High-frequency switching power supply monitoring method and device for transformer substation relay protection

Abstract

The invention discloses a high-frequency switching power supply monitoring method and device for transformer substation relay protection. The method comprises the following steps: acquiring a first audio signal of a high-frequency switching power supply based on an ultrasonic microphone; collecting first sequence data corresponding to the first audio signal, and calculating a second audio signal according to the first sequence data; respectively extracting an upper envelope line and a lower envelope line from the second audio signal, and calculating a characteristic value signal; inputting the characteristic value signal into a trained recurrent neural network to obtain a prediction characteristic; and performing two-norm calculation on the prediction features and the sample features, and determining early warning information of the high-frequency switching power supply according to a calculation result. The high-frequency switching power supply is monitored in real time by collecting the sound wave signal of the high-frequency switching power supply and carrying out sound wave monitoring, so that the full life cycle of the switching power supply for relay protection of the transformer substation is monitored, early warning of relay protection faults is greatly brought forward, and enough maintenance windows are reserved for maintenance.

Description

Technical Field

[0001] This application belongs to the technical field of signal monitoring, and particularly relates to a method and device for monitoring high-frequency switching power supplies for substation relay protection. Background Art

[0002] Relay protection in a substation generally means that when the insulation of a transmission line or a power transformation and distribution device is damaged, the load is short-circuited, resulting in a decrease in impedance and a corresponding increase in current. The relay protection then disconnects (cuts off) the short-circuited transmission line or power transformation and distribution device. If not dealt with in time, it is likely to cause equipment damage, fire, threaten personal safety, and even disrupt the stable operation of the power supply system.

[0003] As a key component for driving the relay protector to work, the high-frequency switching power supply ensures that the switching mechanism can be cut off when a fault occurs. However, if the high-frequency switching power supply has a slight abnormality usually and the circuit cannot detect it, it may not be able to drive the switching mechanism during a fault, thus causing significant losses. Therefore, it is crucial to monitor the high-frequency switching power supply.

[0004] Currently, the ultra-high frequency technical means are mainly used to monitor the high-frequency switching power supply. When the high-frequency switching power supply has a serious abnormality, it will generate an ultra-high frequency electromagnetic field during partial discharge, which is collected and analyzed by using an analyzer. However, in the actual process, when the switching power supply has partial discharge, it indicates that it has a serious fault, and its maintenance window is very short, which is not conducive to the overall planned maintenance plan of the substation, and there is a risk that the fault may not be repaired in time. Summary of the Invention

[0005] To solve the above technical problems that when the switching power supply has partial discharge, it indicates that it has a serious fault, its maintenance window is very short, which is not conducive to the overall planned maintenance plan of the substation, and there is a risk that the fault may not be repaired in time, this application proposes a method and device for monitoring high-frequency switching power supplies for substation relay protection, and its technical solutions are as follows: In the first aspect, an embodiment of this application provides a method for monitoring a high-frequency switching power supply for substation relay protection, including: Obtaining a first audio signal of the high-frequency switching power supply based on an ultrasonic microphone; Collecting first sequence data corresponding to the first audio signal based on at least two microphones, and calculating a second audio signal according to the first sequence data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; Respectively extracting an upper envelope line and a lower envelope line from the second audio signal, and calculating an eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; When it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint conditions, the eigenvalue signal is input into the trained recurrent neural network to obtain a predicted feature; Calculate the two-norm of the predicted feature and the sample feature, and determine the warning information of the high-frequency switching power supply according to the calculation result.

[0006] In an alternative scheme of the first aspect, calculating a second audio signal according to the first sequence data includes: Calculate the average value of each subsequence data in the first sequence data to obtain a second sequence data; Calculate the covariance matrix according to the first sequence data and the second sequence data; Determine at least two eigenvalues of the covariance matrix, and construct a singular value matrix based on the at least two eigenvalues; Calculate the second audio signal according to the covariance matrix, the singular value matrix, and the preset weight ratio.

[0007] In another alternative scheme of the first aspect, extracting the upper envelope and the lower envelope from the second audio signal respectively includes: Based on the second audio signal and the preset signal expression, construct a band-pass signal, and perform filtering processing on the band-pass signal expression to obtain a filtered signal; Construct a composite signal according to the band-pass signal and the filtered signal; Substitute the composite signal into the preset upper envelope expression to obtain the upper envelope; and Substitute the composite signal into the preset lower envelope expression to obtain the lower envelope.

[0008] In another alternative scheme of the first aspect, calculating the eigenvalue signal according to the second audio signal, the upper envelope, and the lower envelope includes: Calculate the average value of the upper envelope and the lower envelope to obtain an average envelope; Calculate the difference between the second audio signal and the average envelope to obtain the eigenvalue signal.

