Rapid harmonic wave identification and monitoring method and device based on broadband electrical parameters

Abstract

The embodiment of the invention provides a harmonic wave rapid identification and monitoring method and device based on broadband electrical parameters, and relates to the technical field of signal processing. The specific implementation mode of the method comprises the following steps: acquiring a to-be-detected analog signal; preprocessing the analog signal to be detected to obtain a preprocessed signal; performing time-frequency analysis on the preprocessed signal to obtain a time-frequency analysis feature; inputting the time-frequency analysis characteristics into a disturbance signal identification model, and identifying the disturbance type and disturbance percentage of the disturbance signal according to the output of the disturbance signal identification model; wherein the disturbance signal identification model adopts a fuzzy SOM neural network. According to the embodiment, monitoring and identification of harmonic waves and disturbance can be accurately carried out in real time, an identification result is visualized, the accuracy of disturbance signal identification is improved, a disturbance source can be conveniently and visually positioned, a power system is further controlled in real time, the reliability, stability and efficiency of the power system are improved, normal operation of equipment is guaranteed, and the power system is protected. And the potential risk and loss are reduced.

Description

Technical Field

[0001] The present disclosure relates to the field of signal processing technology, and in particular to a method and device for rapid harmonic identification and monitoring based on broadband electrical parameters. Background Art

[0002] With the rapid development of AC power grid, electricity has entered thousands of households. Due to the emergence of various safety issues in the process of electricity use, qualified "power quality" has become the top priority in the development of power technology. Qualified power quality means that the power provided to sensitive equipment and the grounding system set are suitable for the normal operation of the equipment. However, in actual applications, there are a large number of high-power, nonlinear, and unbalanced loads, which lead to grid pollution and cause a series of unqualified power quality problems such as harmonics, interharmonics, three-phase imbalance, voltage fluctuations and flickers, voltage drops and short-term interruptions.

[0003] In the power system, the presence of harmonics may cause problems such as overload, damage or unstable operation of power equipment. Harmonic identification technology is often used in existing power systems. By judging whether the power system meets regional standards and specifications, and analyzing the spectrum characteristics and waveform changes of harmonics, the source and source of potential harmonics (for example, it may be a certain power equipment, or the interconnection between the power system and other power systems, etc.), and fault diagnosis and location, etc., appropriate measures can be taken to adjust and optimize, ensure the normal operation of power equipment, reduce harmonic interference, and promptly eliminate system faults, reduce maintenance costs, and ensure the safety and stability of the power system.

[0004] However, due to the variety of actual operating environments of the power system, it may be necessary to adapt to voltages or currents of multiple frequencies, that is, the applicable range of power equipment is wide-band voltage or wide-band current (for example, the motor drive system needs to operate within the frequency range of 50Hz to 1000Hz to adapt to different usage requirements), and the existing harmonic identification technology has a single recognition frequency and cannot accurately identify harmonics in a wide frequency range, resulting in poor reliability and stability of the power system and high operating risks. Summary of the invention

[0005] In view of this, the embodiments of the present disclosure provide a method for rapid harmonic identification and monitoring based on wide-band electrical parameters, which can solve the problem that the existing harmonic identification technology has a single identification range and cannot accurately identify harmonics in a wide frequency range, resulting in poor reliability and stability of the power system and high operating risks.

[0006] To achieve the above objectives, according to one aspect of the present disclosure, a method for rapid harmonic identification and monitoring based on broadband electrical parameters is provided, comprising:

[0007] Collect the analog signal to be tested;

[0008] Preprocessing the analog signal to be tested to obtain a preprocessed signal;

[0009] Performing time-frequency analysis on the preprocessed signal to obtain time-frequency analysis features;

[0010] The time-frequency analysis features are input into a disturbance signal identification model, and the disturbance type and disturbance percentage of the disturbance signal are identified according to the output of the disturbance signal identification model; wherein the disturbance signal identification model adopts a fuzzy SOM neural network.

