High-voltage circuit breaker bolt loosening fault diagnosis method and related equipment
By obtaining and processing bolt loose sound signals in a high-voltage circuit breaker, building a full-pole model and spectrum-enhanced voiceprint characteristics, forming a voiceprint prototype, solving the problem of huge sample data in the existing technology and achieving efficient bolt loosening fault diagnosis.
Patent Information
- Application Number
- CN202411647071.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-18
- Publication Date
- 2025-02-28
AI Technical Summary
The existing fault diagnosis method for bolt loosening of high-voltage circuit breaker requires a large amount of sample data, making it difficult to implement application in actual projects.
By obtaining the occurrence and termination time of the bolt loose sound signal, dividing it into frame segments, building an all-pole model, solving the all-pole filter coefficients, calculating the group delay function, performing discrete cosine transformation to form spectrally enhanced voiceprint features, and forming a voiceprint prototype in the embedded space. Finally, fault diagnosis is achieved by calculating the distance between the voiceprint to be diagnosed and the voiceprint prototype.
The fault diagnosis of high-voltage circuit breaker bolt loosening in the case of insufficient actual samples is realized, which improves the accuracy and efficiency of diagnosis and avoids relying on a huge sample data set.
Smart Images
Figure CN119541541A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of high-voltage circuit breaker fault diagnosis, and specifically relates to the field of high-voltage circuit breaker bolt loosening fault diagnosis, and in particular to a high-voltage circuit breaker bolt loosening fault diagnosis method and related equipment. Background Art
[0002] Generator circuit breakers are key electrical equipment in the busbar tunnel of a pumped storage power station. Due to the large operating power of their mechanisms and the fast opening and closing speeds, it is very easy to cause strong vibrations between the arc extinguishing chamber body and the entire transmission structure. The vibration can cause bolt loosening failures, which will eventually lead to catastrophic failures in the entire circuit system. Therefore, it is crucial to develop a fault diagnosis method for circuit breaker bolt loosening and timely evaluate the looseness of circuit breaker bolts for the safe operation of the electrical system in the busbar tunnel.
[0003] For the fault diagnosis of large equipment structures, the on-site installation of signal generating devices such as exciters is difficult to achieve; in addition, the on-site installation and maintenance of sensors will bring huge economic expenditures and manpower consumption. However, the dynamic characteristics of bolts under different degrees of looseness make the sound signals they excite different. Experienced experts can distinguish the degree of looseness of bolts by the slight difference in sound; and the circuit breaker bolt loosening fault diagnosis method based on sound signals does not rely on signal generating devices and sensors. Therefore, the voiceprint recognition technology used to characterize the loose state of bolts is in urgent need of development. At present, bolt loosening voiceprint recognition technology is increasingly dependent on deep learning technology represented by convolutional neural networks; deep learning generally requires a large amount of data to build a data set, but due to the large number of bolts on the circuit breaker, it is obviously unrealistic to establish a huge data set for each bolt of the circuit breaker; that is, it is difficult to use the big data sample mode to carry out bolt loosening voiceprint recognition in engineering practice.
[0004] It can be seen that the existing theoretical methods for high-voltage circuit breaker loose bolt fault diagnosis usually require the use of a large amount of sample data. However, in actual engineering, it is difficult to implement and apply because of the need to collect huge amounts of data for each bolt. Summary of the invention
[0005] The purpose of the present invention is to provide a method and related equipment for diagnosing loose bolt faults of high-voltage circuit breakers, so as to solve the technical problem that the existing theoretical method for diagnosing loose bolt faults of high-voltage circuit breakers usually requires the use of a large amount of sample data, but in actual engineering, it is difficult to implement the application due to the need to collect huge data for each bolt.
[0006] In order to achieve the above object, the present invention adopts the following technical solution: A method for diagnosing a loose bolt fault of a high-voltage circuit breaker, comprising: Obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; Constructing an all-pole model based on the prediction analysis results of the frame segments; The group delay function is obtained by solving the coefficients of the all-pole filter based on the all-pole model; The discrete cosine transform result based on the group delay function forms the spectrum enhanced voiceprint feature; Based on the spectrum-enhanced voiceprint features, the corresponding voiceprint prototype in the embedding space is obtained; Based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is obtained, so as to realize the loose bolt fault diagnosis of the high-voltage circuit breaker according to the probability.