[0009] In another alternative scheme of the first aspect, after extracting the upper envelope and the lower envelope from the second audio signal respectively, and calculating the eigenvalue signal according to the second audio signal, the upper envelope, and the lower envelope, before inputting the eigenvalue signal into the trained recurrent neural network to obtain a predicted feature when it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint conditions, it further includes: Judge whether the difference between the number of extreme points and the number of zero-crossing points in the characteristic curve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference is within a preset number interval, it is determined whether the average value between the envelope line formed by all the maximum points at any moment and the envelope line formed by all the minimum points in the feature curve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, it is determined that the feature curve corresponding to the eigenvalue signal satisfies the constraint condition.

[0010] In another alternative solution of the first aspect, before inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features, it further includes: Calculating the difference between the second audio signal and the eigenvalue signal, and determining whether the difference calculation result is a monotonic function; Inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features, including: When it is determined that the difference calculation result is a monotonic function, inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features.

[0011] In another alternative solution of the first aspect, determining the warning information of the high-frequency switching power supply according to the calculation result includes: When it is detected that the calculation result is within a preset threshold interval, generating warning information of the high-frequency switching power supply according to the damage information corresponding to the preset threshold interval.

[0012] In a second aspect, an embodiment of the present application provides a monitoring device for a high-frequency switching power supply of substation relay protection, including: An audio acquisition module, configured to obtain a first audio signal of the high-frequency switching power supply based on an ultrasonic microphone; A first processing module, configured to collect first sequence data corresponding to the first audio signal based on at least two microphones, and calculate a second audio signal according to the first sequence data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; A second processing module, configured to respectively extract an upper envelope line and a lower envelope line from the second audio signal, and calculate an eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; A model prediction module, configured to input the eigenvalue signal into the trained recurrent neural network to obtain the predicted features when it is determined that the feature curve corresponding to the eigenvalue signal satisfies the constraint condition; A warning module, configured to calculate the two-norm of the predicted features and the sample features, and determine the warning information of the high-frequency switching power supply according to the calculation result.

[0013] In an alternative solution of the second aspect, the first processing module includes: Calculating the average value of each subsequence data in the first sequence data to obtain a second sequence data; Calculate the covariance matrix according to the first sequence data and the second sequence data; Determine at least two eigenvalues of the covariance matrix, and construct a singular value matrix based on the at least two eigenvalues; Calculate the second audio signal according to the covariance matrix, the singular value matrix and the preset weight ratio.

[0014] In another alternative solution of the second aspect, the second processing module includes: Construct a band-pass signal based on the second audio signal and a preset signal expression, and perform filtering processing on the band-pass signal expression to obtain a filtered signal; Construct a composite signal according to the band-pass signal and the filtered signal; Substitute the composite signal into a preset upper envelope expression to obtain an upper envelope; and Substitute the composite signal into a preset lower envelope expression to obtain a lower envelope.

[0015] In another alternative solution of the second aspect, the second processing module further includes: Perform a mean value calculation on the upper envelope and the lower envelope to obtain an average envelope; Perform a difference calculation on the second audio signal and the average envelope to obtain an eigenvalue signal.

[0016] In another alternative solution of the second aspect, the device further includes: After respectively extracting the upper envelope and the lower envelope from the second audio signal, and calculating the eigenvalue signal according to the second audio signal, the upper envelope and the lower envelope, when it is determined that the eigencurve corresponding to the eigenvalue signal satisfies the constraint condition, before inputting the eigenvalue signal into the trained recurrent neural network to obtain a predicted feature, Judge whether the difference between the number of extreme points and the number of zero-crossing points in the eigencurve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference is within the preset number interval, judge whether the average value between the envelope formed by all the maximum points and the envelope formed by all the minimum points at any moment in the eigencurve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, determine that the eigencurve corresponding to the eigenvalue signal satisfies the constraint condition.

[0017] In another alternative solution of the second aspect, the device further includes: Before inputting the eigenvalue signal into the trained recurrent neural network to obtain a predicted feature, Perform a difference calculation on the second audio signal and the eigenvalue signal, and judge whether the difference calculation result is a monotonic function; Input the eigenvalue signal into the trained recurrent neural network to obtain predicted features, including: When it is determined that the difference calculation result is a monotonic function, input the eigenvalue signal into the trained recurrent neural network to obtain predicted features.

[0018] In another alternative solution of the second aspect, the warning module includes: When it is detected that the calculation result is within the preset threshold interval, generate a warning message for the high-frequency switching power supply according to the damage information corresponding to the preset threshold interval.

[0019] In a third aspect, an embodiment of the present application further provides a monitoring device for a high-frequency switching power supply of substation relay protection, including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to implement the method for monitoring a high-frequency switching power supply of substation relay protection provided in the first aspect or any implementation manner of the first aspect of the embodiments of the present application.

