Abnormal arc detection method and device based on ultrasonic waves, electronic equipment and medium

Abstract

The invention relates to the technical field of arc detection, and particularly discloses an abnormal arc detection method and device based on ultrasonic waves, electronic equipment and a medium, and the detection method comprises the steps: collecting ultrasonic data; frequency spectrum characteristic values are extracted according to the ultrasonic data, a data set is obtained, the data set comprises at least part of a training set before to-be-detected time and a to-be-detected data set comprising the to-be-detected time, and the training set comprises a normal training set obtained based on the ultrasonic data in a normal state; obtaining an abnormal arc training set based on the ultrasonic data in the abnormal arc state; training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model; and inputting the to-be-detected data set into the arc detection model to obtain a detection result. Compared with the prior art, the problems of accuracy reduction and high labor cost caused by manual intervention can be reduced, the detection algorithm is simple, the accuracy of the detection result is higher, and the detection efficiency is high.

Description

Technical Field

[0001] The present invention relates to the field of arc detection technology, and in particular to an abnormal arc detection method, device, electronic equipment and medium based on ultrasonic wave. Background Art

[0002] At present, abnormal arc detection in power systems is mostly based on electrical parameters, that is, identification through electrical parameters such as current and voltage, and the detection purpose is achieved by extracting the corresponding electrical parameter features to train the detection model. However, this single mode of electrical parameters can be affected by external interference and lead to misjudgment, resulting in poor accuracy.

[0003] Compared with electrical parameter modalities, ultrasonic technology can be used to detect ultrasonic signals in power systems to identify abnormal arcs, which can reduce external interference. However, existing ultrasonic detection solutions usually down-convert the collected ultrasonic signals, which are then recognized by technicians through listening. This has poor accuracy and requires high technical and experience skills from technicians, making it difficult to widely use. Summary of the invention

[0004] The purpose of the present invention is to provide an abnormal arc detection method, device, electronic equipment and medium based on ultrasonic wave in view of the existing technical status.

[0005] Compared with the prior art, the abnormal arc detection method of the present invention can reduce the problems of decreased accuracy and high labor costs caused by manual intervention, and the detection algorithm is simple, the detection result is more accurate, and the detection efficiency is high.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme:

[0007] First, the present invention provides an abnormal arc detection method based on ultrasonic wave, comprising:

[0008] Collect ultrasonic data;

[0009] Extracting frequency spectrum feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs;

[0010] Training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model;

[0011] The data set to be detected is input into the arc detection model to obtain a detection result.

[0012] In some embodiments, the step of extracting spectral feature values ​​according to the ultrasonic data to obtain a data set includes:

[0013] Obtain M frequency feature points within a range of 30kHz to 40kHz to obtain a frequency feature set;

[0014] A frequency spectrum feature value is obtained according to the frequency feature set, and the corresponding data set is obtained based on the frequency spectrum feature value.

[0015] In some embodiments, the step of acquiring M frequency feature points within a range of 30kHz to 40kHz includes:

[0016] Performing a fast Fourier transform on the ultrasonic data;

[0017] The frequency characteristic points within the range of 30kHz to 40kHz are intercepted according to the interception formula, and the interception formula is:

[0018] D cut =D fft(i) i=cut-start, cut-start+1,..., cut-start+(M-1),

[0019] in, freq-interval is the frequency interval, D cut is the frequency feature set, D fft(i) is the result of fast Fourier transform.

[0020] In some embodiments, the step of performing a fast Fourier transform on the ultrasonic data comprises:

[0021] D fft(i) = | FFT(x) |

[0022] Furthermore, the positive frequency portion of the ultrasonic data obtained after fast Fourier transformation is 0 kHz to 42 kHz.

[0023] In some embodiments, the step of performing a fast Fourier transform on the ultrasonic data comprises:

[0024] Performing data conversion calibration on the ultrasonic data to obtain floating-point ultrasonic data;

[0025] A fast Fourier transform is performed on the floating point ultrasonic data.

