A method for detecting dissolved gas in oil of a low-oil equipment based on online membrane degassing and multi-dimensional heterogeneous sensing array

Abstract

This invention relates to the field of power monitoring equipment and discloses a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array. This invention aims to solve the problems of difficulty in sensing the insulation status and delayed fault early warning in low-oil equipment. This invention achieves accurate full-spectrum sensing of fault characteristic gases through a cloud-edge collaborative architecture that deeply integrates multidimensional heterogeneous data. Its core lies in applying an evaluation model based on environmental matrix effect correction and fault fingerprint matching to perform multi-source data fusion and differentiated risk calculation, rapidly generating monitoring reports. This invention constructs a closed-loop system of non-destructive sampling, trace detection, and intelligent diagnosis, significantly improving the sensitivity and accuracy of early fault monitoring in low-oil equipment and enhancing its proactive defense capabilities and intelligence level.

Description

Technical Field

[0001] This invention relates to the field of power monitoring equipment, and more specifically, to a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array. Background Technology

[0002] Oil-based power transmission and transformation equipment, such as high-voltage capacitive bushings, current transformers, and voltage transformers, are key node devices in power systems that connect conductors and grounding components, enabling power metering and protection. These devices are internally filled with an oil-paper insulation system, utilizing the oil-paper composite insulation structure to withstand the stress of the high-voltage electric field. During equipment operation, the composition and concentration of dissolved gases in the insulating oil are the most direct and sensitive indicators reflecting the internal insulation health of the equipment. Acetylene, as a characteristic decomposition product of insulating oil under arc discharge or high-temperature overheating, often indicates irreversible and destructive faults within the equipment. Regular detection and analysis of dissolved gases in the oil are crucial technical means to ensure the safe operation of the power grid and prevent sudden equipment accidents.

[0003] However, existing oil-dissolved gas monitoring technologies face significant challenges when applied to equipment with low oil content. Traditional laboratory chromatographic analysis methods require periodic extraction of oil samples from the equipment. However, the total oil volume inside low-oil equipment such as high-voltage bushings is extremely small; frequent sampling can lead to a drop in oil level, damaging the insulation structure and causing secondary faults, thus hindering high-frequency monitoring. Furthermore, most existing mainstream online monitoring devices are designed for large transformers, and their sampling mechanisms are ill-suited to the unique operating conditions of static oil flow within low-oil equipment. Simultaneously, existing methods generally suffer from low signal-to-noise ratios and insufficient detection limits when detecting trace characteristic gases, making effective early identification at the nascent stage of faults difficult and failing to meet the stringent requirements of proactive defense against latent risks in low-oil equipment. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, which solves the problems of difficulty in sensing the insulation status and delayed fault warning in existing low-oil equipment.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, comprising the following steps: S1. Download the fault characteristic data and associated risk assessment matrix of the device model to be tested from the cloud database, and then load them into the running memory; S2. Send gas path control commands to the online membrane degassing unit to establish the permeation equilibrium state of oil-gas separation, control the carrier gas flow through the parallel detection array composed of three heterogeneous sensors, and acquire trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. S3. Based on the multidimensional heterogeneous sensing dataset, perform environmental matrix effect compensation on the trace acetylene photoacoustic spectral signal to obtain the corrected acetylene concentration value; perform calibration and analysis on the multidimensional heterogeneous sensing dataset to generate standardized gas state data. S4. Input the gas state data into the fault characteristic data and perform a comprehensive assessment in conjunction with the risk assessment matrix; generate an equipment monitoring report based on the comprehensive assessment results.

[0007] As a preferred embodiment of the method for detecting dissolved gas in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor array described in this invention, the process of downloading data from the cloud database in step S1 specifically includes: searching the cloud database based on the input equipment model and operating data including voltage level, insulation structure type, and oil-gas volume ratio parameters; The fault characteristic data corresponding to the device under test includes the acetylene concentration alarm threshold, hydrogen production rate threshold and carbon-oxygen ratio fault criteria for the type of device under test. The risk assessment matrix includes the hydrogen concentration response signal, the trace acetylene photoacoustic spectral signal, and the weighting coefficients of background gas component data in the fault determination process. For capacitive bushing equipment, the acetylene concentration data in the risk assessment matrix has the highest weighting coefficient and the lowest trigger threshold; For current transformer equipment, the weighting coefficients of hydrogen and total hydrocarbon data in the risk assessment matrix are increased, and a ratio weighting for solid insulation aging is introduced.

[0008] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor arrays described in this invention, the process of establishing the permeation equilibrium state of oil-gas separation in step S2 specifically includes: Read the temperature and pressure data of the online membrane degassing unit under its current state, and set the initial isothermal heating target range and carrier gas circulation flow rate value based on the preset temperature compensation curve; Start the constant temperature heating module and carrier gas circulation pump integrated on the online membrane degassing unit; dynamically adjust the heating power during the heating process until the operating temperature of the membrane module is within the constant temperature target range to maintain a constant gas permeability; The carrier gas circulation pump is controlled to drive the carrier gas to flow on the gas side of the membrane module at the carrier gas circulation flow rate value. The pressure fluctuation of the gas side flow induces micro-convective disturbance on the oil side surface of the membrane module, which destroys the gas-depleted boundary layer in the stagnant oil layer and accelerates the permeation and diffusion of dissolved gas in the oil to the gas side. Continuously monitor the total pressure change rate in the gas path. If the total pressure change rate is lower than the judgment threshold within the preset time window, it indicates that the oil-gas separation has reached the permeation equilibrium state.

[0009] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor array according to the present invention, wherein: the parallel detection array composed of the three heterogeneous sensors in step S2 includes a photoacoustic spectroscopy detection path, a hydrogen sensing detection path, and a background monitoring path. The photoacoustic spectroscopy detection path uses a tunable laser diode as a light source to lock the near-infrared characteristic absorption spectral lines of acetylene gas, drives the laser to perform intensity modulation at a preset frequency, collects the acoustic wave signal generated in the photoacoustic cell due to the photothermal effect through a high-sensitivity microphone, and uses a lock-in amplifier to extract the signal component with the same frequency as the laser modulation frequency, as the trace acetylene photoacoustic spectral signal. The hydrogen sensing detection circuit utilizes a micro hot plate semiconductor sensor and an electrochemical sensor to contact the flowing carrier gas, measure the changes in conductivity and current caused by the hydrogen redox reaction, and generate a hydrogen concentration response signal. The background monitoring path utilizes a non-dispersive infrared sensor, employing a broadband infrared light source in conjunction with a filter of a specific wavelength, to measure the infrared absorption intensity of the carrier gas for the characteristic bands of carbon monoxide, carbon dioxide, and total hydrocarbons, thereby generating background gas composition data. The hydrogen concentration response signal is packaged with the background gas component data to form the multidimensional heterogeneous sensing dataset.

[0010] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor arrays described in this invention, the process of obtaining the acetylene correction concentration value in step S3 specifically includes: The multidimensional heterogeneous sensing dataset was analyzed to extract the hydrogen concentration, carbon monoxide concentration, carbon dioxide concentration, and the proportion of carrier gas background components. Using the extracted concentration values ​​of each component, the equivalent relaxation time, sound velocity, and thermal conductivity of the current mixed gas are calculated based on the molecular dynamics of the mixed gas, and then the cell constant and Grindelwald constant of the photoacoustic detection are updated. A photoacoustic inversion equation was constructed using the updated cell constant and the Grindelsen constant. The trace acetylene photoacoustic spectral signal was substituted into the equation to eliminate the signal gain drift caused by the background gas components and to calculate the acetylene compensation concentration value after environmental matrix effect compensation. The current temperature and pressure data of the detection chamber are read, the acetylene correction concentration value is normalized to standard state, and the final acetylene correction concentration value is output.

[0011] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor arrays described in this invention, the process of generating standardized gas state data in step S3 specifically includes: The zero-point historical records of the hydrogen sensing detection path and the background monitoring path are called to calculate the current baseline drift of the sensor. The baseline drift is then subtracted from the hydrogen concentration response signal and the background gas component data to obtain the net response value. The net response value is then converted into the corresponding preliminary physical concentration value by fitting the pre-stored full-range response characteristic curve of the sensor. By combining the real-time temperature and pressure data of the detection chamber, the acetylene correction concentration value and the preliminary physical concentration value are uniformly converted into standard concentration units. The normalized concentration data of acetylene, hydrogen, carbon monoxide, and carbon dioxide, along with the calculated total hydrocarbon value, are arranged according to a preset time sequence and dimension to generate standardized gas state data.

[0012] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor arrays described in this invention, the comprehensive evaluation process in step S4 specifically includes: The concentrations of each gas in the gas state data are compared with the acetylene concentration alarm threshold and the hydrogen production rate threshold, and a primary over-limit flag is generated if the threshold is exceeded. The gas state data is weighted and summed using the weighting coefficients to calculate a comprehensive risk index that reflects the overall degree of equipment abnormality. Based on the aforementioned carbon-oxygen ratio fault criterion, the ratio of carbon dioxide to carbon monoxide and the ratio of characteristic gases are calculated to identify solid insulation aging or moisture defects inside the equipment. In addition, if the object to be detected is a capacitor-type bushing device, and the acetylene correction concentration value is greater than the minimum effective detection limit set by the system, and the comprehensive risk index is greater than the preset safety baseline, then the device status will be determined as a high-risk discharge fault state. If the object of detection is a current transformer, the equipment status is classified and determined based on the numerical range of the comprehensive risk index and the results of the fault type pattern recognition.