[0020] In a fourth aspect, an embodiment of the present application provides a computer storage medium, and the computer storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, the method for monitoring a high-frequency switching power supply of substation relay protection provided in the first aspect or any implementation manner of the first aspect of the embodiments of the present application can be implemented.

[0021] In the embodiments of the present application, during the process of monitoring the high-frequency switching power supply, a first audio signal of the high-frequency switching power supply can be obtained based on an ultrasonic microphone; a first sequence of data corresponding to the first audio signal is collected based on at least two microphones, and a second audio signal is calculated according to the first sequence of data; an upper envelope line and a lower envelope line are respectively extracted from the second audio signal, and an eigenvalue signal is calculated according to the second audio signal, the upper envelope line, and the lower envelope line; when it is determined that the eigencurve corresponding to the eigenvalue signal satisfies the constraint condition, the eigenvalue signal is input into the trained recurrent neural network to obtain predicted features; a two-norm calculation is performed on the predicted features and the sample features, and a warning message for the high-frequency switching power supply is determined according to the calculation result. By collecting the acoustic wave signal of the high-frequency switching power supply and performing acoustic wave monitoring to perform real-time monitoring of the high-frequency switching power supply, not only the full life cycle of the monitoring switching power supply of substation relay protection is realized, but also the warning of relay protection faults is greatly advanced, and thus enough maintenance windows are left for maintenance. Description of the Drawings

[0022] To more clearly illustrate the technical solutions in the embodiments of the present application, the accompanying drawings required for use in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0023] Figure 1 This is the overall flowchart of a high-frequency switching power supply monitoring method for substation relay protection provided by an embodiment of the present application; Figure 2 This is the structural schematic diagram of a high-frequency switching power supply monitoring device for substation relay protection provided by an embodiment of the present application; Figure 3 This is the structural schematic diagram of another high-frequency switching power supply monitoring device for substation relay protection provided by an embodiment of the present application. Detailed implementation manners

[0024] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application.

[0025] In the following description, the terms "first" and "second" are only for the purpose of description and cannot be construed as indicating or implying relative importance. The following description provides multiple embodiments of the present application. Different embodiments can be replaced or combined. Therefore, the present application can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present application should also be considered to include embodiments containing all other possible combinations of A, B, C, and D, even though such an embodiment may not be explicitly described in the following content.

[0026] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes can be made to the functions and arrangements of the described elements without departing from the scope of the content of the present application. Various processes or components can be appropriately omitted, substituted, or added to each example. For example, the described methods can be executed in a different order from the described order, and various steps can be added, omitted, or combined. In addition, the features described in some examples can be combined into other examples.

[0027] Please refer to Figure 1 , Figure 1 which shows the overall flowchart of a high-frequency switching power supply monitoring method for substation relay protection provided by an embodiment of the present application.

[0028] As shown in Figure 1As shown, the high-frequency switching power supply monitoring method for substation relay protection may at least include the following steps: Step 102: Obtain a first audio signal of the high-frequency switching power supply based on an ultrasonic microphone.

[0029] In the embodiment of the present application, the high-frequency switching power supply monitoring method for substation relay protection can be applied to a control terminal. The control terminal can collect the audio signal emitted by the high-frequency switching power supply based on an ultrasonic microphone set near the high-frequency switching power supply. Then, an ultrasonic acquisition card is used to obtain the audio signal collected by the ultrasonic microphone and perform signal processing on the audio signal. It can be understood that when it is known from the result of the signal processing that there is a risk in the high-frequency switching power supply, the control terminal can also immediately send a warning message to the operation and maintenance computer room for maintenance by the staff. Among them, the distance between the ultrasonic microphone and the high-frequency switching power supply can be a preset distance to ensure the accuracy of the audio signal collected by the ultrasonic microphone.

[0030] Specifically, during the process of monitoring the high-frequency switching power supply, when the control terminal determines that the high-frequency switching power supply is in a working state, it can control the ultrasonic collector to obtain the first audio signal emitted by the high-frequency switching power supply, and feedback the first audio signal to the control terminal through the ultrasonic acquisition card.

[0031] Step 104: Collect first sequence data corresponding to the first audio signal based on at least two microphones, and calculate a second audio signal according to the first sequence data.

[0032] Specifically, after obtaining the first audio signal, the control terminal can, but is not limited to, control at least two microphones to collect the first audio signal. Each microphone can collect the sequence data corresponding to the first audio signal, and the set of the sequence data collected by each microphone can be used as the first sequence data. Among them, the first subsequence data in the first sequence data can, but is not limited to, be the sequence data collected by the first microphone, the second subsequence data in the first sequence data can, but is not limited to, be the sequence data collected by the second microphone, and the nth subsequence data in the first sequence data can, but is not limited to, be the sequence data collected by the nth microphone. It can be understood that the collection duration of each microphone is kept consistent so that the sequence data collected by each microphone is consistent in length, thereby improving the consistency and effectiveness of the data. Of course, for a segment of audio signal, the control terminal can also divide the audio signal into multiple segments of audio signals according to the collection duration, and each microphone can respectively collect the sequence data corresponding to each segment of audio signal, thereby obtaining multiple first sequence data.