[0011] One or more technical solutions provided in the embodiments of the present application collect signals from signal sources that need to be analyzed and monitored, convert the collected analog signals to be measured into digital signals to be measured, perform denoising on the digital signals to be measured, use generalized S transform to perform harmonic analysis, draw and display time-frequency diagrams and spectrum diagrams, obtain the distribution of the effective values ​​of each frequency over time and the effective values ​​of the harmonic signals at each frequency, and intuitively monitor abnormal points; then use the square of the time amplitude, the mean, and the characteristic frequency points obtained after the generalized S transform as input samples, use the fuzzy SOM neural network for identification, and determine the disturbance type and disturbance percentage of the disturbance signal; comprehensively consider the time-frequency diagram, the spectrum diagram, the disturbance type and the disturbance percentage of the disturbance signal to control the power equipment to be measured, so as to avoid unnecessary noise points caused by noise signals during harmonic analysis, affect the accuracy of harmonic analysis and monitoring, improve the accuracy of disturbance signal identification, and thus improve the reliability, stability and efficiency of the power system, ensure the normal operation of the equipment, and reduce potential risks and losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Further details, features and advantages of the present disclosure are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which:

[0013] Figure 1 A flow chart of a method for rapid harmonic identification and monitoring based on broadband electrical parameters according to an exemplary embodiment of the present disclosure is shown;

[0014] Figure 2 A flow chart showing a method for preprocessing a signal according to an exemplary embodiment of the present disclosure is shown;

[0015] Figure 3 A flow chart showing a method for determining a signal denoising model according to an exemplary embodiment of the present disclosure is shown;

[0016] Figure 4 A flow chart of a method for extracting time-frequency analysis features according to an exemplary embodiment of the present disclosure is shown;

[0017] Figure 5A flow chart showing a method for determining a disturbance signal identification model according to an exemplary embodiment of the present disclosure is shown;

[0018] Figure 6 A schematic block diagram of a harmonic rapid identification and monitoring device based on broadband electrical parameters according to an exemplary embodiment of the present disclosure is shown;

[0019] Figure 7 A structural block diagram of an exemplary electronic device that can be used to implement the embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0020] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0021] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0022] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to". The term "based on" means "based at least in part on". The term "in the embodiment of the present disclosure" means "at least one embodiment"; the term "another exemplary embodiment" means "at least one other embodiment". Relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0023] It should be noted that the modifications of "one" and "plurality" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0024] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0025] Aspects of the present disclosure are described below with reference to the accompanying drawings.

[0026] Figure 1A flowchart of a method for rapid harmonic identification and monitoring based on broadband electrical parameters according to an exemplary embodiment of the present disclosure is shown. Figure 1 As shown, the harmonic rapid identification and monitoring method based on broadband electrical parameters disclosed in the present invention includes the following steps:

[0027] Step 101, collecting an analog signal to be tested.

[0028] In the disclosed embodiment, the analog signal to be tested of the signal source is collected. The signal source may be a certain circuit in the power equipment or circuit system to be monitored. The analog signal to be tested may be an analog voltage signal or an analog current signal at different frequencies, which can be selectively set according to the actual monitoring scenario. For example, the signal source is a motor drive system, and the analog voltage signal of the motor drive system in the frequency range of 50 Hz to 1000 Hz is collected as the analog signal to be tested.

[0029] Furthermore, the signal source is connected in parallel to the power grid voltage monitor, and the power grid voltage monitor is used to collect the analog signal to be measured of the signal source.

[0030] Step 102: preprocess the analog signal to be tested to obtain a preprocessed signal.

[0031] In the embodiment of the present disclosure, in order to facilitate the analysis of the analog signal to be measured, it is necessary to perform pre-processing such as digital conversion and denoising on the analog signal to be measured, so as to extract effective information and further accurately identify the harmonic disturbance.

[0032] Furthermore, if Figure 2 As shown, the signal preprocessing method disclosed in the present invention comprises the following steps:

[0033] Step 201: convert the analog signal to be tested into a digital signal to be tested.

[0034] In the disclosed embodiment, the conversion of digital signals can be achieved through an analog-to-digital converter (A / D). The collected analog signal to be measured is input into the analog-to-digital converter, the amplitude of the analog signal to be measured is graded, and the level encoding of each sampling point is quantized to obtain the corresponding signal sequence of the digital signal to be measured, thereby realizing the conversion of analog signals to digital signals. The sequence values ​​of the digital signal are discrete in time and amplitude.

[0035] Step 202: input the digital signal to be tested into a signal denoising model to obtain a preprocessed signal.

[0036] Noise signals are usually generated by electrical signals transmitted through circuits, electronic devices, etc., and are evenly distributed on each frequency. Therefore, in the embodiments of the present disclosure, a signal denoising model is used to denoise the digital signal to be measured to prevent the noise signal in the signal to be measured from continuously propagating and expanding subsequent signal processing errors. Accordingly, the denoised preprocessed signal is a(p) (p=0,T,2T,...,(N-1)T); wherein T is the sampling period, and each sampling period includes N sampling points.

