In-situ detection method and device for dissolved gas in transformer oil based on photoacoustic spectrometry technology

Abstract

The invention relates to an in-situ detection method and device for dissolved gas in transformer oil based on a photoacoustic spectrometry technology, and relates to the technical field of power equipment state monitoring and detection.The in-situ detection method comprises the following steps that temperature and humidity data in a detection cavity are obtained, and the current gas sound velocity is calculated according to an analytical model of temperature, humidity and gas sound velocity change; calculating a resonance frequency priori value according to the gas sound velocity; and executing narrowband frequency sweeping by taking a resonance frequency priori value as a center, collecting an amplitude response curve and a phase response curve of the detection cavity, and identifying and calibrating characteristic fingerprints of a main formant and an adjacent parasitic mode. The resonant frequency prior value is obtained through temperature and humidity collection and model calculation, a main peak and a parasitic peak are calibrated in combination with narrow-band frequency sweep, frequency is locked in a self-adaptive mode, a parasitic mode is suppressed, sound velocity is regulated and stabilized by heating and dry gas, reference absorption line closed-loop calibration is introduced, the long-term stability and the anti-interference capability of photoacoustic detection are improved, and the detection accuracy is improved. And DGA false signals and misjudgment risks are reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power equipment state monitoring and detection, in particular to a transformer oil dissolved gas in-situ detection method and device based on photoacoustic spectroscopy. BACKGROUND

[0002] The transformer oil dissolved gas in-situ detection based on photoacoustic spectroscopy refers to using the principle of photoacoustic spectroscopy to build an oil-gas exchange channel between the transformer oil and the detection cavity through a microphone combined with a water-resistant gas-permeable membrane (such as fluorinated polymer PTFE, PVDF or silicon-based coating), so that the trace gas components dissolved in the transformer oil achieve rapid oil-gas balance on both sides of the membrane, and effectively block the liquid oil and moisture from entering the detection cavity, ensuring the purity and stability of the measurement environment. Subsequently, the detection light source irradiates into the cavity with a specific wavelength, and the measured gas absorbs light energy and produces periodic thermal expansion to form an acoustic signal. The high-sensitivity microphone collects the sound pressure changes in real time and converts them into an electrical signal, which is analyzed by a photoacoustic spectroscopy data processing algorithm to obtain gas type and concentration information, thereby realizing online, non-sampling, real-time in-situ detection of dissolved gases in transformer oil, and providing a fast and accurate basis for transformer operation state evaluation and fault diagnosis.

[0003] In the process of gas detection based on photoacoustic spectroscopy, the detection cavity is usually designed as an acoustic resonant structure to utilize the energy accumulation effect of acoustic waves at the resonant frequency of the cavity to improve detection sensitivity. However, when the temperature and humidity in the detection cavity fluctuate slightly, it will cause a slight change in the sound speed of the gas in the cavity, resulting in a shift in the inherent resonant frequency of the cavity. When the frequency of the modulated light source fails to match the changed resonant frequency, the acoustic signal received by the microphone will mismatch the designed main resonant mode. In a high-quality factor (high Q value) cavity, this frequency mismatch easily triggers abnormal amplification of adjacent resonant modes or parasitic modes, forming false enhanced signals unrelated to the target gas absorption, causing significant deviations in gas concentration calculation results, and even leading to incorrect judgments of transformer fault states, causing unnecessary shutdown or maintenance risks. SUMMARY

[0004] The present application provides a transformer oil dissolved gas in-situ detection method based on photoacoustic spectroscopy, which includes the following steps: Obtain the temperature and humidity data in the detection cavity, calculate the current gas sound speed based on the analytical model of temperature and humidity changes, and calculate the resonant frequency prior value from the gas sound speed; Perform narrowband frequency sweeping centered on the resonant frequency prior value, collect the amplitude response curve and phase response curve of the detection cavity, identify and calibrate the characteristic fingerprints of the main resonant peak and adjacent parasitic modes; Calculate the frequency offset and phase offset of the main resonance peak based on the characteristic fingerprint, generate the main resonance peak credibility score and mismatch risk level; According to the mismatch risk level, the frequency of the modulation light source is locked to the position of the main resonance peak under the double constraints of amplitude and phase, and the parasitic mode gain is dynamically suppressed; After frequency locking, local heating control of the detection cavity is implemented to address the change in sound speed caused by temperature changes, and dry gas on the membrane side is introduced to adjust the sound speed caused by humidity changes, so as to stabilize the sound speed and fix the position of the main resonance peak; Under the condition of stable sound speed and main resonance peak position, a reference absorption line with known concentration is introduced for online verification. When the deviation between the real-time inversion gas concentration and the reference concentration exceeds the preset threshold, the frequency locking parameters are returned to be re-swept and corrected, and the closed-loop control of gas detection accuracy is completed.

[0005] Preferably, in the step of acquiring temperature and humidity data in the detection cavity and calculating the resonance frequency prior value, the following process is included: Multiple measurement points are set in the detection cavity, each of which is equipped with a platinum resistance thermometer with a measurement accuracy of not less than ±0.05 degrees Celsius and a thin film capacitive humidity sensor with a measurement accuracy of not less than ±1% relative humidity, and temperature and humidity data at different positions are collected; The temperature and humidity data of each measurement point are checked for validity and the arithmetic mean values are taken respectively, and are sent to the temperature and humidity corresponding relationship table established by standard gas calibration to obtain the current gas sound speed; The gas sound speed is combined with the physical size of the detection cavity, the shape of the end structure, the acoustic reflection characteristics and the energy loss correction value to determine the main resonance frequency prior value, and is fine-tuned in combination with the latest narrowband sweep result; The updated resonance frequency prior value is output to the frequency sweep control unit, and the last valid prior value is called as a temporary reference when the temperature and humidity change exceeds the preset threshold, until the new data is replaced by the new prior value after it is stable.

[0006] Preferably, in the step of performing narrowband sweep centered on the resonance frequency prior value, collecting the amplitude response curve and phase response curve of the detection cavity and calibrating the main resonance peak and the characteristic fingerprint of the parasitic mode, the following process is included: The resonance frequency prior value is set as the center frequency, the narrowband sweep start and end range is set as the prior value minus twice the main peak half-width to the prior value plus twice the main peak half-width, the frequency step value is not greater than one percent of the main peak half-width, and the detection cavity gas is replaced by low flow rate dry carrier gas under normal pressure; The light source with wavelength stability better than ±0.01 nanometers emits a modulated light beam and collimates it into the detection cavity. The amplitude and phase values of the sound wave at each frequency point are synchronously collected by the capacitive microphone to form the amplitude response curve and phase response curve data set. The main resonance peak is identified according to the amplitude peak position and phase mutation characteristics, and the peak frequency, amplitude, half-width, and phase mutation characteristics are recorded. The parasitic modes on both sides of the main peak are also identified and their characteristic parameters are recorded to establish a characteristic fingerprint library of the main peak and parasitic modes. When the frequency offset of the main peak exceeds ±5 Hz or the amplitude of the parasitic mode exceeds 30% of the amplitude of the main peak, the peak mismatch risk judgment result is output, and the frequency compensation value and the parasitic mode suppression weight are generated.