[0033] Further, after obtaining the first sequence data, the control terminal can calculate the average value of each subsequence data in the first sequence data. The method of calculating the average value can be to divide each subsequence data by the sum of all subsequence data respectively, and the set of the results of calculating the average value of each subsequence data can be used as the second sequence data.

[0034] Here, taking the first sequence data that can be expressed as as an example, the method of calculating the average value of the nth subsequence data in the first sequence data can be but not limited to referring to the following: Combined with the above formula, the second sequence data can be obtained and expressed as 。

[0035] Further, after obtaining the second sequence data through average value calculation, the control terminal can calculate the covariance matrix according to the first sequence data and the second sequence data. Among them, in the process of calculating the covariance matrix, it can be but not limited to first combine the first sequence data and the second sequence data to obtain the normalized sequence data, and the normalized sequence data can be expressed as follows: In the above formula, can correspond to the normalized sequence data, can correspond to the first sequence data, can correspond to the second sequence data.

[0036] Then, the control terminal can calculate the covariance matrix according to the normalized sequence data, and the covariance matrix can be but not limited to expressed as follows: In the above formula, can correspond to the covariance matrix, n can correspond to the number of subsequence data in the first sequence data, can correspond to the normalized sequence data, can be understood as the transpose matrix of.

[0037] Further, after calculating the covariance matrix, the control terminal can combine the covariance matrix with the expression of the eigenvalue to determine multiple eigenvalues corresponding to the covariance matrix. Among them, the expression of the covariance matrix and the eigenvalue can be but not limited to expressed as follows: In the above formula, can correspond to the covariance matrix, I can be understood as the identity matrix, can be understood as the eigenvalue, and there can be multiple solutions for this eigenvalue (that is, the eigenvalue can be expressed as )

[0038] It can be understood that after obtaining multiple eigenvalues corresponding to the covariance matrix, corresponding eigenvector matrices can be obtained respectively according to each eigenvalue. Among them, the expression corresponding to the eigenvector matrix and the eigenvalue can be but is not limited to the following: Here, V can be understood as the eigenvalue The corresponding eigenvector matrix, and for different eigenvector matrices , can be understood as the corresponding eigenvalue The fundamental solution system in the homogeneous equation system of the above formula.

[0039] It should be noted that in order to implement the dimensionality reduction processing of the audio signal, a new covariance matrix with different matrix dimensions can also be constructed according to the above-mentioned covariance matrix. The matrix dimension of the newly constructed covariance matrix can be but is not limited to being expressed as n*n, and the matrix dimension of the above-mentioned covariance matrix can be but is not limited to being expressed as m*m, where m is greater than n. Among them, the newly constructed covariance matrix can be but is not limited to being expressed as the following formula: Then, referring to the expressions of the above-mentioned covariance matrix and eigenvalue, and the expressions corresponding to the eigenvector matrix and the eigenvalue, multiple eigenvalues of the newly constructed covariance matrix and the corresponding eigenvector matrix can be obtained respectively. It should be noted that the multiple eigenvalues of the above-mentioned covariance matrix are consistent with the multiple eigenvalues of the newly constructed covariance matrix here, and the eigenvector matrices corresponding to the multiple eigenvalues mentioned above are different in dimension from the eigenvector matrices corresponding to the multiple eigenvalues here.

[0040] Furthermore, after determining the multiple eigenvalues of the above-mentioned covariance matrix, the control terminal can also substitute the multiple eigenvalues into a preset singular value matrix expression to obtain a singular value matrix. Among them, the preset singular value matrix expression can be but is not limited to the following: In the above formula, Can correspond to the singular value matrix, Further, after obtaining the singular value matrix, the control terminal can obtain the second audio signal according to the singular value matrix, the covariance matrix mentioned above, the newly constructed covariance matrix, and the preset weight ratio. Among them, it can, but is not limited to, first substituting the singular value matrix, the eigenvector matrix corresponding to the covariance matrix mentioned above, and the eigenvector matrix corresponding to the newly constructed covariance matrix into a preset signal component distribution expression, and its expression is as follows: In the above formula, A can correspond to the signal component distribution, can correspond to the singular value matrix, can correspond to the i-th eigenvector matrix corresponding to the newly constructed covariance matrix, can correspond to the i-th eigenvector matrix corresponding to the covariance matrix mentioned above.

