Electro-optical-acoustic cooperative partial discharge detection device and method

Abstract

The invention discloses an electro-optical-acoustic cooperative partial discharge detection device and method, and belongs to the field of equipment discharge fault detection. The device integrates a high-frequency current detection module, a photoelectric coupling detection module and an acoustic emission detection module; a high-frequency pulse current signal of a cable grounding loop, a pulse electric field of an insulating surface and an ultrasonic signal in a solid medium can be collected at the same time; all the collected signals are sent to the signal integrated processing module for alignment, noise reduction and multi-signal homologous conformity test; after detection, the signal integration processing module extracts combined feature vectors from the signals and inputs the combined feature vectors to the classification discriminator; finally, the discriminator outputs the probability of partial discharge, and if the probability exceeds a preset threshold value, it is judged that partial discharge exists in the cable. Therefore, by implementing the method and the device, the problems that single-mode detection in the prior art is poor in anti-interference capability in a complex field environment, and false alarm or missing alarm is easy to occur can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment discharge fault detection, and in particular to an electro-optical-acoustic coordinated partial discharge detection device and method. Background Art

[0002] Partial discharge (PD) is a weak electrical discharge that occurs when the insulation structure of high-voltage power equipment (such as high-voltage cables, transformers, and switchgear) is exposed to a strong electric field. It is one of the most important precursors to insulation degradation and potential insulation failure. Therefore, accurate and reliable online detection of PD is of irreplaceable importance for ensuring the safe and stable operation of power systems, avoiding major accidents, and implementing condition-based maintenance of equipment. With the increasing reliability requirements of smart grids and critical infrastructure, the development of highly sensitive and reliable online PD detection technology has become a research hotspot and an urgent need in this field.

[0003] At present, the partial discharge detection technologies used in engineering sites mainly include high-frequency current method, acoustic emission method and photoelectric sensing method. These technologies are achieved by detecting electromagnetic pulses, ultrasonic waves or electric field disturbances associated with partial discharge. However, most of the existing technical solutions only rely on single-dimensional measurement of one of the above physical signals. This single-modal detection method has inherent limitations when facing multiple interference sources (such as electromagnetic noise, mechanical vibration, etc.) in complex field environments such as substations and cable tunnels. The reason is that due to the single signal source, it lacks sufficient information redundancy. When the signal is subject to its corresponding type of interference, it is difficult for the device to effectively distinguish between the real discharge signal and the false trigger interference signal through the mutual verification of multiple physical phenomena, resulting in insufficient overall reliability of the detection results, and prone to problems such as false alarms or missed alarms. Summary of the Invention

[0004] The embodiments of the present invention provide an electro-optical-acoustic coordinated partial discharge detection device and method, which can solve the problem in the prior art that single-mode detection has poor anti-interference ability in complex field environments and is prone to false alarms or missed alarms.

[0005] An embodiment of the present invention provides an electro-optical-acoustic coordinated partial discharge detection device, comprising: a high-frequency current detection module, a photoelectric coupling detection module, an acoustic emission detection module, a signal integration processing module, and a support base;

[0006] The support base is used to carry the high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module, and is installed on the cable to be tested;

[0007] The high-frequency current detection module is used to collect a first high-frequency pulse current signal of the ground loop portion of the cable to be tested;

[0008] The photoelectric coupling detection module is used to collect a first local pulse electric field on the insulation surface of the cable to be tested, and modulate the first local pulse electric field into an optical signal to generate a first pulse optical signal;

[0009] The acoustic emission detection module is used to collect the first ultrasonic signal propagating in the solid medium of the cable to be tested;

[0010] The signal integration processing module is used to perform a multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, a combined feature vector is extracted from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and the combined feature vector is input into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test based on the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test; otherwise, it is determined that no partial discharge exists in the cable under test.

[0011] Furthermore, it also includes: a protective housing;

[0012] The protective shell covers the outside of the supporting base, the high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module.

[0013] Furthermore, the performing of a multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal includes:

[0014] Based on a preset sampling frequency, sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal are respectively extracted;

[0015] Recording the moment corresponding to the amplitude peak in each sampling data of the first high-frequency pulse current signal as the first peak moment;

[0016] Recording the moment corresponding to the amplitude peak in each sampling data of the first pulse light signal as the second peak moment;

[0017] Recording the moment corresponding to the amplitude peak in each sampling data of the first ultrasonic signal as the third peak moment;

[0018] Calculate the absolute difference between the first peak time and the third peak time to generate a first arrival time difference;

[0019] Calculating the absolute difference between the second peak time and the third peak time to generate a second arrival time difference;

[0020] Calculating, based on the sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, a first waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first pulse light signal, a second waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first ultrasonic signal, and a third waveform amplitude cross-correlation coefficient between the first pulse light signal and the first ultrasonic signal;

[0021] If the first arrival time difference is not greater than the preset time threshold, the second arrival time difference is not greater than the preset time threshold, the first waveform amplitude correlation coefficient is not less than the preset similarity threshold, the second waveform amplitude correlation coefficient is not less than the preset similarity threshold, and the third waveform amplitude correlation coefficient is not less than the preset similarity threshold, then the multi-signal homology conformity test is determined to have passed; otherwise, the multi-signal homology conformity test is determined to have failed.

[0022] Furthermore, the signal integration processing module, before performing the multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, further includes:

[0023] The first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal are aligned and denoised to generate an updated first high-frequency pulse current signal, an updated first pulse light signal and an updated first ultrasonic signal.