[0037] Furthermore, the model algorithm of the signal denoising model can be selectively set as needed. For example, the model algorithm of the signal denoising model adopts an LMS (Least Mean Square) adaptive filtering algorithm, such as Figure 3 As shown, the method for determining the signal denoising model disclosed in the present invention includes the following steps:

[0038] Step 301: Acquire a sample digital signal and a corresponding expected signal.

[0039] In the embodiment of the present disclosure, a plurality of sample digital signals X(n)=[x(n), x(n-1), ..., x(nL)] are obtained from a pre-constructed sample data set. T , where x(n) represents the code value of the nth sampling point and the corresponding expected signal is d(n).

[0040] Step 302: input the sample digital signal into a signal denoising training model, take the error signal between the output of the signal denoising training model and the corresponding expected signal as the minimum as the training target, and iteratively train the signal denoising training model using a gradient descent method.

[0041] In the disclosed embodiment, the sample digital signal is used as the input of the signal denoising model to be trained, and the corresponding expected signal is used as the output of the signal denoising model to be trained. The signal denoising model to be trained is iteratively trained until the difference between the output of the signal denoising model to be trained and the expected signal is minimized, thereby obtaining the corresponding target weight vector.

[0042] Step 3021, initializing the weight vector of the signal denoising model to be trained.

[0043] In the embodiment of the present disclosure, the model function of the signal denoising model to be trained is shown in the following formula (1):

[0044]

[0045] In the above formula, W(n) is the weight vector, where W(n)=[w0(n),w1(n),...,w L (n)] T , wi (n) is the weight coefficient of the LMS adaptive filter, and L is the order of the LMS adaptive filter.

[0046] Furthermore, the weight coefficients w of the weight vector W(n) are randomly i (n) Assign initial values ​​to initialize the signal denoising model to be trained.

[0047] Step 3022: input the sample digital signal into the initialized signal denoising model to be trained, and determine the output signal corresponding to the sample digital signal according to the actual output of the initialized signal denoising model to be trained.

[0048] In the embodiment of the present disclosure, the output signal is represented by y(n), y(n)=W T (n)X(n).

[0049] Step 3023: Calculate the difference between the expected signal and the output signal to obtain an error signal.

[0050] In the embodiment of the present disclosure, the error signal e(n) is expressed as follows:

[0051] e(n)=d(n)-y(n)=d(n)-W T (n)X(n) (2)

[0052] Step 3024, determine whether the mean square value of the error signal is the minimum value, if yes, go to step 3025; if no, go to step 3026.

[0053] In the embodiment of the present disclosure, the signal denoising model to be trained is based on the mean square value E[e 2 (n)] is minimized as the objective function, as shown in the following formula (3):

[0054] min J(n)=E[e 2 (n)] (3)

[0055] Among them, E[e 2 (n)] is the mean of the sum of squares of the error signals at each sampling point.

[0056] Step 3025: determine the weight vector as the target weight vector.

[0057] In the embodiment of the present disclosure, the change of J(n) over time characterizes the convergence speed and convergence accuracy of the LMS adaptive filter. When J(n) is minimum, the weight vector of the LMS adaptive filter is optimal. That is, when J(n) is minimum, the obtained weight vector is the target weight vector W opt (n).

[0058] Step 3026: Use the gradient descent method to adjust the weight vector of the signal denoising model to be trained.

[0059] In the embodiment of the present disclosure, when the mean square value J(n) of the error signal is not the minimum value, the gradient descent method is used to obtain the performance surface. The weight vector W(n) is adjusted in the negative gradient direction to search for the minimum point of the performance surface. The weight vector adjustment amount each time is As shown in the following formula (4):

[0060]

[0061] Further, according to the weight vector adjustment amount The weight vector W′(n) after each adjustment is determined as shown in the following formula (5):

[0062]

[0063] In the above formula, μ is the step size factor, which can be selectively set as needed, where μ>0.

[0064] Step 3027, input the sample digital signal into the adjusted signal denoising to be trained model to obtain an updated output signal, and go to step 3023.

[0065] In the disclosed embodiment, the sample digital signal is input into the adjusted signal denoising model to be trained to obtain an updated output signal, and the error comparison is performed again until the minimum value of J(n) is obtained.

[0066] Step 303: Determine the signal denoising model according to the target weight vector obtained by the iterative training.

