Intelligent recognition and early warning system for ultrasonic partial discharge of power transformation equipment based on voiceprint diagnosis

Abstract

The invention discloses a power transformation equipment ultrasonic partial discharge intelligent identification and early warning system based on voiceprint diagnosis, and relates to the technical field of partial discharge detection. Comprising a data acquisition unit, a data preprocessing unit and a data processing unit. According to the intelligent recognition and early warning system for the ultrasonic partial discharge of the power transformation equipment based on voiceprint diagnosis, filtering and noise reduction are performed on ultrasonic signals through the data preprocessing unit, time-frequency analysis is performed on the ultrasonic signals of the power transformation equipment by introducing a matrix beam method, the frequency resolution of the ultrasonic signals is improved, multi-source monitoring data are fully utilized, and the accuracy of recognition and early warning of the ultrasonic partial discharge of the power transformation equipment is improved. Through combination with an advanced machine learning algorithm, comprehensive analysis and evaluation of partial discharge signals are realized, and the technical problem that other ultrasonic interference existing in the substation equipment partial discharge ultrasonic monitoring process is difficult to evaluate in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of partial discharge detection, and specifically to an intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on acoustic fingerprint diagnosis. Background Art

[0002] As a key component of the power system, the operating state of substation equipment is directly related to the safety and stability of the entire power grid. During the operation of substation equipment, partial discharge is the most important insulation defect, which will gradually deteriorate the insulation performance of the equipment and even cause equipment failures and accidents. Therefore, the real-time monitoring and analysis of the internal partial discharge signals of substation equipment are of great significance for preventing and reducing the risk of equipment failures and maintaining the safe operation of the power grid.

[0003] The invention patent with the publication number CN118503843A discloses a method, medium and system for interference evaluation of partial discharge monitoring of substation equipment, including: obtaining the operating parameters of substation equipment in real time, denoted as current curve, voltage curve and frequency curve respectively; collecting in real time the ultrasonic signals inside the substation equipment collected by the ultrasonic array sensors arranged inside the substation equipment; calculating the occurrence position and occurrence time of the ultrasonic signals; extracting the features of the ultrasonic signals to obtain ultrasonic features; calculating the difference between the characteristic root of each clustering center and the average characteristic root as the partial discharge signal deviation degree; using a pre-trained harmful and harmless partial discharge judgment model to judge whether each clustering center corresponds to harmful partial discharge or harmless partial discharge to obtain the partial discharge hazard category corresponding to the clustering center; using a pre-trained partial discharge monitoring interference evaluation model, inputting the partial discharge signal deviation degree and partial discharge hazard category corresponding to each clustering center, to obtain the partial discharge monitoring interference degree and output.

[0004] The existing methods for interference evaluation of partial discharge monitoring of substation equipment adopt a pre-trained deep clustering algorithm, and use the method of supervised learning to train a harmful and harmless partial discharge judgment model and a partial discharge monitoring interference evaluation model, realizing the intelligent evaluation of the harm degree of partial discharge signals and the overall impact on the equipment. However, in addition to defect conditions, factors such as voltage level, rain, bird calls, and on-load tap changer switching will also generate corresponding ultrasonic signals. This method has a low sampling frequency, and at the same time, the traditional time-frequency domain transformation method is limited by the Nyquist sampling theorem and can represent a small frequency range. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides an intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on acoustic fingerprint diagnosis, which solves the existing problems.

[0006] To achieve the above object, the present invention is realized through the following technical solutions: An intelligent identification and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis, comprising: a data acquisition unit for real-time acquisition of ultrasonic signals when the substation equipment discharges; a data preprocessing unit including a filtering and noise reduction module, an information extraction module and a communication module, for filtering and noise reduction processing of the collected ultrasonic signals, improving the signal quality, and transmitting the preprocessed data to the data processing unit; a data processing unit for extracting voiceprint features, establishing a voiceprint database, and performing comparison and analysis on the voiceprint features.

[0007] Preferably, the data acquisition unit includes a plurality of ultrasonic sensors arranged in a matrix, and multiple groups of data are continuously acquired by using the ultrasonic sensor array for the ultrasonic signals generated during the operation of the substation equipment at different voltage levels.

[0008] Preferably, the ultrasonic sensor uses a probe with a frequency of 20 - 100 kHz, and the resonance frequency is 40 kHz, and the monitoring sensitivity is 2 mV.

[0009] Preferably, the filtering and noise reduction module uses a band-pass filter for filtering, and the filtering effect is evaluated by calculating the signal-to-noise ratio. The formula for the signal-to-noise ratio is: ; where is the signal power, is the noise power.

[0010] Preferably, the information extraction module performs time-frequency spectrum analysis on the sound signal by the matrix pencil method to extract the time-domain and frequency-domain features of the signal.

