It is an M×N dimensional perception matrix.

[0074] The dimension M of the observed signal projection value u is much lower than the dimension N of the original signal, that is, the observed projection value u realizes the projection of high-dimensional data f (N×1 dimension) to low-dimensional data u (M×1 dimension), realizing the data compression process. If u contains enough reconstructed signal information, the projection matrix satisfies the constrained equidistance condition, and the reconstruction algorithm is used to obtain the measured value u and the projection matrix. It can reconstruct the original signal and complete the decompression process of compressed sampling data.

[0075] In a specific embodiment, in the power usage scenario, taking power quality data as an example, the sparse representation problem of the signal, that is, how to find a certain orthogonal basis The original signal is represented by as few basis functions as possible under this transformation basis. Consider a one-dimensional finite-length discrete-time real signal f, which can be expressed as R N N×1-dimensional vector in space, i.e. f=[f1 f2 … fn] T ∈R N According to the theory of harmonic analysis, R N Any signal in space can be used by Wiki function For simplicity, we assume that is a standard orthogonal basis, Then the signal f∈R N In the orthogonal basis The following can be expressed as:

[0076]

[0077] In the formula, x=[x1 x2 … xn T , Obviously, f and x are two equivalent representations of a signal in different domains. f is a representation in the time domain or spatial domain, and x is a representation in the domain.

[0078] When the signal f can be linearly represented by K basis functions (K < N), then f is K-sparse, that is, only K coefficients of x in the above formula are non-zero, and the remaining N - K coefficient values are zero. If there are a small number of large coefficient values and a large number of small coefficient values in the above formula, then the signal is compressible. When the signal f ∈ R i is compressible under a certain orthogonal basis N , then x is the sparse representation of the signal f in the domain.

[0079] According to the above definition, it can be seen that the power quality signal is not sparse in the time domain, but it may become a sparse signal through a certain transform domain. The reconstructed signal not only effectively denoises the original signal, but also can represent most of the information of the original signal with a small number of large coefficients. Only 4.6% of the number of coefficients needs to be saved, which is the number of the original data, greatly compressing the data volume. Among them, wavelet transform can complete the compression of the original signal by performing sparse transform on the original signal.

[0080] In a specific embodiment, in step 1), whether the power system power quality signal can be sparsely represented is the key to whether the compressive sensing theory can be applied. The most common power quality disturbances in the power system include voltage sags, voltage swells, voltage interruptions, transient oscillations, transient pulses, voltage spikes, voltage notches, voltage flickers, and harmonics, etc. The simulation formulas for normal voltage and various disturbance signals are as follows:

[0081] Normal voltage: y(t) = sin(ω0t), where ω0 is the fundamental angular frequency;

[0082] Voltage swell: y(t) = {1 + A[u(t - t1) - u(t - t2)]}sin(ω0t); where A = 0.1 - 0.8, is the swell amplitude; u(t) is the unit step function; t1 is the disturbance occurrence time, t2 is the disturbance end time, T ≤ t2 - t1 ≤ 9T, and T is the period;

[0083] Voltage sag: y(t) = {1 - A[u(t - t1) - u(t - t2)]}sin(ω0t), where A = 0.1 - 0.9, is the sag amplitude, and T ≤ t2 - t1 ≤ 9T;

[0084] ​​Voltage interruption: y(t) = {1-A[u(t-t1)-u(t-t2)]}sin(ω0t), where A = 0.9-1.0, is the interruption amplitude, T ≤ t2-t1 ≤ 9T;

[0085] Voltage flicker: y(t) = (1 + Asinβω0t) sin(ω0t), where A = 0.05 ~ 0.1, is the flicker amplitude; β = 0.1 ~ 0.5, is the multiple of the flicker fluctuation fundamental frequency;

[0086] Voltage oscillation: y(t) = sin(ω0t) + Ae -c(t-t1) sinβω0t[u(t-t1)-u(t-t2)], where A=0.1~0.8 is the oscillation amplitude; c is the attenuation coefficient; β=0.1~0.5 is the multiple of the oscillation fundamental frequency; 0≤t2-t1≤2T;

[0087] Voltage pulse: y(t)={1-[u(t-t1)-u(t-t2)]}sin(ω0t), where T / 20≤t2-t1≤T / 10 is the duration of the transient pulse.

