Method and system for identifying partial discharge of GIS equipment based on ultra-high frequency signal

Abstract

The application discloses a GIS device partial discharge identification method and system based on a UHF signal, and relates to the technical field of power equipment detection. The method comprises the following steps: collecting a UHF pulse signal; establishing signal interaction between a high-frequency sensor and an analog computing array; the analog computing array receives the UHF pulse signal and performs parallel processing based on rising edge time, oscillation frequency and energy integration to determine a multi-dimensional characteristic voltage; the multi-dimensional characteristic voltage is introduced into a defect detection array for defect type analysis and discharge severity level evaluation to determine a discharge evaluation result; and the discharge evaluation result is visualized on a terminal to perform partial discharge early warning management of the GIS device. The application solves the technical problem that, in the prior art, cumulative delay is introduced due to the fact that the signal needs to pass through multiple processing links in the GIS device partial discharge detection process, which makes it difficult to realize real-time identification of the partial discharge pulse level, and achieves the technical effect of fast response and real-time identification of the partial discharge characteristics.

Description

Technical Field

[0001] This invention relates to the field of power equipment testing technology, specifically to a method and system for identifying partial discharge in GIS equipment based on ultra-high frequency signals. Background Technology

[0002] Gas-insulated switchgear (GIS), as a critical high-voltage device in power systems, directly affects the safety and stability of power grid operation due to its insulation condition. Partial discharge, as an early indicator of insulation degradation and fault occurrence in GIS equipment, is characterized by its suddenness, short duration, and wide spectral range. Efficient and accurate detection and identification of partial discharge is crucial for equipment condition assessment and fault early warning. Current technologies typically use ultra-high frequency sensors to collect pulse signals generated by partial discharge, convert the analog signals to digital signals, and transmit them to a back-end processor where software algorithms extract features and identify the type of the waveform. However, this process involves multiple steps, including analog-to-digital conversion, data transmission, and software computation. Each step introduces a significant time delay, and this delay accumulates further with increasing data volume and algorithm complexity. This makes it difficult to respond promptly to transient discharge pulse characteristics, hindering real-time pulse-level identification and rapid early warning, thus limiting the real-time performance and practicality of online partial discharge monitoring systems for GIS equipment. Summary of the Invention

[0003] This application provides a method and system for identifying partial discharge in GIS equipment based on ultra-high frequency signals. It solves the technical problem in the prior art that the partial discharge detection process of GIS equipment is difficult to achieve real-time identification of partial discharge at the pulse level due to the cumulative delay introduced by the signal having to go through multiple processing links such as analog-to-digital conversion, data transmission and software processing.

[0004] The first aspect of this application provides a method for identifying partial discharge in GIS equipment based on ultra-high frequency signals, the method comprising:

[0005] A high-frequency sensor is deployed on the outside of the insulating flange of the GIS equipment to collect high-frequency pulse signals generated by partial discharge inside the GIS. A signal interaction is established between the high-frequency sensor and the analog computing array. The analog computing array receives the high-frequency pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine a multi-dimensional characteristic voltage. The multi-dimensional characteristic voltage is then imported into a defect detection array. Based on the multi-dimensional characteristic voltage, defect type analysis and discharge severity assessment are performed to determine the discharge assessment result. The defect detection array is a cross-switch architecture, with each cross node configured with a transimpedance amplifier after initializing the conductance value. The discharge assessment result is visualized on the terminal to perform partial discharge early warning management of the GIS equipment.

[0006] A second aspect of this application provides a partial discharge identification system for GIS equipment based on ultra-high frequency signals, the system comprising:

[0007] Signal Acquisition Module: Deploys a UHF sensor on the outside of the insulating flange of the GIS equipment to acquire UHF pulse signals generated by partial discharge inside the GIS; Signal Interaction Processing Module: Establishes signal interaction between the high-frequency sensor and the analog computing array. The analog computing array receives the UHF pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multi-dimensional characteristic voltages; Evaluation Module: Imports the multi-dimensional characteristic voltages into a defect detection array, performs defect type analysis and discharge severity assessment based on the multi-dimensional characteristic voltages, and determines the discharge assessment results. The defect detection array is a cross-switch architecture, with each cross node configured with a transimpedance amplifier after initializing the conductance value; Early Warning Management Module: Visualizes the discharge assessment results on the terminal and performs partial discharge early warning management for the GIS equipment.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, a high-frequency sensor is deployed on the outside of the insulating flange of the GIS equipment to collect high-frequency pulse signals generated by partial discharge inside the GIS. Next, a signal interaction is established between the high-frequency sensor and an analog computing array. The analog computing array receives the high-frequency pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multi-dimensional characteristic voltages. Then, the multi-dimensional characteristic voltages are imported into a defect detection array. Based on these voltages, defect type analysis and discharge severity assessment are performed to determine the discharge assessment results. The defect detection array uses a cross-switch architecture, with each cross node configured with a transimpedance amplifier initialized with a conductivity value. Finally, the discharge assessment results are visualized at the terminal, enabling partial discharge early warning management of the GIS equipment. This method solves the technical problem in existing technologies where the signal undergoes multiple processing stages (analog-to-digital conversion, data transmission, and software processing), leading to cumulative delays and hindering real-time pulse-level identification of partial discharge. It achieves the technical effects of shortening the signal processing chain, enabling rapid response and real-time identification of partial discharge characteristics, and improving the timeliness of online early warnings. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1A schematic flowchart of a partial discharge identification method for GIS equipment based on ultra-high frequency signals provided in an embodiment of this application;

