Direct current GIS partial discharge signal multi-dimensional joint fault diagnosis method and system

Abstract

The invention discloses a DC GIS partial discharge signal multi-dimensional joint fault diagnosis method and system, and belongs to the technical field of power equipment fault detection. The method comprises the following steps: acquiring a typical defect partial discharge UHF signal of the high-voltage direct-current GIS; extracting time domain and frequency domain characteristics of the signal; extracting time-frequency domain features by using empirical wavelet transform and multi-scale permutation entropy; performing normalization processing on the signal to obtain a time domain sequence and a frequency domain sequence; and constructing a double-input neural network, taking the time domain, frequency domain and time-frequency domain features as first input, taking the time domain sequence and the frequency domain sequence as second input, and performing feature fusion and classification to realize fault diagnosis. According to the method, deep information of the partial discharge signals is fully mined and fused through multi-domain feature combination and a double-input deep learning model, the problems that a direct-current GIS partial discharge fault mode is difficult to recognize and low in precision are solved, and accurate, reliable and rapid diagnosis of typical insulation defects is achieved.

Description

Technical Field

[0001] This invention belongs to the field of DC GIS fault detection technology, specifically relating to a multi-dimensional joint fault diagnosis method and system for partial discharge signals of DC GIS. Background Technology

[0002] With the rapid development of power grid construction and the continuous growth of power transmission demand, traditional high-voltage AC transmission technology, due to its high line losses, susceptibility to instability, and high line construction costs, is ill-suited to the current needs of modern large-capacity power grid interconnection, severely limiting the construction and development of the power grid. High-voltage direct current (HVDC) transmission offers significant advantages such as strong grid interconnectivity, large transmission capacity, and rapid regulation. In long-distance power transmission, it can effectively compress line corridors, reduce line losses and construction costs, and improve primary energy utilization, meeting the requirements of modern smart grids and clean power systems. It has gradually become an important direction for future power transmission system design. Transmission lines with DC voltages below 800 kV have been successfully put into operation, and pilot projects below 1000 kV have also been launched. Regarding gas-insulated GIS, Siemens is currently preparing to use ±550 kV clean air insulated GIS in offshore wind power projects; domestically, Xi'an Kaifa Co., Ltd.'s domestically produced ±550 kV DC GIS has just achieved engineering application, and its operation and maintenance methods are still lacking. The key to putting high-voltage DC gas-insulated GIS into operation is ensuring the reliability of each system component. To achieve reliable and safe power distribution, it is necessary to monitor and diagnose the overall status of high-voltage DC GIS / GIL.

[0003] Partial discharge (PD) measurement plays a crucial role in insulation monitoring and diagnosis of gas-insulated switchgear (GIS) and gas-insulated lines (GIL). PD not only indicates the presence of potential defects in the insulation system, but its characteristic parameters also help engineers identify defect types, assess the health of insulation materials, and even predict the risk of dielectric breakdown. Therefore, PD measurement has become an indispensable part of power equipment maintenance and management.

[0004] Partial discharge detection under AC voltage has been extensively studied and standardized. The international standard IEC 60270 provides engineers with a systematic approach to assess the performance of insulating materials. These methods include phase-resolved partial discharge (PRPD) spectrum analysis, a technique that effectively identifies different types of defects, such as bubbles, cracks, or impurities, and can locate defects by analyzing the amplitude, frequency, and phase information of the PD signal, thus providing crucial information for equipment maintenance and repair. However, with the development of high-voltage direct current (HVDC) technology, especially in renewable energy integration and long-distance power transmission, the demand for partial discharge monitoring in HVDC systems is increasing. In HVDC environments, the partial discharge phenomenon exhibits different characteristics from AC systems because the DC electric field significantly affects the movement and behavior of electrons in the gaseous insulating medium. Therefore, current standards and methods for AC PD measurement are not entirely applicable to HVDC equipment. Standard measurement procedures and evaluation methods are lacking for HVDC equipment, particularly for GIS and GIL. Therefore, research on partial discharge signal detection and identification technology under different insulation defect types of high-voltage DC GIS / GIL is urgently needed to lay a theoretical foundation for the reliable monitoring of partial discharge during the operation of high-voltage DC GIS / GIL equipment. Summary of the Invention

[0005] The technical problem to be solved by this invention is to address the shortcomings of the prior art by providing a multi-dimensional joint fault diagnosis method and system for partial discharge signals in DC GIS. This method addresses the issue that the physical characteristics of partial discharge (PD) in high-voltage direct current (HVDC) GIS equipment differ significantly from those in AC systems, making existing standard partial discharge detection and evaluation methods based on AC systems inapplicable. Consequently, there is a lack of effective means for identifying DC partial discharge fault modes, which in turn restricts the safe and reliable operation and condition-based maintenance of HVDC GIS. This invention aims to accurately, reliably, and quickly identify the types of partial discharge fault modes in HVDC GIS.