[0067] In the embodiment of the present disclosure, the target weight vector W is used opt (n) Update the model parameters of the signal denoising model to be trained to obtain a trained signal denoising model.

[0068] In the embodiment of the present disclosure, through the method for determining the signal denoising model of the present disclosure, the signal denoising model to be trained is continuously trained using a sample data set until the difference between the model output and the expected signal is minimized and the optimal weight vector is obtained, and the signal denoising model of the present disclosure can be determined. In the subsequent use process, the signal denoising model can be used to pre-process the digital signal to obtain a high-quality signal after denoising, and then the denoised signal can be analyzed to accurately identify the disturbance signal in the analog signal to be measured.

[0069] Step 103: Perform time-frequency analysis on the preprocessed signal to obtain time-frequency analysis features.

[0070] In the embodiments of the present disclosure, by performing time-frequency analysis on the preprocessed signal, on the one hand, the time-frequency analysis results are visualized, so that the monitoring personnel can quickly locate the harmonic points; on the other hand, the time-frequency analysis features are extracted from the time-frequency analysis results so as to be subsequently input into the disturbance signal identification model to accurately identify the type of disturbance signal and other information, so that the monitoring personnel can control the power system by integrating various information to ensure the safe and stable operation of the power system.

[0071] Furthermore, if Figure 4 As shown, the time-frequency analysis feature extraction method disclosed in the present invention includes the following steps:

[0072] Step 401, performing time-frequency analysis on the preprocessed signal to obtain a time-frequency analysis result; wherein the time-frequency analysis result includes time-frequency analysis values ​​at multiple times and multiple frequencies.

[0073] In the embodiment of the present disclosure, the analysis method of time-frequency analysis can adopt normal S transform or fast Fourier S transform (also called S transform of fast FT algorithm), which can be selected according to actual computing resources and requirements.

[0074] Further, the normal S transform result S obtained by performing a normal S transform on the preprocessed signal a(p) is T [m,q] is shown in the following formula (6):

[0075]

[0076] In the above formula, N is the number of sampling periods; m is time; q is frequency, m,q=0,1,...,N-1; S T [m,q] represents the normal S-transform result at the mth time and qth frequency.

[0077] Alternatively, the fast Fourier S transform result S obtained by performing the fast Fourier S transform method on the preprocessed signal a(p) is FT [m,q] is shown in the following formula (7):

[0078]

[0079] In the above formula, A(l) is the sequence obtained by performing discrete Fourier transform on a(p); wherein A(0) is the magnitude of the DC signal component in a(p), A(1) is the magnitude of the fundamental frequency signal in a(p), A(2) is the magnitude of the fundamental frequency signal at 2 times, A(3) is the magnitude of the fundamental frequency signal at 3 times, and so on. In order to explain more clearly, the above formula (7) includes steps ①-⑦, specifically:

[0080] Since q is used as the denominator in the above formula (7), it is necessary to explain q=0 and q≠0 separately.

[0081] Calculate S when q = 0 FT The first column of [m,q]

[0082] ① Calculate the discrete Fourier transform A(l) of a(kT) when q≠0: A(l)=A[k / (NT)]. In which, the first column k is the coefficient of frequency q, the fundamental frequency is 1 / (NT), k / (NT) is the frequency value of each fundamental frequency component, and A[k / (NT)] is the amplitude of each fundamental frequency component (that is, the signal strength of each k / (NT)). Different symbols are used to distinguish them. The original signal is l, and the transformed signal is m.

[0083] ②Translate A[k / (NT)] to A[(k+m) / (NT)].

[0084] ③Calculate the frequency domain Fourier transform of the Gaussian window function

[0085] ④ Calculate the product A[(k+m) / (NT)]·W[m,q] of the frequency domain extended dimension array A[(k+m) / (NT)] with m=0,1,...,N-1 and the frequency domain Gaussian window function Fourier transform W[m,q].

[0086] ⑤Perform discrete inverse Fourier transform on A[(k+m) / (NT)]·W[m,q].

[0087] ⑥ Repeat steps ④ and ⑤ until all frequency points are calculated.

[0088] Furthermore, S T [m,q]=S FT [m,q], therefore, the time-frequency analysis result S[m,q]=S T [m,q]=S FT [m,q], including the frequency components q at different times m, that is, S[m,q] is a two-dimensional matrix, the columns of the two-dimensional matrix are time and behavior frequencies, each column vector is the size of the component at different frequencies at the same time, and each row vector is the size of the component at different times at the same frequency.