[0088] Voltage spikes: Where A = 0.1 ~ 0.4, 0 ≤ t2-t1 ≤ 2T;

[0089] Voltage Notch: Where A = 0.1 ~ 0.4, 0 ≤ t2-t1 ≤ 2T;

[0090] Harmonics: y(t)=sin(ω0t)+A3 sin(3ω0t)+A5 sin(5ω0t)+A7 sin(7ω0t), where A3, A5, A7=0.05~0.3.

[0091] Under ideal conditions, the voltage and current signals of the power system are ideal sine waves, so the signals are sparse in the frequency domain, which meets the prerequisite for the application of compressed sensing theory. In reality, due to the application of a large number of nonlinear loads in the power system, the problem of power quality pollution is becoming increasingly serious. From the above definition, it can be seen that the power quality signal is not sparse in the time domain, but it is possible to become a sparse signal through a certain transformation domain. Taking Fourier transform as an example, the sparsity of the power quality signal is examined.

[0092] Figure 2-11 The time-frequency domain signals of normal voltage and various power quality disturbance signals. The figure (a) of each amplitude is the time domain representation, and the figure (b) is the Fourier transform frequency domain representation, which are two equivalent representations of signals in different domains. Figure 2-11 The various power quality disturbance signals represented by are not sparse in the time domain, but have only a few large coefficient values ​​and a large number of small coefficient values ​​in the frequency domain. Figure 3Taking the voltage swell signal as an example, near the base frequency, the signal spectrum amplitude is a large non-zero value. As the frequency increases, the spectrum amplitude drops sharply, and most of the spectrum amplitudes are approximately zero. The amplitude-frequency characteristics of other power quality signals are different, but they all have a trend of a sharp drop in spectrum amplitude as the frequency increases. In other words, the frequency domain signals all have a large number of small coefficient values. Therefore, the power quality signal can be compressed, which meets the definition of sparse signals and meets the prerequisites for the application of compressed sensing theory.

[0093] In a specific embodiment, in step 3), a random Gaussian measurement matrix is ​​used to perform signal sampling, and the matrix construction method is: the measurement matrix φ∈R M×N Each element of is independently subject to zero mean and variance Gaussian distribution, that is:

[0094]

[0095] In a specific embodiment, in step 4), the ROMP algorithm performs signal reconstruction, and the specific process is:

[0096] S41, calculate the initial margin r0 = u, the sparsity is K, the number of iterations t = 1, the index value set

[0097] S42, calculate the correlation coefficient V, and find the index value j corresponding to the K maximum values ​​from V and store it in J;

[0098]

[0099] S43, regularize the correlation coefficients of the atoms corresponding to the index values ​​in J, and save the regularized results in the set J0. The correlation coefficients of the atoms in the set satisfy the following formula:

[0100] |V(i)|≤2|V(j)|,i,j∈J

[0101] S44, Update support set Where Λ=Λ∪J0;

[0102] S45. Apply the following formula to obtain the reconstructed value of the sparse signal: And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new

[0103]

[0104]

[0105] S46. If |Λ|≥2K, stop the iteration, otherwise let r=r new, iteration number t=t+1, go to step S42.

[0106] From the above algorithm steps, it can be seen that only when the sparsity of the signal is known or properly estimated can the signal be further accurately reconstructed. Experiments show that if the estimated value of the sparsity of the signal has a large error, the iteration stop condition cannot be reached after multiple iterations, and the original sparse signal cannot be accurately reconstructed, resulting in a large reconstruction error.

[0107] On the basis of the regularization process of the ROMP algorithm, if the adaptive idea can be combined, the limitations of the regularization process can be overcome in the sparsity problem, and the number of selected atoms can be automatically adjusted in the iteration process to reconstruct the signal with unknown sparsity. The number of atoms is gradually increased by the conversion stage, and the same iterative process is divided into multiple stages. The variable step size is set instead of the number of selected atoms, and the size difference of the support set corresponding to the two adjacent stages is the current step size. With the increase of the step size and the increase of the support set, it is achieved that under the premise of unknown sparsity, the step size gradually approaches the true sparsity K, thereby accurately reconstructing the original sparse signal.