[0006] The present invention adopts the following technical solution: A multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS includes the following steps: S1. Obtain UHF signal data of partial discharge of typical defects in high voltage DC GIS; S2. Measure the partial discharge initiation voltage PDIV of different typical defects, and gradually increase the voltage starting from the voltage level corresponding to PDIV to collect the partial discharge UHF signal of each typical defect under different DC voltages. S3. Based on the acquired partial discharge UHF signal, draw the time-domain envelope of the UHF pulse signal and extract the time-domain features of the UHF signal; S4. Perform a fast Fourier transform on the acquired partial discharge UHF signal to obtain the spectrum, and extract the frequency domain features of the UHF signal. S5. Perform empirical wavelet transform on the acquired partial discharge UHF signal to obtain different modal components, and calculate the multi-scale permutation entropy (MPE) eigenvalues ​​of each modal component as time-frequency domain features. S6. Normalize the acquired partial discharge UHF signal to obtain a time-domain sequence, and perform a fast Fourier transform on the time-domain sequence to obtain a frequency-domain sequence. S7. Construct a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification, thereby realizing partial discharge fault diagnosis.

[0007] Preferably, in step S2, the determination of the partial discharge initiation voltage PDIV of different typical defects includes: starting from 1kV, increasing the voltage stepwise, increasing the voltage by 1kV each time, holding the voltage for 30s, observing whether there is partial discharge, until three PD pulses are detected within 30 seconds, at which point the voltage value is PDIV; then starting from PDIV, increasing the voltage by 10% at each step until breakdown, and collecting UHF signals at different voltages.

[0008] Preferably, in step S3, the time-domain features include at least one of wavefront time, pulse rise time, pulse fall time, discharge pulse width, apparent discharge duration, signal mean, signal variance, skewness, steepness, and signal envelope area.

[0009] Preferably, in step S4, the frequency domain features include at least one of the following: average frequency, second-order frequency moment, centroid frequency, standard deviation frequency, root mean square frequency, dominant frequency band position index 1, dominant frequency band position index 2, dimensionless index, third-order frequency domain moment, fourth-order frequency domain moment, and half-order frequency domain moment.

[0010] Preferably, in step S5, the empirical wavelet transform includes: The signal spectrum is extracted using Fast Fourier Transform (FFT); the signal frequency band is segmented; a wavelet filter bank is constructed; and the signal is reconstructed to obtain the modal components.

[0011] Preferably, the calculation of the multi-scale permutation entropy (MPE) includes: The modal components are reconstructed in phase space to obtain the reconstruction matrix; the permutation entropy (PE) is calculated; and coarsening is performed based on the scale factor to obtain the MPE as follows:

[0012] in, It is a discrete sequence. To delay time, For embedded dimensions, For permutation entropy, This is the time series after coarse-graining. is the scale factor.

[0013] Preferably, in step S6, the normalization preprocessing includes normalizing the time-domain signal to eliminate amplitude influence.

[0014] Preferably, in step S7, the dual-input neural network includes: The first branch is used to process the time-domain features, frequency-domain features, and time-frequency-domain features of the first input for feature extraction; The second branch is used to process the time-domain sequence and frequency-domain sequence of the second input by performing convolution, activation, and pooling operations in sequence; The fusion layer connects the features extracted from the first and second branches. Before connecting, the extracted features are flattened so that the final features are all in one-dimensional order.

[0015] Preferably, the dual-input neural network uses a convolutional neural network as a deep learning model, and the feature extraction includes convolution, activation, and pooling operations.

[0016] Secondly, embodiments of the present invention provide a multi-dimensional joint fault diagnosis system for partial discharge signals of DC GIS, comprising: The signal module acquires UHF signal data of partial discharge of typical defects in high-voltage DC GIS. The measurement module measures the partial discharge initiation voltage (PDIV) of different typical defects, and gradually increases the voltage from the voltage level corresponding to the PDIV to collect the partial discharge UHF signals of each typical defect under different DC voltages. The time-domain module, based on the acquired partial discharge UHF signal, plots the time-domain envelope of the UHF pulse signal and extracts the time-domain features of the UHF signal; The frequency domain module performs a fast Fourier transform on the acquired partial discharge UHF signal to obtain the spectrum and extracts the frequency domain features of the UHF signal. The transformation module performs empirical wavelet transform on the acquired partial discharge UHF signal to obtain different modal components, and calculates the multi-scale permutation entropy (MPE) eigenvalues ​​of each modal component as time-frequency domain features. The preprocessing module performs normalization preprocessing on the acquired partial discharge UHF signal to obtain a time-domain sequence, and performs fast Fourier transform on the time-domain sequence to obtain a frequency-domain sequence. The network module constructs a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification, thereby realizing partial discharge fault diagnosis.

[0017] Thirdly, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for multidimensional joint fault diagnosis of partial discharge signals in DC GIS.

[0018] Fourthly, embodiments of the present invention provide a computer-readable storage medium including a computer program, which, when executed by a processor, implements the steps of the above-described method for multidimensional joint fault diagnosis of partial discharge signals in DC GIS.

