Multi-mode frequency mixing partial discharge fusion system

Abstract

The invention relates to the technical field of partial discharge monitoring, and discloses a multi-mode frequency mixing partial discharge fusion system, which comprises the following steps of: obtaining an event triggering time sequence by taking super current, voltage and electromagnetic waves as main triggering; determining relative time delay based on the electric signal channel waveform and the optical signal channel waveform, and performing compensation processing on the optical signal channel waveform according to the relative time delay to generate a re-aligned optical signal envelope; calculating an event phase according to the power frequency reference waveform; extracting an electric signal event amplitude and an optical signal event amplitude based on a preset phase window and the event phase, and performing normalization processing to obtain a standard electric signal amplitude and a standard optical signal amplitude; pairing the standard electric signal amplitude and the standard optical signal amplitude according to the event trigger sequence moment in the realignment optical signal envelope to form a cross-channel amplitude pair sequence; and calculating an inverse correlation principal axis slope, a correlation coefficient, a closed-loop directed area and normalized loop strength based on the cross-channel amplitude pair sequence, and outputting a partial discharge judgment result through robustness test.

Description

Technical Field

[0001] This invention relates to the field of partial discharge monitoring technology, and more specifically, to a multi-mode mixing partial discharge fusion system. Background Technology

[0002] In high-voltage transmission systems, cable joints are areas of concentrated electrical and mechanical stress. During long-term operation, due to construction defects, material aging, or thermal cycling, micro-gaps may form at the internal insulation interface of the joint. These micro-gaps are prone to partial discharge under high voltage, and long-term accumulation can lead to insulation degradation and even breakdown. To effectively diagnose defects in their early stages, the industry is increasingly adopting various sensing methods for online monitoring. Among these, electrical signal detection can capture nanosecond-level electromagnetic radiation pulses, while optical signal detection can capture mechanical vibrations or specific wavelengths of light radiation caused by discharge. Simultaneous use of dual modes can improve the accuracy of discharge identification and condition analysis.

[0003] Specifically, the instantaneous movement of charge during discharge generates a steep current pulse with a spectrum ranging from hundreds of MHz to several GHz. These high-frequency components can penetrate the connector sheath or gaps and be received by UHF sensors. However, UHF amplitude is susceptible to various factors:

[0004] 1. Frequency-selective attenuation formed by the internal conductor and shielding structure of the connector;

[0005] 2. The path length and structure from the source to the sensor cause the amplitude to decrease rapidly with distance;

[0006] 3. Inconsistent reception caused by differences in sensor deployment locations.

[0007] Therefore, under the same discharge energy, the measured amplitude of UHF may show significant differences.

[0008] Specifically, optical signal detection commonly takes two forms: one utilizes the sensitivity of optical fibers to discharge impulses or sound waves, converting them into phase or intensity changes in optical signals and demodulating them into waveforms. The other involves the excitation of molecular transitions by gas discharges, such as the radiation at 337 nm from the dual positive bandgap system of air / nitrogen discharges, which can be acquired using photodetectors. The amplitude of optical signals is affected by the following factors:

[0009] 1. The composition and pressure of the gas in the air gap (e.g., oxygen or water has a quenching effect on specific radiation);

[0010] 2. Discharge channel morphology and duration;

[0011] 3. Absorption and scattering characteristics of materials.

[0012] The slow changes in these factors result in a gradual drift in the optical signal.

[0013] Specifically, in practical dual-mode monitoring, it is usually desirable to determine the discharge strength by comparing the amplitude changes of the two channels. However, the following phenomena are commonly observed in actual measurements:

[0014] 1. During certain periods, UHF signals are strong while optical signals are weak;

[0015] 2. At other times, UHF signals are weak while optical signals are strong;

[0016] 3. Changes are often related to variations in the air gap environment, temperature and humidity, surface temperature, and discharge morphology.

[0017] In summary, the amplitude relationship is not fixed, and direct comparison of the amplitudes of the two channels lacks consistency. Existing fusion analysis commonly uses direct normalization or ratio calculation, but the two channels have large differences in response time scale and environmental sensitivity, resulting in technical problems that cannot eliminate the instability of the amplitude relationship. Summary of the Invention

[0018] This invention provides a multi-mode mixing partial discharge fusion system, which solves the technical problems mentioned in the background art.

[0019] This invention provides a multi-mode mixing partial discharge fusion system, comprising:

[0020] The data acquisition module is used to acquire power frequency reference waveforms, electrical signal channel waveforms, and optical signal channel waveforms, and obtains the event trigger time sequence based on the electrical signal channel as the main trigger.

[0021] The data compensation module is used to determine the relative time delay based on the waveforms of the electrical signal channel and the optical signal channel, and to perform compensation processing on the optical signal channel waveform according to the relative time delay to generate a realigned optical signal envelope;

[0022] The phase window construction module is used to calculate the event phase based on the power frequency reference waveform;

[0023] The data extraction module is used to extract the amplitude values ​​of electrical signal events and optical signal events based on a preset phase window and event phase, and perform normalization processing to obtain standard electrical signal amplitude values ​​and standard optical signal amplitude values.

[0024] The data pairing module is used to pair the standard electrical signal amplitude and the standard optical signal amplitude according to the event trigger sequence time in the realigned optical signal envelope to form a cross-channel amplitude pair sequence;

[0025] The fusion determination module is used to calculate the anticorrelation principal axis slope, correlation coefficient, closed-loop directional area and normalized loop strength based on cross-channel amplitude pair sequences, and output the partial discharge determination result through robustness test.

[0026] Furthermore, the electrical signal channel is the primary trigger to obtain the event trigger time sequence, including:

[0027] Generate an electrical signal envelope from the waveform of the electrical signal channel;

[0028] Within a preset noise statistical interval, the electrical signal threshold is calculated based on the mean and standard deviation of the electrical signal envelope; where the electrical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation.

[0029] When the envelope value of the electrical signal at time i is greater than or equal to the threshold value of the electrical signal, the time period from time i-1 to time i is marked as a discharge event, and time i is taken as the trigger time of the discharge event and time i-1 is taken as the starting time of the discharge event.

[0030] An event trigger sequence is formed based on all discharge events in chronological order.

[0031] Furthermore, the relative time delay is determined based on the waveforms of the electrical signal channel and the optical signal channel, and the optical signal channel waveform is compensated according to the relative time delay to generate a realigned optical signal envelope, including:

[0032] Generate an optical signal envelope from the optical signal channel waveform;

[0033] Within a preset noise statistics interval, the optical signal threshold is calculated based on the mean and standard deviation of the optical signal envelope; whereby the optical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation.