[0041] Next, after calculating the signal component distribution, the control terminal can calculate the second audio signal by combining the signal component distribution and the preset weight ratio. Among them, the data volume of the second audio signal is smaller than that of the first audio signal, thereby reflecting the dimensionality reduction processing effect on the audio signal.

[0042] It can be understood that the preset weight ratio mentioned above can correspond to 90%, and it is not limited to this here.

[0043] Step 106: Respectively extract the upper envelope line and the lower envelope line from the second audio signal, and calculate the eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line.

[0044] Specifically, after obtaining the second audio signal, the control terminal can substitute the second audio signal into a preset signal expression to construct a band-pass signal, which can be, but is not limited to, expressed as follows: In the above formula, can correspond to the band-pass signal, corresponds to the second audio signal, can correspond to the band-pass frequency constant, and t can correspond to the time variable.

[0045] Next, the control terminal can perform filtering processing on the band-pass signal to obtain a filtered signal, which can be, but is not limited to, expressed as follows: Further, after obtaining the band-pass signal and the filtered signal respectively, the control terminal can substitute the band-pass signal and the filtered signal into a composite signal expression to obtain a composite signal, which can be, but is not limited to, expressed as follows: In the above formula, can correspond to a composite signal, can correspond to a band-pass signal, can correspond to a filtered signal, and J can be understood as the imaginary unit.

[0046] Furthermore, the control terminal can substitute the composite signal into a preset upper envelope expression to obtain the upper envelope. Among them, the upper envelope can be but is not limited to being expressed as follows: The control terminal can also substitute the composite signal into a preset lower envelope expression to obtain the lower envelope. Among them, the lower envelope can be but is not limited to being expressed as follows: Furthermore, after obtaining the upper envelope and the lower envelope, the control terminal can calculate the mean value of the upper envelope and the lower envelope to obtain the average envelope, and its calculation expression can be but is not limited to being expressed as follows: After obtaining the mean value of the upper envelope and the lower envelope, the control terminal can calculate the difference value of the mean value of the upper envelope and the lower envelope according to the second audio signal mentioned above to obtain the eigenvalue signal, and its calculation expression can be but is not limited to being expressed as follows: As an option of the embodiment of the present application, after respectively extracting the upper envelope and the lower envelope from the second audio signal, and calculating the eigenvalue signal according to the second audio signal, the upper envelope and the lower envelope, before inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature, it further includes: judging whether the difference between the number of extreme points and the number of zero-crossing points in the feature curve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference is within the preset number interval, judging whether the average value between the envelope formed by all the maximum points and the envelope formed by all the minimum points at any moment in the feature curve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, it is determined that the feature curve corresponding to the eigenvalue signal satisfies the constraint condition.

[0047] Among them, the preset number range here can be from 0 to 1. That is to say, when the difference between the number of extreme points and the number of zero-crossing points is 0 or 1, it can be further determined whether the average value between the envelope line formed by all the maximum points and the envelope line formed by all the minimum points at any moment in the characteristic curve corresponding to the eigenvalue signal is zero. And when it is continuously detected that the average value is zero, it can be determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition.

[0048] It can be understood that when it is determined that the characteristic curve corresponding to the eigenvalue signal does not satisfy the constraint condition, the upper envelope line and the lower envelope line can be respectively extracted from the eigenvalue signal, and a new eigenvalue signal can be calculated based on the eigenvalue signal, the upper envelope line, and the lower envelope line. Then, it can be continuously determined whether the new eigenvalue signal satisfies the constraint condition. Among them, the process of calculating the new eigenvalue signal and the process of determining whether it satisfies the constraint condition can refer to the above-mentioned embodiments, which will not be elaborated here.

[0049] It should be noted that after it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition, the eigenvalue signal can be stored. Then, the upper envelope line and the lower envelope line can be respectively extracted from the above-mentioned average envelope line. A new eigenvalue signal can be calculated based on the average envelope line, the upper envelope line, and the lower envelope line. Then, it can be continuously determined whether the new eigenvalue signal satisfies the constraint condition. If it satisfies, the new eigenvalue signal can be stored, and this process can be repeated until the difference between the finally obtained new eigenvalue signal and the signal used to extract the upper envelope line and the lower envelope line is a monotonic function, and all the stored eigenvalue signals can be used as the final eigenvalue signals.

[0050] Step 108: When it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition, input the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature.

[0051] Specifically, when it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition, the control terminal can input the eigenvalue signal into the trained recurrent neural network to predict the next characteristic signal. Among them, here it is assumed that the eigenvalue signal can be expressed as For example, it specifically includes n eigenvalue features such as average deviation, skewness, kurtosis, and spectral entropy ratio. After inputting the eigenvalue signal into the trained recurrent neural network, it can predict, but not limited to, the data of n eigenvalue features such as average deviation, skewness, kurtosis, and spectral entropy ratio in the next 1 day.