[0007] Preferably, after outputting the peak mismatch risk judgment result and generating the frequency compensation value and the parasitic mode suppression weight, the following processes are included: The frequency compensation value is applied to the frequency control of the modulated light source, so that the modulated light source frequency is locked to the compensated main resonance peak position under the amplitude and phase double constraint conditions. At the same time, the parasitic mode suppression weight is applied to the photoacoustic signal corresponding to the parasitic mode frequency, and the parasitic mode signal amplitude is dynamically reduced to prevent the parasitic mode from being misidentified as the main resonance peak by the frequency locking process, thereby ensuring that the subsequent gas concentration inversion is based on the real photoacoustic signal of the main resonance peak.

[0008] Preferably, in the step of calculating the frequency offset and phase offset of the main resonance peak based on the characteristic fingerprint and generating the main resonance peak credibility score and mismatch risk level, the following processes are included: The real-time frequency offset is calculated by differentiating the current measured main resonance peak frequency and the reference frequency in the reference characteristic fingerprint library; The phase data within the half-width range of the current main peak center are compared with the reference phase curve point by point to calculate the phase offset; The frequency offset and phase offset are normalized and combined into a main peak credibility score according to the preset weight, and the low risk, medium risk, or high risk level is determined according to the score interval; In the low risk, the frequency is locked, in the medium risk, the frequency is fine-tuned and the parasitic mode suppression is enhanced, and in the high risk, the narrowband sweep is returned to reacquire and update the characteristic fingerprint library.

[0009] Preferably, returning to narrowband sweep reacquisition and updating the characteristic fingerprint library under high risk level includes the following processes: Set the narrowband sweep range centered on the latest calculated resonance frequency prior value, and replace the detection cavity gas with low flow rate dry carrier gas at normal pressure; Re-acquire the amplitude response curve and phase response curve of the detection cavity, identify the peak frequency, peak amplitude, half-height width and phase mutation characteristics of the main resonance peak and all parasitic modes; Replace the corresponding data in the reference feature fingerprint library with all the newly acquired characteristic parameters, and verify the stability of the new feature fingerprint under the current environmental temperature and humidity conditions; Use the updated feature fingerprint library as the reference data for subsequent frequency offset and phase offset calculations to ensure that the latest acoustic characteristics of the detection cavity can be accurately reflected when the risk judgment is performed again.

[0010] Preferably, the modulation frequency adaptive tracking control according to the mismatch risk level determination result includes the following processes: When the risk level is low risk, the current modulation light source frequency is kept unchanged; when the risk level is medium risk, fine scanning is performed within ±1 Hz with 0.01 Hz step to find the frequency point with maximum amplitude and consistent phase transition direction as the locking point; when the risk level is high risk, return to narrowband sweep to re-acquire the feature fingerprint; After the locking frequency is established, according to the parasitic mode peak position and half-height width interval in the feature fingerprint library, a suppression frequency band is constructed within 5 Hz around the locking frequency, and the gain compression ratio is determined according to the parasitic mode amplitude ratio, and 1 Hz gradual transition zones are set at both ends of the suppression frequency band to ensure smooth connection of the response.

[0011] Preferably, after frequency locking, the gas sound speed is stabilized and the main resonance peak position is fixed, including the following steps: Locking state acquisition of outer wall temperature, multi-point temperature and humidity in the cavity and execution of ±5Hz, 1Hz step micro-sweep frequency determination of drift source; Paste a resistance heating sheet on the photoacoustic action area, the target temperature is the average of the first locking for 5 minutes, the temperature rise is ≤0.1℃ / min, and the steady-state fluctuation is ≤±0.05℃; Pass nitrogen with a dew point of ≤-40℃ through the back of the waterproof and breathable membrane, the flow is 1-5sccm, and the back pressure is 20-50Pa; Interval 5min, two rounds of ±5Hz re-scan, center frequency difference ≤3Hz, bandwidth difference ≤10%, amplitude difference ≤3% to solidify the parameters, if not up to standard, adjust the temperature by 0.1℃ or the flow by 1sccm and retest.

[0012] Preferably, under the condition that the gas sound speed and the main resonance peak position are stable, in order to ensure detection accuracy and realize closed-loop control, the following steps are included: Under the condition of stable gas sound velocity and main resonance peak position, a reference gas path filled with standard gas is introduced through an optical path bypass, the temperature, pressure and flow of the reference gas are controlled, the reference gas enters the micro absorption cavity for photoacoustic signal collection, the converted reference gas concentration is compared with the real-time inverted sample gas concentration, when the deviation exceeds the preset threshold, narrowband sweep frequency measurement is performed, the main resonance peak position is recalibrated and the modulation frequency locking parameter is updated, until the concentration deviation returns to within the threshold.

[0013] A transformer oil dissolved gas in-situ detection device based on photoacoustic spectroscopy technology, comprising a temperature and humidity driven resonance frequency prior value calculation module, a narrowband sweep and feature peak calibration module, a resonance peak shift analysis and mismatch risk assessment module, an adaptive frequency locking and parasitic mode suppression module, a sound velocity stabilization and resonance peak fixing control module, and a reference absorption line closed-loop calibration and precision maintenance module. The temperature and humidity driven resonance frequency prior value calculation module obtains temperature and humidity data in the detection cavity, calculates the current gas sound velocity based on the analytical model of the change of temperature and humidity and gas sound velocity, and calculates the resonance frequency prior value from the gas sound velocity. The narrowband sweep and feature peak calibration module performs narrowband sweep centered on the resonance frequency prior value, collects the amplitude response curve and phase response curve of the detection cavity, and identifies and calibrates the feature fingerprint of the main resonance peak and the adjacent parasitic mode. The resonance peak shift analysis and mismatch risk assessment module calculates the frequency shift and phase shift of the main resonance peak based on the feature fingerprint, and generates the main resonance peak credibility score and mismatch risk level. The adaptive frequency locking and parasitic mode suppression module performs adaptive tracking control of the modulation frequency according to the mismatch risk level, locks the modulation light source frequency to the main resonance peak position under the amplitude and phase double constraint conditions, and dynamically suppresses the parasitic mode gain. The sound velocity stabilization and resonance peak fixing control module, after frequency locking, implements local heating control of the detection cavity for sound velocity change caused by temperature change, and implements membrane side dry gas input adjustment for sound velocity change caused by humidity change, to stabilize the gas sound velocity and fix the main resonance peak position. The reference absorption line closed-loop calibration and precision maintenance module, under the condition of stable sound velocity and main resonance peak position, introduces a reference absorption line with known concentration for online verification, when the deviation between the real-time inverted gas concentration and the reference concentration exceeds the preset threshold, returns to re-sweep and corrects the frequency locking parameter, to complete the closed-loop control of gas detection precision.

[0014] The technical solution of the present application to solve the above technical problems is as follows: a transformer oil dissolved gas in-situ detection method and device based on photoacoustic spectroscopy technology.