[0089] In the disclosed embodiment, after obtaining the video analysis results, the time-frequency analysis results can also be used to draw the time-frequency diagram and spectrum diagram of the analog signal to be tested, thereby providing a comprehensive reference for monitoring personnel to locate abnormal points.

[0090] Furthermore, when directly using the time-frequency analysis results to draw the time-frequency diagram and the spectrum diagram, once there is a disturbance signal with a similar frequency, it is very easy to overlap the frequency, which is not convenient for direct observation. Therefore, in order to improve the time-frequency resolution of the signal and ensure the intuitive display effect of the harmonics in the time-frequency diagram and the spectrum diagram, preferably, an adjustable factor is introduced into the fast Fourier S transform to obtain a generalized S transform result as the time-frequency analysis result, as shown in the following formula (9):

[0091]

[0092] In the above formula, when λ>1, the time resolution of the signal is improved;

[0093] When 0<λ<1, the frequency resolution is improved;

[0094] When λ=0, it is equivalent to the fast Fourier S transform.

[0095] Therefore, by adjusting the value of λ, the harmonic distribution of the signal can be more comprehensively reflected, and the results of different factors can be presented in the same spectrum.

[0096] Furthermore, the time-frequency analysis result S[m,q]=S G [m,q].

[0097] Step 402, calculate the average of the sum of squares of the time-frequency analysis values ​​at different times under the same frequency to obtain the mean features under multiple frequencies.

[0098] In the embodiment of the present disclosure, for the same q, the average value of the sum of squares of S[m,q] under different m is calculated to obtain the mean feature under each q, that is, for each column in S[m,q], the average value of the sum of squares of each element is calculated; wherein q is the characteristic frequency point of the mean feature.

[0099] Step 403, calculating the sum of squares of time-frequency analysis values ​​of different frequencies at the same time, and obtaining time amplitude features at multiple times.

[0100] In the embodiment of the present disclosure, for the same m, the average value of the sum of squares of S[m,q] under different q is calculated to obtain the time amplitude feature under each m, that is, for each row in S[m,q], the average value of the sum of squares of each element is calculated; wherein m is the characteristic frequency point of the time amplitude feature.

[0101] Step 404: concatenate the characteristic frequency points and the mean value feature and / or the time amplitude feature to obtain the time-frequency analysis feature.

[0102] In the disclosed embodiment, the corresponding relationship of the mean value features under different q and the corresponding relationship of the time amplitude features under different m are combined to obtain various time-frequency analysis features.

[0103] In the embodiment of the present disclosure, through the time-frequency analysis feature extraction method of the present disclosure, the preprocessed signal is subjected to time-frequency analysis using S transform, and the time-frequency analysis features are extracted from the S transform result, which can effectively characterize the analog signal to be tested, so as to facilitate the accurate identification of the subsequent disturbance signal.

[0104] Step 104: input the time-frequency analysis feature into a disturbance signal identification model, and identify the disturbance type and disturbance percentage of the disturbance signal according to the output of the disturbance signal identification model.

[0105] In the embodiment of the present disclosure, the disturbance signal identification model adopts a fuzzy SOM neural network, including an input layer, a competition layer, a fuzzy recognition layer and an output layer, each layer includes multiple neurons, the number of neurons in the input layer is the same as the number of features of the time-frequency analysis features, which is equal to (m+q), that is, including mean features under different q and time amplitude features under different m; the number of neurons out in the output layer is the same as the number of disturbance types, which can be selectively set as needed; the competition layer and the fuzzy recognition layer are hidden layers, and the number of neurons can be selectively set as needed.

[0106] Furthermore, if Figure 5 As shown, the method for determining the disturbance signal identification model disclosed in the present invention includes the following steps:

[0107] Step 501, determine the number of neurons in the rectangular network of the competition layer, and construct a disturbance signal recognition model to be trained.

[0108] In the disclosed embodiment, the neurons of the competition layer are distributed in a grid, and the number of neurons is equal to the product of the number of row units p1 and the number of column units p2 of the grid, that is, the competition layer includes p2 hidden layers, each hidden layer includes p1 neurons, and p1 and p2 can be the same or different.

[0109] Furthermore, p1 and p2 can be determined using empirical formulas, specifically:

[0110] Empirical formula 1: p1×p2>b, where b is the number of disturbance types, which can be selectively set according to the actual analysis scenario. For example, if the disturbance types include no disturbance, voltage swell, voltage sag, short power outage, pulse transient disturbance, high-frequency oscillation, and voltage flicker, then b=7.

