An internal defect positioning method and system for an extremely cold lightning arrester of acoustic-electric combined imaging

Abstract

The present application relates to the technical field of internal defect positioning of lightning arresters, and specifically discloses a method and system for internal defect positioning of lightning arresters in extremely cold environments through acoustic-electric combined imaging, which synchronously collects acoustic-electric combined signals of the lightning arresters, calculates electromagnetic wave front distortion parameters to evaluate the internal medium state, and selects a dynamic spread threshold value or an adaptive anti-interference positioning mode according to a standard; in the adaptive mode, a suspicious abnormal region space mask is generated to perform weighted screening on acoustic signal data, thereby improving the positioning data quality; finally, the initial time difference data set and the filtered high-weight acoustic data set are used to output the three-dimensional coordinates of the internal defect source through a joint optimization algorithm. The present application can effectively deal with the problem of uneven internal medium of lightning arresters in extremely cold environments, improve the accuracy and reliability of defect positioning, and provide strong support for the safe and stable operation of power grids.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of internal defect positioning of lightning arresters, and relates to an internal defect positioning method and system for lightning arresters in extremely cold environments through acoustic-electric combined imaging. BACKGROUND

[0002] In a high-voltage power system, a lightning arrester is a key overvoltage protection device, and the reliability of the internal insulation state of the lightning arrester is directly related to the safe and stable operation of the entire power grid. Internal manufacturing defects, material aging or adverse environmental factors may cause partial discharge to occur, and if not timely discovered and accurately positioned, the discharge may develop into a through breakdown, causing device failure. Therefore, accurate positioning of the internal partial discharge defect source of the lightning arrester is an important technical prerequisite for realizing device state evaluation and preventive maintenance.

[0003] Patent No. CN102175721B discloses a method for visually detecting internal assembly defects of a metal oxide lightning arrester for electric power, which comprises: using electrical measurement and waveform observation as the main means and X-ray perspective irradiation as the auxiliary means, so that rapid diagnosis of internal assembly defects of the metal oxide lightning arrester for electric power can be realized; voltage waveform monitoring is performed on both ends of a discharge counter of a live lightning arrester through a general oscilloscope, and if high-frequency and high-amplitude noise is contained in the voltage waveform, it indicates that there is a metallic suspended particle defect in the lightning arrester; through the X-ray perspective irradiation means, focus is placed on the bottom of a single lightning arrester, and visual display of the internal structure of the lightning arrester can be realized, so that the defects existing in the lightning arrester can be directly determined.

[0004] The prior art has the following disadvantages: first, the complex influence of extremely cold and other adverse environments on the internal medium state of the lightning arrester is not fully considered, and the fixed criterion used may not be suitable for uneven media. Second, the technical means mainly rely on electrical measurement and offline X-ray perspective, which has limitations in information dimension and real-time performance, and it is difficult to realize deep joint analysis and online adaptive positioning based on electromagnetic wave front distortion and acoustic propagation time delay. SUMMARY

[0005] In view of this, the present application provides an internal defect positioning method and system for lightning arresters in extremely cold environments through acoustic-electric combined imaging.

[0006] The purpose of the present application can be achieved by the following technical solutions: the first aspect of the present application provides an internal defect positioning method for lightning arresters in extremely cold environments through acoustic-electric combined imaging, which comprises: S1, synchronously collecting acoustic-electric combined signals of the lightning arrester, for each electromagnetic pulse waveform collected by a super-high-frequency sensor, calculating the broadening amount of the pulse front, and quantifying the broadening amount as an electromagnetic wave front distortion parameter representing the dielectric loss degree of the signal propagation path.

[0007] S2. Obtain the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compare the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determine the internal dielectric state of the surge arrester based on the comparison result, and select the standard positioning processing mode or the adaptive anti-interference positioning processing mode.

[0008] S3. When the adaptive anti-interference positioning processing mode is selected, the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations is calculated, and a spatial mask of suspicious abnormal regions marking high-loss medium regions is generated accordingly.

[0009] S4. When the adaptive anti-interference positioning processing mode is selected, the initial arrival time difference dataset between piezoelectric acoustic sensors is calculated, the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area is analyzed, and the initial arrival time difference dataset is weighted and filtered to generate a filtered high-weight acoustic dataset.

[0010] S5. Based on the selected localization processing mode, execute the localization solution algorithm using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, and output the three-dimensional coordinates of the internal defect source.

[0011] The second aspect of the present invention provides an acoustic-electric imaging system for locating internal defects in extremely cold surge arresters, comprising: a signal acquisition and parameter extraction module, which synchronously acquires the acoustic-electric signals of the surge arrester, calculates the broadening of the pulse leading edge for each electromagnetic pulse waveform acquired by an ultra-high frequency sensor, and quantifies the broadening into an electromagnetic wavefront distortion parameter characterizing the degree of dielectric loss of the signal propagation path.

[0012] The status determination and mode selection module acquires the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compares the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determines the internal medium status of the surge arrester based on the comparison result, and selects either the standard positioning processing mode or the adaptive anti-interference positioning processing mode.

[0013] The regional spatial mask generation module, when the adaptive anti-interference positioning processing mode is selected, calculates the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations, and generates a spatial mask for suspicious abnormal regions that mark high-loss medium areas.

[0014] The high-weighted acoustic data filtering module, when the adaptive anti-interference positioning processing mode is selected, calculates the initial time difference of arrival dataset between piezoelectric acoustic sensors, analyzes the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area, and performs weighted filtering on the initial time difference of arrival dataset to generate a filtered high-weighted acoustic dataset.

[0015] The localization processing mode solving module executes a localization solving algorithm based on the selected localization processing mode, using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, and outputs the three-dimensional coordinates of the internal defect source.