[0015] The beneficial effects of the present application are: the present application obtains the prior value of the resonance frequency in advance through real-time collection and analysis model calculation of temperature and humidity, and accurately calibrates the main peak and parasitic peak characteristics in combination with narrowband sweep, realizes high credibility identification of the main resonance peak position; in combination with the adaptive frequency tracking of amplitude and phase double constraints, it ensures that the frequency of the modulated light source is always accurately locked to the main peak, and dynamically suppresses the parasitic mode response; at the same time, after frequency locking, through local heating and membrane side dry gas adjustment, the gas sound speed and resonance peak position are stabilized from the source; finally, the known concentration reference absorption line is introduced for online verification, and a closed-loop detection and automatic back-scan correction mechanism of concentration deviation is established. Through such multi-dimensional collaborative control and closed-loop feedback, the present application not only ensures the high consistency of the detection signal and the measured gas absorption characteristics, but also significantly improves the long-term stability and anti-environmental interference ability of the detection, thereby reducing false signals and misjudgment risks in online DGA monitoring, and providing high-precision, traceable detection data support for transformer operating state evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The method flowchart of the present application is a kind of in-situ detection method for dissolved gas in transformer oil based on photoacoustic spectroscopy technology.

[0017] Figure 2 The module schematic diagram of the present application is a kind of in-situ detection device for dissolved gas in transformer oil based on photoacoustic spectroscopy technology. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0019] In the description of the present application, the terms "first", "second" are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0020] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0021] Example 1, the present invention provides as follows Figure 1 The method for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy includes the following steps: The system acquires real-time temperature and humidity data within the detection chamber, calculates the current gas velocity based on a pre-defined analytical correlation model of temperature and humidity changes with gas velocity, and calculates the corresponding prior value of the resonant frequency based on the gas velocity. The specific steps of the method for calculating the gas velocity in the detection cavity and determining the prior value of the resonance frequency based on the relationship table of accurate acquisition of temperature and humidity at multiple measurement points and experimental calibration of sound velocity are as follows: To obtain real-time temperature and humidity data that accurately reflect the gas state inside the detection chamber, multiple measuring points are set up inside the chamber, each equipped with one temperature sensor and one humidity sensor. The temperature sensor is a platinum resistance thermometer with a measurement accuracy of at least ±0.05 degrees Celsius, and the humidity sensor is a thin-film capacitive humidity sensor with a measurement accuracy of at least ±1% relative humidity. The sensors are positioned near the central area of ​​the photoacoustic interaction zone, near the sidewall area of ​​the acoustic wave node, and near the gas inlet and outlet, to simultaneously acquire temperature and humidity information from both the acoustically sensitive area and the airflow exchange zone. Each sensor is fixed with a low-heat-capacity stainless steel bracket, and the probe is protected by a fluoropolymer microporous protective cover with good air permeability and oil and water resistance, ensuring direct contact between the measuring element and the gas inside the chamber and preventing contamination from oil mist and water droplets. The sensor signals are connected to the data acquisition device via shielded twisted-pair cables. Before entering the acquisition device, the signals pass through a low-pass filter and surge suppressor to suppress electromagnetic interference and transient impacts. The sampling frequency of the acquisition device is set to once per second to meet the dynamic response requirements of temperature and humidity changes on the gas sound velocity.

[0022] The temperature data and humidity data obtained from multiple measuring points are checked for validity, and abnormal values obviously beyond the physically reasonable range are removed. The arithmetic mean of the temperature data of each measuring point is taken as the current detection cavity average temperature, and the arithmetic mean of the humidity data of each measuring point is taken as the current detection cavity average humidity. The average temperature and average humidity data are sent to the temperature and humidity-gas sound speed corresponding relationship table established in the laboratory by standard gas calibration. The relationship table is a summary of data measured under different temperature and humidity conditions using an experimental cavity consistent with the structural parameters of the actual detection cavity, and is corrected according to the actual influence of the adiabatic properties, molar mass, heat capacity ratio and water vapor partial pressure of the gas composition. Through table lookup, the sound speed value of the detection cavity gas at the current temperature and humidity can be directly obtained. This process does not involve complex calculations, ensuring that the gas sound speed can be quickly and stably obtained in the field operation.

[0023] After obtaining the current gas sound speed, the sound speed is combined with the actual physical dimensions, end structure shape and acoustic reflection characteristics of the detection cavity to determine the prior value of the main resonance frequency of the detection cavity under the current state. In the determination process, the effective sound path of the detection cavity is corrected by the cavity length, cavity diameter and end face shape coefficient measured during the manufacturing stage, while considering the sound energy transmission and reflection ratio when the gas in the cavity body contacts the membrane interface, and the influence of acoustic energy loss is included in the correction value, so that the deviation of the obtained resonance frequency prior value from the actual main resonance frequency does not exceed one fifth of the detection cavity resonance bandwidth. In addition, during the long-term operation of the equipment, the main resonance peak position measured by the last narrowband sweep is combined to fine-tune the above-mentioned prior value, so that it remains stable and consistent with the long-term changes of the environment and gas composition. The resonance frequency prior value obtained in this way can provide an accurate and reliable center reference point for the subsequent frequency matching process.

[0024] After the determination of the resonance frequency prior value is completed, the value is output in real time to the execution unit for controlling the sweep process, so that the narrowband sweep can directly perform frequency scanning with the prior value as the center. This process is automatically repeated after each temperature and humidity data collection is completed, so that the prediction results of the sound speed and resonance frequency can be updated immediately when the temperature and humidity change. When the change amplitude of the temperature or humidity exceeds the pre-set credible threshold range, the system will automatically call the effective resonance frequency prior value calculated in the last time as a temporary reference, until the new temperature and humidity data are collected stably for multiple times and the sound speed calculation result is updated, and then the temporary reference value is replaced by the new prior value. This method can prevent the frequency prediction from jumping greatly due to sensor transient error or short-time disturbance of the environment, and ensure the stability and continuity of frequency matching, providing a stable frequency reference for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy technology.

[0025] The purpose of this step is to provide an accurate and reliable frequency center reference point for subsequent frequency matching and main resonance peak locking in the process of in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy technology, reducing the risk of frequency shift caused by environmental factors from the source. The main resonance frequency of the detection cavity directly determines the energy accumulation efficiency of the acoustic wave in the cavity, which in turn affects the amplitude and signal-to-noise ratio of the gas absorption signal. However, the speed of sound in the cavity will change with the slight fluctuations of temperature and humidity, causing the resonance frequency to drift. If this drift is not predicted and corrected in advance, the operating frequency of the modulated light source may not match the actual main resonance frequency, resulting in signal attenuation or even false enhancement of parasitic modes, which can cause errors in gas concentration calculation. Therefore, by collecting real-time temperature and humidity data in the detection cavity and combining the temperature and humidity vs. sound speed correspondence table calibrated through experiments, the current gas sound speed is calculated and the resonance frequency prior value is derived, which can actively predict and compensate for frequency drift at an early stage of environmental disturbance. This prior value not only significantly reduces the scanning range of subsequent narrow-band frequency sweeping, improving the speed and accuracy of locating the main resonance peak, but also enhances the stability and anti-interference ability of frequency locking, thereby providing a solid foundation for the detection accuracy and reliability of the whole process.

