Pulse width modulation optical time domain reflection detection device for discharge pressure waves in transformer oil

Abstract

The invention provides a pulse width modulation optical time domain reflection detection device for discharge pressure waves in transformer oil. Based on an optical path of a heterodyne coherent optical time domain reflectometer, three time sequence pulses are adopted to modulate continuous optical signals to enter an optical fiber, backward Rayleigh scattering signals are generated, the optical fiber is placed in transformer oil to measure pressure waves generated by partial discharge, and the pressure waves are measured through the mathematical relationship between the pressure waves and the backward Rayleigh scattering signals. And through wavelet denoising, least square smoothing and Hilbert demodulation phase extraction, the measurement of the partial discharge pressure waves in the transformer is finally realized.

Description

Technical Field

[0001] This invention belongs to the field of ultra-high voltage transformers, and particularly relates to a pulse width modulation optical time domain reflectance detection device for discharge pressure waves in transformer oil. Background Technology

[0002] Large oil-filled equipment such as ultra-high voltage transformers are key energy conversion hubs in ultra-high voltage AC and DC power transmission systems. Their safe and stable operation is crucial for energy security and reliable power supply.

[0003] Currently, the main protections for oil-immersed transformers include instantaneous overcurrent protection, differential protection, and heavy gas protection. The identification of these protections often requires approximately 10 ms. Furthermore, these protective devices have relatively low sensitivity, especially when insulation faults occur between winding turns. Severe transformer faults often originate from discharge defects in the internal insulation. To prevent transformer explosions, it is urgently necessary to conduct sensitive, effective, and reliable detection of discharge signals in the transformer oil. This would allow for protective measures to be taken during the initial stages of discharge, before a severe arc fault occurs, or even at the instant a severe arc fault occurs, thereby preventing equipment damage and reducing economic losses.

[0004] Compared to the aforementioned detection methods, the vibration signals generated by discharge in oil are extremely noticeable. Therefore, using distributed optical fiber technology for vibration monitoring can achieve real-time measurement and identification of arc faults. Currently, a few patents describe optical fiber sensing detection systems utilizing heat or vibration, primarily employing Brillouin and Rayleigh scattering principles. For example, CN201911175481 – A Full-Fiber Optic Online Temperature Monitoring System and Method for Submarine Cables. These patents mainly utilize a specific technology to design related monitoring systems.

[0005] Backscattering of Rayleigh light has the highest intensity among various scattered signals, thus vibration detection based on it has high accuracy. However, there are technical bottlenecks in enhancing the intensity of this scattered light, and the signal-to-noise ratio of identification methods relying solely on intensity analysis is limited. The application of phase-sensitive optical time-domain reflectometry (Φ-OTDR) technology significantly improves the system's detection sensitivity and spatial resolution. However, this technology faces random signal fluctuations caused by coherent fading effects, requiring extensive statistical processing to effectively identify continuous vibration signals. Given the transient nature of arc discharge processes, conventional methods capture a limited number of vibration coupling data samples, leading to a decrease in state identification efficiency.

[0006] Currently, there are very few inventions related to the precise monitoring of discharge in transformer oil. Similar patents mainly focus on monitoring using external methods such as cables, and there are no application methods for precise detection placed inside the transformer. Current similar vibration monitoring methods primarily employ single Brillouin, Raman, and Rayleigh scattering principles, and there are no related technologies that integrate multiple principles.

[0007] Currently, technologies based on Brillouin scattering are mainly used for vibration signal detection. However, the intensity of Brillouin scattered light is extremely low, making effective information capture significantly difficult. This signal is also subject to interference from multiple physical fields such as stress and temperature, and requires a highly coherent light source and a precise demodulation device, resulting in the highest system implementation cost. Current patented technologies based on Raman scattering primarily focus on temperature parameter monitoring. This technology struggles to effectively characterize arc discharge phenomena in oil through temperature variations. Furthermore, the system requires a highly coherent light source and a precise demodulation unit, significantly increasing structural complexity and implementation cost compared to conventional sensing solutions.

[0008] Current patents utilize Rayleigh scattering as a mechanism for vibration parameter monitoring (i.e., optical time-domain reflectometry, OTDR). Rayleigh scattering has the largest scattering cross-section among the three types of intrinsic scattering in optical fibers, but the energy is weak in the initial stage of discharge in oil, significantly limiting the intensity of the induced vibration signal. OTDR relies on backscattered light intensity resolution, resulting in insufficient detection sensitivity. To address the weak signal characteristics, phase-sensitive optical time-domain reflectometry (Φ-OTDR) can simultaneously improve system detection sensitivity and spatial resolution by demodulating backscattered phase changes. However, coherent fading effects cause random signal fluctuations, requiring large-sample statistical analysis to identify continuous vibration modes. While some patents involve this technology, they do not incorporate adaptive optimization based on the transient characteristics of discharge in oil. Given the short duration of the discharge process, conventional methods yield a limited number of effective vibration information samples, leading to reduced state identification efficiency. Summary of the Invention

