In-situ detection method and system for ethyne in oil based on coated hollow core fiber enhancement

Abstract

The application provides an in-situ detection method and system for acetylene in oil based on coated hollow core fiber enhancement, relates to the technical field of optical gas sensing, and specifically captures and enriches acetylene molecules from transformer oil flowing through a selective adsorption film modified on the inner wall of a coated hollow core fiber, so that the local concentration of the acetylene molecules in the coated hollow core fiber sensing probe detection is increased by tens of times, the enriched acetylene molecules are excited by high-frequency modulated mid-infrared pump laser, the periodic photothermal effect generated is sensitively detected by in-phase transmission near-infrared detection laser in the form of phase change, the super-high sensitivity detection can be maintained, the fast response meeting the early fault warning requirement is also satisfied, differential detection is simultaneously performed on the signal generation, temperature compensation, filter adjustment and frequency adjustment are simultaneously performed, the four-stage linkage processing mechanism is utilized to perform hierarchical processing on the signal, so that the whole system can not only resist wide temperature range and strong electromagnetic environment, but also adaptively resist time-varying field interference.

Description

Technical Field

[0001] This invention relates to the field of optical gas sensing technology, specifically to a method and system for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement. Background Technology

[0002] Oil-immersed power transformers are core equipment in power systems, and their unexpected outages can severely impact the operation of the power system. Therefore, fault detection of oil-immersed power transformers is necessary. Internal faults are often accompanied by the decomposition of insulation materials, which produces characteristic gases. Acetylene is a typical marker of arc discharge and high-temperature faults. Accurate and rapid detection of dissolved acetylene gas in transformer oil is of vital importance for preventing major power accidents and ensuring the safe operation of the power grid.

[0003] Existing technologies for detecting acetylene in oil are mainly divided into two categories. One is offline chromatographic analysis, which requires taking oil samples from transformers and completing the detection in the laboratory through complex degassing, injection, and chromatographic separation steps. Although this method has high accuracy, the process is cumbersome and time-consuming (usually several hours to several days), making it impossible to achieve real-time early warning of faults. Furthermore, sampling introduces errors, resulting in untimely detection. The other category is online monitoring technology based on optical principles. Existing technologies in this category use headspace phase separation combined with infrared spectroscopy or photoacoustic spectroscopy. Although online monitoring is achieved, oil-gas separation steps are still required, and the system response time is limited by the gas diffusion equilibrium process, resulting in a lag of tens of minutes. Other studies have attempted to use oil-core photonic crystal fibers to achieve in-situ direct detection without separation. However, light transmission loss in the oil medium is high, and changes in the optical properties of the oil itself can cause serious background interference, leading to poor long-term stability and detection specificity. In addition, the transformer site environment is complex, with temperature fluctuations, mechanical vibrations, and electromagnetic interference, posing a severe challenge to the signal-to-noise ratio and long-term reliability of the optical detection system.

[0004] In summary, the existing technology for acetylene detection in oil-immersed power transformers has the following technical problems:

[0005] Problem 1: Whether it is offline chromatography or online optical detection method that requires oil-gas separation, the response speed is limited by the slow oil-gas separation or gas diffusion equilibrium process, which cannot meet the requirement of minute-level real-time early warning for rapidly developing faults such as transformer arc discharge. On the other hand, direct optical detection schemes that omit the separation step (such as oil-core optical fiber) suffer from severe degradation of detection sensitivity and signal-to-noise ratio due to strong background interference and transmission loss of the oil medium. It is impossible to balance detection response speed with detection sensitivity and anti-interference ability, thus resulting in the inability to meet the requirements for detection sensitivity and response speed.

[0006] Problem 2: Transformer operation sites are subject to a variety of complex interference factors, including continuous temperature cycles, mechanical vibrations from fans and magnetostriction, strong electromagnetic fields, and background changes caused by oil aging. Existing online monitoring devices often lack sufficient environmental robustness in their sensors (such as micro-nano optical fibers and spatial optical paths) or signal processing methods (such as single wavelength and fixed frequency). Their output signals are easily affected by the above-mentioned interferences, resulting in drift or false alarms. This leads to poor long-term monitoring reliability, requiring frequent manual calibration, and makes it difficult to meet the application requirements of unattended and highly reliable smart grids. Summary of the Invention

[0007] To achieve the above objectives, the present invention provides the following technical solution: an in-situ detection method for acetylene in oil based on coated hollow optical fiber reinforcement, the method comprising:

[0008] The coated hollow fiber optic sensing probe was deployed in situ in the oil circuit of the transformer under test to perform system initialization.

[0009] Dual-wavelength differential detection is performed based on the modulated pump laser and probe laser, and the original differential signal is obtained based on the dual-wavelength differential detection. The original differential signal is the difference between the amplitude of the optical signal collected at the measurement wavelength and the reference wavelength.

[0010] The temperature and pressure data of the environment in which the coated hollow fiber sensing probe is located are acquired synchronously, and the original differential signal is compensated for in real time by using a compensation function model to obtain the compensated differential signal.

[0011] Adaptive noise suppression and filtering are performed on the compensated differential signal to obtain the denoised differential signal;

[0012] The filtered differential signal is substituted into the pre-stored concentration calibration curve model to invert and output the acetylene concentration value in the transformer oil circuit in real time.

[0013] Furthermore, the step of performing dual-wavelength differential detection based on the modulated pump laser and probe laser, and obtaining the original differential signal based on the dual-wavelength differential detection, includes:

[0014] Based on the pump laser and probe laser generated by the laser, dual-wavelength differential detection is performed after the pump laser and probe laser are modulated.

[0015] The output wavelength of the pump laser is controlled to periodically switch between a pre-calibrated acetylene characteristic absorption measurement wavelength and a non-absorption reference wavelength;

[0016] During each output wavelength switching cycle, the amplitude of the interference signal generated by the probe after the laser passes through the coated hollow fiber sensing probe at the measurement wavelength and the reference wavelength is collected respectively, and is related to the modulation frequency of the pump laser. These are recorded as the first measurement signal and the second measurement signal.

[0017] The difference between the first measurement signal and the second measurement signal is calculated to obtain the original differential signal.

[0018] Furthermore, the generation of the first measurement signal and the second measurement signal includes:

[0019] The second harmonic amplitude of the interference signal is extracted based on the second harmonic of the modulation frequency of the pump laser, using the lock-in amplifier as the reference frequency.

[0020] The second harmonic amplitude extracted at the measurement wavelength is recorded as the first amplitude, and a first measurement signal is generated based on the first amplitude;

[0021] The second harmonic amplitude extracted at the reference wavelength is recorded as the second amplitude, and a second measurement signal is generated based on the second amplitude.

[0022] Furthermore, the real-time compensation of temperature and pressure for the original differential signal using a compensation function model to obtain the compensated differential signal includes:

[0023] At the same time point as acquiring the raw differential signal, the temperature and pressure data of the current environment are read from the temperature sensor and pressure sensor integrated on the coated hollow fiber sensing probe.