[0026] Perform narrow-band frequency sweeping measurement centered on the resonance frequency prior value, collect the amplitude response curve and phase response curve of the detection cavity, and identify and calibrate the main resonance peak position and the characteristic fingerprint of the adjacent parasitic mode according to the amplitude response curve and phase response curve; The specific steps to achieve accurate acoustic property measurement of the detection cavity are as follows: The resonance frequency prior value calculated through the temperature and humidity analytical correlation model is used as the center frequency, and the start and end frequency range of the narrow-band frequency sweeping is set. The upper and lower boundaries of this start and end frequency range are the prior value minus twice the half-height width of the main resonance peak and the prior value plus twice the half-height width of the main resonance peak, respectively, to ensure that the possible shift range of the main peak is covered and the interference of irrelevant frequency bands is reduced. The frequency step value is set to be no more than one percent of the half-height width of the main peak, for example, when the half-height width of the main peak is 20 Hz, the frequency step value is set to 0.2 Hz. During the frequency sweeping process, the gas pressure in the detection cavity is maintained at a constant pressure state (101.325 kPa) by a pressure stabilizing valve, and the gas composition in the cavity is continuously replaced by dry carrier gas at a low flow rate (not more than 50 mL / min) for not less than 5 minutes before the start of the frequency sweeping, to ensure uniform gas distribution and eliminate measurement errors caused by differences in residual gas composition.

[0027] In the execution of the above-mentioned sweep process, the light source emitting the modulated light beam is strictly corresponding to the target gas characteristic absorption peak, and the wavelength stability of the light source is better than ±0.01 nm. The modulation frequency is sequentially increased from the starting frequency to the terminal frequency, and the step is consistent with the frequency setting. The light beam is collimated by a collimating lens with a focal length of 25 mm and then enters the detection cavity. The light beam interacts with the gas molecules in the cavity to produce a periodic thermal expansion and contraction process, and an acoustic signal is formed in the cavity. The acoustic signal is received by a capacitive microphone with a sensitivity of not less than 10 mV / Pa. The received analog signal is amplified by a low-noise amplification circuit and then input into a data acquisition device. The device synchronously records the amplitude and phase values of the acoustic signal at each frequency point and stores them corresponding to the current sweep frequency, forming a complete set of original data of the amplitude response curve and the phase response curve.

[0028] According to the synchronous change characteristics of the amplitude response curve and the phase response curve, the position of the main resonance peak is identified. The identification of the main resonance peak is based on the peak point of the amplitude response curve, and the phase mutation near the point is used as a confirmation condition. In the confirmation process, the frequency difference of the amplitude drop to half of the peak value on both sides of the peak value is calculated to obtain the half-width value, and whether the peak symmetry and the half-width meet the main mode resonance characteristics is judged. When the amplitude curve is symmetrical and the half-width is within the designed range, and the phase presents a steep transition near the peak center, it is determined that the peak is the main resonance peak. At the same time, in the frequency range on both sides of the main peak, the secondary peaks with relatively enhanced amplitude and different phase change trend are identified, which are confirmed as parasitic modes. For each parasitic mode, its peak frequency, peak amplitude, half-width, phase mutation amplitude and mutation direction are recorded, and the above parameters are combined into a three-dimensional feature set of frequency-amplitude-phase to form a characteristic fingerprint library containing the main resonance peak and all parasitic modes.

[0029] The characteristic parameters of the main resonance peak and the parasitic modes measured at present are compared with the pre-stored reference characteristic fingerprint item by item. When the main peak frequency offset is more than ±5 Hz of the reference frequency, or the peak amplitude of the parasitic mode is more than 30% of the main peak amplitude, it is determined that there is a risk of peak position mismatch. According to the determination result, corresponding frequency compensation values are introduced in the subsequent modulation light source frequency locking process, and dynamic suppression weights are applied to the photoacoustic signals in the parasitic mode frequency range, so as to ensure that the modulation frequency is locked at the real main peak position and avoid that the parasitic mode is incorrectly identified as the main peak. Through this method based on narrow-band high-precision sweep, combined with amplitude and phase double-parameter calibration and establishment of characteristic fingerprint library, stable identification can be maintained under the condition that the main peak frequency drifts slightly and the signal-to-noise ratio decreases, and the parasitic mode can be quickly distinguished and suppressed in the case of being excited by the environment, so as to ensure that the subsequent gas concentration inversion result is based on the real and effective photoacoustic signal, and a stable closed loop from frequency prediction to peak position locking is realized.

[0030] The purpose of this step is to achieve high-precision identification and characteristic parameter calibration of the main resonance peak and adjacent parasitic modes in the gas detection process based on photoacoustic spectroscopy technology, thereby providing reliable basis for subsequent precise locking of modulation frequency and effective suppression of parasitic modes. The sensitivity of photoacoustic detection is highly dependent on the accurate matching of modulation frequency and main resonance frequency of the cavity, while the main peak frequency may drift slightly due to factors such as temperature change, humidity change, and gas component difference, and parasitic modes may be excited due to mechanical vibration, structural defects, or specific acoustic conditions. Once the main peak is not accurately identified or the parasitic mode is mistakenly considered as the main peak, it will cause distortion of signal amplitude and phase characteristics, thereby causing significant deviation in gas concentration inversion, and even incorrect judgment of device operating state. By narrow-band high-resolution sweeping centered on the prior value of the resonance frequency, and simultaneously recording the amplitude response curve and the phase response curve, this step can use the amplitude-phase double characteristics to locate the main peak position, and simultaneously extract the frequency, amplitude, half-width, and phase change characteristics of the parasitic mode, to construct a characteristic fingerprint library of the main peak and parasitic modes. This characteristic fingerprint library not only maintains stable identification of peak position in the case of frequency drift and noise ratio reduction, but also quickly distinguishes real signals from interference signals in complex acoustic environments, ensuring the accuracy and stability of subsequent modulation frequency locking, parasitic mode suppression, and gas concentration calculation process, thereby constituting a key basic link for reliable operation of the detection system.

[0031] According to the main resonance peak and parasitic mode characteristic fingerprints obtained by calibration, the frequency offset and phase offset of the current main resonance peak are calculated, the frequency offset and phase offset are converted into main resonance peak credibility scores, and corresponding mismatch risk level judgment results are generated; To ensure that the modulation light source frequency is stably locked at the true main resonance peak position of the detection cavity, and to prevent the parasitic mode from being mistaken for the main peak, after completing the characteristic fingerprint calibration of the main resonance peak and the parasitic mode, the frequency offset and phase offset of the current main peak are calculated based on real-time measurement results, and both are converted into main peak credibility scores, and finally the mismatch risk level judgment results are generated. Specifically, the following steps are included: After completing the main peak calibration, the current measured main resonance peak frequency is differentially calculated with the reference frequency of the main peak in the reference characteristic fingerprint library, to obtain the real-time frequency offset. The reference frequency is the peak value frequency measured by narrow-band high-precision sweeping under the condition that the detection cavity temperature is stable at 25.0℃±0.2℃, the relative humidity is maintained below 5%RH, and the gas in the cavity is dry air with a purity not less than 99.999%, and the measurement accuracy is not less than ±0.1 Hz. The calculation accuracy of the frequency offset is retained to 0.01 Hz, to ensure that a small drift of less than 0.05% can be detected in a high-Q cavity with a quality factor Q value greater than 300.