[0007] Furthermore, the specific steps of dividing the occurrence time and the termination time into frame segments are as follows: A window function is applied to the occurrence time and the end time of the extracted bolt loosening sound signal to divide it into multiple frame segments.
[0008] Furthermore, the specific steps of constructing the full-pole model based on the prediction analysis results of the frame segments are as follows: Linear prediction analysis is performed on the frame segments to obtain prediction analysis results, and an all-pole model is constructed based on the prediction analysis results; wherein, the formula of the all-pole model is as follows:
[0009] Where H(w) is the phase response, G is the signal-dependent gain, P is the all-pole model order, a(k) is the all-pole filter coefficient, j is the imaginary unit, ω is the frequency, and k is the delay associated with the pole.
[0010] Furthermore, the specific steps of obtaining the group delay function from the all-pole filter coefficient solution based on the all-pole model are as follows: The all-pole filter coefficients of the all-pole model are solved based on the Yule-Walker equation. The specific formula is as follows:
[0011] In the formula, Frame fragment The autocorrelation function of is:
[0012] A phase response formed by the all-pole filter coefficients is calculated according to the all-pole filter coefficients; The negative derivative of the phase response is calculated to obtain a group delay function, wherein the calculation of the negative derivative of the phase response is achieved through the sample difference of the bolt loosening sound signal.
[0013] Furthermore, the specific steps of forming the spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function are as follows: The discrete cosine transform result is obtained by performing discrete cosine transform on the group delay function, and the spectrum enhanced voiceprint feature is formed according to the discrete cosine transform result; wherein, the spectrum enhanced voiceprint feature is formed by taking the discrete cosine transform result of the first thirteen order coefficients excluding the zero order.
[0014] Furthermore, the specific steps of obtaining the corresponding voiceprint prototype in the embedding space based on the spectrum enhanced voiceprint feature are as follows: Construct a convolutional neural network as the embedding space of the framework based on the historical voiceprint prototype and the historical spectrum enhanced voiceprint features; The formed spectrum enhanced voiceprint features are mapped to the embedding space by the convolutional neural network to obtain the voiceprint prototype. The prototype of each category By encoding function Sure, is the learnable parameter of the encoding function, and the formula is as follows:
[0015] In the formula, S k is the sample set of category k, Indicates Sample points; Among them, the convolutional neural network consists of convolutional layers, pooling layers, regularization layers, and activation layers.
[0016] Furthermore, the specific steps of obtaining the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype are as follows: According to the voiceprint to be diagnosed and the voiceprint prototype, the distance between the two is calculated. The specific formula is as follows:
[0017] In the formula, v 1 and v 2 are two vectors whose distances are to be determined; M represents the dimension of the vector, v 1,i and v 2,i Representation vector v 1 and v The i-th dimension component of 2; According to the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is calculated. The specific formula is as follows: .
[0018] A high-voltage circuit breaker bolt loosening fault diagnosis system, comprising: A frame segment division module is used to obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; An all-pole model building module, used to build an all-pole model based on the prediction analysis results of the frame fragments; A group delay function calculation module, used for obtaining a group delay function based on a solution result of an all-pole filter coefficient of an all-pole model; A spectrum enhancement voiceprint feature forming module, used for forming a spectrum enhancement voiceprint feature based on a discrete cosine transform result of a group delay function; A mapping module, used to obtain the corresponding voiceprint prototype in the embedding space based on the spectrum-enhanced voiceprint features; The fault diagnosis module is used to obtain the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the loose bolt fault diagnosis of the high-voltage circuit breaker according to the probability.
[0019] A device comprising: Memory for storing computer programs; A processor is used to implement the steps of the above-mentioned high-voltage circuit breaker loose bolt fault diagnosis method when executing the computer program.
[0020] A computer-readable storage medium stores a computer program, which is used to implement the steps of the above-mentioned high-voltage circuit breaker bolt loosening fault diagnosis method when executed by a processor.