[0034] For the j-th discharge event, extract the corresponding electrical signal envelope segment and optical signal envelope segment, and calculate the corresponding discrete cross-correlation within the preset search delay range;

[0035] The time delay corresponding to the discrete cross-correlation is used as the time delay estimate, and the average of the time delay estimates at all discharge event times is used as the relative time delay.

[0036] The optical signal envelope is shifted relative to the time delay along the time axis to form an overaligned optical signal envelope.

[0037] Furthermore, the event phase is calculated based on the power frequency reference waveform, including:

[0038] A positive zero-crossing detection is performed on the power frequency reference waveform to obtain a positive zero-crossing time sequence;

[0039] The Hilbert transform is applied to the power frequency reference waveform to obtain a complex signal; the imaginary and real parts of the complex signal are extracted to calculate the instantaneous phase, and the instantaneous phase is expanded to obtain the continuous phase;

[0040] Take the last positive zero-crossing moment in the positive zero-crossing moment sequence as the zero point moment;

[0041] The negative value of the continuous phase corresponding to the zero point is calculated as the phase calibration value, and the sum of the phase calibration value and the continuous phase is used as the calibration phase function.

[0042] For each discharge event, the value of the calibration phase function at the trigger time is calculated, and the event phase corresponding to the discharge event is obtained by taking the modulus of 2π.

[0043] Furthermore, based on a preset phase window and event phase, the amplitude values ​​of electrical signal events and optical signal events are extracted and normalized to obtain standard electrical signal amplitude and standard optical signal amplitude, including:

[0044] A phase window is constructed using a preset center phase and a preset phase window half-width; wherein the phase window is centered on the preset center phase and extends to both sides by a preset phase window half-width.

[0045] If the phase of the event corresponding to the discharge event is within the phase window, the corresponding discharge event is marked as a candidate event to form a candidate event set;

[0046] For each candidate event in the candidate event set, the electrical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the first preset time window to obtain the corresponding electrical signal event amplitude.

[0047] For each candidate event in the candidate event set, the overlapping optical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the second preset time window to obtain the corresponding optical signal event amplitude.

[0048] The amplitude values ​​of the electrical signal and optical signal events for each candidate event are normalized to obtain the standard electrical signal amplitude and standard optical signal amplitude for each candidate event.

[0049] Furthermore, in the realigned optical signal envelope, the standard electrical signal amplitude and the standard optical signal amplitude are paired according to the event trigger sequence time to form a cross-channel amplitude pair sequence, including:

[0050] For each candidate event in the candidate event set, with the trigger time corresponding to the candidate event as the center of the second preset time window, search for the local maximum point of the realigned optical signal envelope to obtain the peak time of the optical signal at the local maximum point and the corresponding amplitude of the realigned optical signal envelope.

[0051] The time difference between the peak moment of the optical signal and the trigger moment corresponding to the candidate event is used as the pairing error.

[0052] If the pairing error is less than or equal to the preset alignment error threshold, and the amplitude of the realigned optical signal envelope is greater than or equal to the optical signal threshold, then the candidate event is marked as a valid pairing event.

[0053] For each valid pairing event, the corresponding standard electrical signal amplitude and standard optical signal amplitude are recorded as amplitude pairs, and the amplitude pairs are arranged in ascending order according to the corresponding trigger time to form a cross-channel amplitude pair sequence.

[0054] Furthermore, based on the cross-channel amplitude pair sequences, the slope of the anti-correlation principal axis, correlation coefficient, closed-loop directed area, and normalized loop strength are calculated, including:

[0055] The process of obtaining the closed-loop directed area is as follows:

[0056] By concatenating the first and last pairs of cross-channel amplitude pairs in chronological order as input variables for the shoelace formula, the closed-loop directed area is obtained.

[0057] The process of obtaining the normalized ring strength is as follows:

[0058] The square root of the variance of the standard electrical signal amplitude is taken as the standard deviation of the standard electrical signal amplitude;

[0059] The square root of the variance of the standard optical signal amplitude is the standard deviation of the standard optical signal amplitude;

[0060] The ratio of the absolute value of the directional area of ​​the closed loop to the product of pi and the standard deviation of the amplitude of the standard electrical signal and the standard deviation of the amplitude of the standard optical signal is used as the normalized loop strength.

[0061] Furthermore, the partial discharge determination result is output through robustness checks, including:

[0062] Robustness checks are as follows:

[0063] For a preset search delay range centered on relative delay, the delay is micro-scanned within the corresponding preset search delay range; after temporarily shifting the envelope of the realigned optical signal, the closed-loop directed area is recalculated; the absolute value of the closed-loop directed area is normalized to obtain the delay fluctuation rate.

[0064] Set a phase change value for a preset phase window half-width, and increase or decrease the preset phase window half-width by the phase change value to form a first variable phase window and a second variable phase window, respectively; recalculate the corresponding normalized ring strength based on the first variable phase window and the second variable phase window to obtain the first variable ring strength and the second variable ring strength; use the ratio of the standard deviation of the first variable ring strength, the second variable ring strength and the normalized ring strength to the corresponding average value as the phase window volatility.

[0065] If both the time delay volatility and the phase window volatility are less than or equal to the corresponding preset volatility thresholds, then the robustness test is satisfied, and it is determined that a partial discharge exists.

[0066] The beneficial effects of this invention are as follows: by performing time delay compensation, phase alignment, amplitude normalization, event pairing, and cross-channel feature extraction on electrical signal events and optical signal events, and combining robustness verification methods, it effectively avoids misjudgment or missed judgment caused by single-channel characteristic fluctuations in traditional methods, enhances adaptability and signal identification capabilities in complex noise environments, and realizes a high-precision synchronous pairing and cross-modal fusion identification mechanism between electrical signal channel waveforms and optical signal channel waveforms. Under the premise of ensuring time domain accuracy and phase consistency, it effectively improves the robustness and accuracy of partial discharge event determination. Attached Figure Description

[0067] Figure 1 This is a block diagram of a multi-mode mixing partial discharge fusion system according to the present invention. Detailed Implementation

[0068] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0069] like Figure 1 As shown, a multi-mode mixing partial discharge fusion system includes:

[0070] The data acquisition module is used to acquire power frequency reference waveforms, electrical signal channel waveforms, and optical signal channel waveforms, and obtains the event trigger time sequence based on the electrical signal channel as the main trigger.

