A method and system for detecting partial discharge of a tower based on an acoustic imager

Abstract

The application relates to the technical field of power detection, in particular to a tower partial discharge detection method and system based on an acoustic imager. The method comprises the following steps: acquiring tower acoustic data collected by the acoustic imager; analyzing the tower acoustic data to determine an acoustic characteristic map; analyzing the acoustic characteristic map based on a preset discharge identification algorithm to determine a partial discharge type; matching the partial discharge type with a preset detection mode to obtain a matching relationship; and adjusting a detection frequency band of the acoustic imager according to the matching relationship. The method provides multi-dimensional discrimination basis for partial discharge type identification, matches the partial discharge type with the preset detection mode, improves the synchronization rate of the detection time of intermittent discharge, reduces the missed detection rate, helps to eliminate positioning errors in a high-humidity environment, reduces the equivalent propagation distance calculation error, and meets the high-precision positioning demand in a complex environment.

Description

Technical Field

[0001] This application relates to the field of power detection technology, and in particular to a method and system for detecting partial discharge on power poles based on an acoustic imager. Background Technology

[0002] With the advancement of smart grid construction, pole partial discharge detection technology is gradually transforming from traditional contact-based detection to non-contact intelligent detection. Currently, the mainstream detection method primarily employs drones equipped with acoustic imagers for aerial inspection, utilizing microphone arrays to collect discharge acoustic signals, and employing sound source localization technology to achieve spatial identification of the discharge point.

[0003] However, in the current power grid system, the accurate detection of partial discharge phenomena on poles suffers from a high misjudgment rate of atypical discharge modes. Therefore, how to identify atypical discharge conditions has become a significant shortcoming of existing technologies. Summary of the Invention

[0004] This application provides a method and system for detecting partial discharge on power poles based on an acoustic imager, in order to solve the above-mentioned problems.

[0005] In a first aspect, this application provides a method for detecting partial discharge in power towers based on an acoustic imager, the method comprising:

[0006] Acquire the acoustic data of the tower collected by the acoustic imager; analyze the acoustic data of the tower to determine the acoustic feature spectrum;

[0007] Based on a preset discharge identification algorithm, the acoustic feature spectrum is analyzed to determine the type of partial discharge;

[0008] The partial discharge type is matched with a preset detection mode to obtain a matching relationship; the detection frequency band of the acoustic imager is adjusted according to the matching relationship.

[0009] The acoustic imager is controlled to operate according to the adjusted detection frequency band and acquires compensated acoustic data in real time. Based on the preset acoustic wave attenuation compensation model, the compensated acoustic data is analyzed to determine the location coordinates of partial discharge on the tower.

[0010] This solution acquires acoustic data from the tower using an acoustic imager, helping to eliminate spatiotemporal misalignment caused by changes in flight attitude and improving the accuracy of sound source localization. Analyzing the tower acoustic data and determining the acoustic feature spectrum provides a multi-dimensional basis for identifying partial discharge types, reducing the risk of misjudgment in overlapping feature areas. Based on a preset discharge identification algorithm, analyzing the acoustic feature spectrum and determining the partial discharge type helps reduce the misjudgment rate of atypical discharges and avoids misclassification due to a single feature. Matching the partial discharge type with a preset detection mode to obtain a matching relationship helps improve the synchronization rate of intermittent discharge detection time and reduce the missed detection rate. Adjusting the detection frequency band of the acoustic imager based on the matching relationship helps eliminate random interference caused by wind speed changes and improves the data signal-to-noise ratio. Controlling the acoustic imager to operate according to the adjusted detection frequency band and acquiring compensated acoustic data in real time, based on a preset sound wave attenuation compensation model, analyzes the compensated acoustic data to determine the location coordinates of the partial discharge on the tower, helping to eliminate positioning errors in high humidity environments, reduce equivalent propagation distance calculation errors, and meet the high-precision positioning requirements in complex environments.

[0011] Optionally, adjusting the detection frequency band of the acoustic imager according to the matching relationship includes:

[0012] The detection mode is determined based on the matching relationship;

[0013] Based on the detection mode, the center frequency and bandwidth of the acoustic imager are determined;

[0014] Obtain the initial detection frequency band; determine the current filtering parameters based on the center frequency, the bandwidth, and the initial detection frequency band.

[0015] Optionally, the step of analyzing the acoustic feature spectrum based on a preset discharge identification algorithm to determine the type of partial discharge includes:

[0016] Based on a preset discharge identification algorithm, the acoustic feature spectrum is analyzed to determine the spectral characteristics;

[0017] Analyze the spectral characteristics to determine the dominant frequency components, harmonic distribution, and energy concentration frequency bands;

[0018] The type of partial discharge is determined based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band.

[0019] Optionally, the establishment of the preset detection mode includes:

[0020] Acquire several known detection data; analyze the known detection data to determine the detection characteristics and discharge type;

[0021] Using data statistical algorithms, the detection features are clustered to obtain a set of detection features;

[0022] Retrieve the detection parameters from the known detection data;

[0023] Based on the discharge type, the detection feature set is analyzed to determine the effectiveness of the detection parameters;

[0024] Based on the validity, a preset detection mode is obtained; the preset detection mode includes the discharge type and the detection parameters.

[0025] Optionally, determining the partial discharge type based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band includes:

[0026] Extract the peak frequency of the main frequency component;

[0027] Analyze the harmonic distribution to determine the number of harmonics and the proportion of harmonic energy;

[0028] Calculate the energy concentration based on the energy concentration frequency band;

[0029] The peak frequency, the number of harmonics, the energy concentration, and the proportion of harmonic energy are input into a preset discharge type decision model, and the partial discharge type is output.

[0030] Optionally, the step of analyzing the compensated acoustic data based on a preset acoustic attenuation compensation model to determine the location coordinates of the partial discharge on the tower includes:

[0031] Obtain the current flight environment of the device carrying the acoustic imager;

[0032] Analyze the flight environment to determine the current flight altitude, ambient temperature and humidity, and terrain distribution;

[0033] Based on the terrain distribution, determine the terrain shading coefficient;

[0034] Based on the preset sound wave propagation model, the compensated acoustic data is analyzed to determine the theoretical sound pressure value;

[0035] Analyze the compensated acoustic data to determine the measured sound pressure value;

[0036] Based on a preset sound wave attenuation compensation model, the theoretical sound pressure value and the measured sound pressure value are analyzed to determine the location coordinates of partial discharge on the tower.

[0037] Optionally, the partial discharge type includes intermittent discharge; before controlling the acoustic imager to operate according to the adjusted detection frequency band, the method further includes:

[0038] When an intermittent discharge pattern is detected, the multi-cycle frequency band scanning mode of the acoustic imager is activated, and a periodic travel signal is sent to the carrier device.

[0039] Acquire periodic frequency band scanning data from the acoustic imager; analyze the periodic frequency band scanning data to determine the energy variation trend of characteristic frequency bands within adjacent periods;

[0040] The optimal detection time window is determined based on the energy change trend of characteristic frequency bands within adjacent periods.

[0041] Optionally, the step of analyzing the compensated acoustic data and determining the theoretical sound pressure value based on a preset sound wave propagation model includes:

[0042] Analyze the flight environment to determine the current geographical conditions for detection;

[0043] Analyze the ambient temperature and humidity and the current geographical conditions to determine the weather status;

[0044] Analyze the weather conditions to determine the impact of weather noise;

[0045] Analyze the compensated acoustic data to determine the specific frequency bands where the weather noise is present;

[0046] Based on a preset dynamic filter, weather noise in the specific frequency band is eliminated to obtain actual compensated acoustic data;

[0047] Based on the sound wave propagation model, the actual compensated acoustic data is analyzed to determine the theoretical sound pressure value.