[0024] Furthermore, the high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module are arranged sequentially along the axial direction of the cable to be tested;

[0025] The signal output ends of the high-frequency current detection module, the photoelectric coupling detection module, and the acoustic emission detection module are all connected to the signal input end of the signal integration processing module.

[0026] Furthermore, the high-frequency current detection module includes an openable and closable clamping structure with a built-in high-frequency current transformer coil, the openable and closable clamping structure is used to fix the high-frequency current detection module to the ground loop portion of the cable to be tested, and the built-in high-frequency current transformer coil is used to collect the first high-frequency pulse current signal;

[0027] The photoelectric coupling detection module includes an electro-optical modulator based on the Pockels effect, which is used to sense a first local pulse electric field generated on the insulation surface of the cable to be tested and modulate its intensity change into an optical signal to generate the first pulse optical signal;

[0028] The acoustic emission detection module includes a plurality of piezoelectric sensors arranged along the axial direction of the cable to be tested, and each piezoelectric sensor is used to collaboratively receive a first ultrasonic wave signal propagating in a solid medium of the cable to be tested.

[0029] Furthermore, the construction of the classification discriminator includes:

[0030] Acquire a partial discharge feature sample set; wherein the partial discharge feature sample set includes a plurality of combined feature vectors and corresponding partial discharge labels; the combined feature vector includes at least one feature extracted from the amplitude, rising edge, main frequency, bandwidth, and wavelet energy ratio of the signal;

[0031] Randomly dividing the partial discharge feature sample set into a number of training subsample sets according to a preset batch size;

[0032] For each training sub-sample set, generate a decision tree corresponding to the current sub-sample set according to the current sub-sample set;

[0033] The decision trees corresponding to the training sub-sample sets are combined to generate the classification discriminator.

[0034] Furthermore, the signal integration processing module is further configured to obtain the axial distance between the high-frequency current detection module and the acoustic emission detection module and the propagation speed of the acoustic wave in the cable under test when it is determined that partial discharge exists in the cable under test;

[0035] Calculating the absolute difference between the arrival time of the peak of the first high-frequency pulse current signal and the arrival time of the peak of the first ultrasonic signal to generate an arrival time difference between the first high-frequency pulse current signal and the first ultrasonic signal;

[0036] An axial position where partial discharge occurs in the cable to be tested is determined according to the arrival time difference, the axial distance, and the propagation speed.

[0037] Furthermore, the signal integration processing module is further configured to store the combined feature vector and the corresponding classification result of each determination of the presence of partial discharge in the tested cable in a database;

[0038] After each preset time period, the combined feature vectors and their corresponding classification results in the database are added to the partial discharge feature sample set, and the combined feature vectors and their corresponding classification results in the database are cleared; and the classification discriminator is incrementally trained based on the partial discharge feature sample set.

[0039] Based on the above-mentioned device embodiment, the present invention provides a corresponding method embodiment.

[0040] An embodiment of the present invention provides an electro-optical-acoustic coordinated partial discharge detection method, which is applicable to a signal integration processing module of an electro-optical-acoustic coordinated partial discharge detection device, including:

[0041] Acquire a first high-frequency pulse current signal of a ground loop portion of the cable under test, a first pulse light signal on an insulation surface of the cable under test, and a first ultrasonic signal propagating in a solid medium of the cable under test;

[0042] A multi-signal homology conformity test is performed on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, a combined feature vector is extracted from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and the combined feature vector is input into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test based on the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test; otherwise, it is determined that no partial discharge exists in the cable under test.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] An embodiment of the present invention provides an electro-optical-acoustic coordinated partial discharge detection device and method. The device integrates a high-frequency current detection module, a photoelectric coupling detection module, and an acoustic emission detection module, and can simultaneously collect high-frequency pulse current signals of the cable grounding loop, pulse electric fields on the insulation surface, and ultrasonic signals in solid media. All collected signals are sent to the signal integration processing module for alignment, noise reduction, and multi-signal homology compliance testing. After passing the test, the integrated processing module extracts a combined feature vector from these signals and inputs it into a classification discriminator. Ultimately, the discriminator outputs the probability of partial discharge. If the probability exceeds a preset threshold, it is determined that partial discharge exists in the cable.

[0045] This invention uses a collaborative approach to detect multiple physical signals, including electrical, optical, and acoustic signals, addressing the existing single-modality detection issues of poor interference immunity and proneness to false positives and omissions in complex field environments. Furthermore, by acquiring signals from multiple dimensions and performing signal alignment, noise reduction, and multi-signal homology verification, the invention significantly improves the reliability and accuracy of partial discharge detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a structural schematic diagram of an electro-optical-acoustic coordinated partial discharge detection device provided in one embodiment of the present invention.

[0047] Figure 2 It is a flow chart of a method for detecting partial discharge using electro-optical-acoustic coordination provided by one embodiment of the present invention.