[0108] In another specific embodiment, the adaptive matching pursuit algorithm combines the regularization idea and the adaptive idea of ​​the ROMP algorithm, which ensures global optimization while improving the algorithm speed without the need to predict the sparsity. The algorithm implementation steps are as follows:

[0109] 4.1) Initial margin r0 = u. The initial step size is step_size, stage stage = 1, number of iterations t = 1, index value set

[0110] 4.2) If ||r||2≤ε1, stop the iteration and use the obtained atoms to reconstruct the signal; otherwise, go to step ③;

[0111] 4.3) Calculate the correlation coefficient V, and find the index value j corresponding to the maximum value of step_size from V and store it in J;

[0112]

[0113] 4.4) Regularize the correlation coefficients of the atoms corresponding to the index values ​​in J, and save the regularized results in the set J0. The correlation coefficients of the atoms in this set satisfy the following formula:

[0114] |V(i)|≤2|V(j)|,i,j∈J

[0115] 4.5) Update support set where Λ=Λ∪J0

[0116] 4.6) Apply the following formula to obtain the reconstructed value of the sparse signal And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new

[0117]

[0118]

[0119] 4.7) If ||r new -r||2≤ε2, then set stage=stage+1, step_size=step_size+1, and go to step 4.3), otherwise set r=r new , iteration number t=t+1, go to step 4.2).

[0120] The above adaptive matching pursuit reconstruction algorithm adaptively adjusts the step size by setting two thresholds ε1 and ε2. It does not need to use sparsity K as prior knowledge for signal reconstruction, avoiding the problem of multiple iterations without matching or overmatching due to the deviation of the estimated value of sparsity. ε1 and ε2 control the number of iterations and the stage transition threshold respectively. The regularization process ensures that the support set for accurate signal reconstruction can be obtained after at most K iterations, which greatly shortens the running time while ensuring the detection accuracy.

[0121] like Figure 12-16 As shown in the figure, the compression effect of the compression reconstruction algorithm based on adaptive matching pursuit on various noise-free power quality disturbance data is shown, and the statistical compression ratio, reconstruction error and other indicators are calculated. It should be noted that the three types of power quality data, normal voltage, voltage harmonics and voltage flicker, are approximately stable signals, which can ensure high compression ratio while reconstructing with high precision. The measurement dimension M = 32 is selected, and the other measurement dimensions are M = 128.

[0122] Fig.12 (a) is the original normal voltage signal of 10 cycles (sampling frequency 6400 Hz), with a total of 1280 points. The horizontal axis Time / sec represents the time unit in seconds, and the vertical axis Amp / pu represents the normalized voltage amplitude. Fig.12 (b) is the data compressed by the compressed sensing measurement matrix. The horizontal axis sample out represents the number of sampling points, and the vertical axis Amp / pu represents the amplitude of the compressed data. Fig.12 (c) is the normal voltage reconstruction signal based on the adaptive matching pursuit reconstruction algorithm. Fig.12 (d) is the error between the original signal and the reconstructed signal. After analysis, the amount of data after compressed sensing measurement is reduced from 1280 points to 32 points, the compression ratio is 40, and the reconstruction error is less than 0.01%.

[0123] Fig.13(a) is a 10-cycle voltage harmonic signal (sampling frequency 6400 Hz), with a total of 1280 points. The horizontal axis Time / sec represents the time unit in seconds, and the vertical axis Amp / pu represents the normalized voltage amplitude. Fig.13 (b) is the data compressed by the compressed sensing measurement matrix. The horizontal axis sample out represents the number of sampling points, and the vertical axis Amp / pu represents the amplitude of the compressed data. Fig.13 (c) is the voltage harmonic reconstruction signal based on the adaptive matching pursuit reconstruction algorithm. Fig.13 (d) is the error between the original signal and the reconstructed signal. After analysis, the amount of data after compressed sensing measurement is reduced from 1280 points to 32 points, the compression ratio is 40, and the reconstruction error is less than 0.01%.