[0032] The phase data in the current main peak center frequency position and the left and right half-width range are selected, and the phase curve in the corresponding frequency interval in the reference characteristic fingerprint library is compared point by point, and the phase difference value of each corresponding frequency point is calculated. The arithmetic mean value of the phase difference value is taken as the phase shift of the current main peak, and the unit is degree (°). In the calculation process, the original phase data is smoothed by using the three-point moving average method to reduce the instantaneous phase jump caused by electrical noise and airflow disturbance. The positive and negative directions of the phase shift represent that the phase is advanced or delayed relative to the reference curve, and the absolute value size reflects the relative consistency of the acoustic response and the reference state.

[0033] The real-time frequency shift and the phase shift are normalized respectively, and combined into the main peak reliability score according to the pre-set weight coefficient. The score range is 0 to 100 points. In the detection cavity with Q value of 300, the frequency shift weight is set to 0.60, and the phase shift weight is set to 0.40; when the Q value is different, the weight can be recalibrated according to the peak position sensitivity test result. When the reliability score is greater than or equal to 90 points, it is determined that the matching state is stable and reliable; when the score is more than 70 points but less than 90 points, it is determined that there is a moderate risk in the matching state; when the score is less than 70 points, it is determined that there is a high risk in the matching state, and there may be a locking error or signal distortion.

[0034] The main peak reliability score is corresponded to three risk levels: low risk, medium risk and high risk. In the low risk state, the frequency of the modulated light source remains the current locked value; in the medium risk state, the frequency of the modulated light source needs to be adjusted in the range of ±1 Hz based on the current locked value, and the gain suppression coefficient of the parasitic mode frequency band is increased; in the high risk state, directly return to the narrowband sweep step to reacquire the amplitude response curve and the phase response curve, reposition the main peak and the parasitic mode position, and update the characteristic fingerprint library. Through this risk level determination method based on frequency and phase double parameters, the main peak position can still be accurately identified under the conditions of temperature and humidity fluctuation, gas composition change or cavity mechanical disturbance, the gas concentration calculation error caused by frequency mismatch is significantly reduced, and the stability and reliability of the photoacoustic detection process are ensured.

[0035] The function of this step is to accurately evaluate the reliability of the frequency lock of the modulated light source and timely find the possible peak mismatch risk by real-time comparison and analysis of the main resonance peak and the parasitic modal characteristic fingerprint obtained by calibration in the gas detection process based on photoacoustic spectroscopy technology. The sensitivity and accuracy of photoacoustic detection are highly dependent on the accurate matching of the modulation frequency and the main resonance peak frequency, and the main peak frequency and phase will be slightly shifted due to the influence of temperature fluctuations, humidity changes, gas component drift and mechanical vibration, and parasitic modes may be excited due to structural resonance or environmental disturbance. When the main peak and the modulation frequency are mismatched, it will cause the signal amplitude to decrease or the phase to distort, thereby directly affecting the gas concentration inversion accuracy, and even causing fault type misjudgment. This step converts the real-time frequency offset and phase offset of the main resonance peak into a credibility score according to the weight, and generates a graded mismatch risk judgment result, which not only can trigger the frequency fine-tuning or parasitic modal suppression in time when the main peak drifts slightly, but also can directly return to the narrowband frequency scanning step to update the characteristic fingerprint in the high-risk state, forming a closed-loop control link of prediction, monitoring, feedback and correction. The implementation of this step ensures that the main peak position can still be stably identified in the high-Q value and narrow bandwidth cavity environment, significantly reduces the measurement error and false alarm risk caused by frequency mismatch, and provides key technical support for the stability and reliability of the entire detection process.

[0036] According to the mismatch risk level judgment result, perform adaptive tracking control of the modulation frequency, lock the modulation light source frequency to the main resonance peak position under the condition of amplitude constraint and phase constraint, and dynamically suppress the parasitic modal gain during the locking process; To ensure that the modulation light source frequency is always accurately locked to the main resonance peak position during photoacoustic detection, and effectively suppress the interference response of parasitic modes, based on the mismatch risk level judgment result output by the previous stage, perform adaptive tracking control of the modulation frequency, and combine the dual constraints of amplitude and phase to carry out precise adjustment, while dynamically suppressing the gain of the parasitic modes that may be excited, which specifically includes the following steps: When the mismatch risk level is judged as low risk, keep the current modulation light source frequency unchanged; when the risk level is medium risk, start the fine-tuning process within ±1 Hz range, and take the current main peak center frequency as the reference point, perform fine scanning within the frequency range of 0.1 Hz above and below, with 0.01 Hz as the step, to find the point with the maximum amplitude response and the strongest phase stability; when the risk level is high risk, directly return to the narrowband frequency scanning step to reacquire the characteristic fingerprint. During each fine-tuning scanning, the gas flow rate in the detection cavity is maintained at not higher than 50 mL / min, and the temperature fluctuation is controlled within ±0.2℃, to ensure the stability and repeatability of the acoustic response.

[0037] In the frequency trimming process, each modulation frequency point is collected by the sound wave detection device in real time, and the maximum amplitude value of the corresponding frequency point is recorded. When the amplitude value forms a rising-peak-decreasing characteristic structure in the continuous three frequency points, and the peak value is at least 15% higher than the average amplitude value of the surrounding frequency points, it is determined that it is the optimal position of the main peak frequency response. On this basis, the amplitude response of the current frequency point is recorded as the locking amplitude reference, so as to guarantee the amplitude stability of the main peak position.

[0038] After determining the main peak amplitude response maximum point, the phase response corresponding to the frequency point is further extracted and compared with the phase data within ±0.02 Hz range before and after the frequency point. When the phase change occurs suddenly at the main peak point, and the phase mutation amplitude is not less than 10°, and the phase change direction is consistent with the main peak phase transition direction in the reference characteristic fingerprint, it is confirmed that the frequency point is the main peak locking point. If the amplitude maximum point does not satisfy the above phase transition condition, it is automatically extended to the next frequency point that satisfies the condition, so as to ensure that the final locking point satisfies the main peak characteristics under the conditions of amplitude and phase at the same time, and to avoid the errors caused by single parameter judgment. After the locking frequency is established, the parasitic modal peak position and half-width interval recorded in the characteristic fingerprint library are consulted, the targeted suppression frequency band is constructed within 5 Hz range around the current locking frequency, and the gain of the collected photoacoustic signal is linearly compressed in the frequency band. The specific compression ratio is determined according to the ratio of the parasitic modal amplitude to the main peak amplitude. If the parasitic modal amplitude is between 30% and 50% of the main peak, the gain compression ratio is set to 0.7; if it exceeds 50%, the compression ratio is set to 0.5. In addition, in order to prevent signal distortion, a 1 Hz gradual transition zone is set at both ends of the suppression frequency band to ensure smooth connection of the frequency response. Through the above suppression mechanism, the superimposed interference of the parasitic modal on the main peak signal is effectively reduced, and the acoustic signal source under the final locking frequency is ensured to be pure and specific.