[0021] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides a method for diagnosing a loose bolt fault of a high-voltage circuit breaker. The method sequentially extracts the occurrence time and the termination time of a bolt loosening sound signal, constructs an all-pole model, solves the all-pole filter coefficient, calculates the group delay function, performs discrete cosine transform to form a spectrum enhancement voiceprint feature, and finally forms a voiceprint prototype in an embedding space, and determines the attribution probability of the voiceprint to be diagnosed by calculating the distance between the voiceprint to be diagnosed and the voiceprint prototype, thereby realizing accurate diagnosis of the loose bolt fault; the method does not rely on a large sample data set, but represents different loose bolt states based on the voiceprint feature of the sound signal emitted by the circuit breaker, enhances the resonance peak of the sound signal spectrum, improves the characterization effect of the loose bolt of the high-voltage circuit breaker, and realizes the loose bolt fault diagnosis of the high-voltage circuit breaker when the actual sample is insufficient; the method effectively solves the problem that the existing technology is difficult to implement and apply due to the large sample data, has stronger practicality and operability, and can be widely used in engineering practice.
[0022] Preferably, in the present invention, the bolt loosening sound signal is divided into multiple frame segments by applying a window function, which can effectively reduce the edge effect in signal analysis and improve the accuracy of signal processing; at the same time, this division method also facilitates subsequent more detailed analysis and processing of the signal.
[0023] Preferably, in the present invention, a full-pole model is constructed based on linear prediction analysis of frame segments, which can accurately describe the spectral characteristics of the bolt loosening sound signal; the construction of the full-pole model provides a basis for the subsequent solution of the group delay function and the formation of spectral enhancement soundprint features, which helps to improve the accuracy of fault diagnosis.
[0024] Preferably, in the present invention, the group delay function is obtained by solving the all-pole filter coefficients and calculating the negative derivative of the phase response, which can further reveal the dynamic characteristics of the bolt loosening sound signal; the solution of the group delay function provides important information for the subsequent discrete cosine transform and the formation of spectrum enhancement soundprint features, which helps to extract more accurate fault features.
[0025] Preferably, in the present invention, the group delay function is discrete cosine transformed and the first thirteen-order coefficients are selected to form a spectrum enhanced voiceprint feature, which can effectively extract key information in the bolt loosening sound signal while ensuring accuracy. This feature extraction method helps to improve the efficiency and accuracy of fault diagnosis.
[0026] Preferably, in the present invention, by constructing a convolutional neural network as the embedding space of the framework and mapping the spectrum enhanced voiceprint features to the embedding space to obtain the voiceprint prototype, automatic classification and recognition of bolt loosening sound signals can be achieved. This method not only improves the intelligence level of fault diagnosis, but also enhances the robustness and adaptability of the system.
[0027] Preferably, in the present invention, by calculating the distance between the soundprint to be diagnosed and the soundprint prototype, and calculating the probability that the soundprint to be diagnosed belongs to the soundprint prototype based on the distance, accurate diagnosis of bolt loosening faults can be achieved; this method not only improves the accuracy of diagnosis, but also provides quantitative diagnostic results, which is convenient for subsequent processing and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS
[0028] Figure 1 A flow chart of a method for diagnosing a loose bolt fault of a high-voltage circuit breaker provided by an embodiment of the present invention; Figure 2 A schematic diagram of a spectrum enhanced voiceprint feature provided by an embodiment of the present invention; wherein (a) is the normalized amplitude; (b) is the frequency decibel; (c) is the frequency voiceprint coefficient; Figure 3 A schematic diagram of a fault diagnosis result of a loose bolt of a high-voltage circuit breaker provided by an embodiment of the present invention; Figure 4 A flow chart of a method for diagnosing a loose bolt fault of a high-voltage circuit breaker provided by the present invention; Figure 5 A structural schematic diagram of a high-voltage circuit breaker bolt loosening fault diagnosis system provided by the present invention. DETAILED DESCRIPTION
[0029] Example 1 As described in the background technology, currently, bolt loosening soundprint recognition technology is increasingly relying on deep learning technology represented by convolutional neural networks; deep learning generally requires a large amount of data to build a data set, but due to the large number of bolts on the circuit breaker, it is obviously unrealistic to establish a huge data set for each bolt of the circuit breaker; that is to say, it is difficult to use the big data sample model to carry out bolt loosening soundprint recognition in actual engineering.