[0012] Figure 2 A schematic diagram of the partial discharge identification system for GIS equipment based on ultra-high frequency signals provided in this application embodiment.

[0013] Explanation of reference numerals in the attached diagram: Signal acquisition module 11, Signal interaction and processing module 12, Evaluation module 13, Early warning management module 14. Detailed Implementation

[0014] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0015] Example 1, as Figure 1 As shown, this application provides a method for partial discharge identification of GIS equipment based on UHF signals, wherein the method includes:

[0016] UHF sensors are deployed on the outside of the insulating flange of the GIS equipment to collect UHF pulse signals generated by partial discharge inside the GIS.

[0017] In this embodiment, the installation position on the outer side of the insulating flange is determined based on the structural distribution and electric field concentration area of ​​the GIS equipment. The installation position is preferably located at the gas chamber connection or in an area sensitive to electric field distortion to improve the coupling efficiency of the partial discharge signal. A UHF sensor with a working frequency band covering 300MHz to 3GHz is selected and fixed to the outer surface of the insulating flange in a non-invasive manner. The UHF sensor receives the UHF electromagnetic radiation signal generated by partial discharge inside the GIS via electromagnetic coupling. An impedance matching structure is set between the sensor and the GIS equipment to reduce signal reflection loss and improve signal transmission efficiency. The output terminal of the UHF sensor is connected to the front-end signal access interface via a coaxial cable. A bandpass filter unit is configured in the signal access path to suppress power frequency interference and low-frequency noise, and an amplitude limiting protection unit is configured to prevent high-amplitude impact signals from damaging subsequent circuits. During the acquisition process, the UHF pulse signal is continuously received within a preset sampling time window, and effective partial discharge pulses are captured according to the pulse amplitude threshold triggering mechanism to obtain the corresponding UHF pulse signal sequence, which serves as input data for subsequent analog calculation array processing.

[0018] A signal interaction is established between a high-frequency sensor and an analog computing array. The analog computing array receives ultra-high frequency pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multidimensional characteristic voltages.

[0019] Specifically, the pulse signal output from the ultra-high frequency sensor is connected to the input bus of the analog computing array after impedance matching and buffer amplification at the front end. The pulse signal is then replicated at the same amplitude and synchronously distributed to multiple parallel processing branches via an analog signal distribution network. During signal interaction, a unified triggering reference is set, with the moment the pulse signal exceeds a preset threshold serving as the start mark for full array synchronization, driving each processing branch into parallel operation. The first processing branch performs edge detection on the input pulse signal, extracts the pulse rising edge start point using a high-speed comparator, and generates an analog voltage signal corresponding to the rise time by combining it with a ramp reference voltage. The second processing branch… Zero-crossing detection is performed on the input pulse signal, and the pulse oscillation frequency is characterized by period counting or equivalent charge-discharge process, outputting an analog voltage signal corresponding to the frequency change; the third processing branch performs energy accumulation processing on the input pulse signal, constructs a current integration channel through transconductance amplification and integrating capacitor, integrates and converts the pulse amplitude and duration to obtain an analog voltage signal representing the pulse energy; after feature extraction is completed by each processing branch, the outputs of each branch are synchronously latched and time-aligned through an analog convergence node to form a multi-dimensional feature voltage vector under the same time base, and output to the subsequent defect detection array as the input feature signal.

[0020] Furthermore, the analog computing array includes a first processing branch, a second processing branch, and a third processing branch connected in parallel for analog domain processing; by importing the high-frequency pulse signal into the analog computing array, processing based on rise time, oscillation frequency, and energy integration is performed in parallel to determine the multidimensional characteristic voltage.

[0021] Preferably, the high-frequency pulse signal from the ultra-high frequency sensor is amplified by a pre-stage buffer and then input to the common input terminal of the analog computing array. The high-frequency pulse signal is then split into equal amplitude streams via a low-distortion analog distribution network and directed to the first, second, and third processing branches respectively. Before the signal enters each processing branch, a unified threshold trigger unit determines the pulse arrival time and generates a synchronization trigger signal, enabling the three processing branches to initiate their corresponding analog processing steps at the same time reference. The first processing branch performs time-domain response conversion on the pulse rising edge, and the second processing branch performs pulse oscillation characteristic conversion... The third processing branch performs energy accumulation conversion on the pulse amplitude and duration, and each processing branch independently completes the corresponding feature extraction while maintaining the phase consistency of the input signal. A simulated latch node is set at the output of each processing branch to hold the output results of each branch instantaneously and align them through a unified sampling timing to obtain the first voltage, second voltage and third voltage corresponding to the same pulse. The first voltage, second voltage and third voltage are vectorized and integrated to construct a multi-dimensional feature voltage characterizing the characteristics of a single partial discharge pulse, and output to the subsequent defect detection array as input feature quantity.