[0013] As a preferred embodiment of the method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and multidimensional heterogeneous sensor arrays described in this invention, the process of generating the equipment monitoring report specifically includes: Extract current gas state data, comprehensive risk index and comprehensive assessment results to generate a state snapshot reflecting the current insulation state of the equipment; By accessing historical monitoring data from the equipment, historical trend curves of acetylene, hydrogen, and total hydrocarbon concentrations are plotted to estimate the potential failure development rate within a preset time window. Based on the fault level and type in the comprehensive evaluation results, the corresponding operation and maintenance handling strategies are retrieved from the pre-set expert knowledge base; for equipment determined to be in a high-risk discharge fault state, handling suggestions including shortening the detection cycle, offline retesting, or emergency shutdown need to be matched. The status snapshot, the potential fault development rate, and the handling recommendations are packaged into a structured electronic document and pushed to the designated monitoring terminal through a communication interface.

[0014] This invention also provides a dissolved gas detection system for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, used to perform the above method, specifically including the following modules: The configuration management and storage module is used to store the device model and the corresponding working data type, and downloads and stores the matching fault characteristic data and risk assessment matrix from the cloud database according to the device model to be tested; The permeation balance control module is used to send gas path control commands to the online membrane degassing unit to adjust the temperature and flow rate, and to establish and maintain the permeation balance state of oil-gas separation. The sensor data processing module is used to receive electrical signals generated by the parallel detection array of three heterogeneous sensors, collect and preprocess them, and obtain trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. The acetylene concentration correction module is used to compensate for the environmental matrix effect of the trace acetylene photoacoustic spectral signal based on the background gas component information in the multidimensional heterogeneous sensing dataset, and to invert the acetylene correction concentration value. The gas state analysis module is used to calibrate and normalize the multidimensional heterogeneous sensing dataset to generate standardized gas state data. The fault feature data parsing module is used to compare the gas state data with the various thresholds and criteria in the fault feature data to generate a single index limit-out flag and a pattern recognition result. The comprehensive evaluation module is used to call the risk assessment matrix, combine the single indicator limit-breaking flag and pattern recognition results to perform multi-source weighted calculation, and output the comprehensive evaluation result of the device. The report generation module is used to generate an equipment monitoring report based on the comprehensive assessment results, which includes a status snapshot, potential failure progression rate, and remedial recommendations.

[0015] This invention also provides a dissolved gas detection device for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, used to perform the above method and as a detection carrier for the above system, specifically including: A fixing unit, comprising a housing and a gas detection chamber, wherein a flange is provided at the end of the housing away from the gas detection chamber, and a heating ceramic ring is fixedly installed on the flange; A gas delivery unit includes a carrier gas circulation pump fixedly installed at the bottom of the housing. The carrier gas circulation pump is provided with a dual-channel gas transmission pipe and a gas delivery pipe inside the housing. The end of the dual-channel gas transmission pipe is coaxial with the flange. A membrane degassing probe is fixed to the end of the dual-channel gas transmission pipe. A gas pressure sensor and a gas temperature sensor are provided on the inner wall of the gas outlet passage of the dual-channel gas transmission pipe. A gas detection unit includes a photoacoustic resonant cavity connected to the gas transmission pipe, a laser emitter installed at the top of the photoacoustic resonant cavity, the laser emitter passing through the top of the gas detection chamber, a high-sensitivity microphone installed inside the photoacoustic resonant cavity, a flow chamber connected to the gas outlet of the photoacoustic resonant cavity, a hydrogen probe and an infrared probe installed on the inner wall of the flow chamber, and a return gas channel connected to the gas inlet passage of the dual-channel gas transmission pipe at the gas outlet of the flow chamber. The membrane degassing probe includes a support tube filled with a fiber tube. Sealing nozzles are located at both ends of the support tube, and both ends of the fiber tube pass through the sealing nozzles while remaining open, allowing the inner cavity of the fiber tube to form a circulating gas path connected to the inlet and outlet passages of the dual-channel gas transmission tube. The support tube and the sealing nozzles at both ends together define an exchange chamber. A porous support filler is tightly filled within the exchange chamber and in the gaps between the fiber tubes. This porous support filler restricts the radial displacement of the fiber tubes and provides structural support. An oil passage is provided on the side wall of the support tube, allowing external insulating oil to enter the exchange chamber and contact the outer wall of the fiber tubes.

[0016] The beneficial effects of this invention are as follows: By constructing an intelligent monitoring architecture based on the fusion of online membrane degassing and multidimensional heterogeneous sensing, this invention achieves non-destructive, high-precision online sensing for low-oil electrical equipment. Its core lies in utilizing a parallel array technology combining photoacoustic spectroscopy and heterogeneous sensing to solve the problem of insufficient acetylene capture capability under static oil conditions using traditional methods; and by dynamically loading a fault feature fingerprint model and risk assessment matrix, it breaks through the limitation of fixed parameters in traditional monitoring equipment.

[0017] By constructing a closed-loop diagnostic system that integrates environmental matrix correction, multi-source data fusion, and differentiated comprehensive evaluation, the sensitivity and accuracy of identifying early latent faults in low-oil equipment have been significantly improved. This provides strong technical support for the operation and maintenance decisions of key equipment such as high-voltage bushings and instrument transformers, and reduces the risk of severe sudden failures. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array.

[0020] Figure 2 A flowchart for multi-dimensional data processing.

[0021] Figure 3 Establish a flow chart for penetration balance and feedback control.

[0022] Figure 4 This is a comprehensive fault diagnosis decision logic diagram.

[0023] Figure 5 This is a flowchart of a dissolved gas detection system for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array.

[0024] Figure 6 This is an overall view of the dissolved gas detection device in oil for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array.

[0025] Figure 7 This is a cross-sectional view of a dissolved gas detection device in oil for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array.

[0026] Figure 8 This is a schematic diagram of the internal structure of the membrane degassing probe.

[0027] Figure 9 Figure showing the experimental results of optimizing the amount of fluorinated mesoporous silica added.

[0028] Figure 10 The figure shows the experimental results for optimizing sol concentration. Detailed Implementation

[0029] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0030] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0031] Secondly, the term "one embodiment" or "example" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The appearance of an embodiment in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments.

[0032] Example 1 Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, comprising the following steps: S1. Download the fault characteristic data and associated risk assessment matrix of the device model to be tested from the cloud database, and then load them into the running memory; The process of downloading data from the cloud database involves searching the cloud database based on the input device model and operating data, including voltage level, insulation structure type, and oil-gas volume ratio parameters. The fault characteristic data corresponding to the equipment under test includes the acetylene concentration alarm threshold, hydrogen production rate threshold, and carbon-oxygen ratio fault criteria for the type of equipment under test. The risk assessment matrix includes the weighting coefficients of hydrogen concentration response signal, trace acetylene photoacoustic spectrum signal, and background gas component data in the fault determination process. For capacitive bushing equipment, acetylene concentration data in the risk assessment matrix has the highest weighting coefficient and the lowest trigger threshold. For current transformer equipment, increase the weighting coefficients of hydrogen and total hydrocarbon data in the risk assessment matrix, and introduce a ratio weighting for solid insulation aging.

[0033] The process begins by searching a cloud database based on the input equipment model and operational data, including voltage level, insulation structure type, and oil-gas volume ratio. The voltage level determines the internal electric field distribution and insulation margin of the equipment, directly impacting the initiation voltage of discharge faults. The insulation structure type distinguishes whether the equipment uses oil-paper insulation, adhesive-paper insulation, or an oil-immersed inverted structure, as different structures exhibit significant differences in gas generation mechanisms and fault evolution paths. The oil-gas volume ratio serves as the ratio of the internal insulating oil volume to the gas chamber volume, providing the physical basis for subsequent gas generation rate assessments and Henry's Law conversions. Using this parameter combination as the search key, the system locates the appropriate configuration package for the current equipment within the massive cloud database.

[0034] The fault characteristic data corresponding to the equipment under test encapsulates multi-dimensional criteria for this specific equipment type, specifically including acetylene concentration alarm threshold, hydrogen production rate threshold, and carbon-oxygen ratio fault criteria. The acetylene concentration alarm threshold sets the upper limit of the absolute acetylene concentration allowed by the equipment, serving as a hard indicator for determining whether high-energy discharge exists; the hydrogen production rate threshold defines the allowable growth rate of hydrogen concentration per unit time, used to identify progressive faults caused by partial discharge or moisture; the carbon-oxygen ratio fault criteria include the standard ratio range of carbon dioxide to carbon monoxide, used to qualitatively assess whether solid insulation materials have undergone overheating aging or moisture decomposition.

[0035] The downloaded risk assessment matrix is ​​essentially a decision logic table that defines the weights for multi-source data fusion. This matrix details the weighting coefficients for hydrogen concentration response signals, trace acetylene photoacoustic spectral signals, and background gas component data in the final fault determination process. The local assessment algorithm logic is dynamically reconstructed based on the downloaded matrix parameters to adapt to different equipment characteristics.

[0036] Specifically, for capacitive bushing-type equipment, due to its compact insulation structure and extreme sensitivity to discharge faults, acetylene concentration data has the highest weighting coefficient and the lowest trigger threshold in the applied risk assessment matrix. This means that the monitoring system will prioritize capturing trace acetylene signals, and any effective acetylene detection will dominate the calculation of the comprehensive risk index, reflecting a zero-tolerance strategy for failures in this type of equipment.