[0011] Preferably, the analysis process of the matrix pencil method includes: The ultrasonic signal of the substation equipment can be expressed as: ; where , and respectively represent the amplitude, frequency and initial phase of the th frequency component; is the attenuation coefficient of the th frequency component; is the sampling time interval; Further expressed as an exponential model: ; where ; ; For the sampled signal x(n), construct two (N - L)×L Hankel matrices X1 and X2: ; ; is the matrix pencil parameter. Appropriate selection can reduce the influence of noise. to is a better choice; Express and as: ; ; where, ; ; ; ; then there is: ; Therefore, the generalized eigenvalues of the matrix pencil include the number of sound signal frequencies, frequencies and attenuation coefficients. Therefore, the extraction of the sound signal is transformed into the problem of solving the generalized eigenvalues of the formula: ; In the formula, is the pseudo-inverse of; After obtaining the number of sound signal frequencies, frequencies and attenuation coefficients, the signal amplitude can be obtained by solving the following least squares problem: .

[0012] Preferably, the data processing unit includes a data processing module, a database module, an analysis and comparison module, and an early warning module.

[0013] Preferably, the data processing module, based on the preprocessed signal features, identifies the sound pattern features that can characterize the partial discharge type by constructing a ResNet50 neural network; the database module is used to establish a sound pattern database including normal sound pattern features and various abnormal sound pattern features; the analysis and comparison module is used to compare and analyze the real-time collected sound pattern features with the features in the sound pattern database, and automatically identify the type, intensity and possible location of the partial discharge; the early warning module is used for intelligent early warning of the partial discharge of the substation equipment according to the analysis results.

[0014] Preferably, the data processing unit converts the time-frequency spectrogram obtained by the matrix pencil method into a grayscale image, reshapes the grayscale image into an image with a size of 224*224, and inputs it into the ResNet50 neural network.

[0015] Preferably, the network output size is 11, corresponding to short-circuit impact, partial discharge, DC bias, 110kV normal operation, 220kV normal operation, 500kV normal operation, 800kV normal operation, 1000kV normal operation, bird chirping, rain sound, and on-load tap changer switching state respectively, and generates a sound pattern database. Beneficial effects

[0016] The present invention provides an intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis. Compared with the prior art, it has the following beneficial effects: 1. The intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis filters and denoises ultrasonic signals through a data preprocessing unit, and performs time-frequency analysis on the ultrasonic signals of substation equipment by introducing the matrix pencil method, improving the frequency resolution of ultrasonic signals, making full use of multi-source monitoring data, and combining advanced machine learning algorithms to achieve a comprehensive analysis and evaluation of partial discharge signals, solving the technical problem that it is difficult for the prior art to evaluate other ultrasonic interferences existing in the ultrasonic monitoring of partial discharge of substation equipment. 2. The intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis constructs a residual neural network to achieve state classification of transformer defects, environment, switch switching, voltage level, etc., can automatically identify different types of partial discharge signals contained in ultrasonic signals, and realizes the state diagnosis and fault early warning of substation equipment through the early warning module in the data processing unit, effectively improving the analysis ability of ultrasonic signals for on-line monitoring of substation equipment. Brief Description of the Drawings

[0017] Figure 1 It is a schematic diagram of the system connection of the present invention; Figure 2 It is a schematic diagram of the data processing unit of the present invention; Figure 3 It is a schematic diagram of establishing a classification model by the ResNet50 neural network of the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Refer to Figures 1-3 The present invention provides two technical solutions: The first embodiment: An intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis, including: a data acquisition unit for real-time acquisition of ultrasonic signals generated by substation equipment during discharge. The data acquisition unit includes a plurality of ultrasonic sensors arranged in a matrix, and multiple groups of data are continuously acquired by using an ultrasonic sensor array for the ultrasonic signals generated by the substation equipment under different voltage levels during operation.

[0020] The ultrasonic sensor uses a probe with a frequency of 20 - 100 kHz, and the resonant frequency is 40 kHz, with a monitoring sensitivity of 2 mV. Among them, according to the set sampling period for detection, the uploaded data includes ultrasonic AE and the temperature and humidity of the operating environment. The ultrasonic detection circuit consists of an AD8032 preamplifier and an AD8310 logarithmic amplifier, which accurately converts the electromagnetic wave signal into an equivalent decibel scale value at the DC output at the input, and adjusts and amplifies the signal collected by the sensor. At the same time, the digital temperature and humidity IC outputs digital signals through the processor, and finally uses the 470 MHz LoRa wireless communication technology to transmit the data to the data processing unit through the wireless transceiver SX1268 chip.