[0048] Description of reference numerals:

[0049] 1. High-frequency current detection module; 2. Photoelectric coupling detection module; 3. Acoustic emission detection module; 4. Signal integration and processing module; 5. Support base; 6. Protective housing; 7. Cable to be tested. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] like Figure 1 As shown, to solve the problem that single-mode detection in the prior art has poor anti-interference ability in complex field environments and is prone to false alarms or missed alarms, an embodiment of the present invention provides an electro-optical-acoustic coordinated partial discharge detection device, comprising: a high-frequency current detection module 1, a photoelectric coupling detection module 2, an acoustic emission detection module 3, a signal integration and processing module 4, and a support base 5;

[0052] The support base 5 is used to carry the high-frequency current detection module 1, the photoelectric coupling detection module 2, the acoustic emission detection module 3 and the signal integration processing module 4, and is installed on the cable to be tested 7;

[0053] Specifically, the support base 5 is designed to stably integrate and support the various core functional modules of the detection device, including a high-frequency current detection module 1 for collecting high-frequency pulse current signals, a photoelectric coupling detection module 2 for collecting pulse electric fields and modulating them into optical signals, and an acoustic emission detection module 3 for collecting ultrasonic signals. In addition, a signal integration processing module 4 for integrating and processing these signals is also fixed on the support base 5. The structural design of the support base 5 enables it to be tightly and stably installed on the outer surface of the cable 7 to be tested, providing stable mechanical support and positioning for the entire detection device, ensuring that each detection module can accurately sense and collect relevant signals on the cable. Through this bearing and installation method, the support base 5 effectively integrates the multimodal detection function into a whole, facilitating the deployment and use of the device on site.

[0054] The high-frequency current detection module 1 is used to collect a first high-frequency pulse current signal of the ground loop portion of the cable 7 to be tested;

[0055] Specifically, the module is designed to sense and capture the high-frequency pulse current signals associated with partial discharge events. These high-frequency pulse current signals appear in the ground loop of the cable under test 7. The high-frequency current detection module 1 effectively converts these electrical signals into signals suitable for subsequent processing, providing critical electrical information for preliminary partial discharge detection. By collecting high-frequency pulse current signals, the module provides an important electrical detection method for evaluating the insulation condition of cables.

[0056] The photoelectric coupling detection module 2 is used to collect a first local pulse electric field on the insulation surface of the cable 7 to be tested, and modulate the first local pulse electric field into an optical signal to generate a first pulse optical signal;

[0057] Specifically, the module achieves detection by sensing transient pulse electric field changes caused by partial discharge on the cable insulation surface. The optoelectronic coupling detection module 2 has a built-in electro-optical modulator that operates based on the Pockels effect, sensing changes in the intensity of the local pulse electric field and converting them into optical signals. This electro-optical conversion mechanism avoids the electromagnetic interference issues that may exist in traditional electrical detection and improves the detection's anti-interference capability. By converting electric field information into optical signals, the optoelectronic coupling detection module 2 effectively provides optical information on partial discharge, providing multimodal collaborative detection data for subsequent signal processing and fault diagnosis.

[0058] The acoustic emission detection module 3 is used to collect the first ultrasonic signal propagating in the solid medium of the cable 7 to be tested;

[0059] Specifically, when partial discharge occurs within the cable, it is accompanied by the generation of transient elastic waves, or ultrasonic signals. The acoustic emission detection module 3 is capable of sensing and receiving these ultrasonic waves propagating through the solid medium of the cable. This module contains piezoelectric sensors that convert the received mechanical vibration signals into electrical signals, thereby enabling the collection of acoustic information about partial discharge. By collecting ultrasonic signals, the acoustic emission detection module 3 provides important acoustic information for the detection of partial discharge, which is of great significance for subsequent signal fusion and possible spatial positioning.

[0060] The signal integration processing module 4 is used to perform a multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, a combined feature vector is extracted from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and the combined feature vector is input into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test 7 based on the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test 7; otherwise, it is determined that no partial discharge exists in the cable under test 7.

[0061] Specifically, this module receives signals in three different modes: high-frequency current detection module 1, optocoupler detection module 2, and acoustic emission detection module 3. During multi-signal homology compliance testing, signal integration and processing module 4 evaluates the temporal and amplitude correlations among these signals to determine whether they originate from the same partial discharge event. This verification process effectively eliminates interference signals that may arise during single-mode testing, improving detection reliability.

[0062] When the multi-signal homology conformity test passes, the signal integration processing module 4 extracts combined feature vectors from the three signals. These combined feature vectors contain at least one feature of the signal, such as amplitude, rising edge, main frequency, bandwidth and wavelet energy ratio. The extracted combined feature vectors are then input into a preset classification discriminator. The classification discriminator analyzes these combined feature vectors and outputs a probability representing the presence of partial discharge in the cable under test 7. If the output partial discharge probability is greater than the preset discharge threshold, the signal integration processing module 4 determines that there is partial discharge in the cable under test 7; otherwise, it determines that there is no partial discharge in the cable under test 7. Through this multi-signal fusion and discrimination mechanism, the module effectively improves the accuracy and anti-interference ability of partial discharge detection.

[0063] In a preferred embodiment, the electro-optical-acoustic coordinated partial discharge detection device further comprises: a protective housing 6;

[0064] The protective shell 6 covers the outside of the supporting base 5 , the high-frequency current detection module 1 , the photoelectric coupling detection module 2 , the acoustic emission detection module 3 and the signal integration processing module 4 .

[0065] The protective housing 6 provides a layer of external protection for the entire electro-optical-acoustic coordinated partial discharge detection device. It effectively encapsulates all core components carried on the support base 5, including the high-frequency current detection module 1, the photoelectric coupling detection module 2, the acoustic emission detection module 3, and the signal integration and processing module 4. This encapsulation protects the delicate components within from external environmental factors such as dust, moisture, mechanical shock, and potential electromagnetic interference. By providing an effective physical barrier, the protective housing 6 ensures the reliable operation and service life of the detection device in various field environments.