[0009] To address the above technical problems, this invention proposes to enhance signal information density and suppress random noise and interference effects by adjusting the input optical pulse width and combining it with a multi-pulse photon joint detection scheme, thereby achieving accurate detection of arc discharge inside oil-immersed transformers. The specific technical solution is as follows:

[0010] A pulse width modulated optical time-domain reflectometry detection device for discharge pressure waves in transformer oil, wherein the output end of a narrow linewidth laser is divided into a sensing arm and a reference arm via a first coupler;

[0011] In the sensing arm, the input end of the acousto-optic modulator is optically connected to the pulse light output end of the first coupler, the electric drive end of the acousto-optic modulator is cable-connected to the timing square wave output end of the function generator, the pulse light output end of the acousto-optic modulator is optically connected to the first port of the three-port circulator via an erbium-doped fiber amplifier, the second port of the three-port circulator is optically connected to the sensing fiber laid in the transformer oil, and the third port of the three-port circulator is optically connected to the first input end of the second coupler.

[0012] In the reference arm, the continuous optical output terminal of the first coupler is directly optically connected to the second input terminal of the second coupler;

[0013] The beat frequency output of the second coupler is optically connected to the optical input of the balanced photodetector, and the electrical output of the balanced photodetector is connected to the data acquisition system.

[0014] The function generator produces pulse sequence units with pulse widths of 400 ns, 200 ns, and 100 ns, spaced 50 ns apart from each other;

[0015] The backscattered Rayleigh light generated in the sensing fiber returns through the three-port circulator and beats with the continuous light of the reference arm at the second coupler. The resulting signal is converted into an electrical signal by a balanced photodetector. This electrical signal is then processed to extract the phase change caused by the discharge pressure wave in the transformer oil.

[0016] Preferably, the sensing optical fiber is connected to the second port of the three-port circulator via an FC / APC interface, and the sensing optical fiber is completely immersed in the transformer oil and closely attached to the winding surface.

[0017] Preferably, the function generator outputs 100 sets of pulse sequences.

[0018] Preferably, a delay fiber is inserted between the third port of the three-port circulator and the second coupler.

[0019] Preferably, the electrical signal processing includes: performing wavelet denoising to filter out interference and noise components, obtaining relatively smooth data, sending it to a phase meter, performing waveform statistics and feature extraction based on the correlation between light intensity and phase, and finally completing the partial discharge diagnosis.

[0020] Preferably, each set of measured waveforms is denoised using wavelet transform, and the signal phase information is demodulated using Hilbert transform. For the light intensity and phase data of a set of signals, a sliding window least squares polynomial smoothing technique is used to process the waveform: the mean of the data within the window is calculated, and the window width is 1000 ns. Subsequently, the hundred sets of data are classified, that is, the signal of the fiber core is statistically analyzed.

[0021] Preferably, the feature quantities of each window are extracted in the form of a sliding window, namely the average value, deviation, and maximum slope of the waveform, and the rate of change of each parameter relative to the adjacent windows is compared. If the relative deviation of two of the three parameters exceeds 50%, it is determined to be an abnormal window.

[0022] Preferably, after all waveforms have been processed, an analysis interval of 1 second is used to compare the abnormal time window distribution of the three sets of trajectories within 1 second. If three or more trajectories contain abnormal time windows, it is determined that partial discharge has occurred in that interval.

[0023] Preferably, the pressure wave generated by the discharge in the transformer oil causes micro-strain in the optical fiber, including: refractive index change Δn, length change ΔL, and the relationship between the optical phase shift Δφ and the vibration acceleration a is linearly proportional.

[0024] ;

[0025] Where: a is the vibration acceleration, t is the pressure wave propagation time, L is the fiber length, and n is the fiber refractive index.

[0026] Preferably, a time delay T is set between pulse sequence units, the value of which depends on the length of the sensing fiber and is greater than the round-trip time of the signal in the fiber.

[0027] The present invention has the following beneficial effects:

[0028] This invention adjusts the input pulse beam width and uses multi-pulse beam joint detection to improve signal richness while reducing the influence of signal interference and randomness, thereby enabling the detection of partial discharge inside transformers. Attached Figure Description

[0029] Figure 1 The optical path topology for pulse width modulation optical time-domain reflectometry detection of discharge pressure waves in transformer oil;

[0030] Figure 2 This is a signal timing diagram for the present invention;

[0031] Figure 3 This is a flowchart of electrical signal processing. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other. To achieve the above objectives, this invention adopts the following technical solution.