[0052] It can be understood that the recurrent neural network mentioned in the embodiments of the present application can be, but not limited to, the common RNN neural network, which will not be elaborated here too much.

[0053] Step 110: Calculate the two-norm of the predicted features and the sample features, and determine the warning information of the high-frequency switching power supply according to the calculation results.

[0054] Specifically, after obtaining the predicted features, the control terminal can calculate the two-norm according to the predicted features and the sample features corresponding to the known normal switching power supply, and determine whether the calculation result is within the preset threshold range. It can be understood that when it is detected that the calculation result is within the preset threshold range, the corresponding damage information can be obtained according to the threshold range, and the warning information of the high-frequency switching power supply can be generated according to the loss information, so as to remind the staff to perform maintenance immediately according to the loss information. Among them, the preset threshold range can include at least two threshold ranges, and different threshold ranges can respectively correspond to the damage information of the switching power supply for quick maintenance.

[0055] Please refer to Figure 2 , Figure 2 which shows the structural schematic diagram of a high-frequency switching power supply monitoring device for substation relay protection provided by an embodiment of the present application.

[0056] As Figure 2 shown, the high-frequency switching power supply monitoring device for substation relay protection can at least include an audio acquisition module 201, a first processing module 202, a second processing module 203, a model prediction module 204, and a warning module 205, where: The audio acquisition module 201 is used to obtain the first audio signal of the high-frequency switching power supply based on an ultrasonic microphone; The first processing module 202 is used to collect the first sequence data corresponding to the first audio signal based on at least two microphones, and calculate the second audio signal according to the first sequence data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; The second processing module 203 is used to respectively extract the upper envelope line and the lower envelope line from the second audio signal, and calculate the eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; The model prediction module 204 is used to input the eigenvalue signal into the trained recurrent neural network to obtain the predicted features when it is determined that the eigencurve corresponding to the eigenvalue signal satisfies the constraint conditions; The warning module 205 is used to calculate the two-norm of the predicted features and the sample features, and determine the warning information of the high-frequency switching power supply according to the calculation results.

[0057] In some possible embodiments, the first processing module includes: Calculate the average value of each subsequence data in the first sequence data to obtain the second sequence data; Calculate a covariance matrix according to the first sequence data and the second sequence data; Determine at least two eigenvalues of the covariance matrix, and construct a singular value matrix based on the at least two eigenvalues; Calculate a second audio signal according to the covariance matrix, the singular value matrix, and a preset weight ratio.

[0058] In some possible embodiments, the second processing module includes: Construct a band-pass signal based on the second audio signal and a preset signal expression, and perform filtering processing on the band-pass signal expression to obtain a filtered signal; Construct a composite signal according to the band-pass signal and the filtered signal; Substitute the composite signal into a preset upper envelope expression to obtain an upper envelope; and Substitute the composite signal into a preset lower envelope expression to obtain a lower envelope.

[0059] In some possible embodiments, the second processing module further includes: Perform a mean calculation on the upper envelope and the lower envelope to obtain an average envelope; Perform a difference calculation on the second audio signal and the average envelope to obtain an eigenvalue signal.

[0060] In some possible embodiments, the apparatus further includes: After respectively extracting an upper envelope and a lower envelope from the second audio signal, and calculating an eigenvalue signal according to the second audio signal, the upper envelope, and the lower envelope, when it is determined that the eigencurve corresponding to the eigenvalue signal satisfies a constraint condition, before inputting the eigenvalue signal into a trained recurrent neural network to obtain a predicted feature, Judge whether the difference between the number of extreme points and the number of zero-crossing points in the eigencurve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference is within the preset number interval, judge whether the average value between the envelope formed by all the maximum points and the envelope formed by all the minimum points at any moment in the eigencurve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, determine that the eigencurve corresponding to the eigenvalue signal satisfies the constraint condition.

[0061] In some possible embodiments, the apparatus further includes: Before inputting the eigenvalue signal into a trained recurrent neural network to obtain a predicted feature, Perform a difference calculation on the second audio signal and the eigenvalue signal, and judge whether the difference calculation result is a monotonic function; Input the eigenvalue signal into the trained recurrent neural network to obtain predicted features, including: When it is determined that the difference calculation result is a monotonic function, input the eigenvalue signal into the trained recurrent neural network to obtain predicted features.

[0062] In some possible embodiments, the warning module includes: When it is detected that the calculation result is within a preset threshold range, generate a warning message for the high-frequency switching power supply according to the damage information corresponding to the preset threshold range.

[0063] Those skilled in the art can clearly understand that the technical solutions of the embodiments of the present application can be implemented by means of software and / or hardware. The "units" and "modules" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a Field-Programmable Gate Array (FPGA), an Integrated Circuit (IC), etc.