[0022] Furthermore, parallel processing based on rise time, oscillation frequency, and energy integration is performed to determine multidimensional characteristic voltages, including:

[0023] The first processing branch consists of a high-speed comparator and a ramp generator, which performs rise time detection to determine a first voltage proportional to the rise time; the second processing branch consists of a zero-crossing detector and a timing capacitor, which performs oscillation frequency detection to determine a second voltage proportional to the frequency; the third processing branch consists of a transconductance amplifier and an integrating capacitor, which performs energy integration processing to determine a third voltage characterizing the pulse energy; the first voltage, the second voltage, and the third voltage are integrated to determine the multidimensional characteristic voltage, wherein the multidimensional characteristic voltage is marked with a pulse timestamp and stored in a discharge database.

[0024] The first processing branch consists of a high-speed comparator and a ramp generator. When the input high-frequency pulse signal arrives, the high-speed comparator compares the pulse signal with a preset threshold to extract the rising edge start time. At the same time, the ramp generator is triggered to output a linearly increasing reference voltage signal. When the pulse signal reaches its peak value or a preset termination condition is met, the ramp voltage is sampled and held to obtain a first voltage that is proportional to the rising edge duration.

[0025] The second processing branch consists of a zero-crossing detector and a timing capacitor. The zero-crossing detector performs zero-crossing detection on the oscillation process of the pulse signal, and drives the timing capacitor to perform charge and discharge counting or equivalent time integration between adjacent zero-crossing points, converting the oscillation period information into voltage amplitude, and thus obtaining a second voltage that is proportional to the oscillation frequency.

[0026] The third processing branch consists of a transconductance amplifier and an integrating capacitor. The transconductance amplifier converts the input pulse voltage signal into a corresponding current signal, and the current signal is integrated over time on the integrating capacitor to obtain a voltage output that reflects the combined effect of the pulse amplitude and duration, thereby obtaining a third voltage characterizing the pulse energy.

[0027] After feature extraction is completed in each processing branch, the first voltage, the second voltage, and the third voltage are time-aligned and output stably, and combined and encoded according to a unified data format to form multidimensional feature voltages. At the same time, a pulse timestamp identifier based on the trigger time is added to each group of multidimensional feature voltages, and the multidimensional feature voltages are associated with the corresponding time information and stored in the discharge database to support subsequent discharge behavior analysis and sequence feature mining.

[0028] The multidimensional characteristic voltage is imported into the defect detection array, and the defect type analysis and discharge severity level assessment are performed based on the multidimensional characteristic voltage to determine the discharge assessment result. The defect detection array is a cross-switch architecture, and each cross node is configured with a transimpedance amplifier with an initialized conductance value.

[0029] Specifically, multidimensional feature voltages are used as input feature vectors and loaded onto the row nodes of the defect detection array via an input bus, so that each row node corresponds to a feature voltage of a different dimension. The column nodes of the defect detection array correspond to output channels for different defect types, and a mapping relationship between feature dimensions and defect types is constructed through a row-column cross structure. A transimpedance amplifier is set at each cross node, and its conductance value is configured during the initialization phase based on feature weight coefficients obtained from training with historical samples. This amplifier is used to perform weighted transformation on the corresponding input feature voltage and output the current or voltage contribution. During array operation, each input feature voltage is synchronously applied to each cross node in the corresponding row direction. After weighted processing by a transimpedance amplifier, current convergence or voltage superposition is performed along the column direction to form a comprehensive response signal corresponding to the defect type at the output of each column. By comparing or normalizing the amplitude of the output signals of each column, the defect type corresponding to the column with the largest response value is determined as the partial discharge identification result. At the same time, based on the energy characteristic components in the multidimensional characteristic voltage and their response intensity in the array output, the discharge intensity is quantitatively evaluated in combination with the preset level classification rules, and the discharge development stage is determined according to the changing trend of different characteristic components, thereby determining the discharge severity level. The defect type and the discharge severity level are combined and encoded to form the discharge evaluation result.

[0030] Furthermore, the cross nodes of the defect detection array are expandable; wherein, the row nodes of the defect detection array correspond to the input characteristic voltage, and the column corresponds to the output defect type. The input characteristic voltage includes rise time voltage, oscillation frequency voltage, and pulse energy voltage, and the defect type includes tip discharge, free particle, and air gap discharge; and data interaction between the defect detection array and the simulation calculation array is established.