[0037] For instrument transformers, given their relatively high internal oil content and the long-term aging process of solid insulation, the matrix configuration strategy differs. The weighting coefficients of hydrogen and total hydrocarbon data in the risk assessment matrix are increased to maintain a relatively balanced position with acetylene data, and a ratio weighting for solid insulation aging is introduced. Specifically, the focus is on the contribution of changes in the carbon oxide ratio to the risk index, thereby effectively distinguishing between normal insulation aging and abnormal overheating faults.

[0038] S2. Send gas path control commands to the online membrane degassing unit to establish the permeation equilibrium state of oil-gas separation, control the carrier gas flow through the parallel detection array composed of three heterogeneous sensors, and acquire trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. The process of establishing the permeation equilibrium state of oil-gas separation involves reading the temperature and pressure data of the online membrane degassing unit under its current state, and setting the initial isothermal heating target range and carrier gas circulation flow rate based on the preset temperature compensation curve. Start the isothermal heating module and carrier gas circulation pump integrated on the online membrane degassing unit; dynamically adjust the heating power during the heating process until the operating temperature of the membrane module is within the isothermal target range to maintain a constant gas permeability; The carrier gas circulation pump is controlled to drive the carrier gas to flow on the gas side of the membrane module at the carrier gas circulation flow rate. The pressure fluctuation of the gas side flow induces micro-convective disturbance on the oil side surface of the membrane module, which disrupts the gas-depleted boundary layer in the stagnant oil layer and accelerates the permeation and diffusion of dissolved gas in the oil to the gas side. Continuously monitor the total pressure change rate in the gas path. If the total pressure change rate is lower than the judgment threshold within the preset time window, it indicates that the oil-gas separation has reached the permeation equilibrium state. The parallel detection array, composed of three heterogeneous sensors, includes a photoacoustic spectroscopy detection path, a hydrogen sensing detection path, and a background monitoring path. The photoacoustic spectroscopy detection circuit uses a tunable laser diode as a light source to lock onto the near-infrared characteristic absorption lines of acetylene gas, drives the laser to modulate the intensity at a preset frequency, collects the acoustic wave signal generated in the photoacoustic cell due to the photothermal effect through a high-sensitivity microphone, and uses a lock-in amplifier to extract the signal component with the same frequency as the laser modulation frequency as the trace acetylene photoacoustic spectral signal. The hydrogen sensing detection circuit utilizes a micro-hot plate semiconductor sensor and an electrochemical sensor to contact the flowing carrier gas, measure the changes in conductivity and current caused by the hydrogen redox reaction, and generate a hydrogen concentration response signal. The background monitoring circuit utilizes a non-dispersive infrared sensor, employing a broadband infrared light source in conjunction with a filter of a specific wavelength, to measure the infrared absorption intensity of the carrier gas for the characteristic bands of carbon monoxide, carbon dioxide, and total hydrocarbons, thereby generating background gas composition data. The hydrogen concentration response signal and background gas component data are packaged together to form a multidimensional heterogeneous sensing dataset.

[0039] After the model is loaded and configured, the physical detection stage is entered. This stage involves sending gas path control commands to the online membrane degassing unit, establishing a permeation equilibrium state for oil-gas separation, and controlling the carrier gas flow through a parallel detection array composed of three heterogeneous sensors to obtain trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets.

[0040] Establishing a permeation equilibrium state for oil-gas separation is a crucial prerequisite for ensuring accurate detection of equipment with low oil content under static oil conditions. First, the temperature and pressure data of the online membrane degassing unit at its current state are read. Then, based on the pre-stored membrane material permeability-temperature compensation curve, the initial isothermal heating target range and carrier gas circulation velocity value are calculated and set.

[0041] in, The set target value for constant temperature heating; This indicates the activation energy for the membrane material to permeate with acetylene gas. It is the ideal gas constant (8.314 J / (mol·K)). It is the target penetration rate preset by the system to meet the needs of rapid balancing; It is the pre-exponential factor of the membrane material and is related to the microstructure of the membrane.

[0042] The purpose of setting a constant temperature range is to eliminate the interference of outdoor temperature differences on the solubility coefficient (Henry constant) and diffusion coefficient of gas in the membrane, and to ensure that subsequent concentration inversion is based on a unified physical benchmark.

[0043] The isothermal heating module and carrier gas circulation pump integrated into the online membrane degassing unit are activated. During heating, the heating power is dynamically adjusted until the operating temperature of the membrane module is precisely locked within the set isothermal target range, thereby maintaining a constant gas permeability. Simultaneously, the carrier gas circulation pump is controlled to drive the carrier gas to flow within the hollow fibers on the gas side of the membrane module at a set carrier gas circulation flow rate. This flow not only transports gas but also utilizes the pressure fluctuations of the gas-side flow to transfer it to the oil side through the membrane wall, inducing micro-convective disturbances on the oil-side surface of the membrane module. The micro-convective effect effectively disrupts the gas-depleted boundary layer formed in the stagnant oil layer due to slow gas molecule diffusion, continuously renewing the distant gas-rich oil layer to the membrane surface, thereby significantly accelerating the permeation and diffusion process of dissolved gases from the oil to the gas side, solving the response lag problem caused by the lack of forced oil circulation in low-oil equipment.

[0044] During this process, the rate of change of total pressure in the gas path is continuously monitored. When the rate of change of total pressure is continuously lower than the judgment threshold within the preset time window, it indicates that the gas partial pressure on both sides of the membrane has reached dynamic equilibrium, the oil-gas separation process is completed, and a balance ready indicator is generated.

[0045] The carrier gas in equilibrium is introduced into a parallel detection array consisting of three heterogeneous sensors. This array includes a photoacoustic spectroscopy detection path, a hydrogen sensing detection path, and a background monitoring path, which perform parallel or serial detection for fault gases with different characteristics.

[0046] The photoacoustic spectroscopy detection circuit utilizes a tunable laser diode as the light source. The laser locks onto the near-infrared characteristic absorption line of acetylene gas (e.g., 1532.83 nm) and modulates its intensity at a preset frequency. When acetylene molecules in the carrier gas absorb laser energy, they undergo periodic thermal expansion, thereby exciting sound waves. This weak sound wave signal generated by the photothermal effect within the photoacoustic cell is acquired using a high-sensitivity microphone, and a lock-in amplifier is used to extract the signal component with the same frequency and phase as the laser modulation frequency, which is then used as the trace acetylene photoacoustic spectral signal.

[0047] The hydrogen sensing detection circuit employs a composite sensing strategy, utilizing a micro-hotplate semiconductor sensor and an electrochemical sensor in contact with the flowing carrier gas. The micro-hotplate semiconductor sensor exhibits extremely high response speed to hydrogen, while the electrochemical sensor demonstrates good linearity. Both sensors separately measure the changes in conductivity and current caused by the redox reaction of hydrogen on the sensitive material surface, respectively, and combine these measurements to generate a hydrogen concentration response signal for rapidly capturing low-energy discharge or moisture characteristics.

[0048] Background monitoring utilizes a non-dispersive infrared (NDIR) sensor, employing a broadband infrared light source and filters designed for specific gas characteristic absorption peaks to measure the infrared absorption intensity of the carrier gas for the characteristic wavelengths of carbon monoxide, carbon dioxide, and total hydrocarbons. Since the absorption of infrared light by different gas molecules follows Beer-Lambert's law, background gas composition data can be generated by measuring the light intensity attenuation.

[0049] The hydrogen concentration response signal and background gas component data are packaged to form a multidimensional heterogeneous sensing dataset, and time-stamped with the trace acetylene photoacoustic spectral signal to ensure that all data accurately correspond to the same detection time scale in the time dimension, providing a foundation for subsequent data processing.

[0050] S3. Based on the multidimensional heterogeneous sensing dataset, environmental matrix effect compensation is performed on the trace acetylene photoacoustic spectral signal to obtain the corrected acetylene concentration value; the multidimensional heterogeneous sensing dataset is calibrated and analyzed to generate standardized gas state data. The process of inverting to obtain the corrected acetylene concentration value involves analyzing the multidimensional heterogeneous sensing dataset and extracting the hydrogen concentration value, carbon monoxide concentration value, carbon dioxide concentration value, and the proportion of carrier gas background components. Using the extracted concentration values ​​of each component, the equivalent relaxation time, sound velocity, and thermal conductivity of the current mixed gas are calculated based on the molecular dynamics of the mixed gas, and then the cell constant and Grindelwald constant of the photoacoustic detection are updated. The photoacoustic inversion equation was constructed using the updated cell constant and the Grindelsen constant. The trace acetylene photoacoustic spectral signal was substituted into the equation to eliminate the signal gain drift caused by the background gas composition. The acetylene compensation concentration value after environmental matrix effect compensation was calculated. Read the current temperature and pressure data of the detection chamber, perform standard state normalization on the acetylene correction concentration value, and finally output the acetylene correction concentration value. The process of generating standardized gas state data involves calling the zero-point historical records of the hydrogen sensing detection path and the background monitoring path, calculating the current baseline drift of the sensor, and subtracting the baseline drift from the hydrogen concentration response signal and the background gas component data to obtain the net response value. The net response value is then converted into the corresponding preliminary physical concentration value by fitting the pre-stored full-range response characteristic curve of the sensor. By combining the real-time temperature and pressure data of the detection chamber, the acetylene correction concentration value and the preliminary physical concentration value are uniformly converted into standard concentration units. The normalized concentration data of acetylene, hydrogen, carbon monoxide, and carbon dioxide, along with the calculated total hydrocarbon value, are arranged according to a preset time sequence and dimension to generate standardized gas state data.