[0016] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention can realize the online determination of whether the internal medium is uniform by synchronously collecting acoustic and electrical signals and using electromagnetic wavefront distortion parameters to determine the internal medium state. Electromagnetic wavefront distortion parameters are sensitive to changes in dielectric loss on the signal propagation path, and the presence of abnormal media such as ice crystals will cause significant dielectric loss. By comparing the distortion parameters extracted in real time with the normal threshold dynamically calculated according to the current working conditions, the system can distinguish abnormal distortion caused by non-uniformity of the medium, thereby realizing an effective assessment of the internal physical state of the equipment before performing positioning, and avoiding misuse of the standard positioning algorithm under inapplicable working conditions.

[0017] (2) This invention generates a spatial mask for suspicious anomaly regions based on the spatial distribution of electromagnetic wavefront distortion parameters, and uses this mask to perform adaptive weighted filtering on acoustic signal data, thereby improving the data quality used for localization calculations. The spatial mask can mark physical regions where acoustic characteristics may change in the form of a logic diagram. The system performs spatial correlation analysis between the propagation path of the acoustic signal and the mask, and assigns lower weights or removes arrival time difference data that pass through suspicious anomaly regions. This process can actively identify and suppress data that may introduce large errors due to traversing inhomogeneous regions of the medium, thereby purifying the initial dataset and providing more reliable input for subsequent localization solutions.

[0018] (3) When the presence of medium inhomogeneity is determined, the present invention employs a joint optimization algorithm, which can improve the positioning accuracy under complex working conditions. This algorithm not only uses the three-dimensional coordinates of the defect source as the variable to be solved, but also uses the equivalent sound velocity in the suspected abnormal area as the unknown to be solved. By utilizing the filtered high-weight acoustic dataset, the algorithm can simultaneously optimize the position and sound velocity parameters in one optimization iteration. This mechanism enables the sound velocity model to adaptively make local corrections based on the data inversion results, getting rid of the rigid dependence on the globally uniform sound velocity, realizing compensation for the influence of medium inhomogeneity, and improving the accuracy of the final output three-dimensional coordinates of the internal defect source. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the 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.

[0020] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0021] Figure 2 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0022] 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 embodiments of the present invention, and not all embodiments. 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.

[0023] Please see Figure 1 The first aspect of the present invention provides a method for locating internal defects of an extreme cold surge arrester by combined acoustic and electrical imaging, comprising: S1, synchronously acquiring the combined acoustic and electrical signals of the surge arrester, calculating the broadening of the pulse leading edge for each electromagnetic pulse waveform acquired by an ultra-high frequency sensor, and quantifying the broadening into an electromagnetic wavefront distortion parameter characterizing the degree of dielectric loss of the signal propagation path.

[0024] In a specific embodiment of the present invention, the synchronous acquisition of the acoustic-electric combined signal of the surge arrester includes: using a voltage comparator to monitor the output signal of the ultra-high frequency sensor, and generating a global synchronization trigger moment when any signal is detected to exceed a preset voltage threshold.

[0025] Based on the global synchronization trigger time, the data acquisition of all UHF sensors and piezoelectric acoustic sensors is started synchronously, and the broadband full waveform data within a preset time after the trigger point is recorded.

[0026] In a specific embodiment of the present invention, the step of calculating the broadening of the pulse leading edge and quantifying the broadening as an electromagnetic wavefront distortion parameter characterizing the dielectric loss of the signal propagation path includes: analyzing broadband full waveform data to determine a first preset percentage time and a second preset percentage time when the amplitude of the pulse leading edge rises to the final stable peak value.

[0027] Calculate the time difference between the second preset percentage time and the first preset percentage time, and use the time difference as an electromagnetic wavefront distortion parameter.

[0028] Specifically, step S1 is implemented as follows: A hybrid sensor array is pre-installed along the axial direction outside the surge arrester. This array includes several equally spaced ultra-high frequency (UHF) sensors and piezoelectric acoustic sensors. When a partial discharge occurs inside the surge arrester and generates an electromagnetic pulse, the UHF sensor closest to the discharge point will receive the pulse signal first. A high-speed voltage comparator configured in the system continuously monitors the analog signal output by each UHF sensor and compares it with a preset voltage threshold, which is set to 50 mV. This value is set based on the fact that, through long-term monitoring of background noise, its peak voltage is typically below 20 mV. Setting the threshold to 50 mV ensures sufficient detection sensitivity for real discharge pulses while controlling the false trigger probability to below 1%. Once the signal voltage of any UHF sensor exceeds the preset voltage threshold, the high-speed voltage comparator immediately generates a digital pulse and marks the moment of this digital pulse generation as the global synchronization trigger moment. The global synchronization trigger moment is sent to the trigger input port of a high-precision synchronization acquisition card. Based on the global synchronization trigger time, the synchronization acquisition card, through its internal unified clock source and trigger logic, synchronously starts all acquisition channels to digitally sample the analog signals output by each UHF sensor and piezoelectric acoustic sensor. The sampling process is performed at a sampling rate of 1 GHz to 5 GHz per second, and the signal waveforms within a preset time period after the global synchronization trigger time are recorded with nanosecond-level time accuracy, generating broadband full-waveform data corresponding to each sensor. For the broadband full-waveform data corresponding to the electromagnetic pulse waveform acquired by each UHF sensor, the signal processing program in the host computer first filters and denoises the electromagnetic pulse waveform, and then locates the process of the waveform pulse leading edge rising from a steady state to a peak value. Specifically, the program determines the time point when the amplitude of the waveform leading edge reaches 10% of the final stable peak value. and the time point when the final stable peak is reached (90%) The electromagnetic wavefront distortion parameters are quantized as a broadening factor, which is calculated by time difference. get:

[0029]

[0030] in, This represents the electromagnetic wavefront distortion parameter, or broadening, for a specific ultra-high frequency sensor, measured in nanoseconds. This indicates the moment when the amplitude of the pulse leading edge waveform rises to 90% of its final stable peak value in the broadband full waveform data. This represents the moment when the amplitude of the pulse leading edge rises to 10% of its final stable peak value in the broadband full-waveform data. This calculation is performed based on the independent broadband full-waveform data sequence for each sensor.