[0066] In a preferred embodiment, the support base 5 is integrally formed of an insulation-reinforced composite material and has a semicircular arc surface structure surrounding the cable to be tested 7 .

[0067] Specifically, the support base 5 adopts a one-time molding process and uses a composite material with good insulation properties and mechanical strength, which helps the device to operate safely and stably in a high-voltage environment and provides sufficient structural support. Its semi-circular arc surface design enables the support base 5 to fit tightly with the outer surface of the cable 7 to be tested, forming a stable encircling installation. This structure not only ensures the stable installation of the detection device on the cable, effectively avoiding loosening or deviation, but also helps each detection module (such as the photoelectric coupling detection module 2 and the acoustic emission detection module 3) to maintain a precise coupling relationship with the cable surface, thereby ensuring the effectiveness and accuracy of signal acquisition.

[0068] In a preferred embodiment, the multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal includes:

[0069] Based on a preset sampling frequency, sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal are respectively extracted;

[0070] Recording the moment corresponding to the amplitude peak in each sampling data of the first high-frequency pulse current signal as the first peak moment;

[0071] Recording the moment corresponding to the amplitude peak in each sampling data of the first pulse light signal as the second peak moment;

[0072] Recording the moment corresponding to the amplitude peak in each sampling data of the first ultrasonic signal as the third peak moment;

[0073] Calculate the absolute difference between the first peak time and the third peak time to generate a first arrival time difference;

[0074] Calculating the absolute difference between the second peak time and the third peak time to generate a second arrival time difference;

[0075] Calculating, based on the sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, a first waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first pulse light signal, a second waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first ultrasonic signal, and a third waveform amplitude cross-correlation coefficient between the first pulse light signal and the first ultrasonic signal;

[0076] If the first arrival time difference is not greater than the preset time threshold, the second arrival time difference is not greater than the preset time threshold, the first waveform amplitude correlation coefficient is not less than the preset similarity threshold, the second waveform amplitude correlation coefficient is not less than the preset similarity threshold, and the third waveform amplitude correlation coefficient is not less than the preset similarity threshold, then the multi-signal homology conformity test is determined to have passed; otherwise, the multi-signal homology conformity test is determined to have failed.

[0077] Specifically, when performing a multi-signal homology conformance test, the signal integration processing module 4 first synchronously samples the received first high-frequency pulse current signal, first pulse light signal, and first ultrasonic signal based on a preset sampling frequency, thereby obtaining sampled data for each signal at different sampling points. The module then identifies and records the peak moments of each signal: the moment corresponding to the peak amplitude in each sampled data of the first high-frequency pulse current signal is recorded as the first peak moment; the moment corresponding to the peak amplitude in each sampled data of the first pulse light signal is recorded as the second peak moment; and the moment corresponding to the peak amplitude in each sampled data of the first ultrasonic signal is recorded as the third peak moment.

[0078] Subsequently, the signal integration processing module 4 evaluates the arrival time difference of the signal by calculating the absolute difference between these peak moments. Specifically, it includes: calculating the absolute difference between the first peak moment and the third peak moment to generate a first arrival time difference; calculating the absolute difference between the second peak moment and the third peak moment to generate a second arrival time difference. In addition, in order to evaluate the similarity of the waveforms of different modal signals in amplitude, the module will also calculate the mutual correlation coefficient between them based on the sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal. This includes: the first waveform amplitude mutual correlation coefficient between the first high-frequency pulse current signal and the first pulse light signal, the second waveform amplitude mutual correlation coefficient between the first high-frequency pulse current signal and the first ultrasonic signal, and the third waveform amplitude mutual correlation coefficient between the first pulse light signal and the first ultrasonic signal.

[0079] Finally, based on these calculation results, the signal integration processing module 4 performs a homology conformance check. The multi-signal homology conformance check is considered passed only if the first arrival time difference is no greater than a preset time threshold, the second arrival time difference is no greater than a preset time threshold, the first waveform amplitude correlation coefficient is no less than a preset similarity threshold, the second waveform amplitude correlation coefficient is no less than a preset similarity threshold, and the third waveform amplitude correlation coefficient is no less than a preset similarity threshold. Otherwise, the multi-signal homology conformance check is considered failed. This comprehensive time-domain and amplitude correlation analysis effectively improves the accuracy of identifying true partial discharge events and reduces the false alarm rate.

[0080] In a preferred embodiment, the signal integration processing module 4, before performing the multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, further includes:

[0081] The first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal are aligned and denoised to generate an updated first high-frequency pulse current signal, an updated first pulse light signal and an updated first ultrasonic signal.

[0082] The signal integration processing module 4 will also pre-process these original signals, including alignment processing and noise reduction processing, before performing multi-signal homology compliance inspection on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, so as to generate an updated first high-frequency pulse current signal, an updated first pulse light signal and an updated first ultrasonic signal.

[0083] Alignment processing aims to eliminate time deviations between channel signals that may be caused by differences in propagation paths or acquisition systems, ensuring precise synchronization of signals from different modalities on the time axis. This alignment process can be categorized as pre-alignment or post-alignment.

[0084] Pre-alignment is primarily ensured through hardware-level synchronization mechanisms. Specifically, the signal integration and processing module 4 uses a unified high-precision clock source to ensure synchronized sampling of signals across all channels. For example, synchronized sampling at rates exceeding 10 MS / s is achieved using the same 10 ns voltage-stabilized crystal oscillator to ensure that the timestamp error is less than the half-width of a single discharge packet. This synchronized acquisition mechanism inherently ensures the temporal correlation of multiple signals, providing a reliable time reference for subsequent homology determination.