[0071] The data compensation module is used to determine the relative time delay based on the waveforms of the electrical signal channel and the optical signal channel, and to perform compensation processing on the optical signal channel waveform according to the relative time delay to generate a realigned optical signal envelope;

[0072] The phase window construction module is used to calculate the event phase based on the power frequency reference waveform;

[0073] The data extraction module is used to extract the amplitude values ​​of electrical signal events and optical signal events based on a preset phase window and event phase, and perform normalization processing to obtain standard electrical signal amplitude values ​​and standard optical signal amplitude values.

[0074] The data pairing module is used to pair the standard electrical signal amplitude and the standard optical signal amplitude according to the event trigger sequence time in the realigned optical signal envelope to form a cross-channel amplitude pair sequence;

[0075] The fusion determination module is used to calculate the anticorrelation principal axis slope, correlation coefficient, closed-loop directional area and normalized loop strength based on cross-channel amplitude pair sequences, and output the partial discharge determination result through robustness test.

[0076] In detail, the power frequency reference waveform is a periodic electrical signal waveform reflecting the fundamental frequency of the power grid (such as 50Hz or 60Hz), serving as the phase reference for partial discharge events. The power frequency reference waveform is used to map the discharge moment to a specific phase position within the power frequency period (such as 0~2π), ensuring the comparability of phases for different discharge events. The power frequency reference waveform is typically a sinusoidal or near-sinusoidal real signal, containing clear periodic characteristics; its zero-crossing points, peak points, etc., can serve as key anchor points for phase calculation. The power frequency reference waveform is acquired from the power grid line, equipment casing, or a dedicated reference terminal using a voltage divider sensor.

[0077] In detail, the electrical signal channel waveform is the time-domain waveform of the high-frequency electromagnetic radiation signal generated by charge movement during partial discharge. It reflects the electromagnetic pulse characteristics at the moment of discharge, containing pulse signals with steep nanosecond-level leading edges, and its spectrum covers hundreds of MHz to several GHz (typically 300 MHz to 3 GHz). It can be directly used to trigger discharge event detection. The electrical signal channel waveform is obtained by receiving the high-frequency electromagnetic pulses generated by partial discharge through an electrical signal sensor.

[0078] In detail, the optical signal channel waveform is the time-domain waveform of the optical signal caused by partial discharge. It is used to reflect the optical signal response during the discharge process, such as the signal after demodulation of narrowband optical radiation (e.g., nitrogen radiation at 337nm) or broadband acoustic vibration through optical fiber. The signal duration is usually longer than the electrical signal pulse (in microseconds) and is greatly affected by environmental factors (e.g., gas composition, temperature). It is obtained by receiving the optical signal generated by the discharge through an optical signal sensor.

[0079] In one embodiment of the present invention, the electrical signal channel is used as the main trigger to obtain the event trigger time sequence, including:

[0080] Generate an electrical signal envelope from the waveform of the electrical signal channel;

[0081] Within a preset noise statistical interval, the electrical signal threshold is calculated based on the mean and standard deviation of the electrical signal envelope; where the electrical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation.

[0082] When the envelope value of the electrical signal at time i is greater than or equal to the threshold value of the electrical signal, the time period from time i-1 to time i is marked as a discharge event, and time i is taken as the trigger time of the discharge event and time i-1 is taken as the starting time of the discharge event.

[0083] An event trigger sequence is formed based on all discharge events in chronological order.

[0084] In detail, the raw waveform acquired by the electrical signal channel contains a large number of high-frequency oscillation components, which are easily affected by noise interference when directly used for triggering judgment. Generating the electrical signal envelope involves signal processing techniques, such as Hilbert transform to extract the amplitude of the analytic signal, or smoothing the raw waveform using filtering, peak tracking, and other algorithms, to obtain an envelope curve that reflects the trend of the electrical signal pulse amplitude change. This filters out high-frequency noise glitches and highlights the overall amplitude characteristics of the discharge pulse.

[0085] In detail, the preset noise statistics interval refers to a pre-defined time period during which no partial discharge occurs, such as a stable period when the equipment is operating normally and there is no discharge signal. Within the preset noise statistics interval, the value of the electrical signal envelope is mainly composed of background noise. By statistically analyzing the mean and standard deviation of the electrical signal envelope within the preset noise statistics interval, the level of background noise can be quantified: the mean reflects the average intensity of the noise, and the standard deviation reflects the degree of noise fluctuation. The electrical signal threshold is set as the weighted sum of the mean and standard deviation, where the weighting value is a preset weighting coefficient based on the noise characteristics. For example, the weighting coefficient can be adjusted according to the on-site noise level to balance trigger sensitivity and anti-interference capability. This allows it to adapt to changes in background noise; that is, when the noise increases, the electrical signal threshold increases accordingly to avoid false triggering; when the noise decreases, the weighting coefficient decreases to ensure the ability to capture weak discharge signals. The final electrical signal threshold is the critical value that distinguishes noise from valid discharge signals.

[0086] In detail, when the value of the electrical signal envelope at time i reaches or exceeds the electrical signal threshold, it indicates that a possible discharge pulse has been detected. The time period from time i-1 to time i is marked as a discharge event because time i is the moment when the envelope first meets the threshold condition, and the start of a discharge pulse is usually earlier than this moment. Time i-1 can be approximated as the starting point of the pulse, and this time period can completely cover the rising phase of a discharge pulse, ensuring the integrity of the event. At the same time, time i is defined as the trigger time of the discharge event, and time i-1 is defined as the starting time of the event, making the time boundary of each discharge event clear and traceable.

[0087] In detail, all discharge events that meet the condition of electrical signal envelope value ≥ electrical signal threshold are arranged sequentially according to their trigger times to form an event trigger sequence. The event trigger sequence serves as the time base for all subsequent cross-channel analyses, ensuring that all events are orderly correlated on a unified time axis.

[0088] In one embodiment of the present invention, the relative time delay is determined based on the waveforms of the electrical signal channel and the optical signal channel, and the optical signal channel waveform is compensated according to the relative time delay to generate a realigned optical signal envelope, including:

[0089] Generate an optical signal envelope from the optical signal channel waveform;

[0090] Within a preset noise statistics interval, the optical signal threshold is calculated based on the mean and standard deviation of the optical signal envelope; whereby the optical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation.