[0030] Therefore, processing the sound signal of loose bolts of circuit breakers, effectively extracting the soundprint features of loose bolts, and realizing the fault diagnosis of loose bolts of circuit breakers in the case of small samples are the problems that need to be solved urgently in current engineering practice.
[0031] In order to solve the above problems, the present invention provides a method for diagnosing loose bolt faults of high-voltage circuit breakers. The method can complete the loose bolt fault diagnosis in the case of insufficient samples by enhancing the soundprint of the spectrum.
[0032] like Figure 1 As shown, this embodiment provides a method for diagnosing a loose bolt fault of a high-voltage circuit breaker, that is, a method for diagnosing a loose bolt fault of a high-voltage circuit breaker based on spectrum enhancement voiceprint and small sample network is proposed, and the steps are as follows: S1. Collecting different bolt loosening sound data sample sets of high-voltage circuit breakers, wherein the different bolt loosening sound data sample sets include a bolt non-loosening state sample set and a bolt loosening state sample set. In this embodiment, the bolt loosening sound signal is extracted from the bolt loosening state sample set; S2, extracting the occurrence time and end time of the bolt loosening sound signal, and applying a window function to the sound signal to divide it into frame segments; S3, performing linear prediction analysis on the frame segments to establish an all-pole model; calculating all-pole filter coefficients of the all-pole model S4, let the signal correlation gain be 1, and calculate the full-pole filter coefficient a ( k ) forms a phase response H( ω ); S5. Calculate the phase response H( ω ) to determine the group delay function; S6, performing discrete cosine transform on the group delay function to form a spectrum enhanced voiceprint feature; S7. Calculate the voiceprint prototype of each spectrum-enhanced voiceprint feature in the embedded space, calculate the distance between the voiceprint to be diagnosed and the prototype voiceprint in the embedded space, and further calculate the probability that the voiceprint to be diagnosed belongs to a certain voiceprint prototype; finally, judge the voiceprint to be diagnosed by the calculated probability, so as to complete the fault diagnosis of loose bolts of high-voltage circuit breaker.
[0033] In a further embodiment, in step S2, the interval between the occurrence time and the termination time of the sound signal is 150ms, and the signal is divided into frame segments using a Hanning window with a length of 32ms and a frame shift of 8ms; In a further embodiment, in step S3, the full-pole model is , where G is the signal-related gain, p is the model order, a ( k ) are all-pole filter coefficients; j is an imaginary unit; ω is the frequency; k is the delay associated with the pole; the all-pole filter coefficients are obtained by the following Yule-Walker equation:
[0034] in, For speech segments (frame segments) The autocorrelation function of:
[0035] In a further embodiment, in step S5, the calculation of the negative derivative is achieved by sample difference; In a further embodiment, in step S6, the discrete cosine transform result takes the first thirteen order coefficients excluding the zero order.
[0036] In a further embodiment, in step S7, each voiceprint feature is mapped to an embedding space by a convolutional neural network, and the convolutional neural network structure consists of a convolutional layer, a pooling layer, a regularization layer, and an activation layer; the voiceprint prototype of each category By encoding function Determine, among which are learnable parameters of the encoding function.
[0037]
[0038] In the formula, S k is the sample set of category k, Indicates Sample points; Voiceprint to be diagnosed x Voiceprint Prototype Distance d Calculated by the following formula:
[0039] In the formula, v 1 and v 2 are two vectors whose distances are to be determined; M represents the dimension of the vector, v 1,i and v 2,i Representation vector v 1 and v The i-th dimension component of 2; The probability that the voiceprint to be diagnosed belongs to a certain prototype is calculated by the following formula:
[0040] It can be seen that the method for diagnosing a loose bolt fault of a high-voltage circuit breaker provided by this embodiment has the following advantages over the existing fault diagnosis method: First, the high-voltage circuit breaker bolt loosening fault diagnosis method provided in this embodiment represents different bolt loosening states based on the sound print characteristics of the sound signal emitted by the circuit breaker, enhances the resonance peak of the sound signal spectrum, and improves the characterization effect of high-voltage circuit breaker bolt loosening.