[0021] Specifically, the continuous detection mode is mainly used to quickly obtain the signal characteristics of the device under test, and has the characteristics of intuitive display and fast response speed. It can display the effective value, periodic peak value of the measured signal within one power frequency cycle, and the correlation between the measured signal and the frequency. By combining different parameter values, it can quickly judge whether there is abnormal partial discharge in the substation equipment and the possible discharge type.

[0022] When the frequency component is large in the continuous detection mode, it can enter the phase detection mode, which is mainly used to further confirm the specific phase where the abnormal signal occurs, so as to judge whether the abnormal signal is correlated with the power frequency voltage, and then judge whether the abnormal signal is a discharge signal and the potential discharge type. Since the generation of partial discharge signals is correlated with the power frequency electric field, the power frequency voltage can be used as a reference quantity, and by observing whether there is an aggregation effect in the occurrence phase of the measured signal, it can be judged whether the measured signal is caused by internal discharge of the equipment.

[0023] When the effective value or periodic peak value is large but the frequency component is small in the continuous detection mode, it can enter the pulse detection mode, which is mainly used for further confirmation of free particle defects. Each time a particle collides with the shell, a broadband transient acoustic pulse is emitted, which propagates back and forth in the shell. The acoustic signal of this kind of particle is a mixed signal of partial discharge at the end of the particle and the particle colliding with the shell. The pulse mode can record the time and the generated pulse amplitude each time the particle collides with the shell, and display it in the form of a flight diagram.

[0024] The data preprocessing unit, including a filtering and noise reduction module, an information extraction module, and a communication module, is used to filter and reduce the noise of the collected ultrasonic signals, improve the signal quality, and transmit the preprocessed data to the data processing unit.

[0025] The filtering and noise reduction module uses a band - pass filter for filtering, and evaluates the filtering effect by calculating the signal - to - noise ratio after filtering. The formula for the signal - to - noise ratio is: ; where is the signal power, is the noise power.

[0026] The information extraction module performs time-frequency spectrum analysis on the sound signal by the matrix pencil method to extract the time-domain and frequency-domain features of the signal.

[0027] The data processing unit is used to extract voiceprint features, establish a voiceprint database, and perform comparison and analysis on the voiceprint features. The data processing module identifies the voiceprint features that can characterize the partial discharge type through constructing a ResNet50 neural network based on the preprocessed signal features; the database module is used to establish a voiceprint database containing normal voiceprint features and various abnormal voiceprint features; the analysis and comparison module is used to compare and analyze the voiceprint features collected in real time with the features in the voiceprint database to automatically identify the type, intensity, and possible location of the partial discharge; the warning module is used to perform intelligent warning on the partial discharge of the substation equipment according to the analysis results.

[0028] The data processing unit converts the time-frequency spectrum diagram obtained by the matrix pencil method into a grayscale image, reshapes the grayscale image into an image with a size of 224*224, and inputs it into the ResNet50 neural network; the network output has a size of 11, corresponding to short-circuit impact, partial discharge, DC bias, normal operation at 110 kV, normal operation at 220 kV, normal operation at 500 kV, normal operation at 800 kV, normal operation at 1000 kV, bird song, rain sound, and on-load tap changer switching state respectively, and generates a voiceprint database.

[0029] The second implementation manner: The analysis process of the matrix pencil method includes: The ultrasonic signal of the substation equipment can be expressed as: ; where , and respectively represent the amplitude, frequency, and initial phase of the th frequency component; is the th frequency component's attenuation coefficient; is the sampling time interval; further expressed as an exponential model: ; where ; .

[0030] For the sampled signal x(n), construct two (N-L)×L Hankel matrices X1 and X2: ; ; is the matrix pencil parameter, and proper selection can reduce the influence of noise, to is A better choice; transform , as: ; ; where ; ; ; ; then we have: ; Therefore, the generalized eigenvalues of the matrix pencil include the number of sound signal frequencies, frequencies, and attenuation coefficients. Thus, obtaining the sound signal is transformed into the problem of solving the generalized eigenvalues of the following formula: ; In the formula, is the pseudo-inverse; after obtaining the number of sound signal frequencies, frequencies, and attenuation coefficients, the signal amplitude can be obtained by solving the following least squares problem: .

[0031] By equipping the intelligent front-end board in the ResNet50 neural network with an embedded chip for edge computing, that is, an embedded chip dedicated to artificial intelligence image processing, intelligent video image recognition is performed on the edge side; the ResNet50 neural network establishes a classification model as Figure 2 shown, and the cloud sends the trained and evaluated substation equipment defect image diagnosis model to the cloud-edge collaboration component of the intelligent front-end board through the cloud-edge collaboration component on the cloud side.

[0032] The present invention can be a system, method, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.

[0033] A computer-readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punch card or raised structures in grooves having instructions stored thereon, and any suitable combination of the foregoing. The computer-readable storage medium as used herein is not construed as being a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagated through a waveguide or other transmission medium (e.g., an optical pulse through an optical fiber cable), or an electrical signal transmitted through a wire.