[0111] Empirical formula 2: Wherein, z is a constant between [1,10] and can be selectively set according to the actual analysis scenario.

[0112] Alternatively, p1 and p2 can be determined using an exhaustive method by constructing multiple combinations of p1 and p2, training different ones separately, and selecting the grid distribution with the best model performance.

[0113] Furthermore, the competition layer is determined by using the determined p1 and p2, and is combined with the input layer, the output layer and the fuzzy recognition layer to obtain the disturbance signal recognition model to be trained.

[0114] Step 502: Obtain sample time-frequency analysis features and corresponding expected types and expected percentages.

[0115] In the disclosed embodiment, a plurality of sample time-frequency analysis features, corresponding expected types and expected percentages are obtained from a pre-constructed sample data set.

[0116] Step 503: Input the sample time-frequency analysis features into the disturbance signal identification model to be trained, take the minimum distance between the output of the disturbance signal identification model to be trained and the corresponding expected type and expected percentage as the training goal, and use the gradient descent method to iteratively train the disturbance signal identification model to be trained.

[0117] In the disclosed embodiment, the sample time-frequency analysis features are used as the input of the disturbance signal identification model to be trained, and the corresponding expected type and expected percentage are used as the output of the disturbance signal identification model to be trained. The disturbance signal identification model to be trained is iteratively trained until the distance difference between the output of the disturbance signal identification model to be trained and the expected type and expected percentage is minimized, and the corresponding network parameters are obtained.

[0118] Furthermore, the actual output of the disturbance signal identification model to be trained is g, the corresponding expected type is the i-th type, and the expected percentage is G i , calculate g and G i The Euclidean distance difference between them is taken as the training target, and the gradient descent method is used to continuously perform iterative training. The objective function of the disturbance signal identification model to be trained is shown in the following formula (8):

[0119]

[0120] Where K is the total number of perturbation types.

[0121] Furthermore, the fuzzy recognition layer and the output layer have the same number of neurons, which are used to calculate the distance difference between the output of the competition layer and each disturbance type, and transmit each distance difference to the output layer, which determines the disturbance percentage of each disturbance type according to the distance difference and outputs it in sequence; wherein the sum of the disturbance percentages of each disturbance type is 1.

[0122] Step 504: Determine the disturbance signal identification model according to the network parameters obtained by the iterative training.

[0123] In the disclosed embodiment, the network parameters obtained by iterative training are used to update the network parameters of the disturbance signal identification model to be trained, so as to obtain a trained disturbance signal identification model.

[0124] Furthermore, the pre-constructed verification data set can be used to perform network verification on the disturbance signal identification model to be trained after each iteration to determine whether the accuracy of the model meets the requirements, so as to improve the identification accuracy of the disturbance signal identification model.

[0125] In the embodiment of the present disclosure, through the determination method of the disturbance signal identification model of the present disclosure, the disturbance signal identification model to be trained is continuously trained using a sample data set until the difference between the model output and the expected type is minimized, and the optimal network parameters are obtained to determine the disturbance signal identification model of the present disclosure. Therefore, in the subsequent use process, the disturbance signal identification model can be used to identify the time-frequency analysis features, identify the disturbance signal in the simulated signal to be tested, and qualitatively and quantitatively analyze the disturbance situation in the simulated signal to accurately reflect the operation of the power system.

[0126] In the embodiments of the present disclosure, through the harmonic rapid identification and monitoring method based on broadband electrical parameters of the present disclosure, broadband voltage or current signals in the power system can be monitored and identified in real time, thereby quickly locating problems in the power system, providing accurate identification results for monitoring personnel, and reducing potential risks and losses.

[0127] In the embodiment of the present disclosure, after determining the disturbance type and disturbance percentage of the disturbance signal, the time-frequency diagram and spectrum diagram of the simulated signal to be tested are drawn using the time-frequency analysis result and the disturbance type and disturbance percentage of the disturbance signal, thereby providing a comprehensive reference for the monitoring personnel, so that the monitoring personnel can judge whether there is an abnormality in the power system according to the diagram, so as to control the operation of the power system and ensure the safety of the power system. Accordingly, the method for identifying the disturbance signal of the present disclosure also includes:

[0128] Draw a time-frequency diagram and a spectrum diagram according to the time-frequency analysis results, the disturbance type and the disturbance percentage of the disturbance signal;