[0048] Optionally, analyzing the weather conditions to determine the impact of weather noise includes:

[0049] Analyze the weather conditions to determine the weather noise;

[0050] Analyze the weather noise to determine its time-domain impact characteristics and frequency-domain resonance properties;

[0051] The impact of weather noise is determined based on the time-domain impact characteristics and the frequency-domain resonance characteristics.

[0052] Secondly, this application provides a partial discharge detection system for power towers based on an acoustic imager, the system comprising:

[0053] The data analysis module is used to acquire the acoustic data of the tower collected by the acoustic imager; analyze the acoustic data of the tower, and determine the acoustic feature spectrum;

[0054] The spectrum analysis module is used to analyze the acoustic feature spectrum based on a preset discharge identification algorithm to determine the type of partial discharge.

[0055] The frequency band adjustment module is used to match the partial discharge type with a preset detection mode to obtain a matching relationship; and adjust the detection frequency band of the acoustic imager according to the matching relationship.

[0056] The coordinate determination module is used to control the acoustic imager to work according to the adjusted detection frequency band and acquire compensated acoustic data in real time. Based on the preset sound wave attenuation compensation model, the compensated acoustic data is analyzed to determine the position coordinates of the partial discharge of the tower.

[0057] Optionally, when the frequency band adjustment module adjusts the detection frequency band of the acoustic imager according to the matching relationship, it is used for:

[0058] The detection mode is determined based on the matching relationship;

[0059] Based on the detection mode, the center frequency and bandwidth of the acoustic imager are determined;

[0060] Obtain the initial detection frequency band; determine the current filtering parameters based on the center frequency, the bandwidth, and the initial detection frequency band.

[0061] Optionally, when the spectrum analysis module analyzes the acoustic feature spectrum based on a preset discharge identification algorithm to determine the type of partial discharge, it is used for:

[0062] Based on a preset discharge identification algorithm, the acoustic feature spectrum is analyzed to determine the spectral characteristics;

[0063] Analyze the spectral characteristics to determine the dominant frequency components, harmonic distribution, and energy concentration frequency bands;

[0064] The type of partial discharge is determined based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band.

[0065] Optionally, the tower partial discharge detection system based on acoustic imager further includes a mode determination module, used for:

[0066] Acquire several known detection data; analyze the known detection data to determine the detection characteristics and discharge type;

[0067] Using data statistical algorithms, the detection features are clustered to obtain a set of detection features;

[0068] Retrieve the detection parameters from the known detection data;

[0069] Based on the discharge type, the detection feature set is analyzed to determine the effectiveness of the detection parameters;

[0070] Based on the validity, a preset detection mode is obtained; the preset detection mode includes the discharge type and the detection parameters.

[0071] Optionally, when the spectrum analysis module determines the partial discharge type based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band, it is used for:

[0072] Extract the peak frequency of the main frequency component;

[0073] Analyze the harmonic distribution to determine the number of harmonics and the proportion of harmonic energy;

[0074] Calculate the energy concentration based on the energy concentration frequency band;

[0075] The peak frequency, the number of harmonics, the energy concentration, and the proportion of harmonic energy are input into a preset discharge type decision model, and the partial discharge type is output.

[0076] Optionally, when the coordinate determination module analyzes the compensated acoustic data based on a preset acoustic attenuation compensation model to determine the location coordinates of partial discharge on the tower, it is used for:

[0077] Obtain the current flight environment of the device carrying the acoustic imager;

[0078] Analyze the flight environment to determine the current flight altitude, ambient temperature and humidity, and terrain distribution;

[0079] Based on the terrain distribution, determine the terrain shading coefficient;

[0080] Based on the preset sound wave propagation model, the compensated acoustic data is analyzed to determine the theoretical sound pressure value;

[0081] Analyze the compensated acoustic data to determine the measured sound pressure value;

[0082] Based on a preset sound wave attenuation compensation model, the theoretical sound pressure value and the measured sound pressure value are analyzed to determine the location coordinates of partial discharge on the tower.

[0083] Optionally, the tower partial discharge detection system based on acoustic imager further includes a window determination module, used for:

[0084] When an intermittent discharge pattern is detected, the multi-cycle frequency band scanning mode of the acoustic imager is activated, and a periodic travel signal is sent to the carrier device.

[0085] Acquire periodic frequency band scanning data from the acoustic imager; analyze the periodic frequency band scanning data to determine the energy variation trend of characteristic frequency bands within adjacent periods;

[0086] The optimal detection time window is determined based on the energy change trend of characteristic frequency bands within adjacent periods.

[0087] Optionally, when the coordinate determination module analyzes the compensated acoustic data according to a preset sound wave propagation model to determine the theoretical sound pressure value, it is used for:

[0088] Analyze the flight environment to determine the current geographical conditions for detection;

[0089] Analyze the ambient temperature and humidity and the current geographical conditions to determine the weather status;

[0090] Analyze the weather conditions to determine the impact of weather noise;

[0091] Analyze the compensated acoustic data to determine the specific frequency bands where the weather noise is present;

[0092] Based on a preset dynamic filter, weather noise in the specific frequency band is eliminated to obtain actual compensated acoustic data;

[0093] Based on the sound wave propagation model, the actual compensated acoustic data is analyzed to determine the theoretical sound pressure value.

[0094] Optionally, when the coordinate determination module analyzes the weather conditions and determines the impact of weather noise, it is used for:

[0095] Analyze the weather conditions to determine the weather noise;

[0096] Analyze the weather noise to determine its time-domain impact characteristics and frequency-domain resonance properties;

[0097] The impact of weather noise is determined based on the time-domain impact characteristics and the frequency-domain resonance characteristics. Attached Figure Description

[0098] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0099] Figure 1 This is a schematic diagram illustrating an application scenario provided in one embodiment of this application;

[0100] Figure 2 A flowchart illustrating a method for detecting partial discharge in a tower based on an acoustic imager, provided as an embodiment of this application;

[0101] Figure 3 This is a schematic diagram of a partial discharge detection system for poles based on an acoustic imager, provided as an embodiment of this application. Detailed Implementation

[0102] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0103] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0104] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0105] In the current power grid system, the accurate detection of partial discharge phenomena on poles suffers from a high misjudgment rate due to atypical discharge modes. Therefore, how to identify atypical discharge conditions has become a significant shortcoming of existing technologies.