[0041] Second, the high-voltage circuit breaker bolt loosening fault diagnosis method provided in this embodiment realizes the high-voltage circuit breaker bolt loosening fault diagnosis when the actual sample is insufficient by comparing the distance between the soundprint to be confirmed and the prototype soundprint in the embedded space.
[0042] In order to further verify the diagnostic effect of this fault diagnosis method, this method is now implemented in a specific application scenario, as follows: This fault diagnosis method is used to diagnose the loosening fault of the transmission main shaft bolt of the high-voltage circuit breaker. The bolt size is M10 and the rated pre-tightening torque is 22.5Nm. The torque diagnosed in this case is divided into 10 groups, from 0 to 22.5Nm, with an interval of 2.5Nm. The sound signal of the circuit breaker action is collected 105 times under each torque condition; 50 groups of sound signal data under each torque are divided into training sets, and the remaining 55 groups of sound signals are divided into test sets.
[0043] According to steps S2 to S6 in this embodiment, the voiceprint features of all sound data are extracted; wherein the voiceprint extraction of a certain frame signal is as follows: Figure 2 As shown, specifically Figure 2 As shown in (a), (b) and (c) in Figure 2. Figure 2 (b) and Figure 2 (c) It can be seen that the extracted spectrum enhanced voiceprint features strengthen the spectrum change trend, enhance the resonance peak of the sound signal spectrum, and improve the characterization effect of the loose bolts of the high-voltage circuit breaker. Figure 3As shown, the present embodiment has an overall accuracy rate of 97.8% for diagnosing 10 types of loose bolt faults, thereby improving the accuracy of loose bolt fault diagnosis.
[0044] Example 2 like Figure 4 As shown, this embodiment provides a method for diagnosing a loose bolt fault of a high-voltage circuit breaker, comprising the following steps: Obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; The specific steps of dividing the occurrence time and the termination time into frame segments are as follows: A window function is applied to the occurrence time and the end time of the extracted bolt loosening sound signal to divide it into multiple frame segments.
[0045] The full-pole model is constructed based on the prediction analysis results of the frame fragments. The specific steps are as follows: Linear prediction analysis is performed on the frame segments to obtain prediction analysis results, and an all-pole model is constructed based on the prediction analysis results; wherein, the formula of the all-pole model is as follows:
[0046] Where H(w) is the phase response, G is the signal-dependent gain, P is the all-pole model order, a(k) is the all-pole filter coefficient, j is the imaginary unit, ω is the frequency, and k is the delay associated with the pole.
[0047] The group delay function is obtained by solving the coefficients of the all-pole filter based on the all-pole model. The specific steps are as follows: The all-pole filter coefficients of the all-pole model are solved based on the Yule-Walker equation. The specific formula is as follows:
[0048] In the formula, Frame fragment The autocorrelation function of is:
[0049] A phase response formed by the all-pole filter coefficients is calculated according to the all-pole filter coefficients; The negative derivative of the phase response is calculated to obtain a group delay function, wherein the calculation of the negative derivative of the phase response is achieved through the sample difference of the bolt loosening sound signal.
[0050] The discrete cosine transform result based on the group delay function forms the spectrum enhanced voiceprint feature. The specific steps are as follows: The discrete cosine transform result is obtained by performing discrete cosine transform on the group delay function, and the spectrum enhanced voiceprint feature is formed according to the discrete cosine transform result; wherein, the spectrum enhanced voiceprint feature is formed by taking the discrete cosine transform result of the first thirteen order coefficients excluding the zero order.
[0051] Based on the spectrum enhanced voiceprint features, the corresponding voiceprint prototype in the embedding space is obtained. The specific steps are as follows: Construct a convolutional neural network as the embedding space of the framework based on the historical voiceprint prototype and the historical spectrum enhanced voiceprint features; The formed spectrum enhanced voiceprint features are mapped to the embedding space by the convolutional neural network to obtain the voiceprint prototype. The prototype of each category By encoding function Sure, is the learnable parameter of the encoding function, and the formula is as follows:
[0052] In the formula, S k is the sample set of category k, Indicates Sample points; Among them, the convolutional neural network consists of convolutional layers, pooling layers, regularization layers, and activation layers.