[0024] The pre-stored compensation function model is invoked, and the temperature and pressure data are input into the compensation function model to calculate the first influencing factor under the current environment.

[0025] Obtain predefined standard temperature and standard pressure data and input them into the compensation function model to calculate the second influencing factor under standard conditions;

[0026] The original differential signal is calculated based on the first influence factor and the second influence factor to obtain the compensated differential signal.

[0027] Furthermore, the compensation function model characterizes the coupling influence factors of temperature and pressure on the signal response, and the compensation function model is a multivariate linear function based on temperature data and pressure data as variables.

[0028] The original differential signal is multiplied by the ratio of the second influence factor to the first influence factor to obtain the compensated differential signal.

[0029] Furthermore, the adaptive noise suppression and filtering process performed on the compensated differential signal to obtain the denoised differential signal includes:

[0030] During system operation, the output wavelength of the pump laser is periodically switched to a background wavelength far from the acetylene absorption line, and a background noise signal of a preset duration is collected based on the background wavelength.

[0031] Perform spectral analysis on the background noise signal to obtain the real-time noise spectrum, and determine whether there are interference peaks in the real-time noise spectrum with amplitudes exceeding a preset amplitude threshold within a preset frequency range near the current pump laser modulation frequency.

[0032] If there are interference spectral peaks exceeding the preset amplitude threshold, the modulation frequency of the pump laser will be automatically adjusted to the backup modulation frequency; otherwise, the modulation frequency of the pump laser will remain unchanged.

[0033] Based on the adjusted modulation frequency of the pump laser, the continuous values ​​of the compensated differential signal are used as time-series data and input into an adaptive digital filter for real-time filtering, outputting a noise-filtered differential signal.

[0034] Furthermore, the step of using the continuous values ​​of the compensated differential signal as time-series data and inputting them into an adaptive digital filter for real-time filtering includes:

[0035] Based on an adaptive digital filter, the time-series data of the compensated differential signal is monitored in real time, and the real-time noise statistical characteristics of the time-series signal are estimated.

[0036] Based on real-time noise statistics, the adaptive filter dynamically adjusts its internal filter gain parameters.

[0037] By using the adjusted filter gain parameters, iterative prediction and correction are performed on the random pulse interference and measurement noise in the time-series data of the compensated differential signal, thus completing the real-time filtering process.

[0038] Furthermore, the real-time inversion and output of the acetylene concentration value in the transformer oil circuit includes:

[0039] Obtain the pre-stored concentration calibration curve model, and use the current value of the noise-filtered differential signal obtained in real time as the input variable to substitute into the concentration calibration curve model.

[0040] The calculation is performed based on the mathematical relationship defined by the concentration calibration curve model, and the output is the acetylene concentration value in the oil circuit corresponding to the current value of the noise-filtered differential signal.

[0041] Furthermore, after the acetylene concentration value is output, it also includes:

[0042] Set a first concentration threshold and a second concentration threshold, wherein the second concentration threshold is greater than the first concentration threshold;

[0043] The acetylene concentration value is compared with a first concentration threshold and a second concentration threshold;

[0044] If the acetylene concentration exceeds the first concentration threshold but does not exceed the second concentration threshold, a level one warning signal is triggered.

[0045] If the acetylene concentration exceeds the second concentration threshold, a level two warning signal will be triggered.

[0046] An in-situ acetylene detection system for oil based on coated hollow optical fiber reinforcement, the system comprising:

[0047] The coated hollow fiber sensing probe module includes a section of anti-resonant hollow fiber. The inner wall of the anti-resonant hollow fiber is modified with a thin film that selectively adsorbs acetylene molecules. A microfluidic channel is provided on the side. The anti-resonant hollow fiber is encapsulated in a metal capillary tube with a transparent window for in-situ deployment in the transformer oil to be tested.

[0048] The dual laser modulation and optical coupling module includes a wavelength-tunable pump laser, a probe laser, a laser modulator, and an optical fiber combiner and coupling assembly, used to generate modulated pump laser and probe laser and couple them to a coated hollow fiber sensing probe module.

[0049] The photothermal interference signal demodulation module includes a fiber optic interferometer, a balanced photodetector, a lock-in amplifier, and an embedded processing unit. Based on the embedded processing unit, it performs the process of dual-wavelength differential detection, generating a compensated differential signal, generating a noise-filtered differential signal, and outputting the acetylene concentration value.

[0050] This invention provides a method and system for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement. It has the following beneficial effects:

[0051] 1. This invention employs a thin film on the inner wall of a coated hollow fiber to selectively adsorb acetylene, deeply integrating it with photothermal frequency modulation interferometry detection. This fundamentally resolves the contradiction between response speed and sensitivity, achieving minute-level high-sensitivity in-situ detection without oil-gas separation. By selectively adsorbing the film on the inner wall of the coated hollow fiber, acetylene molecules are specifically captured and enriched from the flowing transformer oil, increasing their local concentration within the coated hollow fiber sensing probe by tens of times. This enhances the intensity of the optical interaction signal, laying the foundation for high sensitivity. This adsorption process occurs rapidly and directly in the oil phase, completely eliminating the time-consuming oil-gas separation step. Simultaneously, a high-frequency modulated mid-infrared pump laser excites the enriched acetylene molecules, and the resulting periodic photothermal effect is sensitively detected by a co-transmitted near-infrared detection laser in the form of phase changes. The synergistic mechanism of adsorption, excitation, and detection inherently provides a rapid response, achieving a rapid response that meets the requirements for early fault warning while maintaining ultra-high sensitivity, thus breaking through the inherent performance bottlenecks of existing technologies.

[0052] 2. This invention employs a hierarchical intelligent adaptive signal processing chain, enhancing the system's long-term stability and anti-interference capabilities in complex industrial environments. At the signal generation level, dual-wavelength differential detection is used to deduct the background absorption of the oil medium and the common fluctuation noise of optical fibers in real time. At the signal processing level, a real-time coupled compensation model for temperature and pressure is established to dynamically correct the combined effects of environmental temperature and pressure changes on gas adsorption balance and optical fiber thermo-optic effects, fundamentally suppressing the most significant signal drift source. Simultaneously, it possesses spectrum sensing and dynamic frequency avoidance capabilities. By periodically analyzing the environmental noise spectrum and automatically adjusting the pump laser modulation frequency to an interference-free band, it actively avoids inherent periodic vibration interference in the field. Finally, an adaptive digital filter performs real-time optimal signal estimation to smooth random noise. Utilizing a four-level linkage processing mechanism for hierarchical signal processing, the entire system can not only withstand wide temperature ranges and strong electromagnetic environments but also adaptively resist time-varying field interference. After long-term operation testing, it can achieve stable continuous monitoring for months without manual intervention, with an extremely low false alarm rate and high accuracy, providing a solid data foundation for reliable transformer condition assessment and intelligent diagnosis. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the steps of the in-situ detection method for acetylene in oil based on coated hollow optical fiber reinforcement according to the present invention.