[0129] According to the time-frequency diagram and the spectrum diagram, the operation of the signal source of the analog signal to be measured is controlled. Specifically:

[0130] Furthermore, the time-frequency diagram can display the frequency components at different times, and set different display colors according to the actual analysis standards to highlight the risk points, so that the time period and frequency segment where the disturbance signal appears can be observed intuitively. For example, according to the different degrees to which the frequency component exceeds the component threshold, it can be set to a warning color of different depths; the spectrum diagram can display the frequency size of each disturbance signal, so that the frequency value of the disturbance signal can be observed intuitively, and by adjusting λ, the phenomenon of poor feature extraction caused by frequency overlap can be avoided to a large extent, so as to accurately locate the abnormal harmonic frequency. Among them, the disturbance type and disturbance percentage of the disturbance signal can be displayed in the time-frequency diagram and the spectrum diagram. For example, the additional pop-up window in the time-frequency diagram and the spectrum diagram shows: the current circuit is normal, or the current circuit has disturbances: oscillation 47%, flicker 53%.

[0131] Furthermore, monitoring personnel can use time-frequency graphs and spectrum graphs to analyze and monitor the power system in real time, thereby locating the source of disturbances, achieving real-time monitoring and analysis of harmonics, and then eliminating anomalies to ensure the safe operation of the power system.

[0132] Figure 6 is a schematic diagram of the main modules of the harmonic rapid identification and monitoring device based on broadband electrical parameters according to an embodiment of the present disclosure, such as Figure 6 As shown, the harmonic rapid identification and monitoring device 600 based on broadband electrical parameters of the present disclosure includes:

[0133] The acquisition module 601 is used to acquire the analog signal to be tested.

[0134] The preprocessing module 602 is used to preprocess the analog signal to be tested to obtain a preprocessed signal.

[0135] The analysis module 603 is used to perform time-frequency analysis on the preprocessed signal to obtain time-frequency analysis features.

[0136] The identification module 604 is used to input the time-frequency analysis features into the disturbance signal identification model, and identify the disturbance type and disturbance percentage of the disturbance signal according to the output of the disturbance signal identification model; wherein the disturbance signal identification model adopts a fuzzy SOM neural network.

[0137] The exemplary embodiment of the present disclosure also provides an electronic device, comprising: at least one processor; and a memory connected to the at least one processor in communication. The memory stores a computer program that can be executed by the at least one processor, and the computer program is used to cause the electronic device to perform the method according to the embodiment of the present disclosure when executed by the at least one processor.

[0138] The exemplary embodiments of the present disclosure also provide a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is used to cause the computer to perform the method according to the embodiments of the present disclosure.

[0139] The exemplary embodiments of the present disclosure further provide a computer program product, including a computer program, wherein when the computer program is executed by a processor of a computer, the computer is used to enable the computer to perform the method according to the embodiments of the present disclosure.

[0140] refer to Figure 7 , a block diagram of an electronic device 700 that can be used as a server or client of the present disclosure will now be described, which is an example of a hardware device that can be applied to various aspects of the present disclosure. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0141] like Figure 7 As shown, the electronic device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the device 700 can also be stored. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0142] A plurality of components in the electronic device 700 are connected to the I / O interface 705, including: an input unit 706, an output unit 707, a storage unit 708, and a communication unit 709. The input unit 706 may be any type of device capable of inputting information to the electronic device 700, and the input unit 706 may receive input digital or character information, and generate key signal inputs related to user settings and / or function control of the electronic device. The output unit 707 may be any type of device capable of presenting information, and may include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 708 may include but is not limited to a disk, an optical disk. The communication unit 709 allows the electronic device 700 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and may include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver, and / or a chipset, such as a Bluetooth™ device, a WiFi device, a WiMax device, a cellular communication device, and / or the like.

[0143] The computing unit 701 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 701 performs the various methods and processes described above. For example, in some embodiments, Figures 1 to 5 The method can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 700 via the ROM 702 and / or the communication unit 709. In some embodiments, the computing unit 701 can be configured to execute by any other suitable means (for example, by means of firmware). Figures 1 to 5 method.

[0144] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0145] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0146] As used in this disclosure, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0147] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0148] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0149] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

Claims

1. A method for rapid identification and monitoring of harmonics based on broadband electrical parameters, characterized in that: include: Collect the analog signal to be tested; Preprocessing the analog signal to be tested to obtain a preprocessed signal; Performing time-frequency analysis on the preprocessed signal to obtain time-frequency analysis features; The time-frequency analysis features are input into a disturbance signal identification model, and the disturbance type and disturbance percentage of the disturbance signal are identified according to the output of the disturbance signal identification model; wherein the disturbance signal identification model adopts a fuzzy SOM neural network.