[0053] Based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is obtained, so as to realize the loose bolt fault diagnosis of the high-voltage circuit breaker according to the probability; The specific steps are as follows: According to the voiceprint to be diagnosed and the voiceprint prototype, the distance between the two is calculated. The specific formula is as follows:
[0054] In the formula, v 1 and v 2 are two vectors whose distances are to be determined; M represents the dimension of the vector, v 1,i and v 2,i Representation vector v 1 and v The i-th dimension component of 2; According to the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is calculated. The specific formula is as follows: .
[0055] like Figure 5As shown, the present embodiment further provides a high-voltage circuit breaker bolt loosening fault diagnosis system, comprising: a frame segment division module, used to obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; a full-pole model construction module, used to construct a full-pole model based on the prediction analysis results of the frame segments; a group delay function calculation module, used to obtain the group delay function based on the full-pole filter coefficient solution result of the full-pole model; a spectrum enhanced voiceprint feature formation module, used to form a spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function; a mapping module, used to obtain the corresponding voiceprint prototype in the embedded space based on the spectrum enhanced voiceprint feature; a fault diagnosis module, used to obtain the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the high-voltage circuit breaker bolt loosening fault diagnosis according to the probability.
[0056] The present invention also provides a device, comprising: a memory for storing a computer program; and a processor for implementing the steps of the method for diagnosing a loose bolt fault of a high-voltage circuit breaker when executing the computer program.
[0057] When the processor executes the computer program, the above-mentioned steps of high-voltage circuit breaker bolt loosening fault diagnosis are implemented, for example: obtaining the occurrence time and end time of the bolt loosening sound signal, and dividing the occurrence time and end time into frame segments; constructing a full-pole model based on the prediction analysis results of the frame segments; obtaining a group delay function based on the solution results of the full-pole filter coefficients of the full-pole model; forming a spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function; obtaining a corresponding voiceprint prototype in the embedded space based on the spectrum enhanced voiceprint feature; obtaining the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the high-voltage circuit breaker bolt loosening fault diagnosis according to probability.
[0058] Alternatively, the processor implements the functions of each module in the above system when executing the computer program, for example: a frame segment division module, used to obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; a full-pole model construction module, used to construct a full-pole model based on the prediction analysis results of the frame segments; a group delay function calculation module, used to obtain the group delay function based on the full-pole filter coefficient solution result of the full-pole model; a spectrum enhanced voiceprint feature formation module, used to form a spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function; a mapping module, used to obtain the corresponding voiceprint prototype in the embedded space based on the spectrum enhanced voiceprint feature; a fault diagnosis module, used to obtain the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the high-voltage circuit breaker bolt loosening fault diagnosis according to the probability.
[0059] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of completing preset functions, and the instruction segments are used to describe the execution process of the computer program in the high-voltage circuit breaker bolt loosening fault diagnosis device. For example, the computer program can be divided into a frame segment division module, a full-pole model construction module, a group delay function calculation module, a group delay function calculation module, a spectrum enhanced voiceprint feature formation module, a mapping module and a fault diagnosis module; the specific functions of each module are as follows: a frame segment division module, used to obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; a full-pole model construction module, used to build a full-pole model based on the prediction analysis results of the frame segments; a group delay function calculation module, used to obtain the group delay function based on the full-pole filter coefficient solution result of the full-pole model; a spectrum enhanced voiceprint feature formation module, used to form a spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function; a mapping module, used to obtain the corresponding voiceprint prototype in the embedded space based on the spectrum enhanced voiceprint feature; a fault diagnosis module, used to obtain the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the high-voltage circuit breaker bolt loosening fault diagnosis according to the probability.