[0031] The ultra-high frequency sensor refers to a broadband antenna sensor with a center frequency in the range of 300 MHz to 3 GHz, used to couple electromagnetic radiation signals generated by partial discharge. The piezoelectric acoustic sensor refers to an acoustic emission sensor based on piezoelectric ceramic materials, with a resonant frequency range of 50 kHz to 500 kHz. The global synchronization trigger time is an absolute timestamp latched by hardware circuitry, serving as the time origin for all subsequent data acquisitions. The nanosecond-level accuracy refers to the time resolution capability of the data acquisition system being better than 10 ns. The preset duration is an acquisition window determined based on the maximum propagation time of the acoustic signal inside the surge arrester, typically ranging from 50 μs to 200 μs. The broadband full-waveform data refers to a discrete time series recorded at a high sampling rate, containing complete time-domain characteristics of the signal. The broadening amount... The rise time of an electromagnetic pulse is a specific physical quantity characterizing the rise time of the pulse's leading edge. A larger value indicates more severe high-frequency attenuation during signal propagation, thus indirectly reflecting the degree of dielectric loss along the path. Reaching 10% and 90% of the final stable peak value are common engineering reference points for defining the stable pulse rise time. These points are set according to the signal parameter measurement specifications of the IEEE standard, or can be obtained through statistical analysis of a large number of known partial discharge waveforms. For example, through statistical analysis of the leading edges of over 200 sets of laboratory standard discharge waveforms, it has been determined that these two points can stably characterize the rise behavior of over 95% of waveforms.

[0032] For example, assume that four UHF sensors, numbered 1 to 4, are axially arranged on the surge arrester. When an internal discharge occurs, the output of sensor number 2 first exceeds the 50 mV threshold. The high-speed voltage comparator captures this moment and marks it as T0 = 0 ns. The synchronous acquisition card is triggered at T0 and acquires data from all sensors for 200 μs at a sampling rate of 2 GHz. The host computer program filters the broadband full-waveform data acquired by sensor number 2 to obtain a waveform sequence. The program analyzes this sequence and finds a stable peak value of 1 V. Subsequently, the program searches for the sampling point where the waveform amplitude first exceeds 0.1 V (10% of the peak value), and the corresponding time... =15.5 ns. Next, the program searches for the sampling point where the waveform amplitude first exceeds 0.9 V (90% of the peak value), corresponding to the time... =28.0 ns. Next, the broadening amount is calculated. The value is 12.5 ns. This 12.5 ns value is extracted as the electromagnetic wavefront distortion parameter corresponding to sensor number 2. The system performs the same processing procedure on sensors 1, 3, and 4, which may yield different results. Values, such as 10.0 ns, 13.8 ns, and 9.5 ns.

[0033] S2. Obtain the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compare the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determine the internal dielectric state of the surge arrester based on the comparison result, and select the standard positioning processing mode or the adaptive anti-interference positioning processing mode.

[0034] In a specific embodiment of the present invention, the step of obtaining the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic widening threshold includes: querying a preset temperature-base widening mapping table based on the current external temperature to obtain the base widening amount.

[0035] It should be noted that the preset temperature-base broadening mapping table is based on statistical analysis of over 500 standard partial discharge tests conducted on the same type of defect-free surge arrester within a temperature range of -60°C to +50°C. A specific example is shown in Table 1 below (corresponding to a specific sensor location):

[0036] Table 1. Example of Temperature-Base Stretch Mapping

[0037] Temperature (°C) Base broadening amount (ns) -60 8.5 -40 7.8 -20 7.2 0 6.8 20 6.5 40 6.3 50 6.2

[0038] Based on the historical operating voltage, the preset voltage-loss coefficient relationship curve is queried to obtain the voltage correction coefficient.

[0039] The preset voltage-loss coefficient relationship curve is generated based on a simulation model of the nonlinear volt-ampere characteristics of the surge arrester MOV varistor, and is used to characterize the slight influence of operating voltage on the equivalent loss angle of the internal medium. A simplified linear approximation is shown below: when the historical average operating voltage... At 0.8 to 1.2 times the rated voltage Within the range, voltage influence coefficient .

[0040] The base widening amount is corrected by using a voltage correction factor, and a preset statistical margin constant is added to obtain the dynamic widening threshold.

[0041] Specifically, the status determination module in the host computer reads the current external temperature in real time from the temperature sensor installed at the arrester base via the communication interface configured in the device, such as the Modbus TCP protocol. The unit is Celsius. Simultaneously, this module accesses a historical database, such as an Oracle database in a SCADA system, to retrieve historical operating voltage data and obtain the average effective voltage value across the surge arrester over the most recent 24 hours. The unit is kilovolt. The dynamic broadening threshold... The calculations are performed in a separate calculation thread. This thread first considers the current external temperature. A base broadening amount is obtained by linear interpolation from a preset temperature-base broadening mapping table. The dynamic stretching threshold τ is ultimately determined by the base stretching amount. Multiply by voltage influence coefficient And superimposed a margin constant determined based on the statistical standard deviation of historical data. This yields a value characterizing the upper limit of the expected broadening under normal dielectric loss. After calculation, the state determination module reads the electromagnetic wavefront distortion parameters corresponding to all UHF sensor channels generated in step S1 from the shared memory, i.e., the broadening. ,in Number the sensor.