[0054] Figure 2 This is a data transmission relationship diagram of the in-situ detection method of acetylene in oil based on coated hollow optical fiber reinforcement according to the present invention;

[0055] Figure 3 This is a schematic diagram of the architecture of the in-situ acetylene detection system in oil based on coated hollow optical fiber reinforcement according to the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] like Figures 1 to 2 As shown, an in-situ detection method for acetylene in oil based on coated hollow optical fiber reinforcement includes the following steps:

[0058] Step S100: Deploy the coated hollow fiber optic sensing probe in situ in the oil circuit of the transformer under test and perform system initialization;

[0059] The coated hollow fiber sensing probe includes a section of anti-resonant hollow fiber with its inner wall modified with a polymer film that specifically selectively adsorbs acetylene molecules. This film is used to enrich acetylene from the flowing transformer oil and increase the local concentration. The anti-resonant hollow fiber has microfluidic channels on its side, allowing the transformer oil to flow continuously and fully contact the film. The entire anti-resonant hollow fiber structure is encapsulated in a metal capillary with an optical window, which provides both mechanical protection and allows the pump laser and probe laser to couple into the anti-resonant hollow fiber through the window. When deployed in situ, the coated hollow fiber sensing probe module of the system can be directly connected in series or parallel to the sampling oil circuit or online monitoring bypass of the transformer through a standard mechanical interface (such as a Swagelok connector) on its metal shell. This allows the coated hollow fiber sensing probe to be immersed in the oil flow without the need for offline extraction or degassing of the oil sample.

[0060] During system initialization, the pump laser and probe laser in the dual laser modulation and optical coupling module are activated and put into preset initial current and temperature states; the electronic units in the photothermal interference signal demodulation module are activated; oil circulation is established so that transformer oil begins to flow through the microfluidic channel of the coated hollow fiber optic sensing probe; the zero point and range of the temperature sensor and pressure sensor integrated on the coated hollow fiber optic sensing probe are read and calibrated by the embedded processing unit; finally, the background wavelength is switched to perform a background noise signal measurement, and the differential signal reference value under the current acetylene absorption-free condition is zeroed to complete the detection baseline calibration.

[0061] Step S200: Perform dual-wavelength differential detection based on the modulated pump laser and probe laser, and obtain the original differential signal based on the dual-wavelength differential detection. The original differential signal is the difference between the amplitude of the optical signal collected at the measurement wavelength and the reference wavelength.

[0062] The measurement wavelength and reference wavelength are pre-calibrated based on high-resolution absorption spectral scanning of pure acetylene gas or acetylene-containing oil samples. The spectrometer identifies the center wavelength of a strong absorption peak of acetylene molecules in the mid-infrared band, which is a specific vibration and rotation, as the measurement wavelength (e.g., around 1530.37 nm). A wavelength point with negligible absorption near this peak is selected as the reference wavelength (e.g., around 1530.80 nm). The pump laser energy is efficiently absorbed by the acetylene molecules based on the measurement wavelength, generating a photothermal effect. The reference wavelength provides a background reference for the measurement wavelength. At the reference wavelength, acetylene absorption is extremely weak or negligible, but the optical properties of the oil medium and the optical fiber itself are basically the same as at the measurement wavelength.

[0063] The amplitude of an optical signal specifically refers to the amplitude of the voltage signal output by a balanced photodetector after photoelectric conversion. The amplitude of the voltage signal directly reflects the intensity of the interference light carried by the probe laser after passing through the coated hollow optical fiber.

[0064] Step S201: Based on the pump laser and probe laser generated by the laser, dual-wavelength differential detection is performed after the pump laser and probe laser are modulated.

[0065] The pump laser is generated by a pump laser, which is a continuously outputting, wavelength-tunable mid-infrared quantum cascade laser. The center wavelength of the pump laser corresponds to the strong absorption line of the acetylene vibrational absorption band, with a tuning range of ≥2nm. The pump laser, through periodic adjustment, excites acetylene molecules adsorbed in the thin film to produce a periodic photothermal effect. The probe laser is generated by a probe laser, which is a narrow-linewidth (<1MHz) distributed feedback laser that outputs continuous light in the near-infrared band (such as around 1550nm). The probe laser acts as a phase probe, and its optical phase is modulated by the periodic refractive index change within the fiber induced by the pump laser.

[0066] The pump laser is modulated by selecting a specific frequency (usually in the range of several hundred hertz to several thousand hertz) that can effectively excite the photothermal effect and facilitate subsequent phase-locked detection. A sinusoidal wave of a specific frequency is applied to the driving current to modulate the output laser intensity according to this sinusoidal law. The probe laser itself is not modulated, but its output wavelength needs to be stabilized near the optimal operating point (i.e., the orthogonal point) of the fiber interferometer through precise temperature control to ensure that the phase change can be linearly converted into the interference intensity change. During control, the tube shell temperature of the probe laser is stabilized at a set value through an independent temperature control circuit to keep its output wavelength constant.

[0067] Step S202: The output wavelength of the pump laser is controlled to periodically switch between a pre-calibrated acetylene characteristic absorption measurement wavelength and a reference wavelength with no or weak absorption. The frequency of the periodic switching is generally a few hertz (e.g., 2-10 Hz). Real-time differential detection is achieved by periodically switching the output wavelength. By alternating the measurement of the absorption signal and the background signal at a relatively fast frequency, interference such as the slow drift of the oil medium itself and the common-mode fluctuation of the laser power can be subtracted in almost real-time during the differential process, thereby improving the stability and anti-interference capability of the system.

[0068] The output wavelength of a pump laser refers to the center wavelength of the pump laser actually output by the pump laser at any given time. It is used to control the output wavelength to switch between the measurement wavelength and the reference wavelength in order to obtain signals under conditions of acetylene absorption and no (weak) acetylene absorption.

[0069] In step S203, during each output wavelength switching cycle, the lock-in amplifier is used to collect the amplitude of the interference signal related to the modulation frequency of the pump laser generated after the laser passes through the coated hollow fiber sensing probe at the measurement wavelength and reference wavelength, respectively. The lock-in amplifier uses the second harmonic of the modulation frequency of the pump laser as the reference frequency to extract the amplitude of the second harmonic of the interference signal.

[0070] The second harmonic amplitude extracted at the measurement wavelength is recorded as the first amplitude, and a first measurement signal is generated based on the first amplitude; the second harmonic amplitude extracted at the reference wavelength is recorded as the second amplitude, and a second measurement signal is generated based on the second amplitude. The original differential signal is obtained by calculating the difference between the first measurement signal and the second measurement signal.

[0071] Among them, the lock-in amplifier is an electronic device that can extract a weak signal synchronized with a specific reference frequency from strong noise. In this embodiment, the lock-in amplifier uses the second harmonic of the modulation frequency of the pump laser as the reference frequency to accurately measure the amplitude of the component in the probe laser interference signal that is synchronized with the modulation frequency of the pump laser, that is, to extract the photothermal interference signal generated by acetylene absorption.