[0026] In some embodiments, the step of obtaining a spectrum feature value according to the frequency feature set and obtaining the corresponding data set based on the spectrum feature value comprises:

[0027] The frequency feature set is divided into T groups, and the average sum of the frequency feature points of each group is calculated respectively, and normalized to obtain T spectrum feature values. The T spectrum feature values ​​are used as the data set, M>T, and M and T are positive integers.

[0028] Second, the present invention provides an abnormal arc detection device based on ultrasonic wave, comprising:

[0029] An acquisition module, used for acquiring ultrasonic data;

[0030] A processing module is used to extract spectrum feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs:

[0031] A training module, used for training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model;

[0032] The detection module inputs the data set to be detected into the arc detection model to obtain and output the detection result.

[0033] In some embodiments, it includes an ultrasonic sensor, a development board and a host computer that are communicatively connected to each other, the acquisition module is located on the ultrasonic sensor, and the training module is located on the host computer.

[0034] Thirdly, the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0035] Fourthly, the present invention provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enables the processor to implement the above method.

[0036] The beneficial effects of the present invention are:

[0037] In the present invention, by obtaining normal ultrasonic data under normal conditions and arc abnormal ultrasonic data under abnormal arc conditions, and extracting their spectral characteristic values, a normal training set and an abnormal arc training set are obtained, and the KNN algorithm model is trained to obtain an arc detection model that can identify whether an abnormal arc exists. Subsequently, the data to be detected can be processed and input into the constructed arc detection model, and the constructed arc detection model can be used to quickly detect and identify whether an abnormal arc exists, and then the detection result is output. Compared with the prior art, the problem of decreased accuracy and high labor cost caused by manual intervention can be reduced, and the detection algorithm is simple, the detection result accuracy is higher, and the detection efficiency is high, which is conducive to realizing more accurate abnormal arc detection on a relatively low-cost development board, and is easier to implement and widely used. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The figure is a flow chart of the abnormal arc detection method based on ultrasonic wave of the present invention.

[0039] Figure 2 It is a schematic diagram of the principle of the abnormal arc detection method based on ultrasonic wave of the present invention.

[0040] Figure 3 It is a structural schematic diagram of the training phase of the abnormal arc detection device based on ultrasonic wave of the present invention.

[0041] Figure 4 It is a structural schematic diagram of the detection stage of the abnormal arc detection device based on ultrasonic wave of the present invention.

[0042] Figure 5 It is a schematic diagram of the process of step S2 of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in further detail below.

[0044] First, see Figure 1 to Figure 2 As shown, the present invention provides an abnormal arc detection method based on ultrasonic wave, comprising:

[0045] S1. Collect ultrasonic data;

[0046] S2. Extracting spectral feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs;

[0047] S3. Training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model;

[0048] S4. Input the data set to be detected into the arc detection model to obtain a detection result.

[0049] Among them, ultrasonic data can be collected by ultrasonic sensors, and the ultrasonic data include ultrasonic data to be detected (hereinafter referred to as "data to be detected"), ultrasonic data under normal conditions (hereinafter referred to as "normal ultrasonic data"), and ultrasonic data under abnormal arc conditions (hereinafter referred to as "arc abnormality ultrasonic data"). Among them, the normal state refers to the normal operation of the power system and the absence of abnormal arcs.

[0050] During the application of the detection method of the present invention, the process is as follows:

[0051] S100. Training phase: see Figure 1 , Figure 2 and Figure 3 As shown, build an arc detection model:

[0052] S11. Collecting ultrasonic data, obtaining normal ultrasonic data under normal conditions and arc abnormal ultrasonic data under abnormal arc conditions;

[0053] S21. Extracting spectral feature values ​​according to the normal ultrasonic data and the arc abnormal ultrasonic data to obtain a normal training set and an abnormal arc training set;

[0054] S31. Train a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model.

[0055] S200. Detection phase: see Figure 1 , Figure 2 and Figure 4 As shown, the data to be detected is collected, and the data to be detected is input into the arc detection model to obtain the detection result:

[0056] S12. Collect ultrasonic data and obtain the data to be tested;

[0057] S22. Extracting spectral feature values ​​according to the data to be detected to obtain a data set to be detected;

[0058] S42. Input the data set to be detected into the arc detection model to obtain a detection result.