[0033] Figure 1This paper demonstrates the optical path topology for pulse-width modulated optical time-domain reflectometry (PWDR) detection of pressure waves generated during discharge in transformer oil. A narrow-linewidth laser is used as the light source to generate continuous light. The continuous light output from the laser is split into two paths via an optical coupler. One path is the sensing arm, where an acousto-optic modulator and a function generator are linked via a BNC interface. The pulsed light is modulated into pulses by three sets of time-series square waves from the acousto-optic modulator, then amplified by an erbium-doped fiber amplifier before entering the sensing fiber. The backscattered signal of the pulsed light passes through a three-port circulator, then into a 2×2 coupler, and finally into port 1 of the balanced photodetector. The other path is the reference arm, which serves as continuous reference light and enters port 2 of the balanced photodetector via a 2×2 coupler. The pulsed light and the continuous reference light form a coherent loop to detect the pressure waves commonly generated by partial discharge inside the transformer. The light in the sensing arm is modulated into time-series probe pulses with widths of 100 ns, 200 ns, and 400 ns by the acousto-optic modulator. This modulation is controlled by the function generator to output the pulsed light from the acousto-optic modulator. After being amplified by an erbium-doped laser amplifier, the pulsed light is injected into the sensing fiber as probe light through an optical circulator. The sensing fiber is laid inside a transformer, and the pulsed probe light generates backscattered Rayleigh light within the sensing fiber. The Rayleigh backscattered light is generated and output through the third port of the circulator. It combines with the local optical signal in an optical coupler to form a beat frequency signal. The beat frequency signal enters a balanced photodetector, which detects the output light of the optical circulator. Finally, the data acquisition system is connected to the balanced photodetector via a BNC interface to collect the electrical signal from the balanced photodetector. After filtering, noise reduction, and phase demodulation, the partial discharge signal is obtained. All optical fibers are connected using FC / APC interfaces.

[0034] (1) Signal processing

[0035] During detection, the incident light is split into two beams: a pulsed beam and a continuous reference beam. The two coherent beat frequency signals enter a balanced photodetector to output an electrical signal. After demodulation, the vibration signal is analyzed. The vibration signal can be considered as the superposition of all Rayleigh scattering electric fields within the pulse width. When using a narrowband light source, the coherence distance is large, and interference will occur between scattered light at different locations. The total power received by the terminal is not a simple sum of powers; its phase relationship is relatively fixed but random. The random superposition of phases causes fluctuations in the terminal's measurement trajectory, containing phase information but also introducing disturbances. Typically, hundreds to thousands of continuous detections are required to statistically determine the pattern and identify continuous vibrations. However, partial discharge is not a persistent vibration, and its location is also random. If a single-channel, single-pulse-width measurement is used, even after numerous detections, the effective information obtained is still insufficient.

[0036] See signal timing Figure 2The sequence is generated by a function generator, driving an acousto-optic modulator. The timing sequence consists of three pulses with widths of 400 ns, 200 ns, and 100 ns, spaced 50 ns apart, combined into a single optical pulse sequence unit. A time delay T is set between units, the value of which depends on the length of the sensing fiber and is greater than the round-trip time of the detection signal within the fiber. A single detection outputs 100 sets of pulse sequences.

[0037] This results in three light pulses in the optical fiber with relatively short time intervals. The discharge in the transformer oil is related to the applied AC voltage and typically occurs on the rising and falling edges with a period of 20ms. When partial discharge occurs, it can be approximated that the three signals can detect the same discharge signal. To reduce equipment costs, this invention uses only one sampling device. Two of the three sets of backscattered light signals are passed through time-delayed optical fibers, one set with a delay of the sequence interval T, and the other set with a delay of 2T.

[0038] After the above configuration, the optical signal is converted into an electrical signal by the balanced photodetector. After electrical signal processing, it is ultimately used for partial discharge determination. The main process is as follows: Figure 3 After the balanced photodetector output is acquired, wavelet denoising is performed to filter out interference and noise components, obtaining relatively smooth data, which is then sent to the phase meter. Based on the synchronous correlation between light intensity and phase, waveform statistics and feature extraction are performed to ultimately complete the partial discharge diagnosis.

[0039] The intensity-time trajectory exhibits a decay pattern, with a high initial value followed by a low subsequent value, while the phase-time trajectory shows a random fluctuation curve. When the external environment is stable, repeated measurements show that the curves largely overlap. If random disturbances occur, abnormal fluctuations appear in the trajectory. Each set of measured waveforms is denoised using wavelet denoising, resulting in relatively smooth data output. The phase information of the signal is demodulated using Hilbert transform. For the intensity and phase data of a set of signals, a sliding window least squares polynomial smoothing technique is used to process the waveform: the mean value of the data within the window is calculated, with a window width of 1000 ns. Subsequently, hundreds of sets of data are classified, i.e., the signal of the fiber core is statistically analyzed. The probe signal is obtained from this, and the characteristic quantities of each window are extracted in the form of a sliding window, namely the mean value, deviation, and maximum slope of the waveform, and the rate of change of each parameter relative to adjacent windows is compared. If the relative deviation of two of the three parameters exceeds 50%, it is determined to be an abnormal time window. After all waveforms are processed, the distribution of abnormal time windows in the three sets of trajectories within 1 second is compared, with 1 second as the analysis interval. If three or more trajectories contain abnormal time windows, it is determined that partial discharge occurs in that interval.