2. The method according to claim 1, characterized in that The performing time-frequency analysis on the preprocessed signal to obtain a time-frequency analysis feature includes: Performing time-frequency analysis on the preprocessed signal to obtain a time-frequency analysis result; wherein the time-frequency analysis result includes time-frequency analysis values ​​at multiple times and multiple frequencies; Calculate the average of the sum of squares of the time-frequency analysis values ​​at different times under the same frequency to obtain the mean characteristics under multiple frequencies; Calculate the sum of squares of the time-frequency analysis values ​​of different frequencies at the same time to obtain the time amplitude characteristics at multiple times; Count each frequency as a characteristic frequency point; The mean feature, the time amplitude feature and the characteristic frequency point are concatenated to obtain the time-frequency analysis feature.

3. The method according to claim 2, characterized in that The analysis method of the time-frequency analysis is generalized S transform.

4. The method according to claim 1, characterized in that Also includes: Draw a time-frequency diagram and a spectrum diagram according to the time-frequency analysis results, the disturbance type and the disturbance percentage of the disturbance signal; The operation of the signal source of the analog signal to be measured is controlled according to the time-frequency diagram and the frequency spectrum diagram.

5. The method according to claim 1, characterized in that The disturbance signal identification model includes an input layer, a competition layer, a fuzzy recognition layer and an output layer; the determination of the disturbance signal identification model includes: Determine the number of neurons in the rectangular network of the competition layer and construct a disturbance signal recognition model to be trained; Obtain sample time-frequency analysis characteristics and corresponding expected types and expected percentages; Inputting the sample time-frequency analysis features into the disturbance signal identification model to be trained, taking the minimum distance between the output of the disturbance signal identification model to be trained and the corresponding expected type and expected percentage as the training goal, and iteratively training the disturbance signal identification model to be trained using the gradient descent method; The disturbance signal identification model is determined according to the network parameters obtained by the iterative training.

6. The method according to claim 5, characterized in that The number of neurons in the rectangular network of the competition layer is determined by using an empirical formula method or an exhaustive method.

7. The method according to claim 1, characterized in that The preprocessing of the analog signal to be tested to obtain a preprocessed signal includes: Converting the analog signal to be tested into a digital signal to be tested; The digital signal to be tested is input into a signal denoising model to obtain a preprocessed signal.

8. The method according to claim 7, characterized in that The determination of the signal denoising model includes: Acquire a sample digital signal and a corresponding expected signal; Input the sample digital signal into a signal denoising training model, take the error signal between the output of the signal denoising training model and the corresponding expected signal as the minimum as the training target, and iteratively train the signal denoising training model using a gradient descent method; The signal denoising model is determined according to the target weight vector obtained by the iterative training.

9. The method according to claim 8, characterized in that The step of inputting the sample digital signal into a signal denoising training model, taking the error signal between the output of the signal denoising training model and the corresponding expected signal as the minimum as the training target, and iteratively training the signal denoising training model using a gradient descent method includes: S1, initializing the weight vector of the signal denoising model to be trained; S2, inputting the sample digital signal into the initialized signal denoising to be trained model, and determining the output signal corresponding to the sample digital signal according to the actual output of the initialized signal denoising to be trained model; S3, calculating the difference between the expected signal and the output signal to obtain an error signal; S4, judging whether the mean square value of the error signal is the minimum value, if yes, go to S5; if no, go to S6; S5, determining the weight vector as a target weight vector; S6, using a gradient descent method to adjust the weight vector of the signal denoising model to be trained; S7, input the sample digital signal into the adjusted signal denoising to be trained model to obtain an updated output signal, and go to S3.

10. A harmonic rapid identification and monitoring device based on broadband electrical parameters, characterized in that: include: An acquisition module, used for acquiring an analog signal to be tested; A preprocessing module, used for preprocessing the analog signal to be tested to obtain a preprocessed signal; An analysis module, used for performing time-frequency analysis on the preprocessed signal to obtain time-frequency analysis features; The identification module is used to input the time-frequency analysis features into a disturbance signal identification model, and identify the disturbance type and disturbance percentage of the disturbance signal according to the output of the disturbance signal identification model; wherein the disturbance signal identification model adopts a fuzzy SOM neural network.

Images