[0059] In the present invention, by obtaining normal ultrasonic data under normal conditions and arc abnormal ultrasonic data under abnormal arc conditions, and extracting their spectral characteristic values, a normal training set and an abnormal arc training set are obtained, and the KNN algorithm model is trained to obtain an arc detection model that can identify whether an abnormal arc exists. Subsequently, the data to be detected can be processed and input into the constructed arc detection model, and the constructed arc detection model can be used to quickly detect and identify whether an abnormal arc exists, and then the detection result is output. Compared with the prior art, the problem of decreased accuracy and high labor cost caused by manual intervention can be reduced, and the detection algorithm is simple, the detection result accuracy is higher, and the detection efficiency is high, which is conducive to realizing more accurate abnormal arc detection on a relatively low-cost development board, and is easier to implement and widely used.

[0060] It is understandable that there can be multiple normal training sets and abnormal arc training sets to improve the accuracy of the arc detection model after training. For example, it can be an electric power system corresponding to different electrical appliance usage conditions. During training, an arc generator is used to simulate abnormal arc conditions.

[0061] In some embodiments, see Figure 1 , Figure 2 and Figure 5 As shown, in step S2, the step of extracting spectrum feature values ​​according to the ultrasonic data to obtain a data set includes:

[0062] S201. Obtain M frequency feature points within the range of 30kHz to 40kHz to obtain a frequency feature set;

[0063] S202. Obtain frequency spectrum feature values ​​according to the frequency feature set, and obtain the corresponding data set based on the frequency spectrum feature values.

[0064] Through experiments, it is found that the frequency characteristics that can distinguish whether there is an abnormal arc in the power system are mainly concentrated in the range of 30kHz to 40kHz. Therefore, by obtaining M frequency feature points in the range of 30kHz to 40kHz, the normal training set and the abnormal arc training set obtained can more accurately characterize the normal situation and the situation where there is an abnormal arc, and the detection accuracy is higher. At the same time, concentrating in the range of 30kHz to 40kHz can reduce the subsequent data processing amount and improve the efficiency of abnormal arc detection.

[0065] It can be understood that step S201 and step S202 are applicable to step S21 and also to step S22. In the operation of step S21, the data sources are normal ultrasonic data and abnormal arc ultrasonic data, respectively. After the operations of step S201 and step S202, the normal training set and the abnormal arc training set are obtained, respectively. In the operation of step S22, the data source is the data to be detected. After the operations of step S201 and step S202, the data set to be detected is obtained.

[0066] In some embodiments, see Figure 2 As shown, in step S201, the step of obtaining M frequency feature points within the range of 30kHz to 40kHz includes:

[0067] S01. Performing a fast Fourier transform on the ultrasonic data;

[0068] S02. The frequency characteristic points in the range of 30kHz to 40kHz are intercepted according to the interception formula, wherein the interception formula is:

[0069] D cut =D fft(i) i=cut-start, cut-start+1,..., cut-start+(M-1),

[0070] in, freq-interval is the frequency interval, D cut is the frequency feature set, D fft(i) is the result of fast Fourier transform.

[0071] Specifically, in step S21, a fast Fourier transform is performed on normal ultrasonic data and abnormal arc ultrasonic data, and in step S22, a fast Fourier transform is performed on the data to be detected.

[0072] The data processing algorithm is simple, which is conducive to achieving more accurate abnormal arc detection on a relatively low-cost development board and improving the efficiency of abnormal arc detection.

[0073] In some embodiments, in step S201, the step of performing fast Fourier transform on the ultrasonic data includes:

[0074] D fft(i) = | FFT(x) |

[0075] Furthermore, the positive frequency portion of the normal ultrasonic data or the arc abnormal ultrasonic data obtained after fast Fourier transformation is 0 kHz to 42 kHz.

[0076] After the ultrasonic data is fast Fourier transformed, the frequency feature points in the range of 30kHz to 40kHz are further intercepted in the positive frequency part of 0kHz to 42kHz, thereby reducing the amount of data processing while ensuring the accuracy of abnormal arc detection.