[0124] Fig.14 (a) is the voltage swell signal (sampling frequency 6400 Hz), with a total of 1280 points. The horizontal axis Time / sec represents the time unit in seconds, and the vertical axis Amp / pu represents the normalized voltage amplitude. The voltage amplitude is 1.8 times the standard value between 0.04 seconds and 0.16 seconds. Fig.14 (b) is the data compressed by the compressed sensing measurement matrix. The horizontal axis sample out represents the number of sampling points, and the vertical axis Amp / pu represents the amplitude of the compressed data. Fig.14 (c) is the voltage swell reconstruction signal based on the adaptive matching pursuit reconstruction algorithm. Fig.14 (d) is the error between the original signal and the reconstructed signal. After analysis, the amount of data after compressed sensing measurement is reduced from 1280 points to 128 points, the compression ratio is 10, and the reconstruction error is 1.38%.

[0125] Fig.15 (a) is the voltage sag signal (sampling frequency 6400 Hz), with a total of 1280 points. The horizontal axis Time / sec represents the time unit in seconds, and the vertical axis Amp / pu represents the normalized voltage amplitude. The voltage amplitude is 0.4 times the standard value between 0.04 seconds and 0.16 seconds. Fig.15 (b) is the data compressed by the compressed sensing measurement matrix. The horizontal axis sample out represents the number of sampling points, and the vertical axis Amp / pu represents the amplitude of the compressed data. Fig.15 (c) is the voltage sag reconstructed signal based on the adaptive matching pursuit reconstruction algorithm. Fig.15 (d) is the error between the original signal and the reconstructed signal. After analysis, the amount of data after compressed sensing measurement is reduced from 1280 points to 128 points, the compression ratio is 10, and the reconstruction error is 2.24%.

[0126] Fig.16(a) is the voltage interruption signal (sampling frequency 6400 Hz), with a total of 1280 points. The horizontal axis Time / sec represents the time unit in seconds, and the vertical axis Amp / pu represents the normalized voltage amplitude. The voltage amplitude drops sharply between 0.04 seconds and 0.16 seconds. Fig.16 (b) is the data compressed by the compressed sensing measurement matrix. The horizontal axis sample out represents the number of sampling points, and the vertical axis Amp / pu represents the amplitude of the compressed data. Fig.16 (c) is the voltage interruption reconstruction signal based on the adaptive matching pursuit reconstruction algorithm. Fig.16 (d) is the error between the original signal and the reconstructed signal. After analysis, the amount of data after compressed sensing measurement is reduced from 1280 points to 128 points, the compression ratio is 10, and the reconstruction error is 3.05%.

[0127] From the above analysis and diagrams, it can be seen that the above method is applicable to various steady-state and disturbed power quality data. The compressed sensing method is efficient, simple and universal. The decompression method avoids the sparsity estimation of power quality disturbance signals and has practical value.

[0128] The embodiment of the present invention further discloses a source-load characteristic signal compression sensing system for a typical power usage scenario, including:

[0129] The first program module is used to obtain source-load characteristic signals of typical power usage scenarios;

[0130] The second program module is used to perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal;

[0131] A third program module is used for sampling the frequency domain signal through the measurement matrix;

[0132] The fourth program module is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal.

[0133] The source-load characteristic signal compression sensing system of the typical power usage scenario of the present invention corresponds to the above-mentioned sensing method and also has the advantages described in the above-mentioned sensing method.

[0134] The embodiment of the present invention also discloses a computer-readable storage medium, on which a computer program is stored, and the computer program executes the steps of the above method when it is run by a processor. The embodiment of the present invention also discloses a computer device, including a memory and a processor, on which a computer program is stored, and the computer program executes the steps of the above method when it is run by the processor. The present invention implements all or part of the process in the above embodiment method, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above method embodiment when it is executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable media include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc. The memory is used to store computer programs and / or modules. The processor implements various functions by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices, etc.