[0091] For the j-th discharge event, extract the corresponding electrical signal envelope segment and optical signal envelope segment, and calculate the corresponding discrete cross-correlation within the preset search delay range;

[0092] The time delay corresponding to the discrete cross-correlation is used as the time delay estimate, and the average of the time delay estimates at all discharge event times is used as the relative time delay.

[0093] The optical signal envelope is shifted relative to the time delay along the time axis to form an overaligned optical signal envelope.

[0094] The detailed formula for calculating discrete cross-correlation is as follows:

[0095]

[0096] Among them, R j (τ) represents the discrete cross-correlation of the j-th discharge event, [-τ S ,τ S ] represents the preset search delay range centered on the trigger time of the j-th discharge event, τ S u represents the time interval from time i-1 to time i. j (τ ' ) represents the envelope segment of the electrical signal at time τ ' The electrical signal envelope value, o j (τ'+τ) represents the optical signal envelope value at time τ'+τ corresponding to the optical signal envelope segment, where τ represents the relative time delay to be calculated.

[0097] In detail, the raw waveform acquired by the optical signal channel may contain high-frequency oscillations or multi-peak interference, making it susceptible to noise interference when used directly for time alignment. Generating the optical signal envelope involves signal processing techniques, such as Hilbert transform to extract the amplitude of the analytic signal, or smoothing the raw waveform using filtering, peak tracking, and other algorithms, to obtain an envelope curve that reflects the amplitude variation trend of the optical signal. This process removes high-frequency noise and highlights the signal characteristics corresponding to the discharge event.

[0098] It should be noted that the process of obtaining the threshold of optical signals is the same as that of obtaining electrical signals.

[0099] In detail, for the j-th discharge event, time difference analysis needs to be performed on the local signal corresponding to that event. Specifically, this involves extracting the local segment corresponding to the event from the electrical signal envelope, i.e., the segment within the electrical signal envelope corresponding to the time period of the j-th discharge event. Similarly, the corresponding local segment is extracted from the optical signal envelope.

[0100] In detail, the preset search delay range is a pre-defined interval based on statistical data, which may contain time differences, preferably ±500 nanoseconds, to ensure coverage of possible time offsets between the two channels. Specifically: the propagation speed of an electromagnetic pulse in air or cable is approximately 3 × 10⁻⁶. 8 While the propagation speed of optical signals (such as optical radiation) is close to the speed of light, the response time of optical signal sensors (such as optical fibers) can reach hundreds of nanoseconds; the propagation speed of acoustic vibrations in solids is approximately 3 × 10⁻⁶ m / s. 3 m / s (m / s, much slower than electromagnetic signals). The maximum possible time delay difference is calculated based on the maximum possible distance between the sensor and the discharge point (e.g., the sensor spacing within a cable joint is typically less than 1 meter). For example, a 1-meter distance corresponds to a propagation delay of approximately 3.3 ns for electromagnetic signals and approximately 333 ns for acoustic signals.

[0101] The discrete cross-correlation is calculated within the preset search delay range. The formula essentially measures the similarity between the electrical signal envelope segment and the optical signal envelope segment under different delays: for each possible delay τ, the optical signal envelope segment is shifted along the time axis by τ, and then multiplied point by point with the electrical signal envelope segment and summed. The larger the discrete cross-correlation result, the higher the matching degree of the two segments under that delay.

[0102] In detail, in the formula, R j (τ) represents the cross-correlation value of the j-th event at time delay τ, [-τ] S ,τ S ] indicates the preset search delay range (τ) S For a search range of half width (preferably 500 nanoseconds), u j (τ ' ) represents the envelope segment of an electrical signal in relative time τ ' The amplitude at o j (τ'+τ) represents the amplitude of the optical signal envelope segment at a relative time τ'+τ after being shifted by τ. This is achieved by calculating R corresponding to different τ values. j (τ) can be used to find the time delay that makes the two segments most similar, which is the time delay estimate of the event.

[0103] In detail, the time delay estimate of a single discharge event may be biased due to local noise; therefore, it is necessary to synthesize the time delay estimates of all discharge events. Taking the average of all event time delay estimates as the relative time delay can offset the random errors of individual events and improve the reliability of the overall time difference estimate. This relative time delay reflects the overall time offset of the optical signal channel relative to the electrical signal channel. The optical signal envelope is shifted along the time axis by this relative time delay, so that the time axis of the optical signal channel is completely aligned with that of the electrical signal channel. The resulting realigned optical signal envelope ensures that the corresponding discharge event time is consistent with the event trigger time of the electrical signal channel.

[0104] In one embodiment of the present invention, calculating the event phase based on a power frequency reference waveform includes:

[0105] A positive zero-crossing detection is performed on the power frequency reference waveform to obtain a positive zero-crossing time sequence;

[0106] The Hilbert transform is applied to the power frequency reference waveform to obtain a complex signal; the imaginary and real parts of the complex signal are extracted to calculate the instantaneous phase, and the instantaneous phase is expanded to obtain the continuous phase;

[0107] Take the last positive zero-crossing moment in the positive zero-crossing moment sequence as the zero point moment;

[0108] The negative value of the continuous phase corresponding to the zero point is calculated as the phase calibration value, and the sum of the phase calibration value and the continuous phase is used as the calibration phase function.

[0109] For each discharge event, the value of the calibration phase function at the trigger time is calculated, and the event phase corresponding to the discharge event is obtained by taking the modulus of 2π.

[0110] In detail, the power frequency reference waveform changes periodically, and a positive zero-crossing refers to the instant when the waveform crosses zero as it transitions from a negative value to a positive value. Positive zero-crossing detection of the power frequency reference waveform involves using signal processing algorithms to identify all such moments, recording their occurrence times, and arranging them chronologically to form a positive zero-crossing moment sequence.

[0111] In detail, the power frequency reference waveform is a real signal, and direct phase calculation is easily affected by waveform distortion. Performing a Hilbert transform on it generates an orthogonal imaginary part signal, which, together with the original real part signal, constitutes a complex signal. The complex signal contains both amplitude and phase information. By extracting the imaginary and real parts of the complex signal and using the arctangent function, the instantaneous phase can be calculated. The instantaneous phase reflects the phase state of the power frequency waveform at each moment in real time, but due to mathematical limitations, its value range is constrained to [-π, π], and it will exhibit jumps with periodic changes. Expanding the instantaneous phase eliminates these jumps through a phase expansion algorithm. For example, when the phase abruptly changes from π to -π, 2π is automatically added to maintain continuity, resulting in a continuous phase that grows linearly with time. The continuous phase fully reflects the total phase change of the power frequency waveform from the initial moment to the current moment.