[0106] Based on this, this application provides a method and system for detecting partial discharge on power poles using an acoustic imager. The method involves acquiring acoustic data of the power pole collected by the acoustic imager; analyzing the acoustic data to determine the acoustic feature spectrum; analyzing the acoustic feature spectrum based on a preset discharge identification algorithm to determine the type of partial discharge; matching the partial discharge type with a preset detection mode to obtain a matching relationship; adjusting the detection frequency band of the acoustic imager according to the matching relationship; controlling the acoustic imager to operate according to the adjusted detection frequency band and acquiring compensated acoustic data in real time; and analyzing the compensated acoustic data based on a preset acoustic attenuation compensation model to determine the location coordinates of the partial discharge on the power pole. Acquiring the acoustic data of the power pole collected by the acoustic imager helps eliminate the spatiotemporal misalignment problem caused by changes in flight attitude, improving the accuracy of sound source localization. Analyzing the acoustic data of the power pole and determining the acoustic feature spectrum provides a multi-dimensional basis for partial discharge type identification, reducing the risk of misjudgment in overlapping feature areas. Analyzing the acoustic feature spectrum based on the preset discharge identification algorithm helps reduce the misjudgment rate of atypical discharges and avoids type misclassification caused by misjudgment of a single feature. Matching partial discharge types with preset detection modes to obtain a matching relationship helps improve the synchronization rate of intermittent discharge detection time and reduce the missed detection rate. Adjusting the detection frequency band of the acoustic imager based on the matching relationship helps eliminate random interference caused by wind speed changes and improves the data signal-to-noise ratio. Controlling the acoustic imager to operate according to the adjusted detection frequency band and acquiring compensated acoustic data in real time, and analyzing the compensated acoustic data based on a preset acoustic attenuation compensation model, helps determine the location coordinates of partial discharges on the tower. This helps eliminate positioning errors in high-humidity environments, reduce errors in equivalent propagation distance calculation, and meet the high-precision positioning requirements in complex environments.

[0107] Figure 1This illustration illustrates an application scenario of this application. When using a drone equipped with an acoustic imager to detect partial discharge on a tower, the method provided in this application is applied. Specifically, the method is applied to any server, where the server interacts with the acoustic imager to acquire the tower's acoustic data collected by the imager. This helps eliminate data spatiotemporal misalignment caused by changes in flight attitude, improving the accuracy of sound source localization. Analyzing the tower's acoustic data determines the acoustic feature spectrum, providing multi-dimensional discrimination criteria for partial discharge type identification and reducing the risk of misjudgment in overlapping feature areas. Based on a preset discharge identification algorithm, analyzing the acoustic feature spectrum to determine the partial discharge type helps reduce the misjudgment rate of atypical discharges and avoids misclassification due to single feature misjudgment. Matching the partial discharge type with a preset detection mode to obtain a matching relationship helps improve the synchronization rate of intermittent discharge detection time and reduce the missed detection rate. Adjusting the detection frequency band of the acoustic imager according to the matching relationship helps eliminate random interference caused by wind speed changes and improves the data signal-to-noise ratio. The acoustic imager is controlled to operate according to the adjusted detection frequency band, and the compensated acoustic data is acquired in real time through the acoustic imager. Based on the preset sound wave attenuation compensation model, the compensated acoustic data is analyzed to determine the location coordinates of partial discharge on the tower. This helps to eliminate positioning errors in high humidity environments, reduce the calculation error of equivalent propagation distance, and meet the high-precision positioning requirements in complex environments.

[0108] For specific implementation details, please refer to the following examples.

[0109] Figure 2 This is a flowchart illustrating a method for detecting partial discharge in power poles based on an acoustic imager, as provided in one embodiment of this application. The method of this embodiment can be applied to servers in the above-described scenario. Figure 2 As shown, the method includes:

[0110] S201. Acquire the acoustic data of the tower collected by the acoustic imager; analyze the acoustic data of the tower and determine the acoustic feature spectrum;

[0111] Acoustic imagers can be unmanned aerial vehicle (UAV)-borne inspection devices equipped with a 128-channel EMS microphone array.

[0112] The acoustic data of the tower can be the raw acoustic signal data acquired by the acoustic imager during the inspection process.

[0113] Acoustic feature maps can be feature representations generated by time-frequency analysis of tower acoustic data.

[0114] Specifically, during drone inspections, acoustic data of the towers is collected in real time using a 128-channel EMS microphone array. A dynamic frequency band division algorithm is constructed: First, the short-time Fourier transform spectrum is calculated based on a sliding time window to detect energy abrupt changes as potential discharge events; second, wavelet packet decomposition is performed on the detected event segments; then, the correlation between each sub-band and the discharge type is evaluated using mutual information entropy, and the sub-band with the highest discriminative power is automatically selected as the feature frequency band; finally, an acoustic feature map containing time-frequency energy distribution, harmonic structure, and pulse repetition rate is generated.

[0115] S202. Based on the preset discharge identification algorithm, analyze the acoustic feature spectrum to determine the type of partial discharge;

[0116] The preset discharge identification algorithm can be a pre-defined algorithm used to decouple mixed discharge characteristics and output discharge type determination results. It is pre-stored in the server and invoked when needed.

[0117] Partial discharge type can be a category of discharge modes caused by insulation defects in electrical equipment.

[0118] Specifically, a deep convolutional neural network, employing the convolutional feature extraction principle from image recognition and the statistical pattern recognition theory from time-frequency analysis, is used to classify acoustic feature maps: First, the network input layer receives the time-frequency domain feature matrix; second, the intermediate layer fuses the statistical features of the discharge pulse phase distribution map; then, the output layer sets up a mixed discharge type determination module; furthermore, when the confidence difference between corona discharge and surface discharge is small, secondary feature extraction is initiated: first, the zero-crossing rate of the rising edge of the discharge pulse and the high-frequency oscillation attenuation coefficient are analyzed, and pattern separation is performed using a support vector machine; finally, the partial discharge type containing probability weights is output.

[0119] S203. Match the partial discharge type with the preset detection mode to obtain the matching relationship; adjust the detection frequency band of the acoustic imager according to the matching relationship;

[0120] The preset detection mode can be a pre-defined set of detection parameters, including discharge type and detection parameters. It is pre-stored in the server and recalled when needed.

[0121] The matching relationship can be a correspondence between partial discharge type and preset detection mode.

[0122] The detection frequency band can be the frequency range in which the acoustic imager is currently operating.

[0123] Specifically, the optimal detection parameter matching relationship is selected from the preset detection mode library: first, a high-frequency narrowband mode for corona discharge; second, a wideband multi-stage mode for surface discharge; and third, a dual-band alternating sampling strategy for hybrid discharge mode. The local oscillator frequency of the acoustic imager is adjusted in real time by a digital down-converter, while the variable gain amplifier is controlled to improve the signal-to-noise ratio of the target frequency band. For the hybrid discharge mode, a multi-band parallel processing architecture is adopted, and the synchronous operation of independent digital filtering channels is realized through the multi-phase filtering parallel architecture in digital signal processing and the FPGA based on the hardware acceleration principle of programmable logic devices.

[0124] S204. Control the acoustic imager to work according to the adjusted detection frequency band and acquire compensated acoustic data in real time. Based on the preset sound wave attenuation compensation model, analyze the compensated acoustic data and determine the location coordinates of partial discharge on the tower.

[0125] The detection frequency band operation can be the state in which the acoustic imager performs acoustic signal acquisition and processing based on the adjusted center frequency and bandwidth parameters.

[0126] Compensated acoustic data can be acoustic data obtained after adjusting the detection frequency band.

[0127] The preset sound wave attenuation compensation model can be a pre-defined dynamic sound pressure attenuation calculation model. It is stored in the server and called upon when needed.

[0128] Partial discharge on power transmission towers can be a non-penetrating discharge phenomenon caused by insulation deterioration on or inside the power transmission tower.

[0129] The location coordinates can be the three-dimensional spatial coordinates of the discharge point.