[0042]

[0043] in, This represents the dynamic broadening threshold, in nanoseconds. Indicates based on the current external temperature The base broadening is obtained by interpolation from the preset temperature-base broadening mapping table, in nanoseconds. Indicates based on historical operating voltage average value The voltage influence coefficient obtained by querying the preset relationship curve is a dimensionless correction multiplier. Indicates the statistical standard deviation based on historical normal data. The set margin constant is usually taken as The unit is nanoseconds, used to improve the robustness of the judgment and avoid false triggering.

[0044] In a specific embodiment of the present invention, the step of determining the internal medium state of the surge arrester based on the comparison results and selecting the standard positioning processing mode or the adaptive anti-interference positioning processing mode includes: if the electromagnetic wavefront distortion parameters corresponding to all ultra-high frequency sensors are less than the dynamic broadening threshold, then the internal medium is determined to be uniform, and the standard positioning processing mode is selected.

[0045] If the electromagnetic wavefront distortion parameter corresponding to any UHF sensor is greater than or equal to the dynamic broadening threshold, it is determined that there is an internal medium inhomogeneity anomaly, and the adaptive anti-interference positioning processing mode is selected.

[0046] Specifically, the module executes a loop comparison logic: comparing each... The calculated dynamic broadening threshold Perform numerical comparisons. If a cyclical check finds that all sensors correspond to... All less than If the state determination module generates a state flag (Flag=0) and writes it to the global state register, it simultaneously instructs the positioning algorithm to load the configuration file for the standard positioning processing mode. This configuration file presets globally unified sound velocity model parameters. If, during the loop comparison, any sensor corresponding to... Greater than or equal to If the loop terminates immediately, the state determination module generates a state flag bit Flag=1 and writes it into the register. At the same time, it instructs the positioning solution algorithm to load the configuration file of the adaptive anti-interference positioning processing mode and triggers the execution flow of step S3.

[0047] The current external temperature refers to the ambient temperature of the surge arrester measured and digitized in real time by a platinum resistance temperature sensor. The measurement range is typically -60°C to +50°C, with an accuracy of ±0.5°C. The historical operating voltage data refers to the time-series data of the surge arrester's operating voltage, periodically recorded and stored from the power grid monitoring system. The dynamic broadening threshold τ is a dynamically calculated scalar value. Its physical meaning is the upper limit of the expected electromagnetic wavefront broadening measured by each sensor should not exceed under the current specific operating conditions (temperature, voltage) when the internal dielectric of the surge arrester is in a uniform and healthy state. The temperature-basic broadening mapping table is a two-dimensional lookup table established through numerous benchmark experiments. The horizontal axis represents temperature, and the vertical axis represents the voltage levels measured by multiple sensors when a defect-free surge arrester generates a standard discharge at that temperature. Typical value of the statistical average. The voltage influence coefficient. This parameter, used to characterize the minute effect of operating voltage on the equivalent dielectric constant and loss angle of the internal dielectric, typically ranges from 0.95 to 1.05. The curve is constructed based on a simulation model of the valve plate's electrical characteristics provided by the surge arrester manufacturer. The margin constant... The setting is based on statistical analysis of the dispersion of benchmark experimental data, for example, assuming the standard deviation of the broadening of 200 sets of normal data. The time interval is 0.8 ns. To cover more than 95% of normal conditions, C can be set to 1.6 ns. The standard positioning processing mode is an algorithm operation state in which the system assumes there is no non-uniform medium interference on the acoustic signal propagation path, and subsequent positioning will use a fixed sound velocity model and all initial time difference data. The adaptive anti-interference positioning processing mode is another algorithm operation state. In this state, the system determines the existence of non-uniform medium anomalies caused by ice crystals, etc., and subsequent positioning will activate a special process that includes data filtering and sound velocity inversion.

[0048] S3. When the adaptive anti-interference positioning processing mode is selected, the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations is calculated, and a spatial mask of suspicious abnormal regions marking high-loss medium regions is generated accordingly.

[0049] In a specific embodiment of the present invention, the step of calculating the spatial variation rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations, and generating a spatial mask for suspicious abnormal regions marking high-loss medium regions, includes: obtaining the axial position coordinates of each ultra-high frequency sensor at different locations.

[0050] The spatial distribution gradient is obtained by calculating the ratio of the difference in electromagnetic wavefront distortion parameters to the difference in position coordinates between two adjacent UHF sensors.

[0051] The spatial distribution gradient is compared with a preset gradient threshold. Regions whose spatial distribution gradient exceeds the preset gradient threshold are marked as abnormal regions, and a binary spatial mask of the suspicious abnormal region is generated.

[0052] Specifically, when the status flag set by the status determination module in step S2 is 1, indicating that the system enters the adaptive anti-interference positioning processing mode, the spatial mask generation module in the host computer is automatically invoked. This module first reads the one-dimensional position coordinates of all UHF sensors along the arrester axis from the system configuration file. ,in The sensors are numbered, and the coordinate origin is set at the bottom of the surge arrester, with units in meters. Simultaneously, the module reads the electromagnetic wavefront distortion parameters, extracted in step S1 and corresponding to each sensor, from shared memory. Next, the module calculates the spatial distribution gradient of the electromagnetic wavefront distortion parameters. This calculation is performed by iterating through all adjacent sensor pairs. For any two axially adjacent sensors… and ,in The module calculates the gradient values ​​between them. This value is defined as the ratio of the difference in the width of the two sensors to the difference in their position coordinates. After calculation, the module will assign each gradient value... The absolute value of and a preset gradient threshold Compare. If the absolute value is greater than or equal to... Then determine the sensor and The axial region between these points represents a region of drastic distortion, indicating a potentially anomalous area where high-loss media may exist along the electromagnetic wave propagation path. Subsequently, the module generates a spatial mask for this anomalous region based on the aforementioned determination. This mask exists as a two-dimensional logic array, with its first dimension corresponding to the discretized position index along the arrester's axial direction and its second dimension corresponding to the discretized position index along the radial direction. However, given that the sensors are arranged in a one-dimensional axial direction, the radial dimension is assumed to have a uniform value in this embodiment. The generation logic is as follows: for an axial discrete grid point, if its coordinates fall within the coordinate range of any pair of adjacent sensors marked as anomalous, the mask value for that grid point and all its corresponding radial points is set to logic 1; otherwise, it is set to logic 0. Finally, the module outputs a complete binary logic diagram, namely the spatial mask for the suspected anomalous region, which is used in subsequent steps to indicate paths where acoustic signal propagation may be obstructed.