[0019] Fifthly, a chip includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for multidimensional joint fault diagnosis of partial discharge signals in DC GIS.

[0020] In a sixth aspect, embodiments of the present invention provide an electronic device including a computer program, wherein when the computer program is executed by the electronic device, it implements the steps of the above-described method for multidimensional joint fault diagnosis of partial discharge signals in DC GIS.

[0021] Compared with the prior art, the present invention has at least the following beneficial effects: A multi-dimensional joint fault diagnosis method for partial discharge signals in DC GIS is proposed. This method, for the first time, systematically combines the time-domain, frequency-domain, and time-frequency-domain artificial features of UHF partial discharge signals from DC GIS with the deep learning features of the original time / frequency domain sequences, fusing and complementing them through a dual-input neural network. This approach overcomes the limitations of single-feature dimension information incompleteness, reliance on expert experience in traditional methods, and the poor interpretability and large data requirements of deep learning models. It achieves comprehensive capture and high-precision identification of complex and weak features of DC partial discharge, providing a new and more reliable technical path for fault diagnosis in DC GIS.

[0022] Furthermore, a standardized and repeatable pressurization acquisition procedure is provided, clarifying the data acquisition specifications for the entire process from the initial voltage to breakdown. This "step-by-step voltage increase + constant voltage observation" method can accurately capture the evolution of discharge characteristics of different defects at different voltage levels, providing high-quality sample data covering different development stages for subsequent feature extraction. This ensures the completeness and effectiveness of the diagnostic model training data and is an important foundation for the reliability of the method.

[0023] Furthermore, ten types of features reflecting pulse waveform morphology (such as wavefront time and pulse width), statistical characteristics (such as mean and variance), and higher-order moment characteristics (such as skewness and kurtosis) were comprehensively extracted from the time-domain envelope of the UHF signal. These features characterize the time-domain behavior of the discharge pulse from different perspectives, effectively distinguishing the differences in rise rate, duration, and energy distribution of discharge pulses of different defect types, providing rich and discriminative time-domain evidence for the classifier.

[0024] Furthermore, after FFT transformation, eleven types of frequency domain indicators reflecting the signal's spectral centroid, distribution range, concentration, and higher-order statistical characteristics were extracted. These features can reveal the "fingerprint" differences in the frequency domain of electromagnetic waves excited by different defect discharges; for example, some defects may mainly excite high-frequency components, while others have a wider spectrum. Combined with time-domain features, a preliminary joint time-frequency domain analysis of the discharge signal was achieved, enhancing the comprehensiveness of the feature set.

[0025] Furthermore, EWT can adaptively divide and decompose frequency bands based on the spectral characteristics of the signal itself, overcoming the shortcomings of fixed basis functions in traditional wavelet transforms. It is particularly suitable for processing non-stationary and nonlinear partial discharge UHF signals. Through EWT, complex original signals can be decomposed into several physically meaningful and relatively stationary modal components, laying a solid foundation for subsequent accurate calculation of the complexity characteristics (MPE) of each component.

[0026] Furthermore, MPE, a powerful tool capable of quantifying the complexity and regularity of time series data, was introduced. MPE analyzes the permutation entropy of modal components across multiple time scales, capturing both the microscopic irregularities of the signal and reflecting changes in its macroscopic dynamics, making it extremely sensitive to the dynamic characteristics of discharge signals. Combining the fine decomposition of EWT with the multi-scale complexity analysis of MPE allows for the extraction of deep-seated, human-eye-insensitive time-frequency domain fault features, significantly enhancing the discriminative power of pattern recognition.

[0027] Furthermore, normalization eliminates amplitude differences between samples caused by factors such as discharge intensity and sensor gain, allowing the model to focus on learning the shape features of the signal waveform and spectrum rather than their magnitude. This step significantly improves the training stability, convergence speed, and generalization ability of the deep learning model, making it highly adaptable and robust to signals acquired under different operating conditions and from different devices.

[0028] Furthermore, a fusion model was designed that integrates manual feature branches and deep learning branches for parallel processing. The manual feature branch takes into account features with clear physical meaning extracted from expert knowledge, resulting in strong interpretability; the deep learning branch, on the other hand, autonomously learns abstract features directly from the original sequence to uncover potential patterns. Finally, information complementarity is achieved through feature flattening and concatenation. This dual-driven architecture of "knowledge + data" combines the reliability of traditional methods with the high performance of deep learning, representing a key innovation for achieving high-precision classification.

[0029] Furthermore, the powerful feature learning capabilities of CNNs are utilized, especially their convolutional operations, which can effectively capture local correlation and translation invariance features in time-domain and frequency-domain sequences. This enables the deep learning branch to automatically learn more complex and optimized feature representations than those designed manually, complementing the manual feature branch and jointly improving the performance and intelligence level of the entire diagnostic system.