[0130] Specifically, based on the atmospheric absorption attenuation model of the International Telecommunication Union (ITU), and combined with the geometric acoustic correction of terrain shading effect by ray tracing, a preset sound wave attenuation compensation model is established for precise positioning by real-time correction of sound velocity gradient by temperature and humidity parameters and Doppler frequency shift by wind speed parameters. First, temperature, humidity, and wind speed data from meteorological sensors are accessed in real time to establish a sound velocity correction model. Second, based on the UAV pose data and the three-dimensional coordinates of the tower, the ray tracing method is used to simulate the reflection or diffraction path of sound waves in complex terrain and calculate the equivalent propagation distance. Finally, an improved TDOA algorithm based on the nonlinear relationship between sound wave propagation delay and path in the wave equation is adopted. Combined with the time difference positioning principle in array signal processing, a propagation path compensation factor is introduced on the basis of traditional time delay estimation. The spatial coordinates of the discharge point are obtained by the Levenberg-Marquardt nonlinear optimization algorithm based on nonlinear least squares optimization theory, and the final position coordinates are output.

[0131] This solution acquires acoustic data from the tower using an acoustic imager, helping to eliminate spatiotemporal misalignment caused by changes in flight attitude and improving the accuracy of sound source localization. Analyzing the tower acoustic data and determining the acoustic feature spectrum provides a multi-dimensional basis for identifying partial discharge types, reducing the risk of misjudgment in overlapping feature areas. Based on a preset discharge identification algorithm, analyzing the acoustic feature spectrum and determining the partial discharge type helps reduce the misjudgment rate of atypical discharges and avoids misclassification due to a single feature. Matching the partial discharge type with a preset detection mode to obtain a matching relationship helps improve the synchronization rate of intermittent discharge detection time and reduce the missed detection rate. Adjusting the detection frequency band of the acoustic imager based on the matching relationship helps eliminate random interference caused by wind speed changes and improves the data signal-to-noise ratio. Controlling the acoustic imager to operate according to the adjusted detection frequency band and acquiring compensated acoustic data in real time, based on a preset sound wave attenuation compensation model, analyzes the compensated acoustic data to determine the location coordinates of the partial discharge on the tower, helping to eliminate positioning errors in high humidity environments, reduce equivalent propagation distance calculation errors, and meet the high-precision positioning requirements in complex environments.

[0132] In some embodiments, a detection mode is determined based on the matching relationship; the center frequency and bandwidth of the acoustic imager are determined based on the detection mode; an initial detection frequency band is obtained; and the current filtering parameters are determined based on the center frequency, bandwidth, and initial detection frequency band.

[0133] The detection mode can be a working strategy that is suitable for the current state after matching the partial discharge type with the preset detection mode.

[0134] The center frequency can be the core frequency point for signal acquisition and analysis by the acoustic imager in the current detection mode.

[0135] Bandwidth can be the effective detection frequency range based on the center frequency.

[0136] The initial detection frequency band can be an adjustable initial value calculated based on the acoustic imager hardware performance and environmental parameters.

[0137] The current filtering parameters can be a set of combined parameters used to control the signal extraction and noise suppression characteristics of the target frequency band in real time.

[0138] Specifically, based on the real-time detected discharge pulse phase distribution map and acoustic wave spectrum characteristics, the discharge type identifier code and environmental parameter vector are input into the preset detection mode. Mode matching degree calculation is performed, the detection mode is output, and the corresponding parameter configuration unit is activated. Then, the center frequency reference value under the target detection mode is read from the parameter configuration unit. Next, frequency band alignment is performed based on the initial detection frequency band, and the center frequency is dynamically corrected using a frequency band offset compensation algorithm to ensure that the target frequency band covers the main energy region of the discharge signal. Subsequently, the wideband fast scanning module of the acoustic imager is activated, acquiring full-band acoustic wave signals through a 128-channel EMS microphone array. Then, a fast Fourier transform of the original signal is performed using synchronous sampling for spectral analysis, extracting the frequency band intervals where the signal energy exceeds the noise baseline, which are defined as the initial detection frequency band. Finally, a parameter fusion algorithm based on a dynamic adjustment mechanism for filter parameters using environmental feedback is employed to weightedly fuse the target center frequency, bandwidth, and initial detection frequency band to generate the current filter parameters.

[0139] This scheme determines the detection mode based on the matching relationship, which helps eliminate the false positive rate of atypical discharges. Determining the center frequency and bandwidth of the acoustic imager based on the detection mode helps improve the effective signal energy capture rate while suppressing raindrop collision noise interference in non-target frequency bands. Obtaining the initial detection frequency band helps eliminate the frequency blind zone problem caused by time window misalignment in intermittent discharges, ensuring instantaneous capture of characteristic frequency bands during active discharge periods and reducing the missed detection rate in areas with changing wind speeds. Determining the current filtering parameters based on the center frequency, bandwidth, and initial detection frequency band helps achieve adaptive contraction and expansion of frequency band selection, avoiding signal fragmentation caused by sound wave reflections in complex terrain.

[0140] In some embodiments, based on a preset discharge identification algorithm, the acoustic feature spectrum is analyzed to determine the spectral features; the spectral features are analyzed to determine the dominant frequency component, harmonic distribution, and energy concentration frequency band; and the partial discharge type is determined based on the dominant frequency component, harmonic distribution, and energy concentration frequency band.

[0141] Spectral characteristics can be the distribution characteristics of a sound wave signal after frequency domain conversion.

[0142] The dominant frequency component can be the set of significant frequency points in the acoustic feature spectrum whose energy percentage exceeds the total energy.

[0143] Harmonic distribution can be defined as the energy amplitude of integer multiples of the dominant frequency component and its relative proportion to the fundamental frequency.

[0144] The energy concentration band can be the largest coverage area in the acoustic feature spectrum where the energy of consecutive frequency points is significantly higher than the environmental noise baseline.

[0145] Specifically, the time-domain acoustic signals acquired by the 128-channel EMS microphone array are input into a preset discharge identification algorithm. A frequency-domain energy distribution map is generated based on Fast Fourier Transform (FFT), and simultaneously, Short-Time Fourier Transform (SFT) is used to extract joint time-frequency features, forming a spectral feature containing a three-dimensional mapping of frequency, time, and energy. Then, based on the spectral feature, an energy threshold screening and peak detection algorithm is established using a dynamic energy baseline based on historical noise samples. This algorithm, combined with gradient change and continuity criteria, improves noise immunity and identifies a set of frequency points with high energy proportions, defined as the dominant frequency component. Based on the dominant frequency component, the energy attenuation slope of integer multiples of the dominant frequency component is identified through the fundamental frequency locking mechanism of the harmonic tracking algorithm, determining the harmonic distribution. Finally, the cumulative energy distribution across the entire frequency band is statistically analyzed, and continuous frequency bands with excessively high energy density are defined as energy concentration bands. The dominant frequency component, harmonic components, and energy concentration frequency band are used to form a feature vector. Typical discharge types are simulated in the laboratory, and acoustic spectrum characteristics are collected. Statistical feature analysis is performed in combination with field measurement data to construct a preset detection mode based on multi-dimensional features of dominant frequency, harmonics, and energy distribution. The cosine similarity algorithm is used to calculate the matching degree between the feature vector and the preset detection mode. Combined with the discharge feature dimension weight optimization, the mode with high similarity is selected as the candidate type. For intermittent discharge signals, the pattern matching results are statistically analyzed across scanning cycles, and the final partial discharge type is determined by a majority voting mechanism.

[0146] This scheme, based on a pre-defined discharge identification algorithm, analyzes acoustic feature maps and determines spectral characteristics, which helps improve the signal-to-noise ratio of the original signal. Analyzing spectral characteristics to determine the dominant frequency component, harmonic distribution, and energy concentration band helps improve the separability of different discharge types within the spectral overlap area, avoiding feature omissions caused by static frequency band division. Based on the dominant frequency component, harmonic distribution, and energy concentration band, the scheme determines the type of partial discharge, effectively distinguishing the superposition characteristics of corona and surface discharge in mixed discharge modes, and eliminating the problem of missed detection of energy-dispersive discharges caused by fixed frequency band analysis.