[0053]

[0054] in, Indicates from sensor To the sensor The spatial distribution gradient is expressed in nanoseconds per meter. Indicates sensor The corresponding electromagnetic wavefront distortion parameters are expressed in nanoseconds. Indicates sensor The corresponding electromagnetic wavefront distortion parameters are expressed in nanoseconds. Indicates sensor The axial position coordinates are in meters. Indicates sensor The axial position coordinates are given in meters. This formula is calculated sequentially for each pair of axially adjacent sensors to capture the rate of change of the width along space.

[0055] Wherein, the spatial distribution gradient It is a vector whose magnitude represents the change in electromagnetic wavefront distortion parameters per unit length, and whose sign indicates the direction of change. The physical meaning of this gradient lies in quantifying the severity of the impact of abrupt changes in local medium properties on electromagnetic wave propagation. The gradient threshold... This is a preset constant used to distinguish between normal gradual changes in the medium and abrupt, abnormal changes. Its setting is based on the analysis of a large amount of simulation or experimental data under known homogeneous medium conditions and ice-containing defect conditions. For example, by statistically analyzing the gradient values ​​of over 100 sets of simulated ice crystal accumulation scenarios, the 95th percentile is taken as the typical threshold value, with a reasonable range likely between 3.0 ns / m and 8.0 ns / m. The spatial mask of the suspected anomaly region is a binary logic diagram, represented in computer memory as a matrix with elements of 0 or 1. A logic value of 1 indicates that the spatial cell corresponding to that location is determined to have a high-loss medium anomaly, which may hinder acoustic signal propagation; a logic value of 0 indicates that the medium state at that location is homogeneous or the degree of anomaly is acceptable. The axial position coordinates are precise spatial information obtained by pre-measuring and calibrating the installation position of each UHF sensor on the arrester housing, with a measurement accuracy typically better than 1 cm. The discretized position index refers to the unique number of each grid after dividing the continuous internal space of the surge arrester into regular grids along the axial and radial directions. The grid size is set according to the positioning accuracy requirements. For example, the axial grid step size can be 0.05 m.

[0056] S4. When the adaptive anti-interference positioning processing mode is selected, the initial arrival time difference dataset between piezoelectric acoustic sensors is calculated, the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area is analyzed, and the initial arrival time difference dataset is weighted and filtered to generate a filtered high-weight acoustic dataset.

[0057] In a specific embodiment of the present invention, the analysis of the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area includes: constructing the acoustic signal straight-line propagation path corresponding to each set of arrival time difference data based on the three-dimensional coordinates of the piezoelectric acoustic sensor.

[0058] The linear propagation path of the acoustic signal is discretized into a set of spatial points, and the logical value corresponding to each spatial point in the spatial mask of the suspected anomaly region is queried.

[0059] If there is a spatial point on the straight propagation path of the acoustic signal that falls into the abnormal area marked by the mask, it is determined that the straight propagation path of the acoustic signal and the mask have spatial overlap. Conversely, if there is no spatial point on the straight propagation path of the acoustic signal that falls into the abnormal area marked by the mask, it is determined that the straight propagation path of the acoustic signal and the mask do not have spatial overlap.

[0060] Specifically, when the system operates in the adaptive anti-interference positioning processing mode triggered in step S2, the filtering and weighting module in the host computer is activated. This module first reads the acoustic waveform data recorded by all piezoelectric acoustic sensors synchronously acquired in step S1 within a preset time period after the global synchronization trigger moment from the data buffer. For any two different piezoelectric acoustic sensors A and B, the module uses a generalized cross-correlation algorithm to calculate the arrival time difference between their acquired waveforms. The specific process is as follows: first, pre-filtering is performed on the two discrete time series to enhance common frequency components; then, their cross-correlation function is calculated, and the peak position with the largest absolute value is searched within the cross-correlation function. The time delay corresponding to this position is calculated as the arrival time difference between sensors A and B. The system iterates through all possible sensor pairs and stores each pair of sensor numbers and their corresponding arrival time difference values ​​in a two-dimensional array, forming the initial arrival time difference dataset. Next, the module performs spatial location correlation analysis. The system configuration file pre-stores the three-dimensional spatial coordinates of all piezoelectric acoustic sensors. For each data set in the initial time difference of arrival dataset, i.e., each pair of sensors, the module calculates the spatial straight-line path connecting the two sensors based on their three-dimensional coordinates. Subsequently, the module discretizes this straight-line path into a series of spatial points and queries the spatial mask of the suspected anomaly region generated in step S3. The query process is as follows: each discrete point is mapped to the corresponding logical grid cell of the spatial mask, and the logical value of that cell is checked to see if it is 1 or 0. If any discrete point falls into a grid cell with a logical value of 1, it is determined that the acoustic signal propagation path overlaps spatially with the spatial mask of the suspected anomaly region; if the logical values ​​of all grid cells corresponding to discrete points are 0, it is determined that there is no overlap.

[0061] In a specific embodiment of the present invention, the step of weighted filtering of the initial time difference of arrival dataset to generate a filtered high-weight acoustic dataset includes: for acoustic signal straight propagation paths that are determined to have spatial overlap with the spatial mask of the suspicious abnormal area, assigning the corresponding time difference of arrival data a preset low weight coefficient or directly removing them.