[0112] In detail, in the positive zero-crossing time sequence, the last positive zero-crossing time is the phase reference point closest to the current analysis period. This time is selected as the zero point because it is closest to the discharge event to be analyzed, which can minimize the impact of power frequency fluctuations or waveform drift on the phase reference and ensure the timeliness and accuracy of phase calibration.

[0113] In detail, the continuous phase corresponding to zero point is the cumulative total phase value at that moment. Because the continuous phase grows linearly with time, it is not necessarily zero. The negative value of the continuous phase is calculated as a phase calibration value, thereby forcibly correcting the phase at zero point to zero through reverse compensation. The phase calibration value is added to the continuous phase to obtain the calibration phase function. The calibration phase function uses zero point as a reference and outputs the phase value at any time in real time, and grows naturally with the power frequency cycle. The calibration phase function eliminates the influence of historical phase accumulation, ensuring that phase calculation always starts from the most recent zero-crossing moment, guaranteeing the comparability of phases at different times.

[0114] In detail, each discharge event is triggered by an electrical signal channel, corresponding to a unique trigger time. For this trigger time, the corresponding phase value can be directly obtained by calibrating the phase function, i.e., the value of the calibration phase function at the trigger time. Since the calibration phase function grows linearly with time, its value may exceed 2π. Therefore, a 2π modulo operation is needed to constrain the phase within the range [0, 2π) to obtain the event phase. The event phase quantifies the specific location of the discharge event within the power frequency cycle; for example, 0 corresponds to a positive zero-crossing, and π corresponds to a negative zero-crossing.

[0115] In one embodiment of the present invention, the amplitude values ​​of electrical signal events and optical signal events are extracted based on a preset phase window and event phase, and normalization processing is performed to obtain standard electrical signal amplitude values ​​and standard optical signal amplitude values, including:

[0116] A phase window is constructed using a preset center phase and a preset phase window half-width; wherein the phase window is centered on the preset center phase and extends to both sides by a preset phase window half-width.

[0117] If the phase of the event corresponding to the discharge event is within the phase window, the corresponding discharge event is marked as a candidate event to form a candidate event set;

[0118] For each candidate event in the candidate event set, the electrical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the first preset time window to obtain the corresponding electrical signal event amplitude.

[0119] For each candidate event in the candidate event set, the overlapping optical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the second preset time window to obtain the corresponding optical signal event amplitude.

[0120] The amplitude values ​​of the electrical signal and optical signal events for each candidate event are normalized to obtain the standard electrical signal amplitude and standard optical signal amplitude for each candidate event.

[0121] In detail, the preset center phase is a power frequency phase point pre-set based on the characteristics of partial discharge, and the preset phase window half-width is the phase range extended to both sides of this center phase. The phase window is constructed by extending the preset phase window half-width to both sides of the preset center phase to form a continuous phase interval. For example, if the center phase is π / 2 and the phase window half-width is π / 6, then the phase window is [π / 2-π / 6, π / 2+π / 6], i.e., [π / 32, π / 3]. This focuses on specific intervals with similar electric field conditions within the power frequency cycle. The intensity and characteristics of partial discharge are often closely related to the power frequency phase. Discharge events within the same phase window are under similar electric field intensities and directions, and their amplitude characteristics are comparable.

[0122] In detail, the event phase refers to the specific phase position of each discharge event within the power frequency cycle. By determining whether the event phase corresponding to a discharge event falls within a phase window, events whose phases meet the criteria are marked as candidate events and aggregated into a candidate event set. This excludes events outside the phase window, as these events may have fundamentally different amplitude characteristics from the target range due to different electric field conditions. Retaining only candidate events ensures that the objects for subsequent amplitude extraction and analysis have a consistent phase background.

[0123] In detail, the electrical signal envelope is a smoothed envelope curve obtained after preprocessing the original waveform of the electrical signal channel, which can stably reflect the amplitude changes of the electrical signal pulse. For each event in the candidate event set, a first preset time window is set with its trigger time as the center. The first preset time window covers the complete duration of the electrical signal pulse. The integral value obtained by integrating the electrical signal envelope within the first preset time window is the amplitude of the electrical signal event. Integration can comprehensively reflect the total energy of the entire pulse, avoiding peak deviations caused by pulse shape or instantaneous noise.

[0124] In detail, the realigned optical signal envelope is the realigned envelope curve obtained after time delay compensation of the optical signal channel, and it is synchronized with the time base of the electrical signal channel. Consistent with the electrical signal amplitude extraction logic, a second preset time window is set for each event in the candidate event set, centered on its trigger time. The second preset time window needs to match the duration of the optical signal, typically longer than the electrical signal range. For example, 1 microsecond before and after the trigger time. Within the second preset time window, the realigned optical signal envelope is integrated, and the resulting integral value is the optical signal event amplitude. This ensures complete capture of the optical signal response triggered by discharge, such as the continuous process of optical radiation or mechanical vibration.

[0125] In detail, the physical meaning and dimensions of electrical signal event amplitudes and optical signal event amplitudes differ. For example, the unit of electrical signal amplitude is voltage-time integral, while that of optical signal amplitude is light intensity-time integral. The absolute numerical ranges of the two differ significantly, making direct comparison meaningless. Therefore, normalization is necessary for both types of amplitudes. Normalization, based on the statistical characteristics of the candidate event set, maps electrical signal event amplitudes and optical signal event amplitudes to a unified dimensionless scale (the interval between 0 and 1). This eliminates the influence of dimensional differences and absolute amplitude ranges, allowing for direct cross-channel numerical comparisons of the processed standard electrical signal amplitudes and standard optical signal amplitudes.

[0126] It should be noted that the preset center phase is preferably selected as π / 2 and 3π / 2, corresponding to the phase position near the voltage peak within the power frequency cycle. The probability of partial discharge is closely related to the electric field strength, and the electric field strength is greatest at the peak of the power frequency voltage at π / 2 and 3π / 2, making it most prone to discharge at defects such as micro-gaps. Therefore, discharge events near this phase are more significant and representative. The preset phase window half-width is preferably selected as π / 60. The phase window half-width must meet two conditions: first, it must be small enough to ensure that all events within the window are under similar electric field conditions; second, it must be large enough to include a certain number of events. π / 60 is the optimal choice to balance these two conditions. That is, for a 50Hz power frequency (20ms cycle), π / 60 corresponds to approximately 1.67ms, which ensures that the difference in electric field strength within the window is less than 5%, while also allowing sufficient events to accumulate under normal discharge intensity.