[0147] In some embodiments, several known detection data are acquired; the known detection data are analyzed to determine detection features and discharge types; the detection features are clustered using a data statistical algorithm to obtain a set of detection features; the detection parameters of the known detection data are retrieved; based on the discharge type, the set of detection features is analyzed to determine the validity of the detection parameters; and a preset detection mode is obtained based on the validity.

[0148] The known detection data can be a sample set of partial discharge acoustic signals.

[0149] The detection feature can be the acoustic wave characteristics in the acoustic wave signal obtained through detection.

[0150] The discharge type can be a partial discharge mode as reflected in the known detection data.

[0151] Data statistical algorithms can be mathematical methods used for feature analysis and pattern classification.

[0152] The detection parameters can be the core parameters that need to be configured when performing discharge detection.

[0153] Effectiveness can be an evaluation index of the performance of detection parameters under discharge type and environmental conditions.

[0154] Specifically, acoustic signal samples containing three discharge types—corona discharge, surface discharge, and internal discharge—as well as mixed discharge modes were obtained through laboratory high-voltage discharge simulation platform and on-site tower measurements. Then, a fast Fourier transform was applied to the acoustic signal samples to generate a frequency domain energy distribution spectrum, and a short-time Fourier transform was simultaneously used to extract a three-dimensional frequency-time-energy mapping to determine the detection characteristics. Subsequently, bandpass filtering and environmental noise suppression were applied to the acoustic signal samples, and the discharge type was labeled. Then, a data statistical algorithm was used, with the center frequency of the dominant frequency component, harmonic components, and bandwidth of the energy concentration band as clustering dimensions. The profile coefficient was determined by the average distance from a sample to other samples in the same cluster and the minimum average distance from a sample to samples in other clusters. Furthermore, by statistically analyzing the profile coefficient distribution of laboratory samples, a profile coefficient threshold was set, dividing the feature space into three categories: corona discharge clusters, surface discharge clusters, and mixed discharge clusters. Finally, for intermittent discharge data, a time dimension analysis was added, the signal was segmented by time windows, and the frequency of occurrence of the dominant frequency component within each window was statistically analyzed. This, combined with hierarchical clustering, identified periodic discharge patterns. Then, three detection parameters are extracted from the historical database for each type of discharge: dynamic frequency range adjustment parameters, acoustic attenuation compensation parameters, and intermittent discharge detection parameters. Next, leave-one-out cross-validation is used to address the bias-variance tradeoff, dividing the detection feature set into training and test sets. The effectiveness of the parameters is verified by type recognition accuracy and positioning error. The effectiveness is then bound to the corresponding feature vector and discharge type to generate a standard template. Finally, based on cross-validation, the pattern matching accuracy is evaluated. Randomly partitioned samples are used as the training set to construct a template library and as the test set to verify the classification accuracy, thereby obtaining the preset detection mode.

[0155] This solution acquires several known detection data points, helping to eliminate misjudgments of atypical discharges due to insufficient data in a single frequency band. Analyzing the known detection data to determine detection characteristics and discharge types helps reduce the misjudgment rate of mixed discharge modes and eliminates feature-position association errors caused by positioning deviations in UAV inspections. Utilizing data statistical algorithms to cluster detection features effectively distinguishes the differences in the core feature distribution of corona, surface, and internal discharge samples, improving mode compatibility under non-standard operating conditions. Retrieving detection parameters from known detection data helps overcome the risk of parameter overfitting caused by noise interference in field measurements. Determining the effectiveness of detection parameters based on discharge type helps reduce the misjudgment rate of atypical discharges. Based on the effectiveness, a preset detection mode is obtained, helping to eliminate mode mismatch problems caused by preset fixed frequency bands.

[0156] In some embodiments, the peak frequency of the dominant frequency component is extracted; the harmonic distribution is analyzed to determine the number of harmonics and the proportion of harmonic energy; the energy concentration is calculated based on the energy concentration frequency band; the peak frequency, the number of harmonics, the energy concentration and the proportion of harmonic energy are input into a preset discharge type decision model, and the partial discharge type is output.

[0157] The peak frequency can be the frequency point with the highest energy value within the main frequency component.

[0158] The number of harmonics can be the total number of frequency components that are integer multiples of the peak frequency.

[0159] The proportion of harmonic energy can be the ratio of the sum of the second and third harmonic energies to the total energy of the signal across the entire frequency band.

[0160] Energy concentration can be the ratio of the bandwidth of the continuous frequency range covering the total energy to the peak frequency.

[0161] The preset discharge type decision model can be a pre-defined classification model for output corona discharge, surface discharge, internal discharge, or mixed discharge types. It is pre-stored in the server and invoked when needed.

[0162] Specifically, the significant frequency components of the energy value, i.e., the dominant frequency, are identified; the frequency corresponding to the maximum energy value is selected as the peak frequency. Using the peak frequency as a benchmark, the frequency points of the second and third harmonics are determined; the energy values ​​at the peak frequency, second harmonic, and third harmonic are extracted; thus, the proportion of each harmonic energy to the fundamental frequency energy is calculated. Centered on the peak frequency, the energy concentration band is defined by extending outwards to the boundary points where the accumulated energy reaches the total energy, i.e., the energy concentration degree. The peak frequency, harmonic quantity, energy concentration, and harmonic energy proportion are combined to form a four-dimensional feature vector. If the four-dimensional feature vector matches the preset discharge type decision model based on the multi-dimensional feature fusion theory in pattern recognition, which maps discharge types through a combination of statistical classification and hierarchical clustering, the corresponding type is directly output. If the matching degree is insufficient, hierarchical clustering analysis is triggered: First, the signal is segmented by time window, and the gradient change features of the time-frequency matrix are extracted. Then, the gradient change features of the time-frequency matrix are used as clustering input, and Ward's minimum variance method is used for secondary clustering to identify periodic discharge patterns. Finally, the initial matching and auxiliary clustering results are combined to output the final partial discharge type.

[0163] This scheme extracts the peak frequency of the dominant frequency component, which helps eliminate the aliasing problem caused by fixed frequency band division. Analyzing the harmonic distribution and determining the number and energy proportion of harmonics helps improve the detection rate of mixed discharges. Calculating the energy concentration based on the energy concentration band helps reflect the characteristics of localized concentrated discharges and characterize the concurrent characteristics of multiple discharge types. Inputting the peak frequency, number of harmonics, energy concentration, and harmonic energy proportion into a preset discharge type decision model and outputting the partial discharge type helps overcome the susceptibility of single fundamental frequency characteristics to environmental interference, while avoiding misjudging high-frequency noise as surface discharge, achieving effective separation of mixed discharges from single discharge types, and reducing the misjudgment rate.

[0164] In some embodiments, the flight environment of the acoustic imager's carrier device at the current moment is acquired; the flight environment is analyzed to determine the current flight altitude, ambient temperature and humidity, and terrain distribution; the terrain shielding coefficient is determined based on the terrain distribution; the theoretical sound pressure value is determined by analyzing the compensated acoustic data according to a preset sound wave propagation model; the measured sound pressure value is determined by analyzing the compensated acoustic data; and the location coordinates of the partial discharge on the tower are determined by analyzing the theoretical sound pressure value, the measured sound pressure value, and the terrain shielding coefficient based on a preset sound wave attenuation compensation model.