[0062] For acoustic signal straight-line propagation paths that are determined not to have spatial overlap with the spatial mask of the suspected abnormal area, their corresponding time difference of arrival data are assigned a preset high weighting coefficient.

[0063] All retained data and their weighting coefficients are aggregated to form a filtered, high-weighted acoustic dataset.

[0064] Specifically, the module finally performs adaptive weighted filtering. For arrival time difference data determined to have overlap, the system reads low-weight values ​​from a preset rule base, or removes the data entry directly from the dataset to be processed according to the configuration. For arrival time difference data determined not to have overlap, the system assigns high-weight values. After traversing, judging, and weighting all data items, the module generates a new dataset containing only the data items assigned high weights, or both high and low-weight data with corresponding weight coefficients. This dataset is defined as the filtered high-weight acoustic dataset and output to shared memory for subsequent steps. The low-weight value is set to 0.2, and the high-weight value is set to 1.0, based on finite element simulation analysis: when simulating sound waves passing through ice crystal regions of different sizes, the additional error of its propagation time delay is approximately linearly related to the path crossing length. Statistical analysis shows that for paths identified as traversing suspicious areas, the probability of their time difference data introducing large errors (>10%) exceeds 70%, so they are given a low weight (0.2) to weaken their impact; for paths that are not traversed, their data reliability is high, so they are given a standard weight (1.0).

[0065] The arrival time difference (OTD) refers to the time difference between the acoustic signal's propagation from the internal defect source to sensor A and to sensor B, typically measured in microseconds. Its value can be positive or negative, depending on which sensor receives the signal first. The spatial location correlation analysis refers to determining, through geometric calculations and logical queries, whether a given acoustic propagation path passes through a three-dimensional spatial voxel marked by the spatial mask of the suspected anomaly region. The low and high weight values ​​are preset empirical coefficients used to measure the reliability of different time difference data in subsequent optimization solutions. Their setting is based on simulation analysis: in numerous scenarios containing simulated ice crystal interference, statistics show that the probability of the calculated arrival time difference data for acoustic paths traversing the interference region exceeding the normal range is higher than 80%, thus assigning it a low weight to weaken its impact; while the probability of path data not traversing the interference region having a smaller error is higher than 90%, hence assigning it a high weight.

[0066] S5. Based on the selected localization processing mode, execute the localization solution algorithm using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, and output the three-dimensional coordinates of the internal defect source.

[0067] In a specific embodiment of the present invention, the step of using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset to execute the localization solution algorithm and output the three-dimensional coordinates of the internal defect source includes: when the selected localization processing mode is the adaptive anti-interference localization processing mode, constructing a joint optimization objective function, and setting the variables to be solved as the three-dimensional coordinates of the internal defect source and the equivalent sound velocity in the area marked by the spatial mask of the suspected abnormal region.

[0068] Using a filtered, high-weight acoustic dataset, an iterative algorithm is employed to minimize the weighted error between the measured time difference of arrival (TDOA) and the theoretical TDOA predicted based on the current variables.

[0069] When the algorithm converges, it outputs the optimized three-dimensional coordinates of the internal defect source.

[0070] When the selected positioning processing mode is the standard positioning processing mode, the initial time difference of arrival dataset is directly called.

[0071] Using a pre-defined global uniform sound velocity model, the three-dimensional coordinates of the internal defect source are calculated and output by solving the time difference of arrival equations.

[0072] Specifically, when the positioning solver module in the host computer starts up, it first queries the status flag bit Flag written to the global status register in step S2. If the query result is Flag=0, it indicates that the system selects the standard positioning processing mode, and the positioning solver loads the configuration parameters of the standard mode. This configuration specifies that all piezoelectric acoustic sensors are used to calculate the initial time difference of arrival dataset, and a globally uniform sound velocity model is adopted, that is, assuming that the medium inside the surge arrester is uniform and the sound velocity is a constant. The constant The type and average temperature of the gas filling inside the surge arrester are obtained through a lookup table. The solver then executes a location algorithm based on the time difference of arrival. This location algorithm uses the three-dimensional coordinates of the defect source. For unknowns, a system of nonlinear equations is constructed using the time difference data of each sensor pair. The least-squares solution of this system is then solved numerically through iteration, ultimately outputting an estimated three-dimensional coordinate of the defect source. If the query result is Flag=1, indicating that the system has selected the adaptive anti-interference localization processing mode, the localization solver loads the configuration parameters of the adaptive mode. This configuration specifies that the filtered, high-weight acoustic dataset generated in step S4 is used as the core input data. The solver then executes the joint optimization solution algorithm, which typically employs a nonlinear least-squares optimizer. In the joint optimization solution algorithm, the variable vector to be solved contains two parts: the first part is the three-dimensional coordinates of the defect source. The second part consists of the equivalent sound velocity within one or more spatial regions defined by the suspected anomaly region spatial mask. Specifically, each connected spatial region with a mask logic value of 1 is assigned an independent equivalent sound velocity to be inverted. The objective function is constructed as the weighted sum of squares of the measured time difference of arrival (TDOA) and the theoretical TDOA predicted based on the current variable vectors (coordinates and sound velocity) for each sensor in the filtered, high-weight acoustic dataset. The weights are the weighting coefficients attached to the dataset. In each iteration, the algorithm adjusts the parameters based on the currently estimated defect source coordinates and the equivalent sound velocity for each region. The algorithm calculates the propagation time of the sound signal from the estimated source point to each sensor. The propagation time calculation needs to consider path segmentation: if the sound wave propagation path traverses different sound speed regions, the total time is equal to the sum of the propagation times in each uniform sound speed segment. After the objective function value is calculated, the optimization algorithm updates the variable vector based on gradients or swarm intelligence until the change in the objective function value is less than the preset tolerance or the maximum number of iterations is reached, at which point the algorithm converges. Finally, the variable vector output by the optimizer contains... The portion refers to the high-precision three-dimensional coordinates of the internal defect source, and simultaneously outputs various... The value is the equivalent sound velocity in the abnormal region obtained by inversion.