[0072] The interaction between the probe laser and the coated hollow fiber sensing probe is as follows: After the modulated pump laser is absorbed by acetylene, it causes periodic small temperature fluctuations in the adsorbed thin film and the surrounding area inside the coated hollow fiber. This causes the refractive index of the coated hollow fiber to change synchronously and periodically. The phase of the continuous probe laser, which is transmitted along the same path as the pump laser, is modulated by this change in refractive index. When this phase-modulated light interferes with a reference light in the fiber interferometer, the phase change is converted into a change in the intensity of the interference light, i.e., the interference signal.

[0073] The modulation frequency f of the pump laser refers to the frequency of the sinusoidal signal applied to its driving current. It provides the frequency in the time domain for the entire photothermal excitation and synchronous detection process, enabling the lock-in amplifier to selectively extract the signal at the modulation frequency of the pump laser, thereby suppressing noise at other frequencies.

[0074] The amplitude of the interference signal related to the modulation frequency of the pump laser specifically refers to the amplitude of the signal whose frequency component is equal to the modulation frequency of the pump laser or an integer multiple (harmonic) of it. Among them, the amplitude of the second harmonic is more sensitive to the nonlinear temperature change caused by acetylene absorption and can further suppress the background interference that may be caused by direct modulation of laser intensity. Therefore, it is chosen as the quantity to be measured.

[0075] The second harmonic refers to a signal component whose frequency is twice the modulation frequency of the pump laser (2f). The interference signal refers to the time-varying voltage signal output by the balanced photodetector, which contains phase modulation information. The amplitude of the second harmonic, 2f, refers to the DC voltage value output after the interference signal is processed by the lock-in amplifier and phase-sensitively detected with the 2f sine wave generated internally as a reference. When acquiring the amplitude of the second harmonic, the lock-in amplifier receives the interference signal from the balanced detector and simultaneously receives the modulation frequency of the pump laser with frequency f provided by the signal source as an external reference. It internally multiplies the frequency to 2f and demodulates the input signal with 2f as the reference frequency, finally outputting a DC voltage proportional to the amplitude of the 2f component.

[0076] At the measurement wavelength, acetylene has strong absorption and produces a significant photothermal effect, resulting in a large second harmonic amplitude (first amplitude) in the detected laser interference signal. At the reference wavelength, acetylene has extremely weak absorption and a weak photothermal effect, resulting in a very small second harmonic amplitude (second amplitude), which is close to the background noise level.

[0077] The first and second amplitudes are the DC voltage values ​​demodulated by the lock-in amplifier during the measurement wavelength period and the reference wavelength period, respectively. Therefore, the DC voltage output by the lock-in amplifier during the measurement wavelength period is directly recorded as the first measurement signal; and the DC voltage output during the immediately following reference wavelength period is directly recorded as the second measurement signal.

[0078] The original differential signal is the instantaneous difference obtained by subtracting the first measurement signal from the second measurement signal at the end of each complete wavelength switching cycle. It contains the net signal that mainly reflects the acetylene concentration information after preliminary differential processing, but has not yet compensated for the influence of ambient temperature and pressure and suppressed random noise. As the input signal for subsequent temperature and pressure compensation and adaptive filtering processing, it is the core intermediate variable for inverting the acetylene concentration.

[0079] Step S300: Simultaneously acquire temperature and pressure data of the environment where the coated hollow fiber sensing probe is located, and use the compensation function model to perform real-time temperature and pressure compensation on the original differential signal to obtain the compensated differential signal.

[0080] Step S301: At the same time point as acquiring the original differential signal, read the current environmental temperature and pressure data measured by the temperature sensor and pressure sensor integrated on the coated hollow fiber sensing probe. The temperature and pressure data are key input parameters for real-time environmental correction of the original differential signal. Temperature changes in the transformer oil circuit directly affect the adsorption equilibrium constant of acetylene molecules on the selective adsorption film within the coated hollow fiber; increased temperature leads to decreased adsorption and also changes the refractive index of the fiber material itself. Pressure changes affect the dissolution equilibrium of gases in the oil and the slight deformation of the fiber. These temperature and pressure changes couple to the final photothermal interference signal, causing the original differential signal to drift. Therefore, simultaneously acquiring temperature and pressure data facilitates the extraction of the signal component caused by pure concentration changes from the original differential signal, eliminating measurement errors caused by environmental fluctuations and ensuring the long-term accuracy of the concentration inversion results.

[0081] Step S302: Invoke the pre-stored compensation function model. The compensation function model characterizes the coupling influence factors of temperature and pressure on the signal response. The compensation function model is a multivariate linear function F based on temperature and pressure data as variables, and the formula is:

[0082] F(T, P) = 1 + α(T - T) ref )+β(P-P ref );

[0083] Where F(T, P) represents the influencing factor under specific temperature and pressure; T represents the temperature value, and P represents the pressure value; T ref and P ref These represent predefined standard temperature and standard pressure data, respectively; α is the temperature influence coefficient, in units of 1 / ℃, used to characterize the relative impact of temperature deviation of 1℃ from the reference value on the signal; β is the pressure influence coefficient, in units of 1 / kPa, used to characterize the relative impact of pressure deviation of 1kPa from the reference value on the signal. α and β were obtained by fixing the acetylene concentration under controlled laboratory conditions, systematically changing the ambient temperature and pressure, recording a series of raw differential signal values, and fitting them through multiple linear regression.

[0084] By using the temperature data T measured in real time in the current environment curr and pressure data P curr The first influencing factor F under the current environment is calculated by inputting the compensation function model into the function F(T, P). curr F curr =1 + α(T) curr -T ref )+β(P curr -P ref The first influence factor is used as a divisor to convert the signal under the current environment back to the equivalent signal under the standard environment.

[0085] Obtain predefined standard temperature data T ref and standard pressure data P ref Input the compensation function model F(T, P), and calculate the second influence factor F under standard conditions. ref F ref =1 + α(T) ref -T ref )+β(P ref -P ref =1; The second influence factor represents the theoretical influence of temperature and pressure on the signal under ideal standard reference conditions. The second influence factor is a divisor and together with the first influence factor constitutes the conversion ratio.

[0086] Among them, standard temperature data T ref and standard pressure data P ref These are the stable reference environmental conditions used during system calibration, typically defined according to the laboratory's calibration environment, for example, T. ref =25℃, P ref =101.3 kPa, T ref and P ref This is used to provide a unified benchmark point, under which all measurement signals from different environments will be converted to the benchmark conditions for comparison, thereby ensuring the effectiveness of the concentration calibration curve model and the consistency of the concentration inversion results.

[0087] Step S303: The original differential signal is recalculated based on the first influence factor and the second influence factor. The original differential signal is multiplied by the ratio of the second influence factor to the first influence factor to obtain the compensated differential signal.

[0088] The original differential signal value obtained in step S203 is denoted as S. raw The first influence factor F calculated in step S302 curr Second Influence Factor F ref Substituting 1 into the formula:

[0089] S comp =S raw *(F ref / F curr ) = S raw / F curr ;

[0090] The original differential signal is divided by the actual influence factor under the current environment, and then converted to the equivalent signal value under the standard environment, which is the compensated differential signal S. comp ;

[0091] Among them, the compensated differential signal S compIt is the original differential signal after real-time compensation and correction for ambient temperature and pressure. It has eliminated the interference introduced by environmental fluctuations to the greatest extent and more purely reflects the photothermal response caused by changes in acetylene concentration in the oil. The compensated differential signal is used as the input for subsequent adaptive noise suppression processing.