[0165] The supporting device can be a drone equipped with an acoustic imager.

[0166] The flight environment can be a set of environmental parameters of the space where the drone is currently located.

[0167] The current flight altitude can be the vertical altitude measured in real time by the drone.

[0168] Ambient temperature and humidity can refer to the air temperature and relative humidity of the environment in which the drone is located.

[0169] Terrain distribution can be the classification result of geographical features within the area detected by the UAV.

[0170] The terrain shielding coefficient can be a correction parameter for quantifying the shielding effect of terrain on the sound wave propagation path.

[0171] The preset sound wave propagation model can be a pre-defined mathematical model describing the attenuation of sound waves as they propagate through the air. It is stored in the server and invoked when needed.

[0172] The theoretical sound pressure level can be the expected sound pressure level.

[0173] The measured sound pressure value can be the sound pressure data that is actually measured by the acoustic imager and then filtered and calibrated.

[0174] Specifically, the current flight environment is obtained through the UAV's onboard navigation equipment. Then, based on this environment, the current flight altitude is obtained via RTK positioning, and ambient temperature and relative humidity are simultaneously collected by temperature and humidity sensors. Next, a geographic information database is accessed to analyze the terrain elevation data around the current coordinate point, generating a terrain distribution. Based on this terrain distribution data, a ray tracing algorithm built using geometric acoustics principles is employed to simulate the sound wave propagation path, statistically analyzing the number of obstacles that truncate the path and the occlusion angle. Based on acoustic occlusion effect theory, and combined with weighted corrections for occlusion angle and obstacle density, the terrain occlusion coefficient is calculated. Then, the sound source intensity and propagation time from the compensated acoustic data are input into a preset sound wave propagation model that uses a classical spherical sound wave diffusion attenuation model, superimposed with atmospheric absorption attenuation and terrain occlusion correction terms. Next, the current ambient sound speed is calculated according to the international standard sound speed formula. Finally, the theoretical sound pressure value is output by fusing the preset sound wave propagation model with the attenuation formula. Subsequently, the compensated acoustic data is processed by time-domain windowing to extract the peak sound pressure level of each microphone channel. The effective sound pressure level is calculated by energy integration using the Hanning window applied to the time-domain signal, and then converted to decibels to determine the measured sound pressure level. Next, the theoretical sound pressure level is compared with the average measured sound pressure level to calculate the sound pressure attenuation deviation. If the sound pressure attenuation deviation is low, the sound source localization result is directly used. If the sound pressure attenuation deviation is high, the sound pressure attenuation deviation is input into the location based on a preset sound wave attenuation compensation model to generate a distance correction. The location coordinates are obtained by superimposing the distance correction based on the microphone array's time-of-arrival localization result.

[0175] This solution obtains the current flight environment of the acoustic imager's carrier equipment, helping to eliminate the shortcomings caused by neglecting the impact of temperature and humidity changes on sound wave propagation speed. Analyzing the flight environment to determine the current flight altitude, ambient temperature and humidity, and terrain distribution helps avoid sound source localization errors caused by altitude measurement errors. Determining the terrain shielding coefficient based on terrain distribution helps eliminate sound pressure attenuation errors caused by ignoring terrain reflection and shielding, significantly improving the localization robustness in complex terrain. Analyzing compensated acoustic data based on a preset sound wave propagation model to determine the theoretical sound pressure value helps overcome the limitations of the attenuation coefficient and eliminate the distortion of theoretical values ​​caused by sudden changes in sound speed in rainy or foggy weather. Analyzing compensated acoustic data to determine the measured sound pressure value avoids sound pressure measurement fluctuations caused by noise pollution. Based on a preset sound wave attenuation compensation model, analyzing theoretical and measured sound pressure values ​​determines the location coordinates of partial discharge on the tower, effectively suppressing abnormal channel data interference caused by multiple reflections or path obstruction, reducing localization errors in complex terrain, and solving the problem of rigid localization weight allocation.

[0176] In some embodiments, when an intermittent discharge mode is detected, the multi-cycle frequency band scanning mode of the acoustic imager is activated, and a periodic travel signal is sent to the carrier device; the periodic frequency band scanning data of the acoustic imager is acquired; the periodic frequency band scanning data is analyzed to determine the energy change trend of the characteristic frequency band within adjacent cycles; and the optimal detection time window is determined based on the energy change trend of the characteristic frequency band within adjacent cycles.

[0177] Intermittent discharge modes can be discontinuous, periodic, or quasi-periodic discharge phenomena.

[0178] Multi-cycle frequency band scanning mode can be a scanning mode that is dynamically configured according to the period of intermittent discharge pulses.

[0179] Periodic travel signals can be uniform displacement control commands sent to the carrying equipment.

[0180] Periodic frequency band scanning data can be time-frequency matrix data generated in a multi-period frequency band scanning mode.

[0181] Adjacent periods can be two consecutive scanning periods in a multi-period frequency band scan in the time series.

[0182] The characteristic frequency band can be the frequency band region within the energy peak frequency point in the periodic frequency band scan data.

[0183] The trend of energy change can be a quantitative fluctuation pattern of the energy ratio of characteristic frequency bands within adjacent periods.

[0184] The optimal detection time window can be a continuous period of time selected based on the energy stability trend.

[0185] Specifically, when an intermittent discharge spectrum pattern is detected, a pattern recognition flag is triggered. Then, based on the pulse period of the intermittent discharge, the multi-cycle frequency band scanning mode of the acoustic imager is dynamically configured: using the discharge center frequency as a reference, the extended bandwidth as the scanning range, and setting the number of consecutive scanning cycles. Subsequently, a periodic travel signal is sent to the carrying device. Finally, within each scanning cycle, full-band sound pressure data is collected and stored as a time-frequency matrix; and the start timestamp of each cycle and the spatial coordinates of the carrying device are marked to generate spatiotemporally aligned periodic frequency band scanning data of the acoustic imager. Then, based on the periodic frequency band scanning data, the energy ratio of the same frequency band in adjacent cycles is calculated. If the energy ratio of the same frequency band is too high and the peak frequency offset of adjacent cycles is low, it is determined to be an energy increasing trend; if the energy ratio of the same frequency band is too low and the peak frequency standard deviation is high, it is determined to be an energy decay trend. Finally, when the energy change trend is increasing, the time interval in which the predicted energy reaches its maximum value in the next scanning cycle is set as the optimal detection time window, so as to use the optimal detection time window to acquire the tower acoustic data in the above embodiment.

[0186] This scheme activates the multi-cycle frequency band scanning mode of the acoustic imager when an intermittent discharge mode is detected, and sends periodic travel signals to the carrier equipment. This helps avoid incorrect detection frequency band settings due to mode misjudgment, while suppressing random noise interference and preventing inaccurate acoustic signal acquisition caused by fluctuations in equipment movement speed or path obstruction, thus improving the spatial alignment accuracy of multi-cycle data. Acquiring the periodic frequency band scanning data of the acoustic imager provides a high-precision, traceable raw data foundation for analyzing energy change trends between adjacent cycles. Analyzing the periodic frequency band scanning data and determining the energy change trends of characteristic frequency bands within adjacent cycles helps eliminate interference from non-target signals on the selection of the detection window. Based on the energy change trends of characteristic frequency bands within adjacent cycles, the optimal detection time window is determined, providing a high-confidence time reference for adjusting the detection frequency band of the acoustic imager.