[0073] Among them, the globally uniform speed of sound in the standard positioning processing mode It is a preset constant, which is set based on the sound velocity-temperature relationship curve of the internal medium (such as SF6 gas or dry air) of the surge arrester under standard operating conditions, for example, in dry air at 20°C. A value of 343 m / s can be taken. The location algorithm based on time difference of arrival is a well-known algorithm in the field of sound source localization, which determines the source location by solving a system of hyperboloid equations. The joint optimization solution algorithm is a multi-parameter coupled optimization method, the innovation of which lies in jointly solving the sound velocity variable characterizing the inhomogeneity of the medium and the coordinates of the defect source within the same framework. The equivalent sound velocity... This is an inversion parameter, not representing the actual physical speed of sound, but rather the overall average propagation speed of sound waves as they pass through the complex medium region marked by the spatial mask of the suspected anomaly area. Its initial value can be set to a value between the speed of sound in healthy medium and the speed of sound in ice, for example, 2000 m / s to 3000 m / s, and the search range can be set to 1500 m / s to 3500 m / s. The preset tolerance of the objective function is a parameter controlling the convergence accuracy of the algorithm, with a typical value of 1e-6. The maximum number of iterations is a parameter to prevent the algorithm from running endlessly, with a typical value of 500.

[0074] Reference Figure 2 The second aspect of the present invention provides a system for locating internal defects of an extreme cold surge arrester using combined acoustic and electronic imaging, comprising: a signal acquisition and parameter extraction module, a state determination and mode selection module, a regional spatial mask generation module, a high-weight acoustic data filtering module, and a location processing mode solving module.

[0075] The signal acquisition and parameter extraction module and the state determination and mode selection module are connected. Both the signal acquisition and parameter extraction module and the state determination and mode selection module are connected to the regional spatial mask generation module. Both the state determination and mode selection module and the regional spatial mask generation module are connected to the high-weight acoustic data filtering module. Both the state determination and mode selection module and the high-weight acoustic data filtering module are connected to the localization processing mode solving module.

[0076] The signal acquisition and parameter extraction module synchronously acquires the combined acoustic and electrical signals of the surge arrester. For each electromagnetic pulse waveform acquired by the ultra-high frequency sensor, it calculates the broadening of the pulse leading edge and quantifies the broadening into an electromagnetic wavefront distortion parameter that characterizes the dielectric loss of the signal propagation path.

[0077] The state determination and mode selection module acquires the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compares the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determines the internal medium state of the surge arrester based on the comparison result, and selects either the standard positioning processing mode or the adaptive anti-interference positioning processing mode.

[0078] When the adaptive anti-interference positioning processing mode is selected, the regional spatial mask generation module calculates the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations, and generates a spatial mask that marks the suspicious abnormal regions of high-loss medium regions accordingly.

[0079] When the adaptive anti-interference positioning processing mode is selected, the high-weight acoustic data filtering module calculates the initial time difference dataset between piezoelectric acoustic sensors, analyzes the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area, and performs weighted filtering on the initial time difference dataset to generate a filtered high-weight acoustic dataset.

[0080] The localization processing mode solving module executes a localization solving algorithm using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, based on the selected localization processing mode, and outputs the three-dimensional coordinates of the internal defect source.

[0081] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for locating internal defects in extreme cold surge arresters using combined acoustic and electronic imaging, characterized in that, include: S1. Synchronously acquire the acoustic and electrical signals of the surge arrester. For each electromagnetic pulse waveform acquired by the ultra-high frequency sensor, calculate the pulse front broadening and quantify the broadening into an electromagnetic wavefront distortion parameter that characterizes the dielectric loss of the signal propagation path. S2. Obtain the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compare the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determine the internal dielectric state of the surge arrester based on the comparison result, and select the standard positioning processing mode or the adaptive anti-interference positioning processing mode. S3. When the adaptive anti-interference positioning processing mode is selected, the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations is calculated, and a spatial mask of suspicious abnormal regions marking high-loss medium regions is generated accordingly. S4. When the adaptive anti-interference positioning processing mode is selected, the initial arrival time difference dataset between piezoelectric acoustic sensors is calculated, the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area is analyzed, and the initial arrival time difference dataset is weighted and filtered to generate a filtered high-weight acoustic dataset. S5. Based on the selected localization processing mode, execute the localization solution algorithm using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, and output the three-dimensional coordinates of the internal defect source.

2. The method for locating internal defects in an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 1, characterized in that, The synchronous acquisition of the combined acoustic and electrical signals of the surge arrester includes: A voltage comparator is used to monitor the output signal of the ultra-high frequency sensor. When any signal exceeds a preset voltage threshold, a global synchronization trigger moment is generated. Based on the global synchronization trigger time, the data acquisition of all UHF sensors and piezoelectric acoustic sensors is started synchronously, and the broadband full waveform data within a preset time after the trigger point is recorded.

3. The method for locating internal defects of an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 2, characterized in that, The calculation of the pulse front broadening and quantification of the broadening into an electromagnetic wavefront distortion parameter characterizing the dielectric loss of the signal propagation path includes: Analyze the wideband full waveform data to determine the first and second preset percentage times when the pulse leading edge amplitude rises to the final stable peak value; Calculate the time difference between the second preset percentage time and the first preset percentage time, and use the time difference as an electromagnetic wavefront distortion parameter.