[0060] The high-voltage circuit breaker bolt loosening fault diagnosis device can be a computing device such as a desktop computer, a notebook, a PDA, and a cloud server. The high-voltage circuit breaker bolt loosening fault diagnosis device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the above is an example of a high-voltage circuit breaker bolt loosening fault diagnosis device, and does not constitute a limitation on the high-voltage circuit breaker bolt loosening fault diagnosis device, and may include more components than the above, or a combination of certain components, or different components. For example, the high-voltage circuit breaker bolt loosening fault diagnosis device may also include input and output devices, network access devices, buses, etc.
[0061] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc. The processor is the control center of the high-voltage circuit breaker loose bolt fault diagnosis, and uses various interfaces and lines to connect various parts of the entire high-voltage circuit breaker loose bolt fault diagnosis device.
[0062] The memory can be used to store the computer program and / or module, and the processor implements various functions of the high-voltage circuit breaker bolt loosening fault diagnosis device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.
[0063] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.
[0064] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for diagnosing a loose bolt fault of a high-voltage circuit breaker are implemented.
[0065] If the module / unit integrated in the high-voltage circuit breaker loose bolt fault diagnosis system 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 storage medium.
[0066] Based on such understanding, the present invention implements all or part of the processes in the above-mentioned method for diagnosing the loose bolt fault of a high-voltage circuit breaker, and can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned method for diagnosing the loose bolt fault of a high-voltage circuit breaker can be implemented. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or preset intermediate form, etc.
[0067] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.
[0068] It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electrical carrier signals and telecommunication signals.
[0069] It can be seen that the present invention provides a method for diagnosing a loose bolt fault of a high-voltage circuit breaker, which has the following advantages over the existing fault diagnosis methods: Firstly, this method accurately obtains the occurrence and termination time of the bolt loosening sound signal and subdivides it into frame segments, which provides a high-quality data basis for subsequent analysis and helps to improve the accuracy of diagnosis; secondly, the full-pole model and group delay function are used to analyze the sound signal, which can deeply explore the sound features, especially the spectrum enhancement soundprint features formed by discrete cosine transform, which effectively highlights the key information of the bolt loosening sound and enhances the sensitivity of diagnosis; thirdly, the embedding space based on convolutional neural network is constructed to map the spectrum enhancement soundprint features to the soundprint prototype. This process not only improves the robustness of feature representation, but also simplifies the complexity of fault diagnosis through prototype matching, making the diagnosis process more efficient; finally, by calculating the distance between the soundprint to be diagnosed and the soundprint prototype, and then determining its attribution probability, the quantitative evaluation of fault diagnosis is realized, which provides a scientific basis for the timely discovery and treatment of the loose bolt fault of the high-voltage circuit breaker, and helps to improve the stability and safety of the power system. In summary, this technical solution can realize the diagnosis of the loose bolt fault of the high-voltage circuit breaker when the actual sample is insufficient, and it has significant advantages in improving the accuracy, sensitivity and efficiency of diagnosis, and has important practical application value.
[0070] The above embodiment is only one of the implementation methods that can realize the technical solution of the present invention. The scope of protection claimed by the present invention is not limited only to this embodiment, but also includes changes, replacements and other implementation methods that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed by the present invention.
Claims
1. A method for diagnosing loose bolt faults in a high-voltage circuit breaker, characterized in that: include: Obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; Constructing an all-pole model based on the prediction analysis results of the frame segments; The group delay function is obtained by solving the coefficients of the all-pole filter based on the all-pole model; The discrete cosine transform result based on the group delay function forms the spectrum enhanced voiceprint feature; Based on the spectrum-enhanced voiceprint features, the corresponding voiceprint prototype in the embedding space is obtained; Based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is obtained, so as to realize the loose bolt fault diagnosis of the high-voltage circuit breaker according to the probability.
2. The method for diagnosing a loose bolt fault of a high-voltage circuit breaker according to claim 1, characterized in that: The specific steps of dividing the occurrence time and the termination time into frame segments are as follows: A window function is applied to the occurrence time and the end time of the extracted bolt loosening sound signal to divide it into multiple frame segments.