[0092] Step S400: Perform adaptive noise suppression and filtering on the compensated differential signal to obtain the denoised differential signal.

[0093] Step S401: During system operation, the output wavelength of the pump laser is periodically switched to a background wavelength far from the acetylene absorption line. Based on the lock-in amplifier, while maintaining the reference frequency at the current pump laser modulation frequency f and second harmonic, the DC voltage sequence of the original output data within a specified time period is collected based on the background wavelength. This DC voltage sequence is the background noise signal.

[0094] The background wavelength is a specific wavelength that the pump laser can be tuned to, far from any absorption line of acetylene. For example, it can be selected to be around 1540 nm. At the background wavelength, the pump laser is hardly absorbed by acetylene. Therefore, the detected interference signal only contains the system background noise and environmental interference, and does not contain acetylene concentration information. This can be used to sample and evaluate the current noise characteristics.

[0095] The background noise signal, used as a sample for analyzing real-time noise characteristics, is a time-domain voltage signal acquired at the background wavelength. It includes interference components caused by the absorption of all non-target gases, such as electronic noise from the photoelectric detection link, environmental vibration, and electromagnetic interference.

[0096] Step S402: Perform spectral analysis on the background noise signal to obtain the real-time noise spectrum, and determine whether there are interference peaks in the real-time noise spectrum with amplitudes exceeding a preset amplitude threshold within a preset frequency range near the current modulation frequency of the pump laser.

[0097] If there are interference spectral peaks exceeding the preset amplitude threshold, the modulation frequency of the pump laser will be automatically adjusted to a preset backup modulation frequency that has been confirmed by spectrum analysis to be free of strong interference; otherwise, the modulation frequency of the pump laser will remain unchanged.

[0098] The real-time noise spectrum is acquired by performing a Fast Fourier Transform (FFT) on the time-series data of the collected background noise signal to obtain a series of discrete frequency points and corresponding signal amplitude values ​​from DC to Nyquist frequency range, thus forming the real-time noise spectrum. The real-time noise spectrum is obtained by performing a Fast Fourier Transform (FFT) on the background noise signal to obtain its amplitude distribution map in the frequency domain, revealing the intensity of noise energy at each frequency point at the current moment, and intuitively showing whether there are strong interference sources that coincide with the pump laser modulation frequency or its harmonic frequencies. Moreover, the process of spectrum analysis using Fourier Transform (FFT) can be directly performed in computer software, such as using software like MATLAB or WinScope.

[0099] The preset frequency range near the current pump laser modulation frequency (e.g., f=1kHz) refers to a frequency band with a certain width extending to both sides of f (e.g., f±20Hz). The preset frequency range is usually defined based on the bandwidth of the system lock-in amplifier and the bandwidth of possible environmental interference (e.g., machine vibration fundamental frequency). It is used to focus the spectral analysis on the frequency band most relevant to signal detection, because only strong interference falling within this range will have a direct impact on the lock-in detection results.

[0100] The preset amplitude threshold is used to determine whether the spectral peaks in the noise spectrum constitute strong interference, and to distinguish weak broadband noise from strong narrowband interference that may drown out the effective signal. The preset amplitude threshold is set based on a number of times (e.g., 3-5 times) the system background noise amplitude measured in a quiet laboratory environment.

[0101] Interference peaks are spikes in the real-time noise spectrum that have an amplitude higher than the surrounding noise floor and appear within a preset frequency range. They originate from specific periodic interference sources in the field, such as transformer vibration, cooling fan rotation, or power supply harmonics. They are used to identify specific frequency points that will seriously interfere with phase-locked loop (PLL) detection, and these specific frequency points are avoided as modulation frequencies. Interference peaks indicate the presence of strong environmental noise at that specific frequency point that coincides with the frequency of the signal under test. If the modulation frequency of the pump laser happens to be set at or near this point, the PLL amplifier will be unable to distinguish between signal and noise, leading to a sharp deterioration or even failure of the measurement signal-to-noise ratio. Therefore, using the presence of interference peaks as a criterion for adjusting the modulation frequency allows the system to actively avoid the current harsh electromagnetic or vibration environment and shift the operating frequency to a relatively quiet frequency band, thereby adaptively maintaining a high signal-to-noise ratio detection and greatly improving the long-term stability and robustness of the system in complex and time-varying industrial environments.

[0102] When setting the backup modulation frequency, during the initial system installation phase, under conditions of normal transformer operation but no fault gas generation, the system is instructed to scan within a wide frequency range (e.g., 100Hz to 2kHz) in certain steps. Background noise is collected at each frequency point, and its spectrum is analyzed. Frequencies without strong interference peaks in their own harmonics and surrounding frequency bands are recorded as backup frequencies. The backup modulation frequency f backup It is a set of modulation frequency values ​​that are preset and archived during system initialization and confirmed to be free of strong interference by spectrum scanning, for example, f backup 1 = 1.1 kHz, f backup 2 = 1.5kHz; The backup modulation frequency is a frequency that can be immediately switched to when the current primary modulation frequency f is subjected to strong interference, so as to ensure that the detection can continue.

[0103] Step S403: Based on the adjusted modulation frequency of the pump laser, the continuous values ​​of the compensated differential signal are used as time-series data and input into an adaptive digital filter for real-time filtering, and the filtered differential signal is output.

[0104] An adaptive digital filter is a real-time signal processing device that can automatically adjust its internal parameters (filter gain) according to the characteristics of the input signal. In this embodiment, the adaptive digital filter includes a recursive state estimator and a gain calculation unit, used to optimally estimate the time-series data of the compensated differential signal, and filter out random measurement noise and occasional pulse interference. The formula for calculating the filtered differential signal in the adaptive digital filter is as follows:

[0105] X est (n)=X pred (n)+K(n)*[S comp (n)-X pred (n)];

[0106] Among them, X est (n) represents the optimal estimate of the filtered differential signal at time n, i.e., the final output of the adaptive digital filter; X pred (n) represents the predicted value of the signal at the current time based on the optimal estimate at the previous time step, derived from a simple state transition model X. pred (n))=X est (n-1) is calculated; S comp K(n) is the measured value of the compensated differential signal at the current time; K(n) is the filter gain parameter (Kalman gain) at the current time, and its value is between 0 and 1.

[0107] When the adaptive digital filter generates the noise-filtered differential signal, it will use the continuous discrete values ​​S of the compensated differential signal output in step S303, which are synchronously acquired at the newly adjusted pump laser modulation frequency.comp (1), S comp (2), ..., S comp (n) Input the adaptive digital filter in chronological order; the adaptive digital filter adjusts the input data points S for each new input data point. comp (n) performs its internal iterative prediction and correction calculations, and finally outputs the noise-filtered estimate X corresponding to that moment. est (n), X est The sequence of (n) is the differential signal after noise filtering.