[0031] Preferably, the defect detection array is constructed using modular cross-switch units. The number of row nodes is expanded horizontally to accommodate new feature dimensions, and the number of column nodes is expanded vertically to accommodate new defect types. Corresponding transimpedance amplifiers are configured at the intersections of new rows and columns to maintain the consistency of the array's computational structure. In terms of mapping, the rise time voltage, oscillation frequency voltage, and pulse energy voltage from the analog computing array are respectively loaded onto the input terminals of the corresponding row nodes, allowing each row node to receive the multi-dimensional feature voltage corresponding to the same pulse at the same time. Parallel response channels for different defect types are formed in the column direction, with the first column corresponding to tip discharge, the second column to free particles, and the third column to air gap discharge. During data interaction, in the analog computing array… An analog interface buffer unit is set between the output of the computational array and the input of the defect detection array to achieve impedance matching and signal isolation, and to achieve synchronous loading of multi-dimensional characteristic voltages through a unified trigger control signal. After loading, the input signals of each row are weighted and converted by transimpedance amplifiers at the cross nodes, and then converged and output along the column direction to achieve the mapping conversion from feature space to defect type space. At the same time, in the case of array expansion, programmable connection switches or multiplexing units are configured to address and activate new nodes to ensure that the array can maintain a unified data interaction timing and stable output after scaling, thereby realizing stable data interaction and scalable computing capabilities between the defect detection array and the analog computational array.

[0032] Furthermore, each cross node is configured with a transimpedance amplifier after initializing its conductance value. The initialization process includes:

[0033] Retrieve three types of sample data based on defect type and mine feature weight coefficients based on defect classification hyperplane; map the feature weight coefficients to the conductance values ​​of transconductance amplifiers and configure the parameters of the defect detection array, wherein the three feature weight coefficients corresponding to tip discharge are encoded into the three transconductance amplifiers in the first column, the three feature weight coefficients corresponding to free particles are encoded into the three transconductance amplifiers in the second column, and the three feature weight coefficients corresponding to air gap discharge are encoded into the three transconductance amplifiers in the third column.

[0034] Preferably, three types of sample data based on defect type are retrieved, and feature extraction is performed on the three types of sample data to obtain corresponding rise time features, oscillation frequency features, and pulse energy features, and a feature vector set is constructed. Based on the feature vector set, a defect classification model is constructed using a supervised classification method. By training on the distribution of samples of different defect types in the feature space, a classification hyperplane for distinguishing different defect types is determined, and the feature weight coefficients corresponding to each feature dimension are obtained by parsing the classification hyperplane. The feature weight coefficients are normalized and amplitude constrained to meet the requirements of the transconductance amplifier's conductance adjustment range and linear operating range. The processed feature weight coefficients are mapped to the equivalent conductance value of the transconductance amplifier, and adjusted by current mirror scaling and resistance. The array configuration or programmable bias voltage control method writes the conductance parameters of the transimpedance amplifiers at each cross node to complete the initial configuration of the defect detection array. Specifically, the three feature weighting coefficients corresponding to tip discharge are loaded into the three transconductance amplifiers corresponding to the first column, the three feature weighting coefficients corresponding to free particles are loaded into the three transconductance amplifiers corresponding to the second column, and the three feature weighting coefficients corresponding to air gap discharge are loaded into the three transconductance amplifiers corresponding to the third column, so that each column output channel forms a weighted response model for the corresponding defect type. After initialization, when the defect detection array receives multi-dimensional feature voltage input, it performs weighted calculations on the input features using the configured conductance values ​​to achieve rapid discrimination of different defect types.

[0035] Furthermore, defect type analysis based on the multidimensional characteristic voltage includes:

[0036] The multidimensional characteristic voltage is imported into the analog calculation array to obtain three confidence voltages, wherein the three confidence voltages correspond one-to-one with the defect type; according to the comparator, the three confidence voltages are compared, and the defect type corresponding to the highest confidence level is selected as the discharge identification result.

[0037] Preferably, multidimensional characteristic voltages are introduced into a defect detection array. In the array, the input characteristic voltages are weighted and converted by transimpedance amplifiers with initialized conductance values ​​at each cross-node, so that each characteristic voltage is applied to each column channel in the corresponding row direction. Current convergence or voltage superposition occurs in the column direction, forming comprehensive response signals corresponding to different defect types. At each column output, after transimpedance amplification and voltage conversion, three confidence voltages are obtained, each corresponding to one of the three defect types: tip discharge, free particle discharge, and air gap discharge. The amplitude of each confidence voltage represents the matching degree or confidence level of the corresponding defect type. The three confidence voltages are input to a comparator, which employs a multi-channel voltage comparison structure to synchronously compare the three confidence voltages. A maximum value selection mechanism is used to output the channel identifier corresponding to the confidence voltage with the largest amplitude. The corresponding defect type is determined based on the channel identifier as the discharge identification result, and this result is output to the subsequent discharge level assessment and early warning processing unit, thereby achieving rapid defect type identification based on multidimensional characteristic voltages.