[0030] It is understood that the beneficial effects of the second to sixth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0031] In summary, this invention comprehensively characterizes discharge features through multi-domain feature extraction and combines the advantages of dual-input neural networks to integrate expert knowledge and deep learning. It solves the problems of complex partial discharge characteristics in DC GIS, the inapplicability of traditional AC methods, and low diagnostic accuracy. It achieves accurate, reliable, and rapid automatic identification of partial discharge fault modes, significantly improving the intelligence level and reliability of DC GIS operation and maintenance.

[0032] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0033] Figure 1 This is a flowchart of the multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS based on multi-domain feature extraction according to the present invention; Figure 2 This is a circuit connection block diagram for the acquisition of UHF signals of partial discharge in typical defects of DC GIS according to the present invention; Figure 3 These are diagrams illustrating the structures of four typical defect partial discharge power sources designed for the experimental platform of partial discharge of typical defects in DC GIS according to the present invention. Figure 4 UHF signal images of four typical defects collected using the DC GIS typical defect partial discharge experimental platform described in this invention; Figure 5 The network architecture diagram for the multi-dimensional joint fault diagnosis method of DC GIS partial discharge signal based on multi-domain feature extraction described in this invention is shown below. Figure 6The classification confusion matrix diagram obtained by applying the multi-domain feature extraction-based multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS described in this invention; Figure 7 A schematic diagram of a computer device provided in an embodiment of the present invention; Figure 8 This is a block diagram of a chip provided according to an embodiment of the present invention.

[0034] Among them, 1. Withstand voltage test control box; 2. Transformer; 3. Silicon stack; 4. Protective resistor; 5. Typical defect model; 6. UHF sensor; 7. Oscilloscope; 60. Computer equipment; 61. Processor; 62. Memory; 63. Computer program; 600. Electronic equipment; 610. Processing unit; 620. Storage unit; 6201. Random access memory unit; 6202. Cache memory unit; 6203. Read-only memory unit; 6204. Program / utility; 6205. Program module; 630. Bus; 640. Display unit; 650. Input / output interface; 660. Network adapter; 700. External device. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0037] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0038] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0039] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0040] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0041] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0042] This invention provides a multi-dimensional joint fault diagnosis method for partial discharge signals in DC GIS. It utilizes the acquired UHF partial discharge signals of typical defects in high-voltage DC GIS to extract their time-domain and frequency-domain features. Furthermore, it combines Empirical Wavelet Transform (EWT) and Multiscale Permutation to extract the time-frequency domain features of the partial discharge signals as artificial feature inputs. Additionally, it uses the signal's time-domain and frequency-domain sequences as deep learning inputs. Finally, it proposes a dual-input neural network architecture combining artificial feature extraction and deep learning to achieve the detection and classification of typical partial discharge signals.

[0043] Please see Figure 1 The present invention provides a multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS, comprising the following steps: S1. Acquire UHF signal data of partial discharge of typical defects in high-voltage DC GIS through partial discharge testing and detection systems. The pressurization system consists of a 50kV high-voltage DC generator and a large... RCThe load consists mainly of a withstand voltage test control box, a test transformer, a rectifier silicon stack, a protective resistor, and a coupling capacitor. Based on the given statistics of common GIS insulation faults, four types of typical GIS insulation defect samples were designed: pinpoint defects, surface contamination defects, metal particle defects, and air gap defects. Proceed to step S2. A partial discharge test and detection system was built using the ultra-high frequency method. The partial discharge signal was received by the ultra-high frequency antenna and the partial discharge signal could be directly measured. The measurement results could be directly output using a high-speed oscilloscope, providing data support for subsequent detection and identification research.

[0044] The partial discharge data acquisition test of the high voltage DC GIS was carried out in a high voltage shielded room, which can effectively shield interference from lights, mobile phone signals, external electromagnetic signals, etc. All equipment and shielded rooms are individually grounded, and the background noise of the partial discharge instrument is less than 0.5pC.

[0045] S2. Measure the PDIV of different typical defects. Select four typical insulation defects sequentially and conduct single typical insulation defect discharge tests. Start with 1kV and increase the voltage in steps, increasing by 1kV each time. After each voltage increase, maintain the voltage for 30s and observe whether there is partial discharge. Continue until three PD pulses are detected within 30 seconds. The voltage value at this time is the PDIV of the defect model. If not, continue to increase the voltage. A single discharge during the voltage increase process is not considered as the first partial discharge. Next, measure the voltage magnitude of the PDIV, increasing by 10% at each step until breakdown. Collect UHF signals of different typical defects under different DC voltages. Proceed to step S3. S3. Using the DC GIS partial discharge UHF signal obtained in step S2, draw the time-domain envelope of the UHF pulse signal, extract time-domain features using the drawn envelope, and proceed to step S4. The time-domain features include ten features such as wavefront time, pulse rise time, pulse fall time, discharge pulse width, apparent discharge duration, signal mean, signal variance, skewness, steepness, and signal envelope area.