[0108] When an adaptive digital filter performs real-time filtering, it includes:

[0109] First, the time-series data of the input compensated differential signal is monitored in real time based on an adaptive digital filter. This is achieved by analyzing the compensated differential signal S over a recent period. comp For a sequence, such as the past 100 points, calculate its deviation from the local trend line (or moving average), and take the mean of the squares of these deviations as an estimate of the current noise variance, i.e., the real-time noise statistical characteristic R(n). The real-time noise statistical characteristic mainly refers to the magnitude of measurement uncertainty in the current measurement environment, and is used to dynamically calculate the filter gain parameter K(n): when the estimated variance of the noise variance is large, the filter gain parameter should be reduced, indicating greater confidence in the predicted value; when the estimated variance of the noise variance is small, the filter gain parameter can be increased, indicating greater confidence in the new measurement value.

[0110] Then, based on the real-time noise statistical characteristics, the adaptive filter dynamically adjusts its internal filter gain parameter; according to the estimated noise variance R(n) and the system's preset model prediction error variance Q (a constant small value), the formula for calculating the filter gain parameter K(n) is as follows:

[0111] K(n)=(P pred (n)+Q) / (P pred (n)+Q+R(n)); where P pred (n) is the predicted covariance of the estimation error at the previous time step, which is updated iteratively; the filter gain parameter K(n) is a scalar coefficient between 0 and 1, which determines the new observation S during iterative correction. comp (n) for the final estimated value X est The contribution weights of (n) achieve the optimal balance between believing the prediction and believing the new measurement;

[0112] Finally, using the adjusted filter gain parameters, iterative prediction and correction are performed on the random impulse interference and measurement noise in the time-series data of the compensated differential signal to complete the filtering and output the noise-filtered differential signal. The iterative prediction and correction process is as follows:

[0113] For each new time step n, a prediction is made based on the optimal estimate X from the previous time step. est (n-1) Obtain the current prediction X pred (n)=X est (n-1);

[0114] Simultaneously predict the estimation error covariance P pred (n)=P est (n-1)+Q;

[0115] Calculate the filter gain parameter K(n), K(n) = P pred (n) / (P pred (n)+R(n));

[0116] Perform corrections and calculate the current optimal estimate X. est (n)=X pred (n)+K(n)*[S comp (n)-X pred [n]; and update the estimated error covariance P. est (n)=(1-K(n))*P pred (n);

[0117] This process repeats itself, smoothing out random noise and mitigating sudden impulse interference, manifesting as S... comp (n) and X pred The large deviation of K(n) is suppressed by the dynamic adjustment of K(n).

[0118] Among them, the noise-filtered differential signal X est (n) is the output sequence of the compensated differential signal after being smoothed by an adaptive digital filter. It suppresses random fluctuations and sudden interference to the greatest extent and presents a smooth curve that can quickly track the changes in the real concentration. As the direct input for concentration inversion, it ensures that the signal value substituted into the concentration calibration curve model is stable and reliable, thereby obtaining an accurate and non-jumping acetylene concentration value.

[0119] In step S500, the noise-filtered differential signal is substituted into the pre-stored concentration calibration curve model to invert and output the acetylene concentration value in the transformer oil circuit in real time.

[0120] Step S501: Obtain the pre-stored concentration calibration curve model. The concentration calibration curve model is established by using multiple standard acetylene oil samples with known concentrations for calibration experiments and fitting the relationship between the noise-filtered differential signal and the acetylene concentration value. The current value of the noise-filtered differential signal obtained in real time is substituted into the concentration calibration curve model as an input variable.

[0121] The concentration calibration curve model is a mathematical model, calibrated experimentally, that describes the quantitative relationship between the noise-filtered differential signal and the acetylene concentration in transformer oil. This model exhibits saturation characteristics, which depend on the adsorption limit of the thin film on the inner wall of the hollow optical fiber. It uniquely and accurately maps the optical detection signal to an acetylene concentration value. The formula for calculating the acetylene concentration value C in the concentration calibration curve model is as follows:

[0122] C = (C max *X est ) / (K d +X est );

[0123] Where C represents the acetylene concentration to be determined, in μL / L; X est It is the current amplitude of the input differential signal after noise filtering; C max These are model parameters, representing the signal plateau value corresponding to the detectable saturation concentration of the system, in μL / L, which physically corresponds to the concentration at which the adsorption film reaches saturation adsorption; K d These are model parameters, representing the signal values ​​corresponding to the half-saturation concentration, which physically reflect the adsorption strength (reciprocal of affinity) of the membrane for acetylene; K d and C max The method originates from measuring multiple standard acetylene oil samples of known concentrations in the laboratory, recording the stabilized, noise-filtered differential signal values ​​at each concentration, and then fitting the above formula with the experimental data points using a nonlinear curve. The optimally fitted parameter value is K. d and C max .

[0124] Step S502: Calculate the acetylene concentration C in the oil circuit corresponding to the current value of the differential signal after noise filtering, based on the mathematical relationship defined by the concentration calibration curve model.

[0125] Step S503: Set a first concentration threshold θ1 and a second concentration threshold θ2, where θ2 > θ1; After the acetylene concentration value C is output, compare C with θ1 and θ2.

[0126] If C≤θ1, no warning is triggered, and monitoring continues.

[0127] If θ1 < C ≤ θ2, then a level one warning signal is triggered;

[0128] If C ≥ θ2, then a level 2 warning signal is triggered.

[0129] The first concentration threshold θ1 and the second concentration threshold θ2 are two levels of alarm limits set based on transformer fault diagnosis experience and industry standards. θ1 corresponds to the attention value, indicating that the acetylene concentration has risen to a level that requires attention; θ2 corresponds to the warning value or fault threshold, indicating that the acetylene concentration is so high that a serious fault is very likely to occur. For newly commissioned or normally operating transformers, the values ​​are generally set as follows: θ1 = 1 μL / L, θ2 = 5 μL / L. The specific values ​​need to be adjusted according to the transformer model and historical data.

[0130] The Level 1 warning signal is used to instruct maintenance personnel to increase the monitoring frequency of the transformer, for example, from once a week to once a day, and to check the associated operating conditions. After the signal is issued, the system will record the event and display a yellow warning light or message on the monitoring interface.

[0131] The Level 2 warning signal is used to indicate the possible latent fault or the development of a fault, requiring immediate diagnosis and intervention. After the signal is issued, the system will record the event, display a flashing red alarm on the monitoring interface, and send alarm information to the remote monitoring center through a communication interface (such as IEC61850). It will also activate the backup cooling system or suggest arranging a power outage for maintenance. The specific actions to be performed after the Level 1 and Level 2 warning signals are issued can be set according to the actual usage.

[0132] like Figure 3 As shown, an in-situ detection system for acetylene in oil based on coated hollow fiber reinforced by an optical fiber is presented. The system includes a coated hollow fiber sensing probe module, a dual laser modulation and optical coupling module, and an optical thermal interference signal demodulation module.