[0077] In some embodiments, in step S201, the step of performing fast Fourier transform on the ultrasonic data includes:

[0078] S011. Perform data conversion and calibration on the ultrasonic data to obtain floating-point ultrasonic data;

[0079] S012. Performing a fast Fourier transform on the floating point ultrasonic data, wherein:

[0080] D fft(i) = | FFT(x) |

[0081] Furthermore, the positive frequency portion of the normal ultrasonic data or the arc abnormal ultrasonic data obtained after fast Fourier transformation is 0 kHz to 42 kHz.

[0082] Firstly convert and calibrate the data and then perform fast Fourier transform, which is beneficial to improve the accuracy of abnormal arc detection.

[0083] In some embodiments, see Figure 2 and Figure 5 As shown, in step S202, the step of obtaining a spectrum feature value according to the frequency feature set and obtaining the corresponding data set based on the spectrum feature value includes:

[0084] The frequency feature set is divided into T groups, and the average sum of the frequency feature points of each group is calculated respectively, and normalized to obtain T spectrum feature values. The T spectrum feature values ​​are used as the data set, M>T, and M and T are positive integers.

[0085] The frequency feature set is grouped, averaged, and normalized to obtain a normal training set or an abnormal arc training set that can identify whether an abnormal arc exists. The data processing algorithm is simple and can improve the accuracy of abnormal arc detection.

[0086] In some embodiments, the step of collecting the data to be detected and inputting the data to be detected into the arc detection model includes:

[0087] Obtaining a characteristic value of a frequency spectrum to be measured according to the data to be detected, and then inputting the characteristic value of the frequency spectrum to be measured into the arc detection model, wherein the step of obtaining the characteristic value of the frequency spectrum to be measured according to the data to be detected comprises:

[0088] Performing fast Fourier transform on the data to be detected;

[0089] Intercepting M frequency feature points to be measured within the range of 30kHz to 40kHz in the positive frequency part after fast Fourier transformation;

[0090] The frequency feature points to be measured are divided into T groups, and the average sum of the frequency feature points to be measured in each group is calculated respectively, and normalized to obtain T frequency spectrum feature values ​​to be measured, M>T, and M and T are positive integers.

[0091] Preferably, M is 3000 and T is 10, which is conducive to improving the accuracy of abnormal arc detection.

[0092] For details, see Figure 2 As shown, the detection phase:

[0093] The ultrasonic sensor collects ultrasonic data and transmits it to the N-point data cache in the ARM development board (N is determined according to the performance of the development board, and N=28800 in this example). Data conversion calibration is performed to obtain accurate floating-point ultrasonic data. N-point FFT is performed on each group of N periodic data. The sampling frequency is 84000, and the positive frequency part obtained is 0-42000 (a total of 14400 frequency points), corresponding to 0Hz-42kHz. Based on the frequency point corresponding to 30kHz, 3000 frequency points are intercepted backward to extract frequency features (this group of frequency points is the frequency points that fall in the range of 30kHz-40kHz after calculation). The group of frequency points is divided into 10 groups in order with 300 points as a group. The sum of squares of each group is calculated and the final 10 spectrum feature values ​​are obtained after normalization. The 10 spectrum feature values ​​are used as the corresponding data set to be detected and input into the trained arc detection model to obtain the detection results.

[0094] Second, see Figures 2 to 4 As shown, the present invention provides an abnormal arc detection device based on ultrasonic wave, comprising:

[0095] An acquisition module, used for acquiring ultrasonic data;

[0096] A processing module is used to extract spectrum feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs:

[0097] A training module, used for training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model;

[0098] The detection module inputs the data set to be detected into the arc detection model to obtain and output the detection result.

[0099] In some embodiments, see Figure 2 As shown, it includes an ultrasonic sensor, a development board and a host computer that are communicatively connected to each other, the acquisition module is located on the ultrasonic sensor, and the training module is located on the host computer.

[0100] Among them, during the training stage, ultrasonic data (normal ultrasonic data / arc abnormal ultrasonic data) can be collected by ultrasonic sensors and saved to the host computer. After processing on the host computer, a training set (normal training set / abnormal arc training set) containing corresponding spectral feature values ​​is obtained. The training set is used to train the KNN algorithm model to obtain the arc detection model.