[0085] Post-event alignment is implemented at the software processing level after the signal is triggered. After noise reduction, the signal integration processing module 4 updates the noise variance of the three signals in real time and sets a self-learning threshold accordingly. When the amplitude of any signal exceeds this threshold, it is identified as a suspected discharge trigger window. At this point, the signal integration processing module 4 latches the signals of the remaining two equal-width windows to ensure that all modal signals are time-aligned at the time of the specific event. This post-event alignment mechanism ensures that different modal signals belonging to the same partial discharge event can be accurately correlated, even with slight differences in signal propagation speeds or system response times.

[0086] Noise reduction processing uses algorithms tailored to the characteristics of each signal modality to suppress background noise and interference, thereby improving the signal-to-noise ratio. For example, for electrical channels, wavelet soft-threshold denoising can be used to effectively suppress electromagnetic interference such as power frequency harmonics. For optical channels, adaptive orthogonal demodulation can be used to offset inherent noise such as micro-light source drift. For acoustic channels, envelope detection can be used to attenuate environmental background noise such as mechanical vibration. These preprocessing steps effectively improve the accuracy of subsequent homologous conformance testing and ensure the reliability of subsequent feature extraction and discriminant analysis.

[0087] In a preferred embodiment, the high-frequency current detection module 1, the photoelectric coupling detection module 2, the acoustic emission detection module 3 and the signal integration processing module 4 are arranged sequentially along the axial direction of the cable to be tested 7;

[0088] The signal output ends of the high-frequency current detection module 1 , the photoelectric coupling detection module 2 , and the acoustic emission detection module 3 are all connected to the signal input end of the signal integration processing module 4 .

[0089] Specifically, the high-frequency current detection module 1, the photoelectric coupling detection module 2, the acoustic emission detection module 3, and the signal integration and processing module 4 are arranged sequentially along the axial direction of the cable under test 7. This axial sequential arrangement allows the individual detection modules to be integrated into the device with a predetermined spatial relationship, facilitating their installation and positioning on the cable under test 7 and facilitating subsequent signal acquisition and determination of partial discharge locations.

[0090] The signal outputs of the high-frequency current detection module 1, the photoelectric coupling detection module 2, and the acoustic emission detection module 3 are all connected to the signal inputs of the signal integration and processing module 4. This means that the raw signals collected by the three main sensing modules (high-frequency current detection module 1, photoelectric coupling detection module 2, and acoustic emission detection module 3) are all aggregated and processed by the signal integration and processing module 4. This connection ensures the efficient transmission of multimodal signals, providing the data foundation for the signal integration and processing module 4 to perform multi-signal homology conformity verification, feature extraction, and ultimately partial discharge determination, thereby achieving comprehensive perception and efficient analysis of partial discharge events.

[0091] In a preferred embodiment, the high-frequency current detection module 1 includes an openable and closable clamping structure with a built-in high-frequency current transformer coil, the openable and closable clamping structure is used to fix the high-frequency current detection module 1 to the ground loop portion of the cable to be tested 7, and the built-in high-frequency current transformer coil is used to collect the first high-frequency pulse current signal;

[0092] The photoelectric coupling detection module 2 includes an electro-optical modulator based on the Pockels effect, which is used to sense the first local pulse electric field generated on the insulation surface of the cable to be tested 7 and modulate its intensity change into an optical signal to generate the first pulse optical signal;

[0093] The acoustic emission detection module 3 includes a plurality of piezoelectric sensors arranged along the axial direction of the cable to be tested 7 , and each piezoelectric sensor is used to cooperatively receive a first ultrasonic wave signal propagating in the solid medium of the cable to be tested 7 .

[0094] Specifically, the high-frequency current detection module 1 includes an openable and closable clamping structure with a built-in high-frequency current transformer coil. The openable and closable clamping structure is used to fix the high-frequency current detection module 1 to the metal shielding layer of the cable to be tested 7 or the grounding loop portion outside the grounding wire. The built-in high-frequency current transformer coil is used to sense and collect the transient high-frequency pulse current signal generated by partial discharge. The coil is connected to the signal integration and processing module 4 via a coaxial shielded wire. This design allows the high-frequency current detection module 1 to be quickly and conveniently installed on the grounding loop of the cable to be tested 7, and effectively capture the high-frequency electrical signal generated by partial discharge.

[0095] The photoelectric coupling detection module 2 includes an electro-optical modulator based on the Pockels effect. This electro-optical modulator senses a first local pulsed electric field generated on the insulation surface of the cable under test 7 and modulates its intensity variation into an optical signal, thereby generating the first pulsed optical signal. This electro-optical modulator utilizes a lithium niobate crystal modulation structure, whose refractive index undergoes linear modulation due to changes in the electric field amplitude, thereby instantly mapping the electrical signal to a jump in light intensity or phase. The photoelectric coupling detection module 2 may also be composed of a miniature light source, an optical fiber transceiver assembly, and a reflector. After photoelectric conversion, the optical signal is transmitted back to the signal integration and processing module 4, providing high-voltage electric field detection capabilities under electromagnetic isolation.