[0133] The coated hollow fiber sensing probe module includes a section of anti-resonant hollow fiber. The inner wall of the anti-resonant hollow fiber is modified with a thin film that selectively adsorbs acetylene molecules, and a microfluidic channel is provided on the side.

[0134] The thin film material is typically a porous material with specific adsorption capabilities for acetylene, such as metal-organic frameworks (e.g., ZIF-8, which has a sieving effect on acetylene / ethylene) or functionalized polymers (e.g., polyimides containing acetylene-specific recognition sites). The microfluidic channels are helical or linear grooves etched into the side of the hollow optical fiber using femtosecond laser micromachining technology, communicating with the fiber core. Their width and depth are approximately tens of micrometers, used to guide transformer oil flow through the outer wall of the optical fiber and force acetylene molecules in the oil to diffuse and fully interact with the inner wall thin film. Contact; Anti-resonant hollow fiber is a photonic crystal fiber whose light guiding mechanism is based on anti-resonant reflection. It is typically set with a hollow core diameter of 50-100 micrometers. The cladding is composed of periodically arranged air holes. The wall thickness is designed to form a low-loss transmission window at the target wavelength (pump light and probe light) for low-loss transmission of mid-infrared pump laser and near-infrared probe laser. It also confines the light field within the hollow core, causing it to interact strongly with the adsorption film coated on the inner wall and the adsorbed acetylene molecules, thereby greatly enhancing the detection sensitivity.

[0135] The coated hollow fiber optic sensing probe module also integrates a temperature sensor and a pressure sensor, which are electrically connected to the embedded processing unit to provide ambient temperature and pressure data.

[0136] The dual laser modulation and optical coupling module includes a wavelength-tunable pump laser, a probe laser, a laser modulator, and an optical fiber combiner and coupling assembly, used to generate modulated pump laser and probe laser, and couple them to the coated hollow fiber sensing probe module.

[0137] Among them, the wavelength-tunable pump laser is used to generate a mid-infrared pump laser whose center wavelength can be rapidly switched between the measurement wavelength, reference wavelength, and background wavelength, serving as an excitation source; the probe laser is used to generate a wavelength-stable near-infrared probe laser, serving as a phase detection probe; the laser modulator is actually the current driver of the pump laser, receiving a sinusoidal signal from the embedded unit and performing sinusoidal modulation of the output intensity of the pump laser at a specific frequency; the fiber bundle and coupling assembly includes a wavelength division multiplexer (WDM) and a precision fiber alignment mechanism, used to efficiently couple the pump laser and the probe laser into the same single-mode fiber, and finally inject them into the input end of the coated hollow fiber with low loss.

[0138] The photothermal interference signal demodulation module includes a fiber optic interferometer, a balanced photodetector, a lock-in amplifier, and an embedded processing unit. Based on the embedded processing unit, the process of performing steps S200 to S5500 is as follows: dual-wavelength differential detection, generation of compensated differential signal, generation of noise-filtered differential signal, and output of acetylene concentration value.

[0139] The fiber optic interferometer is used to interfere the probe laser carrying phase modulation information with a reference beam, converting the phase change into a change in light intensity. The balanced photodetector is used to receive the two complementary output beams of the interferometer and convert them into differential electrical signals, thereby canceling common-mode noise and increasing the signal amplitude. The lock-in amplifier uses the modulation frequency f of the pump laser and its second harmonic 2f as the reference frequency, and synchronously demodulates and extracts the amplitude (DC voltage) of the weak AC component (i.e., the 2f component) that is strictly in phase with the reference frequency from the complex electrical signal output by the balanced detector. The embedded processing unit (such as FPGA or high-speed MCU) is responsible for controlling the laser wavelength switching, generating the modulation signal, reading the lock-in amplifier output, executing all digital signal processing algorithms from steps S200 to S500, and finally outputting the acetylene concentration value and warning signal.

[0140] In practical applications, the main cabinet of this system can be installed in a protective box near the transformer. The coated hollow fiber optic sensor probe is connected to the main unit through an armored optical cable and inserted into the transformer sampling valve. After the system is powered on, it can operate automatically and without human intervention. Users can view the real-time concentration curve, historical data and alarm status through a local touch screen or a remote monitoring center. The system performs automatic zero-point and span calibration once a month (through the built-in micro-flow standard gas verification unit) and on-site maintenance once a year.

[0141] In this embodiment, acetylene is selectively adsorbed by coating the inner wall of a hollow fiber with a thin film, and this is deeply integrated with photothermal frequency modulation interferometry detection. This fundamentally solves the contradiction between response speed and sensitivity, achieving minute-level high-sensitivity in-situ detection without oil-gas separation. The selective adsorption film modified on the inner wall of the hollow fiber specifically captures and enriches acetylene molecules from the flowing transformer oil, increasing its local concentration within the detection range of the hollow fiber sensing probe by tens of times. This enhances the intensity of the optical interaction signal, laying the foundation for high sensitivity. This adsorption process is rapid and direct. Born in the oil phase, it completely eliminates the time-consuming oil-gas separation step. At the same time, it uses a mid-infrared pump laser with high-frequency modulation to excite the enriched acetylene molecules. The resulting periodic photothermal effect is sensitively detected by a near-infrared detection laser transmitted along the same path in the form of phase change. The synergistic mechanism of adsorption, excitation and detection has an intrinsically fast response. Experiments show that the detection limit of acetylene in oil is better than 0.5 μL / L, while the response time is shortened to less than 2 minutes. It achieves a rapid response that meets the requirements of early fault warning while maintaining ultra-high sensitivity, breaking through the inherent performance bottleneck of existing technologies.

[0142] A hierarchical intelligent adaptive signal processing chain is adopted to improve the long-term stability and anti-interference capability of the system in complex industrial environments. At the signal generation level, dual-wavelength differential detection is used to deduct the background absorption of the oil medium and the common fluctuation noise of the optical fiber in real time. At the signal processing level, a real-time coupled compensation model of temperature and pressure is established to dynamically correct the combined effects of environmental temperature and pressure changes on gas adsorption balance and optical fiber thermo-optic effects, thereby suppressing the most important source of signal drift at the root. At the same time, it has spectrum sensing and dynamic frequency avoidance capabilities. By periodically analyzing the environmental noise spectrum and automatically adjusting the pump laser modulation frequency to the interference-free frequency band, it actively avoids the inherent periodic vibration interference in the field. Finally, an adaptive digital filter is used to perform real-time optimal estimation of the signal and smooth random noise. Using a four-level linkage processing mechanism, the system performs hierarchical signal processing, enabling the entire system to not only withstand wide temperature range and strong electromagnetic environment, but also to adaptively resist time-varying field interference. After long-term operation and testing, it can achieve stable continuous monitoring for several months without manual intervention, with an extremely low false alarm rate and high accuracy, providing a solid data foundation for reliable assessment and intelligent diagnosis of transformer status.

[0143] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. The memories store computer-readable code that, when executed by the one or more processors, can perform the in-situ acetylene detection method and system for oil based on coated hollow optical fiber reinforcement as described above.