[0187] In some embodiments, the flight environment is analyzed to determine the current detection geographical conditions; the ambient temperature and humidity and the current detection geographical conditions are analyzed to determine the weather conditions; the weather conditions are analyzed to determine the impact of weather noise; the compensated acoustic data is analyzed to determine the specific frequency bands where weather noise has an impact; the weather noise in the specific frequency bands is eliminated according to a preset dynamic filter to obtain actual compensated acoustic data; and the actual compensated acoustic data is analyzed according to the sound wave propagation model to determine the theoretical sound pressure value.

[0188] The current geographical conditions can be a set of terrain feature parameters of the current location of the UAV.

[0189] Weather conditions can be discrete labels that characterize the acoustic propagation characteristics of the current air medium.

[0190] Weather noise can cause abnormal acoustic signal energy in specific frequency bands.

[0191] A specific frequency band can be the range of frequencies in acoustic data that are affected by weather noise pollution.

[0192] The preset dynamic filter can be a set of frequency domain filtering parameters pre-defined and bound to weather status tags. It is stored in the server and invoked when needed.

[0193] Weather noise can be acoustic signal interference caused by meteorological conditions.

[0194] The actual compensated acoustic data can be the acoustic signal after weather noise has been eliminated by a dynamic filter, preserving the original signal characteristics of the non-noise frequency band.

[0195] A sound wave propagation model can be a physical model used to calculate theoretical sound pressure levels.

[0196] Specifically, the system calls upon a geographic information database and matches preset terrain elevation data with GPS coordinates to obtain the current detection geographic conditions of the UAV's flight environment. Real-time ambient temperature and relative humidity data are collected using a temperature and humidity sensor array. Combined with the current detection geographic conditions, an international standard organization (ISO) airborne sound attenuation model is used to calculate the airborne sound attenuation factor. Based on the relationship matrix between the airborne sound attenuation factor and geographic conditions, the weather status is output. Based on a measured database of meteorological acoustic characteristics, a preset noise spectrum library is generated through statistical classification. Weather noise impact is generated by matching the preset noise spectrum library with the weather status. An FFT transformation is performed on the compensated acoustic data to convert the time-domain acoustic signal into a frequency band energy distribution spectrum. Then, a noise threshold is set based on the average frequency band energy during weather-free periods in historical data and a preset critical value for sound source detection requirements. Frequency bands whose energy exceeds the noise threshold in the weather noise impact are identified as specific frequency bands. Based on the specific frequency band range, a preset dynamic filter with linear phase is applied to perform amplitude suppression on the specific frequency bands in the compensated acoustic data, generating actual compensated acoustic data. The actual acoustic compensation data is input into the sound wave propagation model, and the theoretical sound pressure value is calculated by combining the propagation distance and the sound attenuation factor of the air medium.

[0197] This solution analyzes the flight environment and determines the current geographical conditions for detection, helping to eliminate multipath interference caused by terrain reflection and ensuring that the physical environment basis for acoustic data analysis is consistent with the actual propagation conditions. Analyzing the ambient temperature and humidity and the current geographical conditions for detection determines the weather status, providing physical environment parameter support for identifying the impact of weather noise. Analyzing the weather status determines the impact of weather noise, avoiding effective signal loss due to global filtering. Analyzing the compensated acoustic data identifies specific frequency bands affected by weather noise, enabling noise localization and energy amplitude quantification, providing frequency band boundary conditions for dynamic filter parameter selection. Based on a preset dynamic filter, weather noise in specific frequency bands is eliminated, obtaining actual compensated acoustic data, suppressing noise energy, and ensuring the frequency domain purity of the compensated data. Based on the sound wave propagation model, analyzing the actual compensated acoustic data determines the theoretical sound pressure level, which helps in comparative analysis of sound source localization or anomaly detection and verifies the effectiveness of data compensation.

[0198] In some embodiments, weather conditions are analyzed to determine weather noise; weather noise is analyzed to determine time-domain impact characteristics and frequency-domain resonance characteristics; and the impact of weather noise is determined based on the time-domain impact characteristics and frequency-domain resonance characteristics.

[0199] Temporal impulse characteristics can be the set of temporal parameters of transient impulse events in weather noise.

[0200] Frequency domain resonance characteristics can be defined as a set of specific frequency bands in which weather noise has significantly higher energy than background noise in the frequency domain.

[0201] Specifically, a preset weather state is invoked, and the current weather state type is determined by jointly matching temperature and humidity sensor data with the current geographical conditions, and the corresponding weather noise is indexed. Then, based on the weather state, a preset noise spectrum library is invoked to extract the noise frequency bands and time-domain impact characteristic parameters corresponding to the weather state; peak detection is performed on the time-domain waveform of the compensated acoustic data, and the impact pulse interval and amplitude are statistically analyzed. If the impact pulse interval matches the time-domain impact characteristic parameters, the existence of time-domain impact characteristics of weather noise is determined; simultaneously, an FFT transform is performed on the compensated acoustic data to generate a frequency-domain energy distribution spectrum, identify frequency bands with energy exceeding the noise threshold, and filter out frequency bands that intersect with the noise frequency bands, marking them as frequency-domain resonance characteristic bands. Finally, frequency bands that satisfy both time-domain impact characteristics and frequency-domain resonance characteristics are merged, and non-overlapping frequency bands are eliminated to generate the final weather noise impact.

[0202] This solution analyzes weather conditions, identifies weather noise, and achieves accurate classification of weather types, providing a basis for matching noise characteristics. Analyzing weather noise and determining its temporal impact characteristics and frequency resonance characteristics helps eliminate non-weather interference, reduce the false positive rate, and avoid interference from high-frequency or low-frequency environmental noise. Based on the temporal impact characteristics and frequency resonance characteristics, determining the impact of weather noise helps to precisely define the frequency range of noise to be suppressed, ensuring the targeted and effective noise elimination.

[0203] Figure 3 A schematic diagram of a partial discharge detection system for power poles based on an acoustic imager, provided in one embodiment of this application, is shown below. Figure 3 As shown, the tower partial discharge detection system 300 based on an acoustic imager in this embodiment includes: a data analysis module 301, a spectrum analysis module 302, a frequency band adjustment module 303, and a coordinate determination module 304.

[0204] Data analysis module 301 is used to acquire the acoustic data of the tower collected by the acoustic imager; analyze the acoustic data of the tower, and determine the acoustic feature spectrum;

[0205] The spectrum analysis module 302 is used to analyze the acoustic feature spectrum based on a preset discharge identification algorithm to determine the type of partial discharge.

[0206] The frequency band adjustment module 303 is used to match the partial discharge type with a preset detection mode to obtain a matching relationship; and adjust the detection frequency band of the acoustic imager according to the matching relationship.

[0207] The coordinate determination module 304 is used to control the acoustic imager to work according to the adjusted detection frequency band and acquire compensated acoustic data in real time. Based on the preset sound wave attenuation compensation model, the compensated acoustic data is analyzed to determine the position coordinates of the partial discharge of the tower.

[0208] Optionally, when the frequency band adjustment module 303 adjusts the detection frequency band of the acoustic imager according to the matching relationship, it is used for:

[0209] The detection mode is determined based on the matching relationship;

[0210] Based on the detection mode, the center frequency and bandwidth of the acoustic imager are determined;

[0211] Obtain the initial detection frequency band; determine the current filtering parameters based on the center frequency, the bandwidth, and the initial detection frequency band.