[0135] As shown in the present disclosure and claims, unless the context clearly indicates an exception, the words "a", "an", "a kind" and / or "the" do not specifically refer to the singular, but may also include the plural. The words "first", "second" and similar words used in the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "include" or "comprise" mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0136] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for compressing the source-load characteristic signal of a typical power usage scenario, characterized in that: Includes steps: 1) Obtain source-load characteristic signals of typical power consumption scenarios; 2) Perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal; 3) Sampling the frequency domain signal through the measurement matrix; 4) Perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal; In step 4), an adaptive matching pursuit algorithm is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal; The specific process of the adaptive matching pursuit algorithm is as follows: 4.1) Initial margin r0 = u, initial step size step_size, stage stage = 1, number of iterations t = 1, index value set 4.2) If ||r||2≤ε1, stop the iteration and use the obtained atoms to reconstruct the signal; Otherwise go to step 4.3); 4.3) Calculate the correlation coefficient V, and find the index value j corresponding to the maximum value of step_size from V and store it in J; 4.4) Regularize the correlation coefficients of atoms corresponding to index values ​​in J, and save the regularized results in set J0. The correlation coefficients of atoms in this set satisfy the following formula: |V(i)|≤2|V(j)|,i,j∈J 4.5) Update support set Where Λ=Λ∪J0; 4.6) Apply the following formula to obtain the reconstructed value of the sparse signal And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new 4.7) If ||r new -r||2≤ε2, then set stage=stage+1, step_size=step_size+1, and go to step 4.3), otherwise set r=r new , iteration number t=t+1, go to step 4.2).

2. The method for compressing source-load characteristic signals of a typical power usage scenario according to claim 1 is characterized in that: In step 4), the ROMP algorithm is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal. The specific process is as follows: S41, calculate the initial margin r0 = u, the sparsity is K, the number of iterations t = 1, the index value set S42, calculate the correlation coefficient V, and find the index value j corresponding to the K maximum values ​​from V and store it in J; in is the support set; S43, regularize the correlation coefficients of atoms corresponding to the index values ​​in J, and save the regularized results in a set J0, where the correlation coefficients of atoms in the set satisfy the following formula: |V(i)|≤2|V(j)|,i,j∈J S44, Update support set Where Λ=Λ∪J0; S45. Apply the following formula to obtain the reconstructed value of the sparse signal: And update the redundant value to get the reconstructed value of the sparse signal And update the redundant value r new S46, if | Λ | ≥2K, then stop the iteration, otherwise let r=r new , iteration number t=t+1, go to step S42.

3. The method for compressing source-load characteristic signals of a typical power usage scenario according to any one of claims 1 to 2, characterized in that: In step 1), the source-load characteristic signal includes one or more of voltage sag, voltage swell, voltage interruption, transient oscillation, transient pulse, voltage spike, voltage notch, voltage flicker and harmonic wave.

4. The method for compressing source-load characteristic signals of a typical power usage scenario according to any one of claims 1 to 2, characterized in that: In step 2), the source-charge characteristic signal is sparsely transformed by wavelet transform.

5. The method for compressing source-load characteristic signals of a typical power usage scenario according to any one of claims 1 to 2, characterized in that: In step 3), signal sampling is performed by means of a random Gaussian measurement matrix.

6. A system for compressing the characteristic signal of a source-load in a typical power usage scenario, used to execute the steps of the method for compressing the characteristic signal of a source-load in a typical power usage scenario as claimed in any one of claims 1 to 5, characterized in that: include: The first program module is used to obtain source-load characteristic signals of typical power usage scenarios; The second program module is used to perform sparse transformation on the source-charge characteristic signal to obtain the corresponding frequency domain signal; A third program module is used for sampling the frequency domain signal through the measurement matrix; The fourth program module is used to perform adaptive matching pursuit reconstruction on the sampled frequency domain signal to obtain a reconstructed signal.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program executes the steps of the method according to any one of claims 1 to 5 when executed by a processor.

8. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: The computer program executes the steps of the method according to any one of claims 1 to 5 when executed by a processor.

Images