[0096] The acoustic emission detection module 3 includes several piezoelectric sensors arranged axially along the cable under test 7. These piezoelectric sensors are used to collaboratively receive a first ultrasonic signal propagating through the solid medium of the cable under test 7. When a partial discharge occurs, it excites ultrasonic waves in the form of Lamb wave packets. These piezoelectric sensors, such as PVDF piezoelectric films or piezoelectric ceramic sensors, are able to convert the received ultrasonic signals into voltage signals. The layout of the acoustic emission detection module 3 and the collaborative operation of multiple sensors ensure the effective capture of the acoustic signals associated with partial discharge, providing important acoustic information for subsequent precise diagnosis and spatial positioning.

[0097] In a preferred embodiment, the construction of the classification discriminator includes:

[0098] Acquire a partial discharge feature sample set; wherein the partial discharge feature sample set includes a plurality of combined feature vectors and corresponding partial discharge labels; the combined feature vector includes at least one feature extracted from the amplitude, rising edge, main frequency, bandwidth, and wavelet energy ratio of the signal;

[0099] Randomly dividing the partial discharge feature sample set into a number of training subsample sets according to a preset batch size;

[0100] For each training sub-sample set, generate a decision tree corresponding to the current sub-sample set according to the current sub-sample set;

[0101] The decision trees corresponding to the training sub-sample sets are combined to generate the classification discriminator.

[0102] In this embodiment, the construction process of the preset classification discriminator first requires obtaining a local discharge feature sample library. This sample library is constructed in advance in a controlled experimental environment by simulating and collecting electrical, optical, and acoustic signals of various known typical local discharges (such as tip discharge, surface discharge, or suspended discharge), extracting combined feature vectors from them, and then associating each set of feature vectors with its known local discharge label. The combined feature vector may include at least one item, such as the amplitude representing the signal energy, the rising edge representing the steepness of the signal, the main frequency and bandwidth representing the frequency domain distribution of the signal, and the wavelet energy ratio representing the time-frequency energy distribution of the signal. After obtaining the sample library, in order to construct a robust classification discriminator, an ensemble learning strategy is adopted, that is, according to a preset batch size, the sample library is divided into several different training sub-sample sets by random sampling.

[0103] Next, for each training sub-sample set, a corresponding decision tree is independently generated. Specifically, the process of generating a decision tree is a top-down, recursive node splitting process. The process starts with a root node that contains all the samples in the current training sub-sample set. For any node that needs to be split, the algorithm will traverse each feature in the combined feature vector contained in the node, and select an optimal splitting feature and splitting point based on a preset partitioning standard that aims to maximize the purity of the node and make the local discharge labels in the child nodes after the split as single as possible. Subsequently, based on the optimal splitting feature, the samples in the current node are divided into the newly generated child nodes. This splitting process will be repeated on the new child nodes until the node meets the preset termination condition, such as reaching the preset tree depth. At this time, the node will no longer split and become a leaf node representing the final classification result.

[0104] Finally, all the decision trees independently generated for all training subsample sets are combined to form a random forest model containing multiple decision trees. This model is the preset classification discriminator described in this embodiment. By integrating multiple different decision trees and making a comprehensive judgment, the overfitting problem that may exist in a single decision tree can be effectively avoided, making the final classification discriminator more stable and reliable when facing unknown signals.

[0105] In a preferred embodiment, the signal integration processing module 4 is further configured to obtain the axial distance between the high-frequency current detection module 1 and the acoustic emission detection module 3 and the propagation speed of the acoustic wave in the cable 7 under test when it is determined that partial discharge exists in the cable 7 under test;

[0106] Calculating the absolute difference between the arrival time of the peak of the first high-frequency pulse current signal and the arrival time of the peak of the first ultrasonic signal to generate an arrival time difference between the first high-frequency pulse current signal and the first ultrasonic signal;

[0107] The axial position where the partial discharge occurs in the cable 7 to be tested is determined according to the arrival time difference, the axial distance, and the propagation speed.

[0108] Specifically, when the signal integration and processing module 4 determines that partial discharge (PD) exists in the cable under test 7, it initiates an internal discharge source location program. This location function is enabled by the core physical principle that the electrical and acoustic signals generated simultaneously by a partial discharge event have vastly different propagation speeds within the cable medium. The first high-frequency pulse current signal propagates at nearly the speed of light, and its arrival at the detection module can be considered the instantaneous benchmark for the event. The first ultrasonic signal, a mechanical wave, propagates much more slowly within the cable's solid dielectric medium. Therefore, the time difference between the two signals arriving at the detection device directly carries information related to the distance between the discharge point and the sensor.

[0109] When performing the positioning calculation, the signal integration processing module 4 first retrieves two preset key parameters from its internal memory: the known fixed axial distance between the high-frequency current detection module 1 and the acoustic emission detection module 3, and the acoustic wave propagation velocity, pre-calibrated or set for the model of the cable 7 under test. Simultaneously, the processing module accurately calculates the arrival times of the peak amplitudes of the collected first high-frequency pulse current signal and the first ultrasonic signal, and finds the absolute difference between these two times, which is the arrival time difference. Finally, the processing module substitutes this real-time arrival time difference, along with the two preset parameters (sensor axial distance and acoustic wave propagation velocity), into a preset positioning algorithm model to determine the specific axial position of the partial discharge event on the cable.

[0110] The core positioning algorithm model used by this device is based on the Time Difference of Arrival (TDOA) method. The underlying physical principle of this method is that a localized discharge event simultaneously stimulates multiple physical signals propagating at different speeds at its point of occurrence. This device exploits the significant speed difference between high-frequency pulsed current signals (electromagnetic waves that propagate at nearly the speed of light) and ultrasonic signals (mechanical waves that propagate through the cable medium at the speed of sound). While electrical signals reach the detection device almost instantaneously, acoustic signals require a considerable amount of time to arrive. The magnitude of this time difference directly reflects the distance between the discharge point and the acoustic sensor.