4. The method for locating internal defects in an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 1, characterized in that, The process of acquiring the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold includes: Based on the current external temperature, query the preset temperature-basic expansion mapping table to obtain the basic expansion amount; Based on the historical operating voltage, query the preset voltage-loss coefficient relationship curve to obtain the voltage correction coefficient; The base widening amount is corrected by using a voltage correction factor, and a preset statistical margin constant is added to obtain the dynamic widening threshold.

5. The method for locating internal defects in an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 4, characterized in that, The step of determining the internal dielectric state of the surge arrester based on the comparison results and selecting either the standard positioning processing mode or the adaptive anti-interference positioning processing mode includes: If the electromagnetic wavefront distortion parameters corresponding to all UHF sensors are less than the dynamic broadening threshold, the internal medium is determined to be uniform, and the standard positioning processing mode is selected. If the electromagnetic wavefront distortion parameter corresponding to any UHF sensor is greater than or equal to the dynamic broadening threshold, it is determined that there is an internal medium inhomogeneity anomaly, and the adaptive anti-interference positioning processing mode is selected.

6. The method for locating internal defects of an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 1, characterized in that, The calculation of the spatial variation rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations, and the generation of a spatial mask to mark suspicious anomaly regions in high-loss dielectric areas, includes: Obtain the axial position coordinates of each UHF sensor at different locations; The spatial distribution gradient is obtained by calculating the ratio of the difference in electromagnetic wavefront distortion parameters to the difference in position coordinates between two adjacent UHF sensors. The spatial distribution gradient is compared with a preset gradient threshold. Regions whose spatial distribution gradient exceeds the preset gradient threshold are marked as abnormal regions, and a binary spatial mask of the suspicious abnormal region is generated.

7. The method for locating internal defects of an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 1, characterized in that, The analysis of the spatial relationship between the acoustic signal propagation path and the spatial mask of the suspected anomaly area includes: Based on the three-dimensional coordinates of the piezoelectric acoustic sensor, construct the straight-line propagation path of the acoustic signal corresponding to each set of arrival time difference data; The straight-line propagation path of the acoustic signal is discretized into a set of spatial points, and the logical value corresponding to each spatial point in the spatial mask of the suspected anomaly region is queried. If there is a spatial point on the straight propagation path of the acoustic signal that falls into the abnormal area marked by the mask, it is determined that the straight propagation path of the acoustic signal and the mask have spatial overlap. Conversely, if there is no spatial point on the straight propagation path of the acoustic signal that falls into the abnormal area marked by the mask, it is determined that the straight propagation path of the acoustic signal and the mask do not have spatial overlap.

8. The method for locating internal defects of an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 7, characterized in that, The step of weighting and filtering the initial time difference of arrival dataset to generate a filtered, high-weight acoustic dataset includes: For acoustic signal straight-line propagation paths that are determined to have spatial overlap with the spatial mask of the suspicious anomaly area, their corresponding time difference of arrival data are assigned a preset low weight coefficient or are directly removed. For acoustic signal straight propagation paths that are determined not to have spatial overlap with the spatial mask of the suspicious anomaly area, the corresponding time difference of arrival data is assigned a preset high weighting coefficient. All retained data and their weighting coefficients are aggregated to form a filtered, high-weighted acoustic dataset.

9. The method for locating internal defects of an extreme cold surge arrester using combined acoustic and electro-optical imaging according to claim 1, characterized in that, The localization algorithm, which utilizes the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, outputs the three-dimensional coordinates of the internal defect source, including: When the selected positioning processing mode is the adaptive anti-interference positioning processing mode, a joint optimization objective function is constructed, and the variables to be solved are set as the three-dimensional coordinates of the internal defect source and the equivalent sound velocity in the area marked by the spatial mask of the suspected abnormal region. Using a filtered, high-weight acoustic dataset, an iterative algorithm is employed to minimize the weighted error between the measured time difference of arrival (TDOA) and the theoretical TDOA predicted based on the current variables. When the algorithm converges, it outputs the optimized three-dimensional coordinates of the internal defect source; When the selected positioning processing mode is the standard positioning processing mode, the initial time difference of arrival dataset is directly called. Using a pre-defined global uniform sound velocity model, the three-dimensional coordinates of the internal defect source are calculated and output by solving the time difference of arrival equations.

10. A system for locating internal defects in an extreme cold surge arrester using combined acoustic and electronic imaging, characterized in that, include: The signal acquisition and parameter extraction module synchronously acquires the acoustic and electrical signals of the surge arrester. For each electromagnetic pulse waveform acquired by the ultra-high frequency sensor, it calculates the broadening of the pulse leading edge and quantifies the broadening into an electromagnetic wavefront distortion parameter that characterizes the dielectric loss of the signal propagation path. The status determination and mode selection module acquires the current external temperature and historical operating voltage data of the surge arrester to calculate the dynamic broadening threshold, compares the electromagnetic wavefront distortion parameter with the dynamic broadening threshold, determines the internal medium status of the surge arrester based on the comparison result, and selects the standard positioning processing mode or the adaptive anti-interference positioning processing mode. The regional spatial mask generation module, when the adaptive anti-interference positioning processing mode is selected, calculates the spatial change rate of electromagnetic wavefront distortion parameters corresponding to ultra-high frequency sensors at different locations, and generates a spatial mask for suspicious abnormal regions that mark high-loss medium regions. The high-weighted acoustic data filtering module, when the adaptive anti-interference positioning processing mode is selected, calculates the initial time difference of arrival dataset between piezoelectric acoustic sensors, analyzes the spatial positional relationship between the acoustic signal propagation path and the spatial mask of the suspected abnormal area, performs weighted filtering on the initial time difference of arrival dataset, and generates a filtered high-weighted acoustic dataset. The localization processing mode solving module executes a localization solving algorithm based on the selected localization processing mode, using the initial time difference of arrival dataset and the filtered high-weight acoustic dataset, and outputs the three-dimensional coordinates of the internal defect source.

Images