[0038] Furthermore, if the three confidence voltages are less than the pre-set confidence level built into the comparator, the discharge identification result is the first standard case; if the discharge identification result is any type of defect, a discharge level assessment instruction is generated, and discharge intensity quantification and discharge development stage assessment based on multi-dimensional characteristic voltage are performed to determine the discharge severity level and add it to the discharge identification result.

[0039] Preferably, a confidence threshold voltage is preset in the comparator. When the maximum value among the three confidence voltages is lower than the threshold voltage, it is determined that the current input feature does not meet the matching condition with any defect type. This state is marked as the first condition, and the corresponding invalid identification mark or pending confirmation mark is output. When there is a voltage signal among the three confidence voltages that is greater than or equal to the preset confidence level, the corresponding defect type is determined according to the maximum value selection mechanism, and a discharge level assessment instruction is generated simultaneously. The multi-dimensional feature voltage is input to the level assessment processing channel. During the level assessment process, the amplitude of the feature voltage representing pulse energy is extracted or analog-to-digital converted to obtain the discharge intensity quantification value. Combined with the characteristic voltage change trend representing rise time and oscillation frequency, the development stage of the discharge process is determined. The rise time feature is used to reflect the change in the steepness of the discharge, and the oscillation frequency feature is used to represent the evolution of the insulation state. According to the discharge intensity quantification value and the discharge development stage, the discharge severity level is determined according to the preset level classification rules. The defect type and the discharge severity level are combined and encoded to form a discharge identification result containing type and level information for subsequent early warning management and status assessment.

[0040] Furthermore, a discharge severity assessment is conducted, including:

[0041] The third voltage is converted from analog to digital to obtain the discharge intensity quantification value; based on the first voltage and the second voltage, the discharge development stage is evaluated, wherein the first voltage shows a decreasing trend as the discharge develops, and the second voltage shows a bandwidth broadening trend as the insulation deteriorates; based on the discharge intensity quantification value and the discharge development stage, the discharge severity level is determined.

[0042] Preferably, the third voltage is input to the analog-to-digital conversion unit, and the digital quantization result is obtained according to the preset sampling accuracy. The digital quantization result is then mapped to the corresponding discharge energy amplitude through a preset calibration curve, which is used as the discharge intensity quantization value. The first voltage is normalized to reflect the changing trend of the discharge steepness. As the discharge development process intensifies, the first voltage shows an overall decreasing trend. The second voltage is mapped with frequency band characteristics to characterize the change in oscillation frequency distribution. As insulation deterioration deepens, the frequency response corresponding to the second voltage shows a frequency band broadening trend. A two-dimensional development stage judgment space is constructed based on the first voltage and the second voltage, dividing different discharge states into the initial discharge stage, the developing discharge stage, and the severe discharge stage. The current discharge state is assigned to a range through a preset stage judgment threshold. According to the discharge intensity quantization value and the discharge development stage, a discharge severity level judgment rule is constructed. The discharge intensity quantization value is used as the amplitude dimension indicator, and the discharge development stage is used as the state dimension indicator for joint evaluation. The final discharge severity level is determined through weighted fusion or hierarchical mapping, and the corresponding level identifier is output for subsequent early warning management and state assessment.

[0043] The discharge assessment results are visualized on the terminal, and partial discharge early warning management is implemented using GIS equipment.

[0044] Specifically, the defect type, discharge severity level, and corresponding pulse timestamp information in the discharge assessment results are analyzed, and a multi-dimensional visualization interface is generated according to preset display rules. The partial discharge state is then presented graphically on the terminal display unit, including a discharge trend curve based on time series, a distribution statistical chart based on defect type classification, and a level distribution chart based on discharge intensity. Simultaneously, the real-time collected discharge assessment results are correlated with a historical discharge database to construct a two-dimensional discharge spectrum with time as the vertical axis, power frequency phase as the horizontal axis, and characteristic voltage or discharge intensity as color scales, thereby achieving periodic characteristics of discharge behavior. The system demonstrates and analyzes evolution trends. During the early warning management process, multiple early warning thresholds are set according to the severity level of the discharge. When the severity level of the discharge reaches or exceeds the corresponding threshold, an early warning output is triggered. The early warning methods include terminal interface alarm prompts, audible and visual alarms, and remote communication alarm information push. Furthermore, when the frequency of the same defect type occurring within a continuous time window exceeds a preset number or the discharge intensity shows a continuous upward trend, it is determined to be a state of deterioration aggravation, the early warning level is increased, and maintenance suggestion information is generated. Through the aforementioned terminal visualization and early warning management mechanism, real-time monitoring, trend analysis, and risk early warning output of the partial discharge status of GIS equipment are realized.