[0051] The process involves acquiring the raw detection signal and then performing data processing. This includes compensating for environmental matrix effects on the trace acetylene photoacoustic spectral signal based on the multidimensional heterogeneous sensing dataset, and retrieving the corrected acetylene concentration value. Simultaneously, the multidimensional heterogeneous sensing dataset is calibrated and analyzed to generate standardized gas state data. This step aims to eliminate the physical interference of the complex dissolved gas environment in oil on measurement accuracy and restore the true gas concentration.

[0052] The intensity of the photoacoustic effect depends not only on the concentration of the target gas (acetylene) but also significantly on the physical properties of the background gas (such as thermal conductivity, sound velocity, and molecular relaxation time). When the background gas contains different concentrations of components such as hydrogen and carbon dioxide, the physical parameters of the mixed gas will change, causing gain drift in the photoacoustic signal. Therefore, it is necessary to first analyze the multidimensional heterogeneous sensing dataset to extract the hydrogen concentration, carbon monoxide concentration, carbon dioxide concentration, and the proportion of background components in the carrier gas measured by the second and third detection paths.

[0053] Using the extracted concentration values ​​of each component, and based on a molecular dynamics model of the mixed gas, the equivalent relaxation time, sound velocity, and thermal conductivity of the current mixed gas are calculated. Changes in these physical quantities directly determine the energy transfer efficiency and sound wave propagation characteristics during photoacoustic signal generation. Based on this, the cell constant and Grindelwald constant for photoacoustic detection are dynamically updated.

[0054] in, This represents the equivalent thermal conductivity of the current gas mixture. The first in the mixed gas The mole fraction of each component; Indicates the first Thermal conductivity of a pure component gas; Indicates the first The molar mass of each component gas; For the first The dynamic viscosity of the component gas.

[0055] An adaptive photoacoustic inversion equation was constructed using the updated cell constant and the Grindelwald constant. The original acquired trace acetylene photoacoustic spectral signal was substituted into this equation, and the signal gain drift caused by changes in background gas composition was eliminated through mathematical calculations, thereby obtaining the acetylene compensation concentration value after environmental matrix effect compensation.

[0056] in, This is the compensated acetylene concentration value after compensation; It is the raw signal (voltage value) of the trace acetylene photoacoustic spectrum collected; The excitation power of the laser; The geometric structure factor of the photoacoustic cell (a part of the cell constant); It is the quality factor of the photoacoustic resonator; This is the Grignard Eisen constant of the current gas mixture, which is dynamically updated as the background composition changes; The coefficient of thermal expansion of the gas mixture; It is the speed of sound in the gas mixture; This is the isobaric specific heat capacity of the gas mixture.

[0057] Although the concentration value obtained at this point eliminates background interference, it is still limited by the current state of the gas chamber. Therefore, it is necessary to further read the current temperature and pressure data of the detection gas chamber, use the ideal gas law to perform standard state normalization on the acetylene compensation concentration value, and finally output the accurate acetylene correction concentration value.

[0058] in, This is the final output acetylene correction concentration value (concentration under standard conditions). To detect the current pressure and temperature of the air chamber; The standard pressure (101.325 kPa) and standard temperature (293.15 K, i.e. 20 °C) are given.

[0059] The zero-point historical records of the hydrogen sensing detection path and the background monitoring path are retrieved to calculate the baseline drift caused by long-term operation or environmental changes. This baseline drift is then subtracted from the original hydrogen concentration response signal and background gas component data to obtain an accurate net response value. By retrieving the pre-stored full-range response characteristic curve of the sensor (usually a polynomial fitting curve or lookup table), the net response value is converted into the corresponding preliminary physical concentration value, thus resolving the sensor's nonlinear response problem.

[0060] Finally, combining the real-time temperature and pressure data from the detection chamber, the obtained corrected acetylene concentration value and the preliminary physical concentration values ​​of hydrogen, carbon monoxide, and carbon dioxide are uniformly converted into standard concentration units (e.g., μL / L at 20℃ and 101.325 kPa). The normalized acetylene, hydrogen, carbon monoxide, and carbon dioxide concentration data, along with the total hydrocarbon value calculated based on carbon composition, are then arranged according to a preset time sequence and dimension to assemble standardized gas state data. This data vector eliminates various errors from the environment, background, and the sensor itself, and can serve as an input source for subsequent model evaluation.

[0061] S4. Input the gas state data into the fault characteristic data and perform a comprehensive assessment in conjunction with the risk assessment matrix; generate an equipment monitoring report based on the comprehensive assessment results. The comprehensive evaluation process compares the concentrations of various gases in the gas state data with the acetylene concentration alarm threshold and the hydrogen production rate threshold, and generates a primary over-limit flag if the threshold is exceeded. By using weighting coefficients to perform weighted summation on the gas state data, a comprehensive risk index reflecting the overall degree of equipment anomaly is calculated. Based on the carbon-oxygen ratio fault criterion, the ratio of carbon dioxide to carbon monoxide and the ratio of characteristic gases are calculated to identify solid insulation aging or moisture defects inside the equipment. In addition, if the object of detection is a capacitor-type bushing device, and the acetylene correction concentration value is greater than the minimum effective detection limit set by the system and the comprehensive risk index is greater than the preset safety baseline, then the device status will be judged as a high-risk discharge fault state. If the object of the test is a current transformer, the equipment status is classified and determined based on the numerical range of the comprehensive risk index and the results of fault type pattern recognition. The process of generating equipment monitoring reports involves extracting current gas state data, comprehensive risk index, and comprehensive assessment results to generate a state snapshot reflecting the current insulation status of the equipment. By accessing historical monitoring data from the equipment, historical trend curves of acetylene, hydrogen, and total hydrocarbon concentrations are plotted to estimate the potential failure development rate within a preset time window. Based on the fault level and type in the comprehensive assessment results, the corresponding operation and maintenance handling strategies are retrieved from the pre-set expert knowledge base; for equipment determined to be in a high-risk discharge fault state, handling suggestions including shortening the detection cycle, offline retesting, or emergency shutdown need to be matched. Status snapshots, potential fault progression rates, and handling recommendations are packaged into structured electronic documents and pushed to designated monitoring terminals via communication interfaces.

[0062] After standardizing the data, the process involves inputting gas state data into fault characteristic data, combining it with a risk assessment matrix for comprehensive evaluation, and generating an equipment monitoring report based on the comprehensive evaluation results.

[0063] The comprehensive evaluation process is a multi-level, multi-dimensional logical operation. First, the concentrations of each gas in the gas state data are compared one by one with the applied fault characteristic data (i.e., the acetylene concentration alarm threshold and the hydrogen production rate threshold). Once any indicator exceeds the corresponding threshold, the system generates a primary over-limit flag as a signal to trigger the subsequent advanced evaluation.

[0064] Subsequently, the weighting coefficients defined in the risk assessment matrix are used to perform a weighted summation operation on the gas state data. Different types of fault gases (acetylene, hydrogen, carbon monoxide, etc.) are uniformly mapped to a dimensionless comprehensive risk index, thereby intuitively reflecting the overall severity of equipment anomalies.

[0065] in, This is a comprehensive risk index; It is a gas component index (such as acetylene, hydrogen, total hydrocarbons); The first one defined in the risk assessment matrix Weighting coefficients for each gas; It is the first Standardized measured concentrations of the gases; The first one defined in the fault feature fingerprint model Alarm thresholds for various gases (used for normalization to eliminate differences in magnitude between different gases).

[0066] Simultaneously, based on the carbon-oxygen ratio fault criterion, the ratio of carbon dioxide to carbon monoxide and other characteristic gas ratios are calculated. Through ratio analysis, it is possible to effectively identify whether there are overheating aging or moisture-induced hydrolysis defects in solid insulating materials (such as insulating paper and cardboard) inside the equipment.

[0067] After completing the basic calculations, a differentiated final status determination is performed based on the equipment type. If the object being tested is a capacitor-type bushing device, given its zero-tolerance characteristic for discharge faults, a stringent high-risk determination logic needs to be implemented. As long as the detected acetylene correction concentration value is greater than the minimum effective detection limit set by the system (i.e., acetylene is confirmed to be present), or the comprehensive risk index is greater than the preset safety baseline, the equipment status will be directly determined as a high-risk discharge fault state, indicating a possible risk of breakdown.

[0068] If the device being tested is a current transformer, a more detailed classification and judgment logic is executed. Based on the numerical range of the calculated comprehensive risk index (such as normal zone, attention zone, warning zone), and combined with the results of the aforementioned fault type pattern recognition (such as whether it is overheating or discharge), the equipment status is classified as normal operation, attention zone, or abnormal alarm, in order to adapt to the complex fault evolution patterns of such equipment.

[0069] Extract all key information at the current moment, including gas state data, comprehensive risk index, and comprehensive assessment results, to generate a snapshot reflecting the current insulation health level of the equipment. Utilize historical monitoring data from the equipment and time series analysis algorithms to plot historical trend curves for acetylene, hydrogen, and total hydrocarbon concentrations, and based on this, estimate the potential fault development rate within a preset time window (e.g., the next 24 hours or 7 days).

[0070] in, Potential fault development rate (concentration / unit time); This represents the number of historical data points used for fitting. Indicates the first Timestamps of historical data points; Indicates the first The gas concentration values ​​corresponding to each historical data point.