[0064] Please refer to Figure 3 , Figure 3 which shows a schematic structural diagram of another high-frequency switching power supply monitoring device for substation relay protection provided by the embodiments of the present application.

[0065] As Figure 3 shown, the high-frequency switching power supply monitoring device 300 for substation relay protection may include at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302.

[0066] Among them, the communication bus 302 can be used to realize the connection and communication of the above-mentioned various components.

[0067] Among them, the user interface 303 may include buttons, and the optional user interface may further include a standard wired interface and a wireless interface.

[0068] Among them, the network interface 304 can include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0069] Among them, the processor 301 may include one or more processing cores. The processor 301 is connected to various parts within the high-frequency switching power supply monitoring device 300 of the substation relay protection through various interfaces and lines. By running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling the data stored in the memory 305, it executes various functions of the high-frequency switching power supply monitoring device 300 of the substation relay protection and processes data. Optionally, the processor 301 may be implemented in at least one hardware form of DSP, FPGA, or PLA. The processor 301 may integrate one or a combination of several of CPU, GPU, and modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor 301 and may be implemented separately through a single chip.

[0070] Among them, the memory 305 may include RAM and may also include ROM. Optionally, the memory 305 includes a non-transitory computer-readable medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store the data involved in the above-mentioned various method embodiments. Optionally, the memory 305 may also be at least one storage device located far from the aforementioned processor 301. As Figure 3 shown, the memory 305 as a computer storage medium may include an operating system, a network communication module, a user interface module, and a high-frequency switching power supply monitoring application program for substation relay protection.

[0071] Specifically, the processor 301 may be used to call the high-frequency switching power supply monitoring application program stored in the memory 305 and specifically perform the following operations: Obtain a first audio signal of the high-frequency switching power supply based on the ultrasonic microphone; Collect a first sequence of data corresponding to the first audio signal based on at least two microphones, and calculate a second audio signal according to the first sequence of data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; Extract the upper envelope line and the lower envelope line from the second audio signal respectively, and calculate an eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; When it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint conditions, input the eigenvalue signal into the trained recurrent neural network to obtain a predicted feature; Calculate the two-norm of the predicted feature and the sample feature, and determine the warning information of the high-frequency switching power supply according to the calculation result.

[0072] In some possible embodiments, calculating a second audio signal according to the first sequence data includes: Calculate the average value of each subsequence data in the first sequence data to obtain a second sequence data; Calculate the covariance matrix according to the first sequence data and the second sequence data; Determine at least two eigenvalues of the covariance matrix, and construct a singular value matrix based on the at least two eigenvalues; Calculate the second audio signal according to the covariance matrix, the singular value matrix, and the preset weight ratio.

[0073] In some possible embodiments, extracting an upper envelope and a lower envelope from the second audio signal respectively includes: Based on the second audio signal and a preset signal expression, construct a band-pass signal, and perform filtering processing on the band-pass signal expression to obtain a filtered signal; Construct a composite signal according to the band-pass signal and the filtered signal; Substitute the composite signal into the preset upper envelope expression to obtain the upper envelope; and Substitute the composite signal into the preset lower envelope expression to obtain the lower envelope.

[0074] In some possible embodiments, calculating the eigenvalue signal according to the second audio signal, the upper envelope, and the lower envelope includes: Calculate the average value of the upper envelope and the lower envelope to obtain an average envelope; Calculate the difference between the second audio signal and the average envelope to obtain the eigenvalue signal.

[0075] In some possible embodiments, after extracting the upper envelope and the lower envelope from the second audio signal respectively, and calculating the eigenvalue signal according to the second audio signal, the upper envelope, and the lower envelope, before inputting the eigenvalue signal into the trained recurrent neural network to obtain a predicted feature when it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint conditions, it further includes: Judge whether the difference between the number of extreme points and the number of zero-crossing points in the characteristic curve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference value is within a preset number range, it is determined whether the average value between the envelope line formed by all the maximum points at any moment and the envelope line formed by all the minimum points in the characteristic curve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition.

[0076] In some possible embodiments, before inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features, it further includes: Calculating the difference between the second audio signal and the eigenvalue signal, and determining whether the difference calculation result is a monotonic function; Inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features, including: When it is determined that the difference calculation result is a monotonic function, inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted features.

[0077] In some possible embodiments, determining the warning information of the high-frequency switching power supply according to the calculation result includes: When it is detected that the calculation result is within a preset threshold range, generating the warning information of the high-frequency switching power supply according to the damage information corresponding to the preset threshold range.

[0078] This application also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the above method are implemented. Among them, the computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, micro drives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nano-systems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0079] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0080] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0081] In several embodiments provided in the present application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling, or communication connection between each other can be through some service interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical or other form.