[0045] Furthermore, after implementing partial discharge early warning management for GIS equipment, it includes:

[0046] Based on the discharge database, a periodic two-dimensional spectrum is constructed with pulse occurrence time as the vertical axis, power frequency phase as the horizontal axis, and multi-dimensional characteristic voltage as color marks. Through spatiotemporal clustering, pulse sequences of the same discharge source are identified in the two-dimensional spectrum, and multiple pulse sequences corresponding to different discharge sources are determined. For the multiple pulse sequences, sequence feature extraction and judgment are performed to generate deterioration warning information for sporadic discharges.

[0047] Preferably, by mapping the timestamp of each pulse to the corresponding time coordinate and projecting the pulse onto the phase coordinate axis based on the synchronously acquired power frequency phase information, and using the energy component or fused feature value in the multidimensional characteristic voltage as the color mark intensity, a two-dimensional discrete dot matrix map characterizing the distribution characteristics of partial discharge is generated, and the map is superimposed and updated over multiple power frequency cycles to form a periodic evolution spectrum. A spatiotemporal distance metric is constructed based on the proximity of pulses on the time axis and phase axis, and a multidimensional clustering criterion is constructed in combination with the similarity of characteristic voltages. Pulse points are grouped by density clustering or threshold-based region growing methods, and the pulse sets with continuous spatial distribution and similar characteristics in the clustering results are determined to be pulse sequences corresponding to the same discharge source. Sequence feature parameters, including pulse occurrence frequency, phase concentration, energy fluctuation amplitude, and time interval distribution, are extracted for each pulse sequence. The evolution of the pulse sequence within a continuous time window is evaluated by constructing sequence stability index and growth trend index. When a pulse sequence is detected to exhibit a characteristic pattern of low frequency occurrence but sudden energy increase, or intermittent occurrence and gradual convergence of phase distribution, it is determined to be an occasional discharge degradation trend, and corresponding degradation warning information is generated.

[0048] In summary, the embodiments of this application have at least the following technical effects:

[0049] First, a high-frequency sensor is deployed on the outside of the insulating flange of the GIS equipment to collect high-frequency pulse signals generated by partial discharge inside the GIS. Next, a signal interaction is established between the high-frequency sensor and an analog computing array. The analog computing array receives the high-frequency pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multi-dimensional characteristic voltages. Then, the multi-dimensional characteristic voltages are imported into a defect detection array. Based on these voltages, defect type analysis and discharge severity assessment are performed to determine the discharge assessment results. The defect detection array uses a cross-switch architecture, with each cross node configured with a transimpedance amplifier initialized with a conductivity value. Finally, the discharge assessment results are visualized at the terminal, enabling partial discharge early warning management of the GIS equipment. This method solves the technical problem in existing technologies where the signal undergoes multiple processing stages (analog-to-digital conversion, data transmission, and software processing), leading to cumulative delays and hindering real-time pulse-level identification of partial discharge. It achieves the technical effects of shortening the signal processing chain, enabling rapid response and real-time identification of partial discharge characteristics, and improving the timeliness of online early warnings.

[0050] Example 2 is based on the same inventive concept as the GIS equipment partial discharge identification method based on UHF signals in the previous examples, such as... Figure 2 As shown, this application provides a partial discharge identification system for GIS equipment based on ultra-high frequency signals, wherein the system includes:

[0051] Signal acquisition module 11: Deploys a UHF sensor on the outside of the insulating flange of the GIS equipment to acquire UHF pulse signals generated by partial discharge inside the GIS; Signal interaction processing module 12: Establishes signal interaction between the high-frequency sensor and the analog computing array. The analog computing array receives the UHF pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multidimensional characteristic voltages; Evaluation module 13: Imports the multidimensional characteristic voltages into the defect detection array, performs defect type analysis and discharge severity assessment based on the multidimensional characteristic voltages, and determines the discharge assessment results. The defect detection array is a cross-switch architecture, and each cross node is configured with a transimpedance amplifier after initializing the conductance value; Early warning management module 14: Visualizes the discharge assessment results on the terminal and performs partial discharge early warning management for the GIS equipment.

[0052] Furthermore, the signal interaction processing module 12 is used to perform the following methods:

[0053] The analog computing array includes a first processing branch, a second processing branch, and a third processing branch connected in parallel for analog domain processing; by importing the high-frequency pulse signal into the analog computing array, processing based on rise time, oscillation frequency, and energy integration is performed in parallel to determine multidimensional characteristic voltages.

[0054] Furthermore, the signal interaction processing module 12 is used to perform the following methods:

[0055] By comparing the real-time inventory with the multiple demand forecasts, multiple replenishment gaps are obtained. Based on the multiple replenishment gaps, during the dynamic replenishment process of the various board specifications and categories in the multiple flexible storage units, the wear-resistant QR code tags of the replenished board are updated and written with identity process data, and the storage information of the multiple RFID storage location identification tags is updated synchronously.