[0034] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to respective computing / processing devices, or can be downloaded to an external computer or an external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include a copper transmission cable, an optical fiber transmission, a wireless transmission, a router, a firewall, a switch, a gateway computer, and / or an edge server. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in each computing / processing device.

[0035] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages such as Smalltalk, C++, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or, alternatively, may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, by using the status information of the computer-readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field-programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer-readable program instructions to implement various aspects of the present disclosure. Meanwhile, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited, and conventional devices can be used.

[0036] It should be noted that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0037] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The ultrasonic partial discharge intelligent identification and early warning system for substation equipment based on voiceprint diagnosis is characterized by: include: A data acquisition unit, used for real-time acquisition of ultrasonic signals of the substation equipment during discharge; The data preprocessing unit includes a filtering and noise reduction module, an information extraction module and a communication module, which are used to filter and reduce the noise of the collected ultrasonic signals, improve the signal quality, and transmit the preprocessed data to the data processing unit; the data processing unit is used to extract voiceprint features, establish a voiceprint database, and compare and analyze the voiceprint features.

2. According to claim 1, the ultrasonic partial discharge intelligent identification and early warning system for substation equipment based on voiceprint diagnosis is characterized by: The data acquisition unit includes a plurality of ultrasonic sensors arranged in a matrix, and uses an ultrasonic sensor array to continuously acquire ultrasonic signals generated when the substation equipment is in operation at different voltage levels to obtain a plurality of data sets.

3. According to claim 2, the ultrasonic partial discharge intelligent identification and early warning system for substation equipment based on voiceprint diagnosis is characterized by: The ultrasonic sensor uses a probe with a frequency of 20 to 100 kHz, a resonance frequency of 40 kHz, and a monitoring sensitivity of 2 mV.

4. According to claim 1, the intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis is characterized by: The filtering and noise reduction module uses a bandpass filter for filtering. After filtering, the filtering effect is evaluated by calculating the signal-to-noise ratio. The formula for the signal-to-noise ratio is: ;in, is the signal power, is the noise power.

5. According to claim 1, the intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis is characterized by: The information extraction module performs time-frequency spectrum analysis on the sound signal through a matrix bundle method to extract the time domain and frequency domain features of the signal.

6. According to claim 5, the ultrasonic partial discharge intelligent identification and early warning system for substation equipment based on voiceprint diagnosis is characterized by: The analysis process of the matrix beam method includes: the ultrasonic signal star of the substation equipment can be expressed as: ;in, , and Respectively represent The amplitude, frequency and initial phase of each frequency component; For the Attenuation coefficient of frequency components; is the sampling time interval; Further expressed as an exponential model: ;in, ; ; For the sampled signal x(n), construct two (NL)×L Hankel matrices X1 and X2: ； ; is the matrix bundle parameter; , It is expressed as: ; ;in, ； ； ; ; then: ; Therefore, the matrix bundle The generalized eigenvalues ​​include the number of sound signal frequencies, frequency and attenuation coefficient, so the solution of the sound signal is transformed into the problem of solving the generalized characteristics: ; In the formula, for The pseudo-inverse of ; After obtaining the number of sound signal frequencies, frequency and attenuation coefficient, the signal amplitude can be obtained by solving the following least squares problem: 。 7. According to claim 1, the ultrasonic partial discharge intelligent identification and early warning system for substation equipment based on voiceprint diagnosis is characterized by: The data processing unit includes a data processing module, a database module, an analysis and comparison module and an early warning module.

8. The intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis according to claim 7 is characterized by: The data processing module identifies the soundprint features that can characterize the type of partial discharge by constructing a ResNet50 neural network based on the preprocessed signal features; The database module is used to establish a voiceprint database containing normal voiceprint features and multiple abnormal voiceprint features; the analysis and comparison module is used to compare and analyze the voiceprint features collected in real time with the features in the voiceprint database, and automatically identify the type, intensity and possible location of partial discharge; the early warning module is used to provide intelligent early warning of partial discharge of substation equipment based on the analysis results.

9. According to claim 1, the intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis is characterized by: The data processing unit converts the time-frequency spectrum obtained by the matrix bundle method into a grayscale image, reshapes the grayscale image into an image of size 224*224, and inputs it into the ResNet50 neural network.

10. The intelligent recognition and early warning system for ultrasonic partial discharge of substation equipment based on voiceprint diagnosis according to claim 9 is characterized by: The network output size is 11, which respectively corresponds to short circuit impact, partial discharge, DC bias magnetism, 110kV normal operation, 220kV normal operation, 500kV normal operation, 800kV normal operation, 1000kV normal operation, bird singing, rain sound and on-load switch switching state to generate a voiceprint database.

Images