[0039] The function of this step is to adaptively and dynamically adjust the frequency of the modulated light source during photoacoustic spectral detection according to the previously determined mismatch risk level, to ensure that it is accurately locked at the main resonance peak position of the detection cavity, and to simultaneously suppress the real-time response of the possible excited parasitic modes, thereby improving the signal purity, stability and gas concentration inversion accuracy of the detection system. Since the photoacoustic cavity is a high-quality factor structure, its resonance response is extremely sensitive to frequency changes. Once the modulated light source fails to accurately align with the main resonance peak, even if the shift is less than 1 Hz, it will cause significant attenuation or phase mismatch of the acoustic signal, and then cause serious deviation of the concentration judgment result. At the same time, under the condition of complex acoustic or structural disturbance, parasitic modes are easily excited and mixed with the main mode, forming "false peak" interference, which makes the system mistakenly lock the parasitic peak as the main peak, resulting in detection distortion. This step introduces amplitude and phase double constraint mechanism during modulation, and only when the signal amplitude reaches a significant peak and the phase undergoes a characteristic transition, it is locked, significantly improving the accuracy of peak identification. In addition, based on the historical fingerprint parameters of parasitic modes, targeted band suppression is implemented, which effectively weakens the interference components without damaging the characteristics of the main peak. This series of control actions constitutes the core link of closed-loop adaptive adjustment of the modulation frequency, not only ensures the specificity of the photoacoustic signal source, but also provides a reliable physical basis for subsequent concentration fitting, fault diagnosis and state prediction, and is a key step to build high-reliability gas in-situ monitoring capability.

[0040] After completing the modulation frequency locking, analyze the physical reasons for the resonance frequency drift, perform local heating control of the detection cavity for the sound velocity change caused by temperature change, and perform film side dry gas input adjustment for the sound velocity change caused by humidity change, to stabilize the gas sound velocity and fix the main resonance peak position; In the process of photoacoustic spectral-based in-situ detection of dissolved gases in transformer oil, even if the modulation frequency locking is completed, if the temperature or humidity in the detection cavity fluctuates, it may still cause the gas sound velocity to change, resulting in resonance frequency drift. Frequency drift will reduce the matching degree of sound waves and light modulation, causing signal amplitude attenuation, and even triggering the response of parasitic modes, affecting the accuracy of concentration inversion. Therefore, after frequency locking, the source of drift is identified and targeted temperature and humidity control measures are taken to stabilize the sound velocity from the source and fix the main resonance peak position. The specific steps are as follows: The drift source is determined. In the modulation frequency lock state, the detection cavity outer wall temperature, the gas temperature of the detection cavity internal multiple distribution measurement points, the relative humidity of the detection cavity internal multiple distribution measurement points, and the micro-sweeping frequency data in the lock state are continuously collected. The micro-sweeping frequency range is set to be 5 Hz above and below the lock center frequency, with a step of 1 Hz, and a single point stays for 50 ms. In the collection period (preferably 10 minutes), if the center frequency changes consistently with the outer wall temperature and the internal humidity changes less than ±2%RH, it is determined that the drift is mainly caused by temperature changes; if the center frequency is sensitive to humidity changes but not sensitive to outer wall temperature changes, it is determined that the drift is mainly caused by humidity changes; if both are affected, the priority is determined according to the sensitivity.

[0041] Local heating control is implemented for temperature-induced sound speed changes. Polyimide insulation layer resistive heating sheets are pasted on the detection cavity outer wall photoacoustic action area 10 mm above and below the position, the heating sheet power density is not less than 1 W / cm², and the outer cover is 3 mm thick aerogel insulation layer, and an aluminum foil reflection layer is added outside the insulation layer to reduce radiation heat dissipation. Thin film thermocouple sensors are attached at the center position of the heating sheet, and temperature data is collected once per second. The heating target temperature is set to the average temperature of the region for 5 minutes at the first frequency lock, the temperature rising speed is controlled to be not more than 0.1 ℃ / min, and the stability criterion is that the temperature fluctuation is less than ±0.05 ℃ within 5 minutes. When the temperature is stable, if the center frequency deviation is less than 10% of the resonance bandwidth and the signal amplitude change is not more than 3%, it is confirmed that the temperature source is effectively suppressed; if it does not meet the standard, the target temperature is adjusted by 0.2 ℃ within the safe temperature range and the detection is repeated until it meets the standard.

[0042] Film side dry gas input adjustment is implemented for humidity-induced sound speed changes. An independent dry gas micro-flow channel is provided on the back of the water-resistant and breathable film, the channel uses a polytetrafluoroethylene tube with an inner diameter of 1 mm and a length of not more than 30 cm, the inlet is connected to a high-purity nitrogen source with a dew point not higher than -40 ℃, and the flow is controlled at 1-5 sccm (standard cubic centimeter per minute), which is accurately controlled by a mass flow meter. The dry gas passes through a constant temperature section of more than 10 cm before entering the channel to make its temperature consistent with the detection cavity outer wall temperature, avoiding the formation of temperature gradient. A micro back pressure valve is provided at the outlet end of the channel to control the back pressure to be higher than the environmental pressure by 20-50 Pa to prevent the detection cavity gas from being sucked. When the humidity decreases to the first lock humidity ±1.5%RH within 10 minutes, and the center frequency returns to the first lock reference value ±5 Hz range, it is confirmed that the humidity source is effectively suppressed; if the conditions are not met, the flow is increased by 1 sccm each time under the premise of stable back pressure and the detection is repeated until it meets the standard or reaches the maximum flow.

[0043] Verify and solidify the control strategy. After the local heating and film-side drying measures are completed, two rounds of independent micro-sweep verification are performed, with a 5-minute interval between each round, each round of scanning ranging from 5 Hz above and below the center frequency, stepping 1 Hz, and recording the center frequency, half-power bandwidth, peak amplitude, and phase transition characteristics. If the difference in center frequency is ≤3 Hz, the difference in bandwidth is ≤10%, and the difference in amplitude is ≤3% between the two rounds, then the current temperature target value, drying gas flow rate, and back pressure value, as well as the corresponding temperature and humidity reference values, are stored as the long-term operation baseline. If any of the indicators exceeds the range, the temperature target value is adjusted by 0.1 ℃ or the drying gas flow rate is adjusted by 1 sccm, and the verification is performed again until the criteria are met. Through the above steps, the influence of temperature and humidity disturbances on gas sound speed can be effectively offset, allowing the main resonance peak to remain stable for a long time after being locked, thereby ensuring the accuracy and long-term repeatability of the photoacoustic detection signal.