[0056] Furthermore, the evaluation module 13 is used to perform the following method:

[0057] The cross nodes of the defect detection array are expandable; wherein, the row nodes of the defect detection array correspond to the input characteristic voltage, and the column corresponds to the output defect type. The input characteristic voltage includes rise time voltage, oscillation frequency voltage and pulse energy voltage, and the defect type includes tip discharge, free particle and air gap discharge; data interaction between the defect detection array and the simulation calculation array is established.

[0058] Furthermore, the evaluation module 13 is used to perform the following method:

[0059] Retrieve three types of sample data based on defect type and mine feature weight coefficients based on defect classification hyperplane; map the feature weight coefficients to the conductance values ​​of transconductance amplifiers and configure the parameters of the defect detection array, wherein the three feature weight coefficients corresponding to tip discharge are encoded into the three transconductance amplifiers in the first column, the three feature weight coefficients corresponding to free particles are encoded into the three transconductance amplifiers in the second column, and the three feature weight coefficients corresponding to air gap discharge are encoded into the three transconductance amplifiers in the third column.

[0060] Furthermore, the evaluation module 13 is used to perform the following method:

[0061] The multidimensional characteristic voltage is imported into the analog calculation array to obtain three confidence voltages, wherein the three confidence voltages correspond one-to-one with the defect type; according to the comparator, the three confidence voltages are compared, and the defect type corresponding to the highest confidence level is selected as the discharge identification result.

[0062] Furthermore, the evaluation module 13 is used to perform the following method:

[0063] If the three confidence voltages are less than the pre-set confidence level built into the comparator, the discharge identification result is the first standard case; if the discharge identification result is any type of defect, a discharge level assessment instruction is generated, and discharge intensity quantification and discharge development stage assessment based on multi-dimensional characteristic voltage are performed to determine the discharge severity level and add it to the discharge identification result.

[0064] Furthermore, the evaluation module 13 is used to perform the following method:

[0065] The third voltage is converted from analog to digital to obtain the discharge intensity quantification value; based on the first voltage and the second voltage, the discharge development stage is evaluated, wherein the first voltage shows a decreasing trend as the discharge develops, and the second voltage shows a bandwidth broadening trend as the insulation deteriorates; based on the discharge intensity quantification value and the discharge development stage, the discharge severity level is determined.

[0066] Furthermore, the early warning management module 14 is used to perform the following methods:

[0067] Based on the discharge database, a periodic two-dimensional spectrum is constructed with pulse occurrence time as the vertical axis, power frequency phase as the horizontal axis, and multi-dimensional characteristic voltage as color marks. Through spatiotemporal clustering, pulse sequences of the same discharge source are identified in the two-dimensional spectrum, and multiple pulse sequences corresponding to different discharge sources are determined. For the multiple pulse sequences, sequence feature extraction and judgment are performed to generate deterioration warning information for sporadic discharges.

[0068] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for identifying partial discharge in GIS equipment based on ultra-high frequency signals, characterized in that, The method includes: A high-frequency sensor is deployed on the outside of the insulating flange of the GIS equipment to collect high-frequency pulse signals generated by partial discharge inside the GIS. A signal interaction is established between a high-frequency sensor and an analog computing array. The analog computing array receives ultra-high frequency pulse signals and performs parallel processing based on rise time, oscillation frequency, and energy integration to determine multidimensional characteristic voltages. The multidimensional characteristic voltage is imported into the defect detection array, and the defect type analysis and discharge severity level assessment are performed based on the multidimensional characteristic voltage to determine the discharge assessment result. The defect detection array is a cross-switch architecture, and each cross node is configured with a transimpedance amplifier after initializing the conductance value. The discharge assessment results are visualized on the terminal, and partial discharge early warning management is implemented using GIS equipment; The cross nodes of the defect detection array are scalable; Among them, the row nodes of the defect detection array correspond to the input characteristic voltage, and the column corresponds to the output defect type. The input characteristic voltage includes rise time voltage, oscillation frequency voltage and pulse energy voltage, and the defect type includes tip discharge, free particle and air gap discharge. Establish data interaction between the defect detection array and the simulation computing array; Each cross node is configured with a transimpedance amplifier after initializing its conductance value. The initialization process includes: Retrieve three types of sample data based on defect type and mine the feature weight coefficients based on the defect classification hyperplane; The feature weighting coefficients are mapped to the conductance values ​​of the transconductance amplifiers, and the defect detection array is configured with parameters. Specifically, the three feature weighting coefficients corresponding to tip discharge are encoded into the three transconductance amplifiers in the first column, the three feature weighting coefficients corresponding to free particles are encoded into the three transconductance amplifiers in the second column, and the three feature weighting coefficients corresponding to air gap discharge are encoded into the three transconductance amplifiers in the third column.

2. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 1, characterized in that, The simulation computing array includes a first processing branch, a second processing branch, and a third processing branch connected in parallel for simulation domain processing. By importing the high-frequency pulse signal into the analog computing array, parallel processing based on rise time, oscillation frequency, and energy integration is performed to determine the multidimensional characteristic voltage.

3. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 2, characterized in that, Parallel processing based on rise time, oscillation frequency, and energy integration is performed to determine multidimensional characteristic voltages, including: The first processing branch consists of a high-speed comparator and a ramp generator, which performs rise time detection to determine a first voltage that is proportional to the rise time. The second processing branch consists of a zero-crossing detector and a timing capacitor, which performs oscillation frequency detection and determines a second voltage that is proportional to the frequency; The third processing branch consists of a transconductance amplifier and an integrating capacitor, performs energy integration processing, and determines a third voltage characterizing the pulse energy. By integrating the first voltage, the second voltage, and the third voltage, the multidimensional characteristic voltage is determined, wherein the multidimensional characteristic voltage is identified by pulse timestamp and stored in the discharge database.

4. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 1, characterized in that, Defect type analysis based on the aforementioned multidimensional characteristic voltage includes: The multidimensional characteristic voltage is introduced into the defect detection array to obtain three confidence voltages, wherein the three confidence voltages correspond one-to-one with the defect type; The three confidence voltages are compared using a comparator, and the defect type corresponding to the highest confidence level is selected as the discharge identification result.

5. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 4, characterized in that, If the three confidence voltages are less than the preset confidence level built into the comparator, the discharge identification result is the first standard case; If the discharge identification result is any type of defect, a discharge level assessment instruction is generated, and discharge intensity quantification and discharge development stage assessment based on multi-dimensional characteristic voltage are performed to determine the discharge severity level and add it to the discharge identification result.

6. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 5, characterized in that, Conduct a discharge severity assessment, including: The third voltage is converted from analog to digital and used as the quantized value of the discharge intensity. Based on the first voltage and the second voltage, the discharge development stage is evaluated. The first voltage shows a decreasing trend as the discharge progresses, while the second voltage shows a bandwidth broadening trend as the insulation deteriorates. The severity level of the discharge is determined based on the quantified value of the discharge intensity and the stage of discharge development.

7. The method for partial discharge identification of GIS equipment based on UHF signals as described in claim 3, characterized in that, After implementing partial discharge early warning management for GIS equipment, the following is included: Based on the discharge database, a periodic two-dimensional spectrum is constructed with pulse occurrence time as the vertical axis, power frequency phase as the horizontal axis, and multi-dimensional characteristic voltage as color marks. By using spatiotemporal clustering, pulse sequences from the same discharge source are identified in the two-dimensional spectrum, and multiple pulse sequences corresponding to different discharge sources are determined. For the multiple pulse sequences, sequence feature extraction and judgment are performed to generate deterioration warning information for sporadic discharges.

8. A partial discharge identification system for GIS equipment based on ultra-high frequency signals, characterized in that, The system is used to implement the partial discharge identification method for GIS equipment based on UHF signals according to any one of claims 1-7, the system comprising: Signal acquisition module: UHF sensors are deployed on the outside of the insulating flange of the GIS equipment to acquire UHF pulse signals generated by partial discharge inside the GIS; Signal interaction processing module: Establishes signal interaction between high-frequency sensor and analog computing array. The analog computing array receives ultra-high frequency pulse signal and performs parallel processing based on rise time, oscillation frequency and energy integration to determine multi-dimensional characteristic voltage. Evaluation module: The multidimensional characteristic voltage is imported into the defect detection array, and the defect type analysis and discharge severity level assessment are performed based on the multidimensional characteristic voltage to determine the discharge assessment result. The defect detection array is a cross-switch architecture, and each cross node is configured with a transimpedance amplifier after initializing the conductance value. Early warning management module: Visualizes the discharge assessment results on the terminal and performs partial discharge early warning management for GIS equipment; Furthermore, the evaluation module is used to perform the following methods: The cross nodes of the defect detection array are scalable; Among them, the row nodes of the defect detection array correspond to the input characteristic voltage, and the column corresponds to the output defect type. The input characteristic voltage includes rise time voltage, oscillation frequency voltage and pulse energy voltage, and the defect type includes tip discharge, free particle and air gap discharge. Establish data interaction between the defect detection array and the simulation computing array; Retrieve three types of sample data based on defect type and mine the feature weight coefficients based on the defect classification hyperplane; The feature weighting coefficients are mapped to the conductance values ​​of the transconductance amplifiers, and the defect detection array is configured with parameters. Specifically, the three feature weighting coefficients corresponding to tip discharge are encoded into the three transconductance amplifiers in the first column, the three feature weighting coefficients corresponding to free particles are encoded into the three transconductance amplifiers in the second column, and the three feature weighting coefficients corresponding to air gap discharge are encoded into the three transconductance amplifiers in the third column.

Images