[0046] S4. Using the DC GIS partial discharge UHF signal obtained in step S2, the partial discharge spectrum is obtained by using Fast Fourier Transform (FFT), and frequency domain features are extracted; proceed to step S5. The frequency domain features include 11 features such as average frequency, second-order frequency moment, centroid frequency, standard deviation frequency, root mean square frequency, dominant frequency band position index 1, dominant frequency band position index 2, dimensionless index, third-order frequency moment, fourth-order frequency moment, and half-order frequency moment.

[0047] S5. Using the DC GIS partial discharge UHF signal obtained in step S2, the partial discharge UHF signal is decomposed using empirical wavelet transform to obtain different modal components based on its time-frequency characteristics. This mainly includes four steps: extracting the signal spectrum using FFT, segmenting the signal frequency band, constructing a wavelet filter bank, and signal reconstruction. Next, the MPE eigenvalues ​​of each component are calculated as the time-frequency domain characteristics of the partial discharge signal. Proceed to step S6. Using FFT to extract the signal spectrum, i.e., extracting the signal spectrum... The formula for performing an FFT is as follows:

[0048] Bundle The spectrum is denoted as , .

[0049] The calculation method for frequency band segmentation of a signal is as follows: First of all Marked above The local maxima are arranged in descending order of magnitude. Then take its corresponding Ascending order, i.e. ; Finally, the normalized frequency band Divided into Each sub-band, and set with The sub-band containing the AM-FM component centered on the center is .in, , .

[0050] The calculation method for constructing a wavelet filter bank is as follows: for sub-bands Its scaling function and wavelet function They are respectively

[0051]

[0052] It is a random function. This parameter ensures the minimum overlap area, and its value is determined by the boundary values. In this method, Using the same definitions as in the literature, the corresponding formula is:

[0053] The signal reconstruction calculation method is as follows: using inverse Fourier transform to calculate... and This allows us to obtain the time-domain representation of each frequency band, i.e., the modal component ewt(k).

[0054] The method for calculating MPE eigenvalues ​​is as follows: First, calculate the PE value. Given a discrete sequence... By setting a delay time and embedding dimension For the sequence Phase space reconstruction yields the following formula:

[0055] , , Represents the reconstruction matrix The number of new components formed by the given elements. The new components are arranged in ascending order, and their corresponding columns are represented as follows: The reconstructed matrix The symbol combination is:

[0056] and . It can represent all possible permutations. If... This represents the probability of each permutation, therefore the normalized permutation entropy is...

[0057] MPE adds a scale factor to PE, thus obtaining a coarse-grained time series, the calculation formula of which is:

[0058] From this we can obtain

[0059] S6. Normalize and preprocess the DC GIS partial discharge UHF signal obtained in step S2 as the input of the time domain sequence; extract the spectrum of the normalized time domain signal through fast Fourier transform as the frequency domain sequence; proceed to step S7. S7. Using the time-domain, frequency-domain, and time-frequency-domain features of the partial discharge signal obtained in steps S3, S4, and S5 as one input, and the time-domain and frequency-domain sequences of the partial discharge signal obtained in step S6 as another input, a dual-input neural network combining manual feature extraction and deep learning is constructed to realize the detection and recognition technology of typical types of partial discharge signals.

[0060] The construction of a two-input neural network is as follows: Convolutional neural networks (CNNs) are used as the deep learning model. The key is how to fuse multiple inputs, a process that occurs after feature extraction but before result classification. The two inputs are first subjected to feature extraction operations such as convolution and pooling, respectively. Then, fully connected layers connect the extracted features to construct a fused feature set. Each branch of the fused feature set includes convolutional, activation, and pooling layers. Finally, to ensure that the feature dimensions are the same, the extracted features are flattened so that both sets of features are ultimately one-dimensional sequences for easy feature concatenation.

[0061] In another embodiment of the present invention, a multi-dimensional joint fault diagnosis system for DC GIS partial discharge signals is provided. This system can be used to implement the above-mentioned multi-dimensional joint fault diagnosis method for DC GIS partial discharge signals. Specifically, the multi-dimensional joint fault diagnosis system for DC GIS partial discharge signals includes a signal module, a measurement module, a time domain module, a frequency domain module, a transformation module, a preprocessing module, and a network module.

[0062] Among them, the signal module acquires UHF signal data of partial discharge of typical defects in high voltage DC GIS; The measurement module measures the partial discharge initiation voltage (PDIV) of different typical defects, and gradually increases the voltage from the voltage level corresponding to the PDIV to collect the partial discharge UHF signals of each typical defect under different DC voltages. The time-domain module, based on the acquired partial discharge UHF signal, plots the time-domain envelope of the UHF pulse signal and extracts the time-domain features of the UHF signal; The frequency domain module performs a fast Fourier transform on the acquired partial discharge UHF signal to obtain the spectrum and extracts the frequency domain features of the UHF signal. The transformation module performs empirical wavelet transform on the acquired partial discharge UHF signal to obtain different modal components, and calculates the multi-scale permutation entropy (MPE) eigenvalues ​​of each modal component as time-frequency domain features. The preprocessing module performs normalization preprocessing on the acquired partial discharge UHF signal to obtain a time-domain sequence, and performs fast Fourier transform on the time-domain sequence to obtain a frequency-domain sequence. The network module constructs a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification, thereby realizing partial discharge fault diagnosis.