[0082] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0083] In addition, each functional unit in various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0084] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of the present application. And the aforementioned memory includes: USB flash drive, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk, or optical disk and other media that can store program codes.

[0085] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory. The memory can include: flash drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc.

[0086] The above are only exemplary embodiments of the present disclosure, and the scope of the present disclosure cannot be limited thereby. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. Those skilled in the art will readily conceive of other embodiments of the present disclosure after considering the specification and practicing the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure, which follow the general principles of the present disclosure and include well-known common general knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A monitoring method for high-frequency switching power supplies of substation relay protection, characterized in that, Including: Obtaining a first audio signal of a high-frequency switching power supply based on an ultrasonic microphone; Collecting first sequence data corresponding to the first audio signal based on at least two microphones, and calculating a second audio signal according to the first sequence data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; Respectively extracting an upper envelope line and a lower envelope line from the second audio signal, and calculating an eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; When it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition, inputting the eigenvalue signal into a trained recurrent neural network to obtain a predicted feature; Performing a two-norm calculation on the predicted feature and a sample feature, and determining a warning message of the high-frequency switching power supply according to the calculation result.

2. The method according to claim 1, wherein The calculating the second audio signal according to the first sequence data includes: Calculating an average value of each subsequence data in the first sequence data to obtain a second sequence data; Calculating a covariance matrix according to the first sequence data and the second sequence data; Determining at least two eigenvalues of the covariance matrix, and constructing a singular value matrix based on the at least two eigenvalues; Calculating a second audio signal according to the covariance matrix, the singular value matrix, and a preset weight ratio.

3. The method according to claim 1, wherein The respectively extracting the upper envelope line and the lower envelope line from the second audio signal includes: Constructing a band-pass signal based on the second audio signal and a preset signal expression, and performing a filtering process on the band-pass signal expression to obtain a filtered signal; Constructing a composite signal according to the band-pass signal and the filtered signal; Substituting the composite signal into a preset upper envelope line expression to obtain an upper envelope line; and Substituting the composite signal into a preset lower envelope line expression to obtain a lower envelope line.

4. The method according to claim 3, wherein The calculating the eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line includes: Calculating an average value of the upper envelope line and the lower envelope line to obtain an average envelope line; Calculating a difference between the second audio signal and the average envelope line to obtain an eigenvalue signal.

5. The method according to claim 1, wherein After respectively extracting the upper envelope line and the lower envelope line from the second audio signal, and calculating the eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line, and before inputting the eigenvalue signal into a trained recurrent neural network to obtain a predicted feature when it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition, it further includes: Judging whether the difference between the number of extreme points and the number of zero-crossing points in the characteristic curve corresponding to the eigenvalue signal is within a preset number interval; When it is detected that the difference is within the preset number interval, judging whether the average value between the envelope line formed by all maximum points and the envelope line formed by all minimum points at any moment in the characteristic curve corresponding to the eigenvalue signal is zero; When it is detected that the average value is zero, it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition.

6. The method according to claim 5, wherein Before inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature, it further includes: Performing a difference calculation on the second audio signal and the eigenvalue signal, and determining whether the difference calculation result is a monotonic function; The step of inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature includes: When it is determined that the difference calculation result is a monotonic function, inputting the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature.

7. The method according to claim 1, characterized in that, The step of determining the warning information of the high-frequency switching power supply according to the calculation result includes: When it is detected that the calculation result is within a preset threshold range, generating the warning information of the high-frequency switching power supply according to the damage information corresponding to the preset threshold range.

8. A high-frequency switching power supply monitoring device for substation relay protection, characterized in that, It includes: An audio acquisition module, configured to obtain a first audio signal of the high-frequency switching power supply based on an ultrasonic microphone; A first processing module, configured to collect first sequence data corresponding to the first audio signal based on at least two microphones, and calculate a second audio signal according to the first sequence data; wherein, the data volume of the second audio signal is lower than that of the first audio signal; A second processing module, configured to respectively extract an upper envelope line and a lower envelope line from the second audio signal, and calculate an eigenvalue signal according to the second audio signal, the upper envelope line, and the lower envelope line; A model prediction module, configured to input the eigenvalue signal into the trained recurrent neural network to obtain the predicted feature when it is determined that the characteristic curve corresponding to the eigenvalue signal satisfies the constraint condition; A warning module, configured to perform a two-norm calculation on the predicted feature and the sample feature, and determine the warning information of the high-frequency switching power supply according to the calculation result.

9. A high-frequency switching power supply monitoring device for substation relay protection, characterized in that, It includes a processor and a memory; The processor is connected to the memory; The memory is configured to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, Instructions are stored in the computer-readable storage medium, and when the instructions run on a computer or a processor, the computer or the processor is caused to execute the steps of the method according to any one of claims 1-7.

Images