[0212] Optionally, when the spectrum analysis module 302 analyzes the acoustic feature spectrum based on a preset discharge identification algorithm to determine the partial discharge type, it is used for:

[0213] Based on a preset discharge identification algorithm, the acoustic feature spectrum is analyzed to determine the spectral characteristics;

[0214] Analyze the spectral characteristics to determine the dominant frequency components, harmonic distribution, and energy concentration frequency bands;

[0215] The type of partial discharge is determined based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band.

[0216] Optionally, the tower partial discharge detection system based on an acoustic imager further includes a mode determination module 305, used for:

[0217] Acquire several known detection data; analyze the known detection data to determine the detection characteristics and discharge type;

[0218] Using data statistical algorithms, the detection features are clustered to obtain a set of detection features;

[0219] Retrieve the detection parameters from the known detection data;

[0220] Based on the discharge type, the detection feature set is analyzed to determine the effectiveness of the detection parameters;

[0221] Based on the validity, a preset detection mode is obtained; the preset detection mode includes the discharge type and the detection parameters.

[0222] Optionally, when the spectrum analysis module 302 determines the partial discharge type based on the dominant frequency component, the harmonic distribution, and the energy concentration frequency band, it is used for:

[0223] Extract the peak frequency of the main frequency component;

[0224] Analyze the harmonic distribution to determine the number of harmonics and the proportion of harmonic energy;

[0225] Calculate the energy concentration based on the energy concentration frequency band;

[0226] The peak frequency, the number of harmonics, the energy concentration, and the proportion of harmonic energy are input into a preset discharge type decision model, and the partial discharge type is output.

[0227] Optionally, when the coordinate determination module 304 analyzes the compensated acoustic data based on a preset acoustic attenuation compensation model to determine the location coordinates of the partial discharge on the tower, it is used for:

[0228] Obtain the current flight environment of the device carrying the acoustic imager;

[0229] Analyze the flight environment to determine the current flight altitude, ambient temperature and humidity, and terrain distribution;

[0230] Based on the terrain distribution, determine the terrain shading coefficient;

[0231] Based on the preset sound wave propagation model, the compensated acoustic data is analyzed to determine the theoretical sound pressure value;

[0232] Analyze the compensated acoustic data to determine the measured sound pressure value;

[0233] Based on a preset sound wave attenuation compensation model, the theoretical sound pressure value, the measured sound pressure value, and the terrain shielding coefficient are analyzed to determine the location coordinates of partial discharge on the tower.

[0234] Optionally, the tower partial discharge detection system based on an acoustic imager further includes a window determination module 306, used for:

[0235] When an intermittent discharge pattern is detected, the multi-cycle frequency band scanning mode of the acoustic imager is activated, and a periodic travel signal is sent to the carrier device.

[0236] Acquire periodic frequency band scanning data from the acoustic imager; analyze the periodic frequency band scanning data to determine the energy variation trend of characteristic frequency bands within adjacent periods;

[0237] The optimal detection time window is determined based on the energy change trend of characteristic frequency bands within adjacent periods.

[0238] Optionally, when the coordinate determination module 304 analyzes the compensated acoustic data according to a preset sound wave propagation model to determine the theoretical sound pressure value, it is used for:

[0239] Analyze the flight environment to determine the current geographical conditions for detection;

[0240] Analyze the ambient temperature and humidity and the current geographical conditions to determine the weather status;

[0241] Analyze the weather conditions to determine the impact of weather noise;

[0242] Analyze the compensated acoustic data to determine the specific frequency bands where the weather noise is present;

[0243] Based on a preset dynamic filter, weather noise in the specific frequency band is eliminated to obtain actual compensated acoustic data;

[0244] Based on the sound wave propagation model, the actual compensated acoustic data is analyzed to determine the theoretical sound pressure value.

[0245] Optionally, when the coordinate determination module 304 analyzes the weather conditions and determines the impact of weather noise, it is used for:

[0246] Analyze the weather conditions to determine the weather noise;

[0247] Analyze the weather noise to determine its time-domain impact characteristics and frequency-domain resonance properties;

[0248] The impact of weather noise is determined based on the time-domain impact characteristics and the frequency-domain resonance characteristics.

[0249] The system in this embodiment can be used to execute the methods of any of the above embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

Claims

1. A method for detecting partial discharge of a tower based on an acoustic imager, characterized in that, The method comprises the following steps: acquiring tower acoustic data collected by the acoustic imager; analyzing the tower acoustic data to determine an acoustic feature map; based on a preset discharge identification algorithm, analyzing the acoustic feature map to determine a partial discharge type, which comprises: based on a preset discharge identification algorithm, analyzing the acoustic feature map to determine a frequency spectrum feature; analyzing the frequency spectrum feature to determine a main frequency component, a harmonic distribution, and an energy concentration frequency band; extracting a peak frequency of the main frequency component; analyzing the harmonic distribution to determine a harmonic number and a harmonic energy proportion; based on the energy concentration frequency band, calculating an energy concentration degree; inputting the peak frequency, the harmonic number, the energy concentration degree, and the harmonic energy proportion into a preset partial discharge type decision model to output a partial discharge type; matching the partial discharge type with a preset detection mode to obtain a matching relationship; and adjusting a detection frequency band of the acoustic imager according to the matching relationship. The establishment of the preset detection mode comprises the following steps: acquiring a plurality of known detection data; analyzing the known detection data to determine a detection feature and a discharge type; using a data statistical algorithm to cluster the detection feature to obtain a detection feature set; calling a detection parameter of the known detection data; based on the discharge type, analyzing the detection feature set to determine the effectiveness of the detection parameter; obtaining a preset detection mode according to the effectiveness; the preset detection mode comprises the discharge type and the detection parameter. controlling the acoustic imager to work according to the adjusted detection frequency band and to acquire compensation acoustic data in real time, analyzing the compensation acoustic data based on a preset sound wave attenuation compensation model to determine a position coordinate of a tower partial discharge, which comprises the following steps: acquiring a flight environment of a bearing device of the acoustic imager at a current time; analyzing the flight environment to determine a current flight height, an environmental temperature and humidity, and a terrain distribution; determining a terrain shielding coefficient according to the terrain distribution; analyzing the compensation acoustic data according to a preset sound wave propagation model to determine a theoretical sound pressure value; analyzing the compensation acoustic data to determine a measured sound pressure value; based on a preset sound wave attenuation compensation model, analyzing the theoretical sound pressure value, the measured sound pressure value, and the terrain shielding coefficient to determine a position coordinate of a tower partial discharge.

2. The method of claim 1, wherein, The adjustment of the detection frequency band of the acoustic imager according to the matching relationship comprises the following steps: determining a detection mode according to the matching relationship; determining a center frequency and a bandwidth of the acoustic imager according to the detection mode; acquiring an initial detection frequency band; and determining a current filtering parameter according to the center frequency, the bandwidth, and the initial detection frequency band.

3. The method of claim 1, wherein, The partial discharge type comprises intermittent discharge; before the control of the acoustic imager to work according to the adjusted detection frequency band, the method further comprises the following steps: when detecting an intermittent discharge mode, activating a multi-cycle frequency band scanning mode of the acoustic imager and sending a periodic marching signal to the bearing device; acquiring periodic frequency band scanning data of the acoustic imager; analyzing the periodic frequency band scanning data to determine an energy change trend of a characteristic frequency band in adjacent cycles; determining an optimal detection time window according to the energy change trend of the characteristic frequency band in adjacent cycles.

4. The method of claim 1, wherein, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

5. The method of claim 4, wherein, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps:

6. An acoustic imager based tower partial discharge detection system characterized by, The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following steps: The method comprises the following

Images