[0063] This invention provides a terminal device comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, graphics processing units (GPUs), tensor processing units (TPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve a corresponding method flow or function. The processor described in this embodiment can be used in the operation of a multi-dimensional joint fault diagnosis method for partial discharge signals in DC GIS, including: Acquire UHF signal data of partial discharge from typical defects in high-voltage direct current GIS; determine the partial discharge initiation voltage (PDIV) of different typical defects, and gradually increase the voltage from the voltage level corresponding to the PDIV to collect partial discharge UHF signals of each typical defect under different DC voltages; based on the acquired partial discharge UHF signals, draw the time-domain envelope of the UHF pulse signal and extract the time-domain features of the UHF signal; perform Fast Fourier Transform on the acquired partial discharge UHF signals to obtain the spectrum and extract the frequency-domain features of the UHF signal; perform Empirical Wavelet Transform on the acquired partial discharge UHF signals to obtain different modal components, calculate the multi-scale permutation entropy (MPE) feature value of each modal component as the time-frequency domain feature; perform normalization preprocessing on the acquired partial discharge UHF signals to obtain the time-domain sequence, and perform Fast Fourier Transform on the time-domain sequence to obtain the frequency-domain sequence; construct a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification to achieve partial discharge fault diagnosis.

[0064] Please see Figure 7The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When executed by the processor 61, the computer program 63 implements the method for estimating the concentration of radioactive iodine species in the containment vessel after an accident, as described in this embodiment. To avoid repetition, these details are not elaborated here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the multi-dimensional joint fault diagnosis system for partial discharge signals of DC GIS, as described in this embodiment. To avoid repetition, these details are not elaborated here.

[0065] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 7 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0066] The processor 61 may be a Central Processing Unit (CPU), or other general-purpose processors, graphics processing units (GPUs), tensor processing units (TPUs), digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0067] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., provided on the computer device 60.

[0068] Furthermore, the memory 62 may include both internal storage units of the computer device 60 and external storage devices. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0069] Please see Figure 8 The terminal device is an electronic device 600, which is manifested in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0070] The storage unit stores program code, which can be executed by the processing unit 610 to perform the steps described in the method section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 610 can perform, as follows: Figure 1 The steps are shown in the figure.

[0071] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access memory (RAM) 6201 and / or cache memory 6202, and may further include a read-only memory (ROM) 6203.

[0072] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0073] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the multiple bus structures.

[0074] Electronic device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 600, and / or with any device that enables electronic device 600 to communicate with one or more other computing devices (e.g., router, modem). This communication can be performed via input / output interface 650. Furthermore, electronic device 600 can also communicate with one or more networks (e.g., local area network, wide area network, and / or public network, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of electronic device 600 via bus 630. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0075] Example 4 This invention also provides a storage medium, specifically a computer-readable storage medium, which is a memory device in a terminal device for storing programs and data. It is understood that the computer-readable storage medium here can include both built-in storage media in the terminal device and extended storage media supported by the terminal device; it can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor, which can be one or more computer programs (including program code). More specific examples of the computer-readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical fiber, portable compact disk read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0076] Computer-readable storage media also include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium can also be any readable medium other than a readable storage medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, radio frequency, etc., or any suitable combination thereof.

[0077] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0078] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS in the above embodiments; one or more instructions in the computer-readable storage medium are loaded and executed by the processor in the following steps: Acquire UHF signal data of partial discharge from typical defects in high-voltage direct current GIS; determine the partial discharge initiation voltage (PDIV) of different typical defects, and gradually increase the voltage from the voltage level corresponding to the PDIV to collect partial discharge UHF signals of each typical defect under different DC voltages; based on the acquired partial discharge UHF signals, draw the time-domain envelope of the UHF pulse signal and extract the time-domain features of the UHF signal; perform Fast Fourier Transform on the acquired partial discharge UHF signals to obtain the spectrum and extract the frequency-domain features of the UHF signal; perform Empirical Wavelet Transform on the acquired partial discharge UHF signals to obtain different modal components, calculate the multi-scale permutation entropy (MPE) feature value of each modal component as the time-frequency domain feature; perform normalization preprocessing on the acquired partial discharge UHF signals to obtain the time-domain sequence, and perform Fast Fourier Transform on the time-domain sequence to obtain the frequency-domain sequence; construct a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification to achieve partial discharge fault diagnosis.