[0144] The methods and systems according to the embodiments of this application can also be implemented using the architecture of the electronic device shown in this application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as a ROM or hard disk, may store the in-situ acetylene detection method and system based on coated hollow fiber reinforcement provided in this application. Furthermore, the electronic device may also include a user interface. Of course, the architecture shown in this application is merely exemplary; when implementing different devices, one or more components in the electronic device shown in this application may be omitted according to actual needs.

[0145] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0146] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement, characterized in that, The method includes: The coated hollow fiber sensing probe is deployed in situ in the oil circuit of the transformer under test for system initialization. The inner wall of the coated hollow fiber is modified with a thin film that has selective adsorption function for acetylene molecules. Dual-wavelength differential detection is performed based on the modulated pump laser and probe laser. The original differential signal is obtained based on the dual-wavelength differential detection. The original differential signal is the difference between the amplitude of the optical signal collected at the measurement wavelength and the reference wavelength. The pump laser and probe laser are generated by the laser. The pump laser and probe laser are modulated and then subjected to dual-wavelength differential detection. The output wavelength of the pump laser is controlled to periodically switch between the pre-calibrated acetylene characteristic absorption measurement wavelength and the non-absorption reference wavelength. In each output wavelength switching cycle, the amplitude of the interference signal related to the modulation frequency of the pump laser generated by the probe laser after passing through the coated hollow fiber sensing probe at the measurement wavelength and the reference wavelength is collected by the lock-in amplifier. These are recorded as the first measurement signal and the second measurement signal. The difference between the first measurement signal and the second measurement signal is calculated to obtain the original differential signal. The temperature and pressure data of the environment in which the coated hollow fiber sensing probe is located are acquired synchronously, and the original differential signal is compensated for in real time by using a compensation function model to obtain the compensated differential signal. The compensated differential signal is subjected to adaptive noise suppression and filtering to obtain a denoised differential signal. The adaptive noise suppression and filtering includes: during system operation, periodically controlling the output wavelength of the pump laser to switch to a background wavelength far from the acetylene absorption line, acquiring a background noise signal of a preset duration based on the background wavelength, performing spectral analysis on the background noise signal to obtain a real-time noise spectrum, and determining whether there are interference peaks in the real-time noise spectrum with amplitudes exceeding a preset amplitude threshold within a preset frequency range near the current modulation frequency of the pump laser. If there are interference peaks exceeding the preset amplitude threshold, the modulation frequency of the pump laser is automatically adjusted to a backup modulation frequency; otherwise, the modulation frequency of the pump laser is kept unchanged. Based on the adjusted modulation frequency of the pump laser, the continuous values ​​of the compensated differential signal are used as time-series data and input into an adaptive digital filter for real-time filtering to output a denoised differential signal. The filtered differential signal is substituted into the pre-stored concentration calibration curve model to invert and output the acetylene concentration value in the transformer oil circuit in real time.

2. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 1, characterized in that, The generation of the first measurement signal and the second measurement signal includes: The second harmonic amplitude of the interference signal is extracted based on the second harmonic of the modulation frequency of the pump laser, using the lock-in amplifier as the reference frequency. The second harmonic amplitude extracted at the measurement wavelength is recorded as the first amplitude, and a first measurement signal is generated based on the first amplitude; The second harmonic amplitude extracted at the reference wavelength is recorded as the second amplitude, and a second measurement signal is generated based on the second amplitude.

3. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 1, characterized in that, The method of using a compensation function model to perform real-time temperature and pressure compensation on the original differential signal to obtain a compensated differential signal includes: At the same time point as acquiring the raw differential signal, the temperature and pressure data of the current environment are read from the temperature sensor and pressure sensor integrated on the coated hollow fiber sensing probe. The pre-stored compensation function model is invoked, and the temperature and pressure data are input into the compensation function model to calculate the first influencing factor under the current environment. Obtain predefined standard temperature and standard pressure data and input them into the compensation function model to calculate the second influencing factor under standard conditions; The original differential signal is calculated based on the first influence factor and the second influence factor to obtain the compensated differential signal.

4. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 3, characterized in that, The compensation function model characterizes the coupling influence factors of temperature and pressure on the signal response, and the compensation function model is a multivariate linear function based on temperature data and pressure data as variables. The original differential signal is multiplied by the ratio of the second influence factor to the first influence factor to obtain the compensated differential signal.

5. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 1, characterized in that, The step of using the continuous values ​​of the compensated differential signal as time-series data and inputting them into an adaptive digital filter for real-time filtering includes: Based on an adaptive digital filter, the time-series data of the compensated differential signal is monitored in real time, and the real-time noise statistical characteristics of the time-series signal are estimated. Based on real-time noise statistics, the adaptive filter dynamically adjusts its internal filter gain parameters. By using the adjusted filter gain parameters, iterative prediction and correction are performed on the random pulse interference and measurement noise in the time-series data of the compensated differential signal, thus completing the real-time filtering process.

6. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 1, characterized in that, The real-time inversion and output of the acetylene concentration value in the transformer oil circuit includes: Obtain the pre-stored concentration calibration curve model, and use the current value of the noise-filtered differential signal obtained in real time as the input variable to substitute into the concentration calibration curve model. The calculation is performed based on the mathematical relationship defined by the concentration calibration curve model, and the output is the acetylene concentration value in the oil circuit corresponding to the current value of the noise-filtered differential signal.

7. The method for in-situ detection of acetylene in oil based on coated hollow optical fiber reinforcement according to claim 6, characterized in that, After the acetylene concentration value is output, it also includes: Set a first concentration threshold and a second concentration threshold, wherein the second concentration threshold is greater than the first concentration threshold; The acetylene concentration value is compared with a first concentration threshold and a second concentration threshold; If the acetylene concentration exceeds the first concentration threshold but does not exceed the second concentration threshold, a level one warning signal is triggered. If the acetylene concentration exceeds the second concentration threshold, a level two warning signal will be triggered.

8. An in-situ acetylene detection system in oil based on coated hollow optical fiber reinforcement, characterized in that, The system is used to implement the in-situ detection method for acetylene in oil based on coated hollow optical fiber reinforcement as described in any one of claims 1-7, and the system comprises: The coated hollow fiber sensing probe module includes a section of anti-resonant hollow fiber. The inner wall of the anti-resonant hollow fiber is modified with a thin film that selectively adsorbs acetylene molecules. A microfluidic channel is provided on the side. The anti-resonant hollow fiber is encapsulated in a metal capillary tube with a transparent window for in-situ deployment in the transformer oil to be tested. The dual laser modulation and optical coupling module includes a wavelength-tunable pump laser, a probe laser, a laser modulator, and an optical fiber combiner and coupling assembly, used to generate modulated pump laser and probe laser and couple them to a coated hollow fiber sensing probe module. The photothermal interference signal demodulation module includes a fiber optic interferometer, a balanced photodetector, a lock-in amplifier, and an embedded processing unit. Based on the embedded processing unit, it performs the process of dual-wavelength differential detection, generating a compensated differential signal, generating a noise-filtered differential signal, and outputting the acetylene concentration value.

Images