[0044] The purpose of this step is to identify and eliminate the temperature and humidity source interference that causes the resonance frequency to drift after the modulation frequency is locked, stabilize the sound speed of the gas in the detection cavity from a physical level, and thus fix the position of the main resonance peak for a long time, ensuring the accuracy and repeatability of the photoacoustic detection signal. In photoacoustic spectroscopy, the main resonance peak frequency is directly related to the gas sound speed, which changes with even small fluctuations in temperature and humidity. This change is particularly sensitive in high-quality factor detection cavities, and even a few hertz of drift can cause the frequency and modulation light source to mismatch, resulting in a decrease in signal amplitude, distortion of phase response, and even triggering of abnormal amplification effects of adjacent parasitic modes, producing false signals unrelated to gas absorption. By monitoring temperature and humidity changes after locking and analyzing their correlation with frequency drift, the source of the drift can be accurately determined. For temperature changes, the temperature is stabilized at the reference value of the first lock by local heating of the outer wall and constant temperature insulation. For humidity changes, the humidity in the cavity is reduced and stabilized to the reference level by introducing constant-temperature drying gas on the film side. Through this source control, not only can the frequency shift caused by environmental disturbances be eliminated, but the need for frequent relocking can also be reduced, the lock retention time can be extended, and the stability and reliability of the detection results can be improved, which is particularly critical for application scenarios that require long-term continuous monitoring of dissolved gases in transformer oil.

[0045] Under the condition of stable gas sound speed and main resonance peak position, a reference absorption line with a known concentration is introduced for online verification, and the real-time inverted gas concentration is compared with the reference concentration. When the concentration deviation exceeds the pre-set threshold, the narrow-band sweep measurement is performed again and the modulation frequency lock parameters are corrected, thereby completing the closed-loop control of gas detection accuracy. To ensure the high accuracy of the photoacoustic spectroscopy detection system in long-term online operation, on the basis of the stable gas sound speed and main resonance peak position in the detection cavity, the known concentration reference absorption line is introduced to check the real-time inversion results online, and the narrowband sweep measurement and frequency locking correction are performed when the deviation exceeds the set range, forming a closed-loop control of detection accuracy. This scheme can offset the cumulative effects of factors such as wavelength drift of the light source, slow changes in the acoustic properties of the detection cavity, and attenuation of sensor sensitivity on the detection results, ensuring the repeatability and long-term stability of the measurement results.

[0046] A hardware path for establishing and introducing a reference absorption line is established. A reference gas path is set up in the optical bypass of the detection device, and a standard gas certified by the national metrology institute is filled in the gas path, which has a clear and long-term stable concentration value, such as 150 ppm acetylene by volume fraction. The outlet of the reference gas bottle is stabilized to 101 ± 1 kPa by a precision pressure reducing valve, heated to 35 ± 0.1 ℃ by a constant temperature jacket, and output at a flow rate of 1–2 sccm by a mass flow controller (flow rate error not more than 1%). The reference gas enters a micro absorption cavity with a volume of ≤2 mL, and the inlet and outlet of the absorption cavity are connected by a collet joint and a polytetrafluoroethylene microtube to ensure airtightness and oil and moisture resistance. The optical path uses a fused taper fiber beam splitter to couple 10%–20% of the incident light power to the absorption cavity, and the absorption signal is collected by a high-sensitivity microphone with a sampling period of 20 s. The zero gas background value is recorded before sampling to calibrate the baseline.

[0047] Reference signal conversion and dual-channel synchronous measurement are performed. The photoacoustic amplitude of the reference absorption line is converted to the concentration of the reference gas according to the calibration curve established in the laboratory, and the calibration curve is obtained by fitting multiple concentration points, covering 50–500 ppm with a repeatability error of ≤2%. At the same time, the sample gas photoacoustic signal output by the main detection cavity is also converted to the concentration of the gas to be measured. The two data are stored synchronously using a unified timestamp, and are linearly normalized based on the output value of the light source monitoring diode once, eliminating the influence of light intensity fluctuations. The temperature, pressure, and flow rate monitoring values of the reference gas path are recorded together with the concentration data, and if any parameter exceeds the set range, the current calibration is marked as invalid and automatically re-measured.

[0048] Concentration deviation judgment and retrace correction are performed. The concentration deviation threshold is set to ±5% of the reference concentration or an absolute value of ±0.1 ppm, whichever is greater. When the deviation of the sample gas concentration from the reference concentration exceeds the threshold, it is determined that there is a drift in the current modulation frequency locking parameter. At this time, the system immediately performs narrowband sweep frequency, with a scan range of 30 Hz above and below the current resonance frequency prior value, a step of 1 Hz, a single point residence of 50 ms, and simultaneous collection of amplitude and phase response curves. According to the curve, the center, half-power bandwidth and phase transition characteristics of the main resonance peak are re-identified, and it is checked whether the parasitic mode is activated; if the parasitic peak amplitude exceeds 3 times the background and the distance from the main peak center is ≤10 Hz, the notch and gain suppression strategy is loaded simultaneously when updating the locking parameter to prevent the locking point from deviating to the parasitic peak.

[0049] Complete closed-loop confirmation and strategy solidification. After updating the locking parameter, three sets of sample gas and reference gas concentration data are continuously collected, with an interval of 20 s between each set. If the deviation of all three sets is within the threshold, it is determined that the correction is effective, and the locking parameter, reference gas path operating parameter, detection cavity outer wall temperature, cavity humidity, and main peak bandwidth are all solidified as the current operating reference. If any one of the deviations exceeds the limit, repeat the retrace and parameter update, with a maximum of 3 repetitions. If it still does not meet the standard after 3 repetitions, issue a maintenance warning, and then check again after adjusting the light source temperature control point by ±0.1 °C. Through the above continuous steps, the system can achieve verifiable maintenance and long-term stable operation of detection accuracy without changing the composition of the sample gas.

[0050] The purpose of this step is to establish a traceable, quantifiable, and self-correcting concentration accuracy guarantee mechanism in the photoacoustic spectroscopy detection system, thereby ensuring the stability and reliability of the detection results in long-term online operation. Although the previous steps have stabilized the acoustic conditions of the detection cavity through temperature and humidity control and frequency locking, in the complex power field environment, the wavelength of the light source may drift slightly due to insufficient temperature control accuracy, the mechanical structure of the acoustic cavity may slowly shift due to long-term vibration or stress changes, and the sensitivity of the microphone and other sensor elements may decrease due to aging. These subtle changes may not affect the frequency locking, but they can accumulate errors in the concentration inversion calculation, causing the results to gradually deviate from the true value. This step introduces a reference absorption line with a known concentration as a measurement reference, synchronously compares it with the real-time detection results of the sample gas, and sets a clear deviation threshold to determine whether there is an accuracy drift. Once the deviation exceeds the threshold, the system immediately reverts to perform narrowband sweep and main peak re-calibration, automatically corrects the modulation frequency locking parameters, and avoids the delay caused by human intervention. This closed-loop process not only detects and eliminates the hidden dangers of accuracy decline in real time, but also ensures that the correction is effective through multiple rounds of verification, and solidifies the corrected operating parameters to form a self-adaptive accuracy maintenance capability. In this way, the detection system can maintain consistent detection accuracy with the standard gas when facing long-term changes in light sources, cavities, sensors, and environmental conditions, providing continuous protection for reliable monitoring of dissolved gas concentrations in transformer oil, reducing the risk of false positives, and improving the accuracy of operation and decision-making.