[0071] Based on the fault level and type in the comprehensive assessment results, the system retrieves corresponding operation and maintenance (O&M) strategies from a pre-built expert knowledge base. For example, for equipment identified as having a high-risk discharge fault state, the system will prioritize matching handling suggestions that include shortening the detection cycle (e.g., increasing the frequency to once per hour), immediately performing DGA offline retesting, or recommending emergency shutdown and maintenance. The aforementioned status snapshot, potential fault development rate, and handling suggestions are packaged into a structured electronic document and pushed in real-time to the designated monitoring terminal or O&M management platform via a communication interface to assist O&M personnel in making rapid decisions.

[0072] Example 2 Reference Figure 2 and Figure 5These are two embodiments of the present invention. This embodiment provides a dissolved gas detection system in oil-poor equipment based on online membrane degassing and a multidimensional heterogeneous sensor array. This system is typically deployed on a high-performance computing server to execute the steps described in Embodiment 1.

[0073] The system's software architecture consists of a set of highly collaborative functional modules, specifically including: The configuration management and storage module is used to store the device model and the corresponding working data type, and downloads and stores the matching fault characteristic data and risk assessment matrix from the cloud database according to the device model to be tested; The permeation balance control module is used to send gas path control commands to the online membrane degassing unit to adjust the temperature and flow rate, and to establish and maintain the permeation balance state of oil-gas separation. The sensor data processing module is used to receive electrical signals generated by the parallel detection array of three heterogeneous sensors, collect and preprocess them, and obtain trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. The acetylene concentration correction module is used to compensate for the environmental matrix effect of trace acetylene photoacoustic spectral signals based on the background gas component information in the multidimensional heterogeneous sensing dataset, and to invert the acetylene corrected concentration value. The gas state analysis module is used to calibrate and normalize multidimensional heterogeneous sensing datasets to generate standardized gas state data. The fault feature data parsing module is used to compare the gas state data with the various thresholds and criteria in the fault feature data, and generate single index over-limit flags and pattern recognition results. The comprehensive evaluation module is used to call the risk assessment matrix, combine the single indicator limit-breaking flag and pattern recognition results to perform multi-source weighted calculation, and output the comprehensive evaluation result of the device. The report generation module is used to generate equipment monitoring reports based on comprehensive assessment results, including status snapshots, potential failure progression rates, and remedial recommendations.

[0074] Upon receiving a monitoring start command for a specific low-oil electrical device, the system first activates the configuration management and storage module.

[0075] The configuration management and storage module first parses the device model and corresponding working data type of the device to be monitored carried in the command, and uses this as an index to connect to the cloud database through an encrypted communication link. After verifying the device identity, it downloads and stores fault characteristic data and risk assessment matrices matching the device model from the cloud, completing the dynamic deployment of the local monitoring strategy. After the configuration is loaded, a ready signal is sent to the penetration balancing control module.

[0076] Upon receiving the readiness signal, the permeation balance control module sends a gas path control command to the online membrane degassing unit, initiating the isothermal heating and microcirculation process. By adjusting the heating power and flow rate in real time, it establishes and maintains a permeation balance state for oil-gas separation. Once the gas path pressure stabilizes and balance is established, a data acquisition trigger signal is generated and sent to the sensor data processing module.

[0077] The sensor data processing module responds to the acquisition trigger signal and activates a three-channel heterogeneous sensor parallel detection array. It continuously receives raw electrical signals generated by the photoacoustic spectroscopy, hydrogen sensing, and background monitoring channels, and performs preprocessing operations such as high-frequency sampling, filtering and noise reduction, and timestamp alignment. Finally, the processed data is packaged and output as a trace acetylene photoacoustic spectral signal and a multi-dimensional heterogeneous sensing dataset containing hydrogen and background components, which are then passed to subsequent modules.

[0078] The acetylene concentration correction module calculates the physical parameters of the mixed gas based on the background gas component information (such as hydrogen, carbon monoxide, and carbon dioxide concentrations) in the received multidimensional heterogeneous sensing dataset, and compensates for environmental matrix effects in the trace acetylene photoacoustic spectral signal. After eliminating background interference through an inversion algorithm, it outputs a high-precision acetylene correction concentration value, which, along with the original multidimensional heterogeneous sensing dataset, is sent to the gas state analysis module.

[0079] The multidimensional data input to the gas state analysis module undergoes zero-point drift correction and nonlinear compensation, and is normalized in conjunction with ambient temperature and pressure data. After processing, standardized gas state data containing the concentrations of all key gases (acetylene, hydrogen, carbon monoxide, carbon dioxide, and total hydrocarbons) is generated and submitted to the fault characteristic data analysis module.

[0080] Based on the previously loaded configuration information, the fault feature data analysis module compares the standardized gas state data with each threshold (such as the acetylene alarm threshold) and criterion in the fault feature data. It generates single-index over-limit flags (indicating which specific indicators exceed the limit) and pattern recognition results (indicating possible fault types), and passes these intermediate results to the comprehensive evaluation module.

[0081] The comprehensive assessment module calls the risk assessment matrix and, combined with the received single-indicator over-limit flags and pattern recognition results, performs multi-source weighted calculations. Based on the different weight configurations for equipment type (bushing or transformer), it calculates the comprehensive risk index, determines the equipment status, and outputs the final comprehensive assessment result for the equipment.

[0082] After receiving the comprehensive assessment results, the report generation module extracts a snapshot of the current status, uses historical data to estimate the potential failure rate, and matches corresponding handling recommendations. Finally, it generates a complete equipment monitoring report containing comprehensive diagnostic information and pushes it to the designated monitoring terminal via the communication interface, completing one full monitoring cycle.

[0083] Example 3 Reference Figures 6-8 These are three embodiments of the present invention. This embodiment provides a dissolved gas detection device for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array. This device is used to perform the method described in Embodiment 1 and serves as the detection carrier of the system described in Embodiment 2.

[0084] Specifically, it includes: The fixing unit 100 includes a housing 101 and a gas detection chamber 102. A flange 103 is provided at one end of the housing 101 away from the gas detection chamber 102. A heating ceramic ring 104 is fixedly installed on the flange 103. A gas delivery unit 200 includes a carrier gas circulation pump 201 fixedly installed at the bottom of the housing 101. The carrier gas circulation pump 201 is provided with a dual-channel gas transmission pipe 202 and a gas delivery pipe 203 inside the housing 101. The end of the dual-channel gas transmission pipe 202 is coaxial with the flange 103. A membrane degassing probe 204 is fixed to the end of the dual-channel gas transmission pipe 202. A gas pressure sensor 205 and a gas temperature sensor 206 are provided on the inner wall of the gas outlet passage of the dual-channel gas transmission pipe 202. A gas detection unit 300 includes a photoacoustic resonant cavity 301 connected to the gas transmission pipe 203. A laser emitter 302 is disposed on the top of the photoacoustic resonant cavity 301 and passes through the top of the gas detection chamber 102. A high-sensitivity microphone 303 is disposed inside the photoacoustic resonant cavity 301. A flow chamber 304 is connected to the gas outlet of the photoacoustic resonant cavity 301. A hydrogen probe 305 and an infrared probe 306 are disposed on the inner wall of the flow chamber 304. A return gas channel 307 is connected to the gas outlet of the flow chamber 304 and is connected to the gas inlet passage 202a of the dual-channel gas transmission pipe 202. The membrane degassing probe 204 includes a support tube 204a, the inside of which is filled with a fiber tube 204b. Sealing nozzles 204c are provided at both ends of the support tube 204a. Both ends of the fiber tube 204b pass through the sealing nozzles 204c and remain open, so that the inner cavity of the fiber tube 204b forms a circulating gas path connected to the inlet passage 202a and the outlet passage 202b of the dual-channel gas transmission pipe 202. The support tube 204a and the two... The sealing head 204c at the end together defines an exchange cavity 204d. The gap between the fiber tubes 204b and the exchange cavity 204d is tightly filled with a porous support filler 204e. The porous support filler 204e is used to restrict the radial displacement of the fiber tubes 204b and provide structural support. The side wall of the support tube 204a is provided with an oil passage hole 204f, so that external insulating oil can enter the exchange cavity 204d and contact the outer wall of the fiber tubes 204b.

[0085] First, the entire structure is sealed and installed at the oil inlet of the low-oil equipment via flange 103. The fixing unit 100 serves as the main support, and the housing 101 and the gas detection chamber 102 on its upper part provide a protective space for the internal precision components.

[0086] During device operation, in order to establish a stable physical detection environment, the heating ceramic ring 104, fixedly mounted on the flange 103, is activated first. Heat is conducted through the flange to the membrane degassing probe 204 below, providing localized constant-temperature heating to the surrounding static insulating oil. At the same time, the carrier gas circulation pump 201 located at the bottom of the housing 101 starts operating, driving the carrier gas to form a closed-loop circulation within the system.

[0087] The carrier gas circulation path begins at the carrier gas circulation pump 201, and is delivered downwards through the inlet passage 202a of the dual-channel gas transmission pipe 202, directly entering the interior of the membrane degassing probe 204. After entering the membrane degassing probe 204, the carrier gas is diverted into the micro-cavities of numerous hollow fiber tubes 204b. Because both ends of the fiber tubes 204b pass through sealing rubber heads 204c and remain open, the carrier gas, while flowing inside the tubes, is separated from the external insulating oil entering the exchange chamber 204d through the oil passages 204f on the side wall of the support tube 204a by only a layer of hydrophobic and breathable membrane wall. During this process, the tightly packed porous support packing 204e within the exchange chamber 204d not only provides radial support for the flexible fiber tubes 204b, preventing deformation under high pressure, but also disperses the oil flow through its porous structure. Dissolved fault gases in the oil permeate into the interior of the fiber tubes 204b under the drive of the partial pressure difference, merging with the carrier gas flow.