[0127] In one embodiment of the present invention, the standard electrical signal amplitude and the standard optical signal amplitude are paired according to the event trigger sequence time in the realigned optical signal envelope to form a cross-channel amplitude pair sequence, including:

[0128] For each candidate event in the candidate event set, with the trigger time corresponding to the candidate event as the center of the second preset time window, search for the local maximum point of the realigned optical signal envelope to obtain the peak time of the optical signal at the local maximum point and the corresponding amplitude of the realigned optical signal envelope.

[0129] The time difference between the peak moment of the optical signal and the trigger moment corresponding to the candidate event is used as the pairing error.

[0130] If the pairing error is less than or equal to the preset alignment error threshold, and the amplitude of the realigned optical signal envelope is greater than or equal to the optical signal threshold, then the candidate event is marked as a valid pairing event.

[0131] For each valid pairing event, the corresponding standard electrical signal amplitude and standard optical signal amplitude are recorded as amplitude pairs, and the amplitude pairs are arranged in ascending order according to the corresponding trigger time to form a cross-channel amplitude pair sequence.

[0132] In detail, the candidate event set is a set of discharge events with consistent phase characteristics after being filtered by a phase window. Each candidate event corresponds to a unique trigger time. A second preset time window is set with this trigger time as the center. The local maximum point within the second preset time window is searched in the realigned optical signal envelope because the peak value of the optical signal envelope usually corresponds to the strongest moment of the optical signal response triggered by the discharge event, and can most accurately match the electrical signal trigger time. The occurrence time of this local maximum point is recorded as the peak moment of the optical signal, and the corresponding envelope value is the amplitude of the realigned optical signal envelope. Thus, by focusing on the strongest response point, the interference of clutter or trailing in the optical signal on pairing is reduced, ensuring that each candidate event corresponds to a unique and reliable optical signal characteristic moment.

[0133] In detail, the pairing error is the time difference between the peak moment of the optical signal and the trigger moment of the candidate event. The pairing error quantifies the degree of synchronization between the optical signal response and the electrical signal trigger. Due to the differences in the physical propagation paths and response speeds of electrical and optical signals, perfect synchronization is difficult to achieve. However, the error needs to be controlled within a reasonable range. That is, if the error is too large, it indicates that the two signals may correspond to different events, or that there is an abnormal delay in the optical signal.

[0134] In detail, a valid pairing event must simultaneously meet two conditions. First, the pairing error ≤ a preset alignment error threshold. The preset alignment error threshold is set according to the system's time synchronization accuracy (preferably 500 nanoseconds) to ensure that the peak value of the optical signal and the triggering of the electrical signal are highly coincident in time, belonging to the same discharge event response. Second, the amplitude of the realigned optical signal envelope ≥ the optical signal threshold. The optical signal threshold is a critical value determined based on noise statistics, ensuring that an optical signal amplitude exceeding the background noise indicates a response triggered by a valid discharge, rather than noise interference.

[0135] In detail, the standard electrical signal amplitude and standard optical signal amplitude are dimensionless characteristic quantities that have been normalized to eliminate the influence of dimensional differences and absolute amplitude fluctuations. For each valid pairing event, its corresponding standard electrical signal amplitude and standard optical signal amplitude are bound as amplitude pairs to ensure a one-to-one correspondence between the cross-channel characteristics of the same event. All amplitude pairs are arranged in ascending order of the candidate event trigger time to form a cross-channel amplitude pair sequence. The significance of ascending order is to preserve the temporal sequence of events, so that the sequence can reflect the dynamic changes in the characteristics of the two channels during the discharge process.

[0136] In one embodiment of the present invention, the calculation of the anticorrelation principal axis slope, correlation coefficient, closed-loop directed area, and normalized loop strength based on cross-channel amplitude pair sequences includes:

[0137] The process of obtaining the closed-loop directed area is as follows:

[0138] By concatenating the first and last pairs of cross-channel amplitude pairs in chronological order as input variables for the shoelace formula, the closed-loop directed area is obtained.

[0139] The process of obtaining the normalized ring strength is as follows:

[0140] The square root of the variance of the standard electrical signal amplitude is taken as the standard deviation of the standard electrical signal amplitude;

[0141] The square root of the variance of the standard optical signal amplitude is the standard deviation of the standard optical signal amplitude;

[0142] The ratio of the absolute value of the directional area of ​​the closed loop to the product of pi and the standard deviation of the amplitude of the standard electrical signal and the standard deviation of the amplitude of the standard optical signal is used as the normalized loop strength.

[0143] In detail, the cross-channel amplitude pair sequence consists of paired data of standard electrical signal amplitude and standard optical signal amplitude arranged in ascending order according to the event trigger time. Each data pair can be regarded as a point on a two-dimensional plane, with the horizontal coordinate being the standard electrical signal amplitude and the vertical coordinate being the standard optical signal amplitude. Connecting these points in chronological order forms a trajectory reflecting the dynamic changes in the amplitudes of the two channels. Due to the periodicity of partial discharge and the time mismatch between the responses of the two channels (e.g., the electrical signal response is faster than the optical signal response), the trajectory often exhibits the characteristic that the outbound and return paths are different. To quantify this closed-loop characteristic, the first and last data pairs need to be connected, that is, the last data pair is connected to the first data pair to form a closed polygon. The shoelace formula is a classic algorithm for calculating the area of ​​a polygon. Its core is to successively accumulate the area of ​​the trapezoid formed by adjacent vertices and take half of the absolute value to obtain the total area. Substituting the vertex coordinates of the closed polygon (i.e., the cross-channel amplitude pairs) into this formula yields the closed-loop directed area. The direction of the directed area is reflected in its sign: if the trajectory is counterclockwise, the area is positive; if it is clockwise, the area is negative. The sign reflects the asymmetry of the amplitude changes of the two channels (such as the order of change of the light signal when the electrical signal increases first and then decreases), while the absolute value represents the size of the region enclosed by the closed-loop trajectory. The larger the value, the more significant the difference in the dynamic response of the two channels.