[0079] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0080] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0081] The steps of the multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS based on multi-domain feature extraction described in this invention are explained in detail below: Step 1: Using the DC GIS typical defect partial discharge experimental platform built by this invention, the initial discharge voltage PDIV of the typical defect model is obtained by step boosting method. Next, the voltage magnitude of PDIV is measured, increasing by 10% at each step until breakdown. UHF signals of different typical defects under different DC voltages are collected using an oscilloscope with a sampling rate of 1GS / s. Then proceed to Step 2 to calculate the time domain characteristics of the partial discharge UHF signal.

[0082] Step 2: Obtain the time-domain envelope of the partial discharge UHF pulse signal from the acquired signal, and extract ten characteristic parameters: wavefront time, pulse rise time, pulse fall time, discharge pulse width, apparent discharge duration, skewness, kurtosis, signal envelope area, signal mean, and signal variance. Proceed to Step 3 to calculate the frequency-domain characteristics of the partial discharge UHF signal.

[0083] Step 3: Perform a Fast Fourier Transform (FFT) on the acquired partial discharge UHF pulse signal to obtain the desired spectrum. Then extract its frequency domain features, including 11 features such as average frequency, second-order frequency moment, centroid frequency, standard deviation frequency, root mean square frequency, dominant band position index 1, dominant band position index 2, dimensionless index, third-order frequency moment, fourth-order frequency moment, and half-order frequency moment. Proceed to Step 4 to calculate the frequency domain features of the partial discharge UHF signal. Proceed to Step 4 to calculate the time-frequency domain features of the partial discharge UHF signal.

[0084] Step 4: Perform empirical wavelet decomposition on the acquired partial discharge UHF pulse signal to obtain decomposed signals of different frequency bands. Then, extract the MPE value for each decomposed signal. Note that the number of frequency bands and the scaling factor must be selected in this step. Proceed to Step 5 to calculate the deep learning input features of the partial discharge UHF signal.

[0085] Step 5: Normalize the acquired partial discharge UHF time-domain signal to eliminate amplitude influence and increase model transferability. Perform FFT on the normalized time-domain sequence to obtain the standardized spectrum, which in turn yields the time-domain and frequency-domain sequences used as inputs for deep learning. Proceed to Step 6 to set up the dual-input neural network architecture.

[0086] Step Six: First, select a suitable loss function and optimizer, and based on experience, choose the number of convolutional channels and convolutional layers of the CNN network to determine the network architecture of the deep learning model. Proceed to Step Seven: Input the extracted artificial feature data dataset and the deep learning feature dataset into the dual-input neural network constructed in this invention for pattern recognition.

[0087] Step 7: Use 80% of the extracted dataset as the training set to train the model, and use the remaining 20% ​​of the data as the test set to evaluate the model performance and obtain the classification accuracy of the four typical defects.

[0088] The practical application of the multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS based on multi-domain feature extraction of this invention is described, such as... Figure 2-6 The diagram illustrates the actual construction process of the multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS based on multi-domain feature extraction according to the present invention.

[0089] Please see Figure 2 Four typical defects (such as) were collected using a UHF antenna. Figure 3 The DC partial discharge UHF signal (shown) is an example of the UHF signal for four types of defects. Figure 4 As shown.

[0090] Subsequently, manual feature extraction and deep learning feature computation were performed on the acquired UHF signals. Manually extracted feature datasets were obtained for the time domain, frequency domain, and time-frequency domain datasets; and deep learning feature datasets for the time domain and frequency domain sequences were obtained and labeled. Regarding the selection of the number of frequency bands and scale factor for time-frequency domain feature extraction, this example analyzed the relationship between the number of frequency bands, the scale factor of permutation entropy, and diagnostic accuracy through extensive experimental data. It was found that if these values ​​were too large, the algorithm's computation time would be too long, making practical applications impossible. A frequency band count of 6 and a scale factor of 5 for permutation entropy were chosen as the most suitable for the model's real-time performance and accuracy in fault diagnosis. 80% of the obtained dataset was used as the training set to train the model, which was then input into a dual-input neural network.

[0091] In summary, this invention provides a multi-dimensional joint fault diagnosis method and system for partial discharge signals in DC GIS. It combines multiple characterizations and feature information of the partial discharge UHF signal, including time-domain, frequency-domain, and time-frequency-domain information, each providing different characteristics of the discharge phenomenon. By comprehensively analyzing these features, the type of partial discharge fault in the GIS equipment can be more accurately determined, thereby improving diagnostic accuracy. Furthermore, this invention combines manual feature extraction technology with deep learning. Through the complementarity between manually created features and deep learning features, more potential patterns in the signal can be identified, thus improving classification performance. Deep learning can also analyze and learn from large amounts of signal data to establish corresponding models and rules. These models and rules enable automatic diagnosis of partial discharge faults in GIS equipment, reducing reliance on human experience, improving diagnostic reliability and consistency, effectively enhancing the operational reliability of current DC GIS, and increasing the economic benefits of DC power system operation.

[0092] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0093] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0094] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0095] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0096] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0097] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0098] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random-access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed 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 media do not include electrical carrier signals and telecommunication signals.