[0088] The carrier gas carrying the faulty gas then flows out from the other end of the fiber tube 204b, converges into the outlet passage 202b of the dual-channel gas transmission pipe 202, and flows back upwards. Along this return path, the gas pressure sensor 205 and the gas temperature sensor 206 collect the airflow status in real time and feed it back to the control system.

[0089] Subsequently, the carrier gas is introduced into the gas detection unit 300 via the gas supply pipe 203. It first enters the photoacoustic resonant cavity 301, where a modulated laser is emitted through the laser emitter 302 at the top of the gas detection chamber 102, exciting sound waves that are picked up by the high-sensitivity microphone 303 within the cavity. After photoacoustic detection, the carrier gas continues to flow into the series-connected circulating gas chamber 304, sequentially passing through the hydrogen probe 305 and the infrared probe 306, completing the scanning of hydrogen and background components.

[0090] Finally, the carrier gas that has completed the full spectrum detection flows out through the return gas channel 307 and is drawn back into the intake passage 202a of the dual-channel gas transmission pipe 202 (or connected to the pump's suction port), thus completing a complete closed-loop cycle and ensuring non-destructive and continuous monitoring of gas under low-oil conditions.

[0091] Example 4 Reference Figure 9 and Figure 10 This is the fourth embodiment of the present invention. This embodiment focuses on the targeted optimization of the material formulation and preparation process of the hollow fiber separation membrane, the core consumable of the online membrane degassing unit. In existing online monitoring applications of power equipment, the conventionally used homogeneous microporous membranes of polytetrafluoroethylene (PTFE) or ordinary polyvinylidene fluoride (PVDF) suffer from performance bottlenecks such as low gas transmission efficiency and insufficient mechanical strength and structural stability of the membrane fibers.

[0092] 4.1 Gas transport efficiency modification and optimization Traditional membrane materials rely primarily on free volumes or complex, tortuous microporous pathways between polymer chains for gas diffusion, resulting in significant resistance to transmembrane transport and low gas transfer efficiency. Particularly under the static oil conditions characteristic of low-oil equipment, an effective concentration gradient is difficult to establish across the membrane, leading to severe lag in response time (often tens of hours) to large molecules such as ethane and ethylene. Furthermore, the membrane is highly susceptible to adsorption of impurities in the oil, further reducing flux and failing to meet the requirements for rapid detection of transient faults.

[0093] 4.1.1 Modification of bulk transport efficiency In hollow fiber membranes, transmembrane transport of gas molecules primarily follows a post-dissolution diffusion mechanism. The core microscopic factors affecting transport efficiency (flux) mainly include two points: first, the free volume within the polymer matrix, i.e., the unoccupied spaces between polymer chain segments, which determines the steric hindrance for gas molecule hopping diffusion; and second, the effective diffusion path length of the dense skin layer on the membrane surface. Traditional PVDF homogeneous membranes, due to the dense packing of polymer chain segments, have a limited free volume and a highly tortuous gas diffusion path, resulting in a slow gas response rate under static oil micro-pressure differential conditions.

[0094] Take 18 parts by weight of PVDF resin, 75 parts by weight of DMAC solvent and 7 parts by weight of PVP pore maker, and mechanically stir at 60℃ for 24 hours until completely dissolved. After standing and degassing, a uniform and transparent base casting solution is obtained, which can be used as a carrier for subsequent filler modification or a precursor for process modification.

[0095] Experimental groups: A1 with 2 parts fluorinated mesoporous silica nanoparticles (F-MSN); A2 with 3 parts zeolite imidazole ester framework-8 (ZIF-8); A3 with 5 parts 1-butyl-3-methylimidazolium hexafluorophosphate ionic liquid; A4 after spinning with pure substrate casting solution, the film was subjected to 200% biaxial mechanical stretching at 120℃; A5 with 3 parts MOF-74 (NI) metal-organic framework; A6 after spinning with pure substrate casting solution, the film was placed in a plasma cleaner and subjected to oxygen plasma surface etching at 100W power for 60 seconds; A7 with 3 parts fumed silica; A8 with 4 parts nano-titanium dioxide; A9 with the solvent replaced by dibutyl phthalate (DBP).

[0096] A constant-volume gas permeation apparatus was used, with the test pressure maintained at 0.2 MPa, to test three pure gases: hydrogen, acetylene, and ethane. The effect of the modification on accelerating the transport of gas molecules with different kinetic diameters was evaluated, particularly to verify whether a sieving effect exists that hinders the transport of large molecules.

[0097] The acetylene permeation curve was analyzed using the time-delay method. The gas transport mechanism was deconstructed to determine whether the performance improvement stemmed from increased kinetic diffusion or enhanced thermodynamic dissolution, thereby verifying whether the modifier's specifications met expectations.

[0098] The test results are as follows: Among them, the zeolite imidazole ester framework group showed the best performance in hydrogen flux but extremely poor performance in ethane flux, exhibiting a strong pore size sieving effect; the ionic liquid group had an advantage in acetylene solubility coefficient, but its diffusion coefficient was significantly low, limiting the overall transport rate; the biaxial stretching process group had the best ethane flux, but its acetylene solubility coefficient was extremely low, indicating that it lost its affinity and selectivity for specific gases; the nickel-based metal-organic framework group had the most significant effect on improving acetylene flux, but also significantly inhibited hydrogen transport efficiency; the plasma etching process group obtained the highest diffusion coefficient, but its solubility coefficient was poor, indicating that surface etching destroyed the adsorption active sites of the material.

[0099] In contrast, fluorinated mesoporous silica maintained high fluxes of hydrogen, acetylene, and ethane, and achieved a good balance between diffusion and solubility coefficients, with no obvious performance shortcomings. Considering the requirements for full-spectrum gas monitoring, fluorinated mesoporous silica was ultimately selected as the modified raw material to improve gas transport efficiency.

[0100] 4.1.2 Optimization of Fluorinated Mesoporous Silica Addition Amount To investigate the optimal addition amount of fluorinated mesoporous silica and balance the contradiction between gas transport efficiency and membrane structure stability, a gradient variable was set from 0.5 parts to 5.0 parts (in increments of 0.5 parts).

[0101] Excessive nanoparticles can easily aggregate within the membrane, forming large, non-selective defects that reduce the membrane's oleophobicity. Using a self-made hydraulic test bench, transformer oil was applied in stages to the outside of the membrane fibers at a rate of 0.01 MPa / min. The pressure at which the first oil droplet appeared on the permeate side was recorded as the transformer oil penetration pressure, thus testing the membrane's ultimate ability to resist liquid oil wetting and permeation.

[0102] High levels of inorganic fillers can disrupt the continuity of polymer molecular chains, leading to embrittlement of the membrane material. A universal testing machine with a tensile rate set to 50 mm / min is also required to record the maximum stress at fracture. Tensile strength is used to determine whether the membrane fibers possess sufficient mechanical reliability during long-term operation.

[0103] Test results are as follows Figure 9 As shown, by fitting the experimental data, the acetylene permeation flux first increases and then decreases with the increase of the addition amount. The peak value shown by the fitting curve is located at 2.88 parts, at which the gas transmission efficiency is the highest. The transformer oil penetration pressure remains stable at low addition amounts, but then shows an exponential decay. Its marginal benefit is optimal at 1.82 parts. The tensile strength exhibits typical characteristics of nano-reinforcement and agglomeration embrittlement, with the maximum value of the fitted curve located at an addition amount of 1.36 parts.

[0104] Taking into account the weights of transmission efficiency, safety, and mechanical lifespan, a weighted average was calculated for the three theoretically optimal points, yielding a comprehensive theoretical optimal addition amount of 2.02 parts. Considering the precision of ingredient preparation and ease of operation in industrial production, this embodiment ultimately selected 2 parts as the optimal addition amount of fluorinated mesoporous silica.

[0105] 4.2 Modification of Membrane Filament Mechanical Strength and Structural Stability Secondly, the membrane fibers suffer from insufficient mechanical strength and structural stability. Low-oil equipment often involves high-voltage electric fields and certain oil pressure fluctuations. Ordinary hollow fiber membranes are highly susceptible to radial collapse (flattening) or creep rupture during long-term immersion and negative pressure pumping. Once the membrane fibers deform, their effective air permeability area is significantly reduced; and if they break, insulating oil will be directly drawn into the gas path, causing irreversible damage to expensive photoacoustic spectroscopy sensors and potentially introducing air bubbles into the equipment itself.

[0106] 4.2.1 Screening for Modification of Membrane Filaments Based on Mechanical Strength and Structural Stability Mechanical failure of hollow fiber membranes during long-term service often stems from chain segment slippage and creep in the amorphous regions of the polymer, while oil contamination arises from the adsorption of organic molecules by high surface energy sites. Improving mechanical strength hinges on limiting the relative displacement of molecular chains (e.g., introducing crosslinking points or crystalline regions), while enhancing oleophobicity relies on reducing the surface energy of the membrane (e.g., introducing fluorine / silicon groups). However, conventional surface densification or coating treatments often clog effective gas transport micropores, causing a sharp decline in permeation flux. Therefore, this experiment aims to find a synergistic modification process that can construct a robust, low-surface-energy protective layer while maximizing the preservation of gas transport channels.