[0111] The algorithm model establishes a mathematical relationship involving the discharge position, sensor position, signal propagation speed, and arrival time difference. In a simplified one-dimensional coordinate system along the cable axis, this mathematical relationship can be expressed as:

[0112]

[0113] Where τ is the arrival time difference calculated by the signal integration processing module 4 by measuring the peak moments of the first high-frequency pulse current signal and the first ultrasonic signal. This is the core real-time input of the algorithm. x is the axial position coordinate of the partial discharge event to be solved. This is the final output of the algorithm. a is the axial position coordinate of the acoustic emission detection module 3 on the support base 5. This is a known, preset structural parameter. e is the axial position coordinate of the high-frequency current detection module 1 on the support base 5. This is also a known, preset structural parameter. a The propagation speed of the ultrasonic signal in the solid medium of the cable under test. This is a parameter that is pre-calibrated or set according to the cable model and material. e The propagation speed of the high-frequency pulse current signal in the cable ground loop (close to the speed of light). This is also a preset parameter.

[0114] Through this positioning method that combines two different modal signals, the device can non-invasively achieve high-precision tracing of partial discharge fault points. Its positioning results can be used to guide subsequent precise troubleshooting and maintenance work, greatly improving the efficiency and accuracy of fault diagnosis.

[0115] In a preferred embodiment, the signal integration processing module 4 is further configured to store the combined feature vector and the corresponding classification result each time the partial discharge is determined to exist in the tested cable 7 into a database;

[0116] After each preset time period, the combined feature vectors and their corresponding classification results in the database are added to the partial discharge feature sample set, and the combined feature vectors and their corresponding classification results in the database are cleared; and the classification discriminator is incrementally trained based on the partial discharge feature sample set.

[0117] The detection device in this embodiment not only possesses factory-set detection capabilities but also possesses the ability to self-learn and adaptively optimize during long-term operation. To achieve this, the signal integration and processing module 4 incorporates a feedback mechanism for recording and learning. Specifically, during field operation, each time the device successfully identifies a true partial discharge event, the signal integration and processing module 4 automatically stores the combined feature vector of that event, along with the specific classification result (i.e., partial discharge label) from the classification discriminator, as a successful "experience case" in its internal local database (e.g., a SQLite database) for accumulation.

[0118] In order to make use of these newly accumulated experiences, the device will automatically perform incremental training tasks at a preset time period (for example, once a day). When each training cycle arrives, the signal integration processing module 4 will extract all the new "experience cases" temporarily stored in the aforementioned database and add them to the more permanent "partial discharge feature sample set" used when building the classification discriminator, thereby realizing the dynamic expansion of the core knowledge base. After the data merging is completed, the temporary database will be cleared to prepare for the next cycle of experience collection. Subsequently, the signal integration processing module 4 will call this expanded feature sample set that contains more actual on-site cases to conduct a new round of training for the preset classification discriminator (i.e., random forest model) so that it can learn new discharge feature patterns.

[0119] Through this periodic incremental training, the classification and recognition capabilities of the classification discriminator will continue to evolve over time. This adaptive optimization mechanism for the classification model enables the detection device to continuously adapt to changes in field conditions during use, continuously improving its detection sensitivity and anti-interference capabilities, and ensuring the stability and reliability of long-term online monitoring.

[0120] like Figure 2 As shown, an embodiment of the present invention provides an electro-optical-acoustic coordinated partial discharge detection method, which includes at least the following steps:

[0121] Step S1 : obtaining a first high-frequency pulse current signal of a ground loop portion of a cable 7 under test, a first pulse light signal on an insulating surface of the cable 7 under test, and a first ultrasonic signal propagating in a solid medium of the cable 7 under test.

[0122] Step S2, performing a multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, extracting a combined feature vector from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and inputting the combined feature vector into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test 7 according to the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test 7; otherwise, it is determined that no partial discharge exists in the cable under test 7.

[0123] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0124] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. An electro-optical-acoustic coordinated partial discharge detection device, characterized in that: include: High-frequency current detection module, photoelectric coupling detection module, acoustic emission detection module, signal integration processing module and supporting base; The support base is used to carry the high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module, and is installed on the cable to be tested; The high-frequency current detection module is used to collect a first high-frequency pulse current signal of the ground loop portion of the cable to be tested; The photoelectric coupling detection module is used to collect a first local pulse electric field on the insulation surface of the cable to be tested, and modulate the first local pulse electric field into an optical signal to generate a first pulse optical signal; The acoustic emission detection module is used to collect the first ultrasonic signal propagating in the solid medium of the cable to be tested; The signal integration processing module is used to perform a multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, a combined feature vector is extracted from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and the combined feature vector is input into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test based on the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test; otherwise, it is determined that no partial discharge exists in the cable under test.

2. The electro-optical-acoustic coordinated partial discharge detection device according to claim 1, characterized in that: Also includes: protective housing; The protective shell covers the outside of the supporting base, the high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module.