[0144] In detail, the standard deviation of the standard electrical signal amplitude is the square root of the variance of the standard electrical signal amplitude sequence. The standard deviation of the standard optical signal amplitude is also the square root of the variance of the standard optical signal amplitude sequence. The standard deviation quantifies the dispersion of the data; a larger standard deviation indicates a wider amplitude fluctuation range for the corresponding channel. The absolute value of the directed area of ​​the closed loop reflects the physical size of the closed-loop trajectory, but the directed area is affected by the amplitude fluctuation range. For example, sequences with generally larger amplitudes naturally have a larger area. Directly comparing the areas of different scenarios has limited meaning. Therefore, normalization is needed to eliminate this scale difference. The calculation logic for normalized loop strength is: divide the absolute value of the directed area of ​​the closed loop by a reference area. The reference area is defined as the product of pi and the standard deviations of the standard electrical signal amplitude and the standard optical signal amplitude, representing the area of ​​an ellipse with the standard deviations of the two channels as semi-axes. This area objectively reflects the scale of the natural fluctuations in the data. The normalized loop strength obtained through normalization is a dimensionless index. The magnitude of the normalized loop strength is determined only by the relative size of the closed-loop trajectory. That is, the larger the normalized loop strength, the more significant the closed-loop characteristics are relative to the natural fluctuations of the data.

[0145] In one embodiment of the present invention, the partial discharge determination result is output through a robustness test, including:

[0146] Robustness checks are as follows:

[0147] For a preset search delay range centered on relative delay, the delay is micro-scanned within the corresponding preset search delay range; after temporarily shifting the envelope of the realigned optical signal, the closed-loop directed area is recalculated; the absolute value of the closed-loop directed area is normalized to obtain the delay fluctuation rate.

[0148] Set a phase change value for a preset phase window half-width, and increase or decrease the preset phase window half-width by the phase change value to form a first variable phase window and a second variable phase window, respectively; recalculate the corresponding normalized ring strength based on the first variable phase window and the second variable phase window to obtain the first variable ring strength and the second variable ring strength; use the ratio of the standard deviation of the first variable ring strength, the second variable ring strength and the normalized ring strength to the corresponding average value as the phase window volatility.

[0149] If both the time delay volatility and the phase window volatility are less than or equal to the corresponding preset volatility thresholds, then the robustness test is satisfied, and it is determined that a partial discharge exists.

[0150] Detailed robustness tests are used to verify the stability of partial discharge characteristics (closed-loop directional area, normalized loop strength) to minor perturbations of key parameters, ensuring that the judgment results are not caused by accidental parameter settings, but truly reflect the physical characteristics of partial discharge. The test is achieved through two dimensions: stability of time alignment (time delay volatility) and stability of phase range (phase window volatility).

[0151] Relative delay is a core parameter for time alignment between electrical and optical signal channels. Even minor deviations can lead to time mismatches in the amplitude pair sequences across channels, resulting in false closed-loop characteristics. The specific process is as follows: Using a determined relative delay as the center, a preset search delay range is set (e.g., ±500 nanoseconds, covering possible minor time deviations). Within this range, the delay is gradually fine-tuned (i.e., temporarily shifting the envelope of the realigned optical signal). After each shift, the directed area of ​​the closed loop (reflecting the degree of closure of the dynamic trajectory of the two channel amplitudes) is recalculated. The absolute value of the directed area of ​​the closed loop under all fine-tuned delays is normalized to obtain the delay volatility. Normalization aims to eliminate differences in the absolute area values ​​under different scenarios, focusing on the volatility ratio. In other words, the delay volatility quantifies the sensitivity of the closed-loop area to minor time alignment deviations; a smaller value indicates more stable time alignment, and the closed-loop characteristics are not affected by accidental time deviations.

[0152] The half-width of the phase window determines the phase range of discharge events included in the analysis. Even small changes in this half-width can alter the selected event set, affecting the normalized ring strength. The specific process is as follows: A small phase change value (e.g., 1°, much smaller than the preset half-width of the phase window) is set. The preset half-width of the phase window is adjusted using this change value; increasing the phase change value yields the first changed phase window, and decreasing it yields the second changed phase window. Based on these two new phase windows, the corresponding normalized ring strengths are recalculated, namely the first changed ring strength and the second changed ring strength. Combining the normalized ring strengths of the phase windows, the ratio of the standard deviation to the mean of these three ring strengths is calculated; this is the phase window volatility. The standard deviation reflects the dispersion of the data, while the mean reflects the overall level. The ratio quantifies the sensitivity of the normalized ring strength to small adjustments in the phase range. A smaller ratio indicates that random changes in the phase range do not affect the ring strength characteristics, resulting in higher stability.

[0153] Based on the above two test results: if the time delay volatility is less than or equal to the preset volatility threshold, and the phase window volatility is also less than or equal to the preset volatility threshold, then the partial discharge characteristics are considered robust to both time alignment deviations and phase range adjustments, eliminating spurious characteristics caused by accidental parameter settings. In this case, the robustness test passes, and partial discharge is determined to exist. True partial discharge characteristics should possess inherent stability and not change significantly due to minor, reasonable perturbations of key parameters; only by simultaneously satisfying time alignment stability and phase range stability can the reliability and repeatability of the determination results be ensured.

[0154] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. A multi-mode mixing partial discharge fusion system, characterized in that, include: The data acquisition module is used to acquire power frequency reference waveforms, electrical signal waveforms, and optical signal channel waveforms, and obtains the event trigger time sequence mainly based on the electrical signal channel. The data compensation module is used to determine the relative time delay based on the electrical signal waveform and the optical signal channel waveform, and to perform compensation processing on the optical signal channel waveform according to the relative time delay to generate a realigned optical signal envelope; The phase window construction module is used to calculate the event phase based on the power frequency reference waveform; The data extraction module is used to extract the amplitude values ​​of electrical signal events and optical signal events based on a preset phase window and event phase, and perform normalization processing to obtain standard electrical signal amplitude values ​​and standard optical signal amplitude values. The data pairing module is used to pair the standard electrical signal amplitude and the standard optical signal amplitude according to the event trigger sequence time in the realigned optical signal envelope to form a cross-channel amplitude pair sequence; The fusion determination module is used to calculate the anticorrelation principal axis slope, correlation coefficient, closed-loop directional area and normalized loop strength based on cross-channel amplitude pair sequences, and output the partial discharge determination result through robustness test.