[0107] Experimental groups: B1 The base membrane was immersed in a 2% (w / w) perfluoropolyether (PFPE) solution for 30 minutes, then dried at 60°C; B2 The base membrane was first self-polymerized and deposited in a dopamine solution for 2 hours, then immersed in a 1H,1H,2H,2H-perfluorooctyltriethoxysilane solution for reaction; B3 The base membrane was immersed in an isopropanol solution containing 5% hexamethylenediamine and chemically crosslinked at 60°C for 24 hours; B4 The base membrane was fixed in length and then heat-treated in an oven at 160°C (close to the melting point of PVDF) for 2 hours; B5 After pretreatment with argon plasma, surface grafting polymerization was carried out by introducing pentafluorostyrene monomer vapor; B6 An additional 10% (w / w) thermoplastic polyurethane elastomer (TPU) relative to the mass of PVDF was added to the casting solution before spinning; B7 A fluorosilane sol precursor was prepared, coated on the membrane surface, and in-situ hydrolytic condensation was carried out at 80°C to form a nano-hybrid network.

[0108] In addition to testing tensile strength and acetylene permeation flux, it is also necessary to test the oil contact angle to determine the oleophobic modification effect.

[0109] The test results are as follows: Among the various membrane materials, the perfluoropolyether impregnation group achieved an extremely high oil contact angle, but its acetylene permeation flux plummeted, indicating that the thick coating severely blocked the gas channels. The diamine chemical crosslinking group significantly improved tensile strength by constructing a three-dimensional network structure, but its improvement on the oil contact angle was almost zero, failing to meet antifouling requirements. While the polyurethane blend group maintained good flux, the introduction of oleophilic components resulted in an oil contact angle lower than the substrate, increasing the risk of oil contamination. Sol-gel in-situ crosslinking exhibited excellent overall performance; therefore, it was ultimately adopted as the final treatment process for the membrane material.

[0110] 4.2.2 Optimization of Sol Concentration The concentration of the sol directly determines the growth kinetics of the in-situ crosslinking layer. If the concentration is too low, a continuous and dense coating layer cannot be formed on the membrane surface, resulting in oleophobic failure and insufficient structural support. If the concentration is too high, the crosslinking layer will become excessively thick, which will not only severely block the nanopores of gas transport, causing a sharp drop in flux, but also introduce large internal stress due to the high proportion of inorganic components, leading to increased brittleness of the membrane fibers. Under stress, the coating is prone to cracking and thus loses its protective function.

[0111] To precisely control the thickness and density of the hybrid network coating, this experiment selected the concentration of fluorinated silane sol as a key process variable (range 0.5%~5.0%, step size 0.5%). While keeping other process conditions constant (crosslinking at 80℃, time 2 hours), the mass fraction and even grouping of the sol precursor solution were adjusted. In addition to conventional indicators, the elongation at break was also measured to quantitatively assess the negative impact of high-concentration coatings on the toughness of the membrane material.

[0112] Test results are as follows Figure 10 As shown, the acetylene permeation flux and elongation at break decrease monotonically with increasing concentration, indicating that the thinner the coating, the lower the gas transport resistance and the better the membrane fiber toughness, and the theoretical optimal point tends to be in the low concentration range; the tensile strength exhibits a parabolic characteristic of first increasing and then decreasing, with the fitted peak located at a concentration of 2.45%, at which point the hybrid network structure is most complete and no brittle cracking has occurred; the transformer oil contact angle tends to level off after the concentration reaches 3.15%, indicating that the surface coverage has reached saturation, and further increasing the concentration only increases the thickness without increasing the surface energy.

[0113] Taking into account the balance between structural strength, oleophobic durability, and transport efficiency, a weighted average calculation of the above key inflection points yielded a theoretically optimal concentration of 2.48%. For ease of industrial formulation, 2.5% was ultimately selected as the optimal sol concentration for the sol-gel in-situ crosslinking process.

[0114] 4.3 Preparation process and final performance verification of modified composite membrane Preparation of S1 modified casting solution: Take 18 parts PVDF resin, 75 parts DMAC solvent, 7 parts PVP pore-forming agent, and 2.0 parts fluorinated mesoporous silica (F-MSN). First, ultrasonically disperse F-MSN in the solvent, then add PVP and PVDF, stir and dissolve at 60℃ for 24 hours, and then degas under vacuum to obtain a uniformly dispersed modified casting solution.

[0115] S2 hollow fiber membrane spinning: Using a dry-wet spinning machine, the modified casting solution is extruded through a spinneret (the core solution is a DMAC / water mixture), passed through a 10cm air bath, and then placed in a 25℃ water coagulation bath to undergo phase inversion and form a membrane. The resulting nascent membrane fibers are soaked in deionized water for 48 hours to remove residual solvent, and then naturally air-dried to obtain the modified base membrane.

[0116] S3 Surface Sol-Gel In-situ Crosslinking Treatment: Prepare a 2.5% (w / w) fluorinated silane sol precursor solution (solvent: ethanol, pH adjusted to 4.0). Immerse the modified base film in the precursor solution and perform a pull coating, then place it in an 80°C constant temperature oven for 2 hours to allow the silane to undergo in-situ hydrolysis condensation and crosslinking curing on the film surface.

[0117] S4 Post-processing: The membrane fibers were removed and ultrasonically cleaned with anhydrous ethanol to remove unreacted monomers. Finally, the membrane was dried in a vacuum oven at 60°C for 12 hours to obtain the final fluorinated mesoporous silica-doped enhanced PVDF mixed matrix composite membrane.

[0118] The prepared modified composite membrane exhibits the following properties: acetylene permeation flux of 46.85 GPU, hydrogen permeation flux of 46.12 GPU, ethane permeation flux of 17.95 GPU, tensile strength of 5.41 MPa, elongation at break of 117.56%, transformer oil contact angle of 130.45°, transformer oil penetration pressure of 0.82 MPa, and acetylene diffusion coefficient D. The solubility coefficient S of acetylene is units.

[0119] In summary, this invention achieves non-destructive, high-precision online sensing for low-oil electrical equipment by constructing an intelligent monitoring architecture based on the fusion of online membrane degassing and multi-dimensional heterogeneous sensing. Its core lies in utilizing a parallel array technology combining photoacoustic spectroscopy and heterogeneous sensing to solve the problem of insufficient acetylene capture capability under static oil conditions using traditional methods; and by dynamically loading a fault feature fingerprint model and a risk assessment matrix, it overcomes the limitations of fixed parameters in traditional monitoring equipment.

[0120] By constructing a closed-loop diagnostic system that integrates environmental matrix correction, multi-source data fusion, and differentiated comprehensive evaluation, the sensitivity and accuracy of identifying early latent faults in low-oil equipment have been significantly improved. This provides strong technical support for operation and maintenance decisions for key nodes such as high-voltage bushings and instrument transformers, and greatly reduces the risk of severe sudden failures.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, characterized in that, Performed by a computer device, including the following steps: S1. Download the fault characteristic data and associated risk assessment matrix of the device model to be tested from the cloud database, and then load them into the running memory; S2. Send gas path control commands to the online membrane degassing unit to establish the permeation equilibrium state of oil-gas separation, control the carrier gas flow through the parallel detection array composed of three heterogeneous sensors, and acquire trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. S3. Based on the multidimensional heterogeneous sensing dataset, perform environmental matrix effect compensation on the trace acetylene photoacoustic spectral signal to obtain the corrected acetylene concentration value; perform calibration and analysis on the multidimensional heterogeneous sensing dataset to generate standardized gas state data. S4. Input the gas state data into the fault characteristic data and perform a comprehensive assessment in conjunction with the risk assessment matrix; generate an equipment monitoring report based on the comprehensive assessment results.

2. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 1, characterized in that, The process of downloading data from the cloud database in step S1 specifically includes: searching the cloud database based on the input device model and operating data including voltage level, insulation structure type, and oil-gas volume ratio parameters; The fault characteristic data corresponding to the device under test includes the acetylene concentration alarm threshold, hydrogen production rate threshold and carbon-oxygen ratio fault criteria for the type of device under test. The risk assessment matrix includes the hydrogen concentration response signal, the trace acetylene photoacoustic spectral signal, and the weighting coefficients of background gas component data in the fault determination process. For capacitive bushing equipment, the acetylene concentration data in the risk assessment matrix has the highest weighting coefficient and the lowest trigger threshold; For current transformer equipment, the weighting coefficients of hydrogen and total hydrocarbon data in the risk assessment matrix are increased, and a ratio weighting for solid insulation aging is introduced.

3. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 1, characterized in that, The process of establishing the permeation equilibrium state of oil-gas separation in step S2 specifically includes: Read the temperature and pressure data of the online membrane degassing unit under its current state, and set the initial isothermal heating target range and carrier gas circulation flow rate value based on the preset temperature compensation curve; Start the constant temperature heating module and carrier gas circulation pump integrated on the online membrane degassing unit; dynamically adjust the heating power during the heating process until the operating temperature of the membrane module is within the constant temperature target range to maintain a constant gas permeability; The carrier gas circulation pump is controlled to drive the carrier gas to flow on the gas side of the membrane module at the carrier gas circulation flow rate value. The pressure fluctuation of the gas side flow induces micro-convective disturbance on the oil side surface of the membrane module, which destroys the gas-depleted boundary layer in the stagnant oil layer and accelerates the permeation and diffusion of dissolved gas in the oil to the gas side. Continuously monitor the total pressure change rate in the gas path. If the total pressure change rate is lower than the judgment threshold within the preset time window, it indicates that the oil-gas separation has reached the permeation equilibrium state.

4. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 2, characterized in that, The parallel detection array composed of the three heterogeneous sensors in step S2 includes a photoacoustic spectroscopy detection path, a hydrogen gas sensing detection path, and a background monitoring path. The photoacoustic spectroscopy detection path uses a tunable laser diode as a light source to lock the near-infrared characteristic absorption spectral lines of acetylene gas, drives the laser to perform intensity modulation at a preset frequency, collects the acoustic wave signal generated in the photoacoustic cell due to the photothermal effect through a high-sensitivity microphone, and uses a lock-in amplifier to extract the signal component with the same frequency as the laser modulation frequency, as the trace acetylene photoacoustic spectral signal. The hydrogen sensing detection circuit utilizes a micro hot plate semiconductor sensor and an electrochemical sensor to contact the flowing carrier gas, measure the changes in conductivity and current caused by the hydrogen redox reaction, and generate a hydrogen concentration response signal. The background monitoring path utilizes a non-dispersive infrared sensor, employing a broadband infrared light source in conjunction with a filter of a specific wavelength, to measure the infrared absorption intensity of the carrier gas for the characteristic bands of carbon monoxide, carbon dioxide, and total hydrocarbons, thereby generating background gas composition data. The hydrogen concentration response signal is packaged with the background gas component data to form the multidimensional heterogeneous sensing dataset.

5. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 4, characterized in that, The process of obtaining the corrected acetylene concentration value through inversion in step S3 specifically includes: The multidimensional heterogeneous sensing dataset was analyzed to extract the hydrogen concentration, carbon monoxide concentration, carbon dioxide concentration, and the proportion of carrier gas background components. Using the extracted concentration values ​​of each component, the equivalent relaxation time, sound velocity, and thermal conductivity of the current mixed gas are calculated based on the molecular dynamics of the mixed gas, and then the cell constant and Grindelwald constant of the photoacoustic detection are updated. A photoacoustic inversion equation was constructed using the updated cell constant and the Grindelsen constant. The trace acetylene photoacoustic spectral signal was substituted into the equation to eliminate the signal gain drift caused by the background gas components and to calculate the acetylene compensation concentration value after environmental matrix effect compensation. The current temperature and pressure data of the detection chamber are read, the acetylene correction concentration value is normalized to standard state, and the final acetylene correction concentration value is output.

6. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 5, characterized in that, The process of generating standardized gas state data in step S3 specifically includes: The zero-point historical records of the hydrogen sensing detection path and the background monitoring path are called to calculate the current baseline drift of the sensor. The baseline drift is then subtracted from the hydrogen concentration response signal and the background gas component data to obtain the net response value. The net response value is then converted into the corresponding preliminary physical concentration value by fitting the pre-stored full-range response characteristic curve of the sensor. By combining the real-time temperature and pressure data of the detection chamber, the acetylene correction concentration value and the preliminary physical concentration value are uniformly converted into standard concentration units. The normalized concentration data of acetylene, hydrogen, carbon monoxide, and carbon dioxide, along with the calculated total hydrocarbon value, are arranged according to a preset time sequence and dimension to generate standardized gas state data.

7. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 6, characterized in that, The comprehensive evaluation process in step S4 specifically includes: The concentrations of each gas in the gas state data are compared with the acetylene concentration alarm threshold and the hydrogen production rate threshold, and a primary over-limit flag is generated if the threshold is exceeded. The gas state data is weighted and summed using the weighting coefficients to calculate a comprehensive risk index that reflects the overall degree of equipment abnormality. Based on the aforementioned carbon-oxygen ratio fault criterion, the ratio of carbon dioxide to carbon monoxide and the ratio of characteristic gases are calculated to identify solid insulation aging or moisture defects inside the equipment. In addition, if the object to be detected is a capacitor-type bushing device, and the acetylene correction concentration value is greater than the minimum effective detection limit set by the system, and the comprehensive risk index is greater than the preset safety baseline, then the device status will be determined as a high-risk discharge fault state. If the object of detection is a current transformer, the equipment status is classified and determined based on the numerical range of the comprehensive risk index and the results of the fault type pattern recognition.

8. The method for detecting dissolved gases in oil in low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array according to claim 7, characterized in that, The process of generating the device monitoring report specifically includes: Extract current gas state data, comprehensive risk index and comprehensive assessment results to generate a state snapshot reflecting the current insulation state of the equipment; By accessing historical monitoring data from the equipment, historical trend curves of acetylene, hydrogen, and total hydrocarbon concentrations are plotted to estimate the potential failure development rate within a preset time window. Based on the fault level and type in the comprehensive evaluation results, the corresponding operation and maintenance handling strategies are retrieved from the pre-set expert knowledge base; for equipment determined to be in a high-risk discharge fault state, handling suggestions including shortening the detection cycle, offline retesting, or emergency shutdown need to be matched. The status snapshot, the potential fault development rate, and the handling recommendations are packaged into a structured electronic document and pushed to the designated monitoring terminal through a communication interface.

9. A dissolved gas detection system for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, characterized in that, The system is used to perform the method according to any one of claims 1 to 8, specifically including: The configuration management and storage module is used to store the device model and the corresponding working data type, and download and store the matching fault characteristic data and risk assessment matrix from the cloud database according to the device model to be tested; The permeation balance control module is used to send gas path control commands to the online membrane degassing unit to adjust the temperature and flow rate, and to establish and maintain the permeation balance state of oil-gas separation. The sensor data processing module is used to receive electrical signals generated by the parallel detection array of three heterogeneous sensors, collect and preprocess them, and obtain trace acetylene photoacoustic spectral signals and multidimensional heterogeneous sensing datasets. The acetylene concentration correction module is used to compensate for the environmental matrix effect of the trace acetylene photoacoustic spectral signal based on the background gas component information in the multidimensional heterogeneous sensing dataset, and to invert the acetylene correction concentration value. The gas state analysis module is used to calibrate and normalize the multidimensional heterogeneous sensing dataset to generate standardized gas state data. The fault feature data parsing module is used to compare the gas state data with the various thresholds and criteria in the fault feature data to generate a single index limit-out flag and a pattern recognition result. The comprehensive evaluation module is used to call the risk assessment matrix, combine the single indicator limit-breaking flag and pattern recognition results to perform multi-source weighted calculation, and output the comprehensive evaluation result of the device. The report generation module is used to generate an equipment monitoring report based on the comprehensive assessment results, which includes a status snapshot, potential failure progression rate, and remedial recommendations.

10. A dissolved gas detection device for low-oil equipment based on online membrane degassing and a multidimensional heterogeneous sensor array, characterized in that, The device is used to perform the method of any one of claims 1 to 8 and serves as a detection carrier for the system of claim 9, specifically comprising: The fixing unit (100) includes a housing (101) and a gas detection chamber (102). A flange (103) is provided at the end of the housing (101) away from the gas detection chamber (102). A heating ceramic ring (104) is fixedly installed on the flange (103). A gas delivery unit (200) includes a carrier gas circulation pump (201) fixedly installed at the bottom of the housing (101). The carrier gas circulation pump (201) is provided with a dual-channel gas transmission pipe (202) and a gas delivery pipe (203) inside the housing (101). The end of the dual-channel gas transmission pipe (202) is coaxial with the flange (103). A membrane degassing probe (204) is fixed at the end of the dual-channel gas transmission pipe (202). A gas pressure sensor (205) and a gas temperature sensor (206) are provided on the inner wall of the gas outlet passage of the dual-channel gas transmission pipe (202). A gas detection unit (300) includes a photoacoustic resonant cavity (301) connected to the gas transmission pipe (203). A laser emitter (302) is installed on the top of the photoacoustic resonant cavity (301). The laser emitter (302) passes through the top of the gas detection chamber (102). A high-sensitivity microphone (303) is installed inside the photoacoustic resonant cavity (301). A circulating gas chamber (304) is connected to the gas outlet of the photoacoustic resonant cavity (301). A hydrogen probe (305) and an infrared probe (306) are installed on the inner wall of the circulating gas chamber (304). A return gas passage (307) is connected to the gas outlet of the circulating gas chamber (304) and is connected to the gas inlet passage (202a) of the dual-channel gas transmission pipe (202). The membrane degassing probe (204) includes a support tube (204a), the inside of which is filled with a fiber tube (204b). Sealing nozzles (204c) are provided at both ends of the support tube (204a). Both ends of the fiber tube (204b) pass through the sealing nozzles (204c) and remain open, so that the inner cavity of the fiber tube (204b) forms a circulating gas path with the inlet passage (202a) and outlet passage (202b) of the dual-channel gas transmission tube (202). The support tube (204a) Together with the sealing rubber heads (204c) at both ends, an exchange cavity (204d) is defined. The gap between the fiber tubes (204b) and the exchange cavity (204d) is tightly filled with porous support filler (204e). The porous support filler (204e) is used to restrict the radial displacement of the fiber tubes (204b) and provide structural support. The side wall of the support tube (204a) is provided with an oil passage hole (204f) so that external insulating oil can enter the exchange cavity (204d) and contact the outer wall of the fiber tubes (204b).

Images