3. The electro-optical-acoustic coordinated partial discharge detection device according to claim 2, characterized in that: The performing multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal includes: Based on a preset sampling frequency, sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal are respectively extracted; Recording the moment corresponding to the amplitude peak in each sampling data of the first high-frequency pulse current signal as the first peak moment; Recording the moment corresponding to the amplitude peak in each sampling data of the first pulse light signal as the second peak moment; Recording the moment corresponding to the amplitude peak in each sampling data of the first ultrasonic signal as the third peak moment; Calculate the absolute difference between the first peak time and the third peak time to generate a first arrival time difference; Calculating the absolute difference between the second peak time and the third peak time to generate a second arrival time difference; Calculating, based on the sampling data of each sampling point of the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, a first waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first pulse light signal, a second waveform amplitude cross-correlation coefficient between the first high-frequency pulse current signal and the first ultrasonic signal, and a third waveform amplitude cross-correlation coefficient between the first pulse light signal and the first ultrasonic signal; If the first arrival time difference is not greater than the preset time threshold, the second arrival time difference is not greater than the preset time threshold, the first waveform amplitude correlation coefficient is not less than the preset similarity threshold, the second waveform amplitude correlation coefficient is not less than the preset similarity threshold, and the third waveform amplitude correlation coefficient is not less than the preset similarity threshold, then the multi-signal homology conformity test is determined to have passed; otherwise, the multi-signal homology conformity test is determined to have failed.

4. The electro-optical-acoustic coordinated partial discharge detection device according to claim 3, characterized in that: The signal integration processing module, before performing the multi-signal homology conformity test on the first high-frequency pulse current signal, the first pulse light signal, and the first ultrasonic signal, further includes: The first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal are aligned and denoised to generate an updated first high-frequency pulse current signal, an updated first pulse light signal and an updated first ultrasonic signal.

5. The electro-optical-acoustic coordinated partial discharge detection device according to claim 4, characterized in that: The high-frequency current detection module, the photoelectric coupling detection module, the acoustic emission detection module and the signal integration processing module are arranged sequentially along the axial direction of the cable to be tested; The signal output ends of the high-frequency current detection module, the photoelectric coupling detection module, and the acoustic emission detection module are all connected to the signal input end of the signal integration processing module.

6. The electro-optical-acoustic coordinated partial discharge detection device according to claim 5, characterized in that: The high-frequency current detection module includes an openable and closable clamping structure with a built-in high-frequency current transformer coil, the openable and closable clamping structure is used to fix the high-frequency current detection module to the ground loop portion of the cable to be tested, and the built-in high-frequency current transformer coil is used to collect the first high-frequency pulse current signal; The photoelectric coupling detection module includes an electro-optical modulator based on the Pockels effect, which is used to sense a first local pulse electric field generated on the insulation surface of the cable to be tested and modulate its intensity change into an optical signal to generate the first pulse optical signal; The acoustic emission detection module includes a plurality of piezoelectric sensors arranged along the axial direction of the cable to be tested, and each piezoelectric sensor is used to collaboratively receive a first ultrasonic wave signal propagating in a solid medium of the cable to be tested.

7. The electro-optical-acoustic coordinated partial discharge detection device according to claim 6, characterized in that: The construction of the classification discriminator includes: Acquire a partial discharge feature sample set; wherein the partial discharge feature sample set includes a plurality of combined feature vectors and corresponding partial discharge labels; the combined feature vector includes at least one feature extracted from the amplitude, rising edge, main frequency, bandwidth, and wavelet energy ratio of the signal; Randomly dividing the partial discharge feature sample set into a number of training subsample sets according to a preset batch size; For each training sub-sample set, generate a decision tree corresponding to the current sub-sample set according to the current sub-sample set; The decision trees corresponding to the training sub-sample sets are combined to generate the classification discriminator.

8. The electro-optical-acoustic coordinated partial discharge detection device according to claim 7, characterized in that: The signal integration processing module is further configured to obtain the axial distance between the high-frequency current detection module and the acoustic emission detection module and the propagation speed of the acoustic wave in the cable under test when it is determined that partial discharge exists in the cable under test; Calculating the absolute difference between the arrival time of the peak of the first high-frequency pulse current signal and the arrival time of the peak of the first ultrasonic signal to generate an arrival time difference between the first high-frequency pulse current signal and the first ultrasonic signal; An axial position where partial discharge occurs in the cable to be tested is determined according to the arrival time difference, the axial distance, and the propagation speed.

9. The electro-optical-acoustic coordinated partial discharge detection device according to claim 8, characterized in that: The signal integration processing module is further used to store the combined feature vector and the corresponding classification result of each determination of the presence of partial discharge in the tested cable in the database; After each preset time period, the combined feature vectors and their corresponding classification results in the database are added to the partial discharge feature sample set, and the combined feature vectors and their corresponding classification results in the database are cleared; and the classification discriminator is incrementally trained based on the partial discharge feature sample set.

10. An electro-optical-acoustic coordinated partial discharge detection method, characterized in that: A signal integration processing module for a partial discharge detection device using electro-optical-acoustic coordination, comprising: Acquire a first high-frequency pulse current signal of a ground loop portion of the cable under test, a first pulse light signal on an insulation surface of the cable under test, and a first ultrasonic signal propagating in a solid medium of the cable under test; A multi-signal homology conformity test is performed on the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal; if the multi-signal homology conformity test passes, a combined feature vector is extracted from the first high-frequency pulse current signal, the first pulse light signal and the first ultrasonic signal, and the combined feature vector is input into a preset classification discriminator, so that the classification discriminator outputs a probability representing the presence of partial discharge in the cable under test based on the combined feature vector; if the probability of partial discharge is greater than a preset discharge threshold, it is determined that partial discharge exists in the cable under test; otherwise, it is determined that no partial discharge exists in the cable under test.

Images