[0099] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0100] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0102] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS, characterized in that, Includes the following steps: S1. Obtain UHF signal data of partial discharge of typical defects in high voltage DC GIS; S2. Measure the partial discharge initiation voltage PDIV of different typical defects, and gradually increase the voltage starting from the voltage level corresponding to PDIV to collect the partial discharge UHF signal of each typical defect under different DC voltages. S3. Based on the acquired partial discharge UHF signal, draw the time-domain envelope of the UHF pulse signal and extract the time-domain features of the UHF signal; S4. Perform a fast Fourier transform on the acquired partial discharge UHF signal to obtain the spectrum, and extract the frequency domain features of the UHF signal. S5. Perform empirical wavelet transform on the acquired partial discharge UHF signal to obtain different modal components, and calculate the multi-scale permutation entropy (MPE) eigenvalues ​​of each modal component as time-frequency domain features. S6. Normalize the acquired partial discharge UHF signal to obtain a time-domain sequence, and perform a fast Fourier transform on the time-domain sequence to obtain a frequency-domain sequence. S7. Construct a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification, thereby realizing partial discharge fault diagnosis.

2. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S2, the determination of the partial discharge initiation voltage PDIV of different typical defects includes: starting from 1kV, the voltage is increased stepwise by 1kV each time, and the voltage is maintained for 30s. The presence of partial discharge is observed until three PD pulses are detected within 30 seconds. The voltage value at this time is PDIV. Then, starting from PDIV, the voltage is increased by 10% at each step until breakdown occurs, and UHF signals at different voltages are collected.

3. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S3, the time-domain features include at least one of wavefront time, pulse rise time, pulse fall time, discharge pulse width, apparent discharge duration, signal mean, signal variance, skewness, steepness, and signal envelope area.

4. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S4, the frequency domain features include at least one of the following: average frequency, second-order frequency moment, centroid frequency, standard deviation frequency, root mean square frequency, dominant frequency band position index 1, dominant frequency band position index 2, dimensionless index, third-order frequency domain moment, fourth-order frequency domain moment, and half-order frequency domain moment.

5. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S5, the empirical wavelet transform includes: The signal spectrum is extracted using Fast Fourier Transform (FFT); the signal frequency band is segmented; a wavelet filter bank is constructed; and the signal is reconstructed to obtain the modal components.

6. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 5, characterized in that, The calculation of the multi-scale permutation entropy (MPE) includes: The modal components are reconstructed in phase space to obtain the reconstruction matrix; the permutation entropy (PE) is calculated; and coarsening is performed based on the scale factor to obtain the MPE as follows: in, It is a discrete sequence. To delay time, For embedded dimensions, For permutation entropy, This is the time series after coarse-graining. is the scale factor.

7. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S6, the normalization preprocessing includes normalizing the time-domain signal to eliminate amplitude effects.

8. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 1, characterized in that, In step S7, the dual-input neural network includes: The first branch is used to process the time-domain features, frequency-domain features, and time-frequency-domain features of the first input for feature extraction; The second branch is used to process the time-domain sequence and frequency-domain sequence of the second input by performing convolution, activation, and pooling operations in sequence; The fusion layer connects the features extracted from the first and second branches. Before connecting, the extracted features are flattened so that the final features are all in one-dimensional order.

9. The multi-dimensional joint fault diagnosis method for partial discharge signals of DC GIS according to claim 8, characterized in that, The dual-input neural network uses a convolutional neural network as a deep learning model, and the feature extraction includes convolution, activation, and pooling operations.

10. A multi-dimensional joint fault diagnosis system for partial discharge signals of DC GIS, characterized in that, include: The signal module acquires UHF signal data of partial discharge of typical defects in high-voltage DC GIS. The measurement module measures the partial discharge initiation voltage (PDIV) of different typical defects, and gradually increases the voltage from the voltage level corresponding to the PDIV to collect the partial discharge UHF signals of each typical defect under different DC voltages. The time-domain module, based on the acquired partial discharge UHF signal, plots the time-domain envelope of the UHF pulse signal and extracts the time-domain features of the UHF signal; The frequency domain module performs a fast Fourier transform on the acquired partial discharge UHF signal to obtain the spectrum and extracts the frequency domain features of the UHF signal. The transformation module performs empirical wavelet transform on the acquired partial discharge UHF signal to obtain different modal components, and calculates the multi-scale permutation entropy (MPE) eigenvalues ​​of each modal component as time-frequency domain features. The preprocessing module performs normalization preprocessing on the acquired partial discharge UHF signal to obtain a time-domain sequence, and performs fast Fourier transform on the time-domain sequence to obtain a frequency-domain sequence. The network module constructs a dual-input neural network, using the time-domain features, frequency-domain features, and time-frequency-domain features as the first input, and the time-domain sequence and frequency-domain sequence as the second input, to perform feature fusion and classification, thereby realizing partial discharge fault diagnosis.

Images