[0101] During the detection stage, ultrasonic data (data to be detected) can be collected by ultrasonic sensors and transmitted to the ARM development board for detection. The ARM development board calls the trained arc detection model, processes the data to be detected, and outputs the detection results.

[0102] The KNN algorithm model is trained using the training set in the host computer to build an arc detection model. The ARM development board can call the trained arc detection model to realize abnormal arc detection. Therefore, more accurate abnormal arc detection can be achieved on a low-cost development board, which is easy to implement and widely used.

[0103] Thirdly, the present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0104] Fourthly, the present invention also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enables the processor to implement the above method.

[0105] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with the field can make some changes or modify the technical contents suggested above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the solution of the present invention.

Claims

1. An abnormal arc detection method based on ultrasonic wave, characterized in that: include: Collect ultrasonic data; Extracting frequency spectrum feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs; Training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model; The data set to be detected is input into the arc detection model to obtain a detection result.

2. The abnormal arc detection method based on ultrasonic wave according to claim 1, characterized in that: The step of extracting spectrum feature values ​​according to the ultrasonic data to obtain a data set comprises: Obtain M frequency feature points within a range of 30kHz to 40kHz to obtain a frequency feature set; A frequency spectrum feature value is obtained according to the frequency feature set, and the corresponding data set is obtained based on the frequency spectrum feature value.

3. The abnormal arc detection method based on ultrasonic wave according to claim 2, characterized in that: The step of obtaining M frequency characteristic points within the range of 30kHz to 40kHz comprises: Performing a fast Fourier transform on the ultrasonic data; The frequency characteristic points within the range of 30kHz to 40kHz are intercepted according to the interception formula, and the interception formula is: D cut =D fft(i) i=cut-start,cut-start+1,...,cut-start+(M-1), in, freq-interval is the frequency interval, D cut is the frequency feature set, D fft(i) is the result of fast Fourier transform.

4. The abnormal arc detection method based on ultrasonic wave according to claim 3, characterized in that: The step of performing fast Fourier transform on the ultrasonic data comprises: D fft(i) = | FFT(x) | Furthermore, the positive frequency portion of the ultrasonic data obtained after fast Fourier transformation is 0 kHz to 42 kHz.

5. The abnormal arc detection method based on ultrasonic wave according to claim 3, characterized in that: The step of performing fast Fourier transform on the ultrasonic data comprises: Performing data conversion calibration on the ultrasonic data to obtain floating-point ultrasonic data; A fast Fourier transform is performed on the floating point ultrasonic data.

6. The abnormal arc detection method based on ultrasonic wave according to claim 2, characterized in that: The step of obtaining a frequency spectrum feature value according to the frequency feature set and obtaining the corresponding data set based on the frequency spectrum feature value comprises: The frequency feature set is divided into T groups, and the average sum of the frequency feature points of each group is calculated respectively, and normalized to obtain T spectrum feature values. The T spectrum feature values ​​are used as the data set, M>T, and M and T are positive integers.

7. An abnormal arc detection device based on ultrasonic wave, characterized in that: include: An acquisition module, used for acquiring ultrasonic data; A processing module is used to extract spectrum feature values ​​according to the ultrasonic data to obtain a data set, wherein the data set includes at least a part of a training set before the time to be detected and a data set to be detected including the time to be detected, and the training set includes a normal training set obtained based on ultrasonic data in a normal state and an abnormal arc training set obtained based on ultrasonic data in a state where an abnormal arc occurs: A training module, used for training a KNN algorithm model based on the normal training set and the abnormal arc training set to obtain an arc detection model; The detection module inputs the data set to be detected into the arc detection model to obtain and output the detection result.

8. The abnormal arc detection device based on ultrasonic wave according to claim 7, characterized in that: It comprises an ultrasonic sensor, a development board and a host computer which are communicatively connected to each other, the acquisition module is located on the ultrasonic sensor, and the training module is located on the host computer.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method described in any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that: Executable instructions are stored thereon, and when the instructions are executed by a processor, the processor implements the method described in any one of claims 1 to 6.

Images