[0040] The pressure wave generated by the discharge in transformer oil causes micro-strain in the optical fiber (change in refractive index Δn, change in length ΔL). The optical phase shift Δφ is linearly proportional to the vibration acceleration a.

[0041] ;

[0042] Where: a: vibration acceleration (measured by the amplitude of φ-OTDR phase split, unit m / s) 2 ), t: pressure wave propagation time (located by vibration signal time delay).

Claims

1. A pulse-width modulated optical time-domain reflectometry (OTRDR) detection device for discharge pressure waves in transformer oil, characterized in that, The output of the narrow linewidth laser is divided into a sensing arm and a reference arm via a first coupler. In the sensing arm, the input end of the acousto-optic modulator is optically connected to the pulse light output end of the first coupler, the electric drive end of the acousto-optic modulator is cable-connected to the timing square wave output end of the function generator, the pulse light output end of the acousto-optic modulator is optically connected to the first port of the three-port circulator via an erbium-doped fiber amplifier, the second port of the three-port circulator is optically connected to the sensing fiber laid in the transformer oil, and the third port of the three-port circulator is optically connected to the first input end of the second coupler. In the reference arm, the continuous optical output terminal of the first coupler is directly optically connected to the second input terminal of the second coupler; The beat frequency output of the second coupler is optically connected to the optical input of the balanced photodetector, and the electrical output of the balanced photodetector is connected to the data acquisition system. The function generator produces pulse sequence units with pulse widths of 400 ns, 200 ns, and 100 ns, spaced 50 ns apart from each other; The backscattered Rayleigh light generated in the sensing fiber returns through the three-port circulator and beats with the continuous light of the reference arm at the second coupler. The resulting signal is converted into an electrical signal by a balanced photodetector. This electrical signal is then processed to extract the phase change caused by the discharge pressure wave in the transformer oil.

2. The apparatus according to claim 1, characterized in that: The sensing fiber is connected to the second port of the three-port circulator via an FC / APC interface, and the entire sensing fiber is immersed in the transformer oil and closely attached to the winding surface.

3. The apparatus according to claim 1, characterized in that: The function generator outputs 100 sets of pulse sequences.

4. The apparatus according to claim 1, characterized in that: A delay fiber is inserted between the third port of the three-port circulator and the second coupler.

5. The apparatus according to claim 1, characterized in that: The electrical signal processing includes: performing wavelet denoising to filter out interference and noise components, obtaining relatively smooth data, sending it to a phase meter, performing waveform statistics and feature extraction based on the correlation between light intensity and phase, and finally completing the partial discharge diagnosis.

6. The apparatus according to claim 5, characterized in that: Each set of measured waveforms was denoised using wavelet transform, and the phase information of the signal was demodulated using Hilbert transform. For the light intensity and phase data of a set of signals, the waveform was processed using sliding window least squares polynomial smoothing technique: the mean of the data within the window was calculated, the window width was 1000 ns, and then the hundred sets of data were classified, that is, the signal of the fiber core was statistically analyzed.

7. The apparatus according to claim 6, characterized in that: The feature quantities of each window are extracted in the form of a sliding window, namely the average value, deviation, and maximum slope of the waveform. The rate of change of each parameter relative to the adjacent window is compared. If the relative deviation of two of the three parameters exceeds 50%, it is judged as an abnormal window.

8. The apparatus according to claim 7, characterized in that: After all waveforms have been processed, the analysis interval is 1 second. The abnormal time window distribution of the three sets of trajectories within 1 second is compared. If there are three or more trajectories that contain abnormal time windows, it is determined that partial discharge has occurred in that interval.

9. The apparatus according to claim 8, characterized in that: The pressure wave generated by the discharge in transformer oil causes micro-strain in the optical fiber, including: refractive index change Δn, length change ΔL, and optical phase shift Δφ. The relationship between this shift and the vibration acceleration a is linearly proportional. ； Where: a is the vibration acceleration, t is the pressure wave propagation time, L is the fiber length, and n is the fiber refractive index.

10. The apparatus according to claim 1, characterized in that: The time delay T between pulse sequence units is set, and its value depends on the length of the sensing fiber, which is greater than the round-trip time of the signal in the fiber.

Images