3. The method for diagnosing a loose bolt fault of a high-voltage circuit breaker according to claim 1, characterized in that: The specific steps of constructing the full-pole model based on the prediction analysis results of the frame segments are as follows: Linear prediction analysis is performed on the frame segments to obtain prediction analysis results, and an all-pole model is constructed based on the prediction analysis results; wherein, the formula of the all-pole model is as follows: Where H(w) is the phase response, G is the signal-dependent gain, P is the all-pole model order, a(k) is the all-pole filter coefficient, j is the imaginary unit, ω is the frequency, and k is the delay associated with the pole.
4. The method for diagnosing a loose bolt fault in a high-voltage circuit breaker according to claim 3, characterized in that: The specific steps of obtaining the group delay function from the all-pole filter coefficient solution based on the all-pole model are as follows: The all-pole filter coefficients of the all-pole model are solved based on the Yule-Walker equation. The specific formula is as follows: In the formula, Frame fragment The autocorrelation function of is: A phase response formed by the all-pole filter coefficients is calculated according to the all-pole filter coefficients; The negative derivative of the phase response is calculated to obtain a group delay function, wherein the calculation of the negative derivative of the phase response is achieved through the sample difference of the bolt loosening sound signal.
5. The method for diagnosing a loose bolt fault of a high-voltage circuit breaker according to claim 1, characterized in that: The specific steps of forming the spectrum enhanced voiceprint feature based on the discrete cosine transform result of the group delay function are as follows: The discrete cosine transform result is obtained by performing discrete cosine transform on the group delay function, and the spectrum enhanced voiceprint feature is formed according to the discrete cosine transform result; wherein, the spectrum enhanced voiceprint feature is formed by taking the discrete cosine transform result of the first thirteen order coefficients excluding the zero order.
6. The method for diagnosing a loose bolt fault in a high-voltage circuit breaker according to claim 1, characterized in that: The specific steps of obtaining the corresponding voiceprint prototype in the embedding space based on the spectrum enhanced voiceprint feature are as follows: Construct a convolutional neural network as the embedding space of the framework based on the historical voiceprint prototype and the historical spectrum enhanced voiceprint features; The formed spectrum enhanced voiceprint features are mapped to the embedding space by the convolutional neural network to obtain the voiceprint prototype. The prototype of each category By encoding function Sure, is the learnable parameter of the encoding function, and the formula is as follows: In the formula, S k is the sample set of category k, Indicates Sample points; Among them, the convolutional neural network consists of convolutional layers, pooling layers, regularization layers, and activation layers.
7. The method for diagnosing a loose bolt fault in a high-voltage circuit breaker according to claim 6, characterized in that: The specific steps of obtaining the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype are as follows: According to the voiceprint to be diagnosed and the voiceprint prototype, the distance between the two is calculated. The specific formula is as follows: In the formula, v 1 and v 2 are two vectors whose distances are to be determined; M represents the dimension of the vector, v 1,i and v 2,i Representation vector v 1 and v The i-th dimension component of 2; According to the distance between the voiceprint to be diagnosed and the voiceprint prototype, the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype is calculated. The specific formula is as follows: 。 8. A high-voltage circuit breaker bolt loosening fault diagnosis system, characterized in that: include: A frame segment division module is used to obtain the occurrence time and end time of the bolt loosening sound signal, and divide the occurrence time and end time into frame segments; An all-pole model building module, used to build an all-pole model based on the prediction analysis results of the frame fragments; A group delay function calculation module, used for obtaining a group delay function based on a solution result of an all-pole filter coefficient of an all-pole model; A spectrum enhancement voiceprint feature forming module, used for forming a spectrum enhancement voiceprint feature based on a discrete cosine transform result of a group delay function; A mapping module, used to obtain the corresponding voiceprint prototype in the embedding space based on the spectrum-enhanced voiceprint features; The fault diagnosis module is used to obtain the probability that the voiceprint to be diagnosed belongs to the voiceprint prototype based on the distance between the voiceprint to be diagnosed and the voiceprint prototype, so as to realize the loose bolt fault diagnosis of the high-voltage circuit breaker according to the probability.
9. A device, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the high-voltage circuit breaker bolt loosening fault diagnosis method as described in any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, it is used to implement the steps of the high-voltage circuit breaker bolt loosening fault diagnosis method according to any one of claims 1 to 7.