2. The multi-mode mixing partial discharge fusion system according to claim 1, characterized in that, The electrical signal channel is the main trigger to obtain the event trigger time sequence, including: Generate an electrical signal envelope from the waveform of the electrical signal channel; Within a preset noise statistical interval, the electrical signal threshold is calculated based on the mean and standard deviation of the electrical signal envelope; where the electrical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation. When the envelope value of the electrical signal at time i is greater than or equal to the threshold value of the electrical signal, the time period from time i-1 to time i is marked as a discharge event, and time i is taken as the trigger time of the discharge event and time i-1 is taken as the starting time of the discharge event. An event trigger sequence is formed based on all discharge events in chronological order.

3. The multi-mode mixing partial discharge fusion system according to claim 2, characterized in that, The relative time delay is determined based on the waveforms of the electrical signal channel and the optical signal channel, and the optical signal channel waveform is compensated according to the relative time delay to generate a realigned optical signal envelope, including: Generate an optical signal envelope from the optical signal channel waveform; Within a preset noise statistics interval, the optical signal threshold is calculated based on the mean and standard deviation of the optical signal envelope; whereby the optical signal threshold is the sum of the weighted values ​​of the corresponding mean and standard deviation. For the j-th discharge event, extract the corresponding electrical signal envelope segment and optical signal envelope segment, and calculate the corresponding discrete cross-correlation within the preset search delay range; The time delay corresponding to the discrete cross-correlation is used as the time delay estimate, and the average of the time delay estimates at all discharge event times is used as the relative time delay. The optical signal envelope is shifted relative to the time delay along the time axis to form an overaligned optical signal envelope.

4. The multi-mode mixing partial discharge fusion system according to claim 3, characterized in that, Calculating the event phase based on the power frequency reference waveform includes: A positive zero-crossing detection is performed on the power frequency reference waveform to obtain a positive zero-crossing time sequence; The Hilbert transform is applied to the power frequency reference waveform to obtain a complex signal; the imaginary and real parts of the complex signal are extracted to calculate the instantaneous phase, and the instantaneous phase is expanded to obtain the continuous phase; Take the last positive zero-crossing moment in the positive zero-crossing moment sequence as the zero point moment; The negative value of the continuous phase corresponding to the zero point is calculated as the phase calibration value, and the sum of the phase calibration value and the continuous phase is used as the calibration phase function. For each discharge event, the value of the calibration phase function at the trigger time is calculated, and the event phase corresponding to the discharge event is obtained by taking the modulus of 2π.

5. The multi-mode mixing partial discharge fusion system according to claim 4, characterized in that, Based on a preset phase window and event phase, the amplitude values ​​of electrical signal events and optical signal events are extracted and normalized to obtain standard electrical signal amplitudes and standard optical signal amplitudes, including: A phase window is constructed using a preset center phase and a preset phase window half-width; wherein the phase window is centered on the preset center phase and extends to both sides by a preset phase window half-width. If the phase of the event corresponding to the discharge event is within the phase window, the corresponding discharge event is marked as a candidate event to form a candidate event set; For each candidate event in the candidate event set, the electrical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the first preset time window to obtain the corresponding electrical signal event amplitude. For each candidate event in the candidate event set, the overlapping optical signal envelope is integrated with the trigger time corresponding to the candidate event as the center of the second preset time window to obtain the corresponding optical signal event amplitude. The amplitude values ​​of the electrical signal and optical signal events for each candidate event are normalized to obtain the standard electrical signal amplitude and standard optical signal amplitude for each candidate event.

6. The multi-mode mixing partial discharge fusion system according to claim 5, characterized in that, In the realigned optical signal envelope, the standard electrical signal amplitude and the standard optical signal amplitude are paired according to the event trigger sequence time to form a cross-channel amplitude pair sequence, including: For each candidate event in the candidate event set, with the trigger time corresponding to the candidate event as the center of the second preset time window, search for the local maximum point of the realigned optical signal envelope to obtain the peak time of the optical signal at the local maximum point and the corresponding amplitude of the realigned optical signal envelope. The time difference between the peak moment of the optical signal and the trigger moment corresponding to the candidate event is used as the pairing error. If the pairing error is less than or equal to the preset alignment error threshold, and the amplitude of the realigned optical signal envelope is greater than or equal to the optical signal threshold, then the candidate event is marked as a valid pairing event. For each valid pairing event, the corresponding standard electrical signal amplitude and standard optical signal amplitude are recorded as amplitude pairs, and the amplitude pairs are arranged in ascending order according to the corresponding trigger time to form a cross-channel amplitude pair sequence.

7. The multi-mode mixing partial discharge fusion system according to claim 6, characterized in that, Based on cross-channel amplitude pairs, the anticorrelation principal axis slope, correlation coefficient, closed-loop directed area, and normalized loop strength are calculated, including: The process of obtaining the closed-loop directed area is as follows: By concatenating the first and last pairs of cross-channel amplitude pairs in chronological order as input variables for the shoelace formula, the closed-loop directed area is obtained. The process of obtaining the normalized ring strength is as follows: The square root of the variance of the standard electrical signal amplitude is taken as the standard deviation of the standard electrical signal amplitude; The square root of the variance of the standard optical signal amplitude is the standard deviation of the standard optical signal amplitude; The ratio of the absolute value of the directional area of ​​the closed loop to the product of pi and the standard deviation of the amplitude of the standard electrical signal and the standard deviation of the amplitude of the standard optical signal is used as the normalized loop strength.

8. The multi-mode mixing partial discharge fusion system according to claim 7, characterized in that, The partial discharge determination result is output through robustness checks, including: Robustness checks are as follows: For a preset search delay range centered on relative delay, the delay is micro-scanned within the corresponding preset search delay range; after temporarily shifting the envelope of the realigned optical signal, the closed-loop directed area is recalculated; the absolute value of the closed-loop directed area is normalized to obtain the delay fluctuation rate. Set a phase change value for a preset phase window half-width, and increase or decrease the preset phase window half-width by the phase change value to form a first variable phase window and a second variable phase window respectively; recalculate the corresponding normalized ring strength based on the first variable phase window and the second variable phase window respectively to obtain the first variable ring strength and the second variable ring strength; use the ratio of the standard deviation of the first variable ring strength, the second variable ring strength and the normalized ring strength to the corresponding average value as the phase window volatility. If both the time delay volatility and the phase window volatility are less than or equal to the corresponding preset volatility thresholds, then the robustness test is satisfied, and it is determined that a partial discharge exists.

Images