[0051] Through the above scheme, the present application can effectively solve the technical problems of resonance frequency drift, mismatching of the modulated light source and the main resonance peak, and abnormal amplification of the parasitic mode caused by the slight fluctuation of the temperature and humidity in the detection cavity during the photoacoustic spectroscopy in-situ detection of the dissolved gas in the transformer oil. The present application obtains the prior value of the resonance frequency in advance through real-time acquisition and analysis of the temperature and humidity and model calculation, and accurately calibrates the main peak and parasitic peak characteristics in combination with narrowband sweep, to realize high-credibility identification of the position of the main resonance peak. Then, in combination with the adaptive frequency tracking with amplitude and phase double constraints, the frequency of the modulated light source is always accurately locked to the main peak, and the parasitic mode response is dynamically suppressed. At the same time, after the frequency is locked, the gas speed and the position of the resonance peak are stabilized from the source through local heating and membrane-side dry gas adjustment. Finally, the online calibration of the reference absorption line with a known concentration is introduced to establish a closed-loop detection and automatic back-scan correction mechanism for concentration deviation. Through this multi-dimensional collaborative control and closed-loop feedback, the present application not only guarantees the high consistency of the detection signal and the absorption characteristics of the measured gas, but also significantly improves the long-term stability and environmental interference resistance of the detection, thereby reducing false signals and false positives in online DGA monitoring, and providing high-precision and traceable detection data support for transformer operation state evaluation.

[0052] The present application provides a method for detecting a dissolved gas in a transformer oil, comprising the steps of Figure 2The device for detecting dissolved gas in transformer oil in situ based on photoacoustic spectroscopy technology shown comprises a temperature and humidity driven resonant frequency prior value calculation module, a narrowband sweep and characteristic peak calibration module, a resonant peak shift analysis and mismatch risk assessment module, an adaptive frequency locking and parasitic mode suppression module, a sound speed stabilization and resonant peak fixing control module, and a reference absorption line closed-loop calibration and precision maintenance module. The temperature and humidity driven resonant frequency prior value calculation module acquires temperature and humidity data in the detection cavity, calculates the current gas sound speed according to an analytical model of the change of the temperature and humidity and the gas sound speed, and calculates the resonant frequency prior value from the gas sound speed. The narrowband sweep and characteristic peak calibration module performs narrowband sweep with the resonant frequency prior value as the center, collects the amplitude response curve and the phase response curve of the detection cavity, identifies and calibrates the characteristic fingerprint of the main resonant peak and the adjacent parasitic mode. The resonant peak shift analysis and mismatch risk assessment module calculates the frequency shift and the phase shift of the main resonant peak based on the characteristic fingerprint, and generates the main resonant peak credibility score and the mismatch risk level. The adaptive frequency locking and parasitic mode suppression module performs adaptive tracking control of the modulation frequency according to the mismatch risk level, locks the modulation light source frequency to the main resonant peak position under the amplitude and phase double constraint conditions, and dynamically suppresses the parasitic mode gain. The sound speed stabilization and resonant peak fixing control module, after frequency locking, implements local heating control of the detection cavity for the sound speed change caused by temperature change, and implements film side dry gas input adjustment for the sound speed change caused by humidity change, so as to stabilize the gas sound speed and fix the main resonant peak position. The reference absorption line closed-loop calibration and precision maintenance module, under the condition that the sound speed and the main resonant peak position are stable, introduces a reference absorption line with a known concentration for online verification, and when the deviation between the real-time inverted gas concentration and the reference concentration exceeds a preset threshold, returns to re-sweep and corrects the frequency locking parameters, to complete the closed-loop control of the gas detection precision.

[0053] The embodiment of the present application provides a kind of dissolved gas in transformer oil in situ detection method based on photoacoustic spectroscopy technology, which is realized by the above-mentioned dissolved gas in transformer oil in situ detection device based on photoacoustic spectroscopy technology, and the specific method and process of a kind of dissolved gas in transformer oil in situ detection device based on photoacoustic spectroscopy technology are described in the above-mentioned embodiment of the dissolved gas in transformer oil in situ detection method based on photoacoustic spectroscopy technology, which will not be repeated here.

[0054] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0055] Those skilled in the art will appreciate that embodiments of the present application can be devised for a variety of applications. It is intended that the present application be limited only by the scope of the appended claims, and it is intended that various modifications and alterations made by those skilled in the art be considered as within the scope of the present application. The embodiments of the present application will be described with reference to the attached drawings, wherein:

[0056] The present application is described in reference to the drawings using a flowchart and / or a block diagram of the method, apparatus (system) and computer program product according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0057] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0058] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 means for carrying out each of the one or more functions specified in the flowchart and / or block diagram block or blocks.

[0059] While the preferred embodiments of the application have been described, additional variations and modifications can be employed by those skilled in the art. Therefore, the appended claims are intended to cover all such variations and modifications as falling within the scope of the present application.

[0060] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy, characterized in that, The method comprises the following steps: Obtain temperature and humidity data in the detection cavity, calculate the current gas sound speed according to the analytical model of temperature and humidity and the change of gas sound speed, and calculate the prior value of the resonance frequency from the gas sound speed; Perform narrowband sweep centered on the prior value of the resonance frequency, collect the amplitude response curve and phase response curve of the detection cavity, identify and calibrate the characteristic fingerprint of the main resonance peak and the adjacent parasitic mode; Calculate the frequency offset and phase offset of the main resonance peak based on the characteristic fingerprint, and generate the main resonance peak credibility score and mismatch risk level; According to the mismatch risk level, perform adaptive tracking control of the modulation frequency, lock the modulation light source frequency to the position of the main resonance peak under the amplitude and phase double constraint conditions, and dynamically suppress the parasitic mode gain; After frequency locking, implement local heating control of the detection cavity for the sound speed change caused by temperature change, and implement membrane side dry gas input adjustment for the sound speed change caused by humidity change, so as to stabilize the gas sound speed and fix the position of the main resonance peak; Under the condition that the sound speed and the position of the main resonance peak are stable, introduce a reference absorption line with a known concentration for online verification, and when the deviation between the real-time inverted gas concentration and the reference concentration exceeds the preset threshold, return to re-sweep and correct the frequency locking parameters to complete the closed-loop control of the gas detection accuracy.

2. The method for detecting dissolved gases in transformer oil in-situ based on photoacoustic spectroscopy according to claim 1, characterized in that, In the step of obtaining temperature and humidity data in the detection cavity and calculating the prior value of the resonance frequency, The method comprises the following processes: Multiple measuring points are arranged in the detection cavity, a platinum resistance thermometer with a measuring accuracy of not less than ±0.05 degrees Celsius and a thin film capacitive humidity sensor with a measuring accuracy of not less than ±1% relative humidity are installed at each measuring point, and temperature and humidity data at different positions are collected; The temperature data and humidity data of each measuring point are subjected to validity check and respectively take the arithmetic mean value, and are sent into the temperature and humidity-gas sound speed corresponding relationship table established through standard gas calibration to obtain the current gas sound speed; The gas sound speed is combined with the physical size of the detection cavity, the shape of the end structure, the acoustic reflection characteristics and the energy loss correction value to determine the prior value of the main resonance frequency, and is fine-tuned in combination with the latest narrowband sweep result; The updated prior value of the resonance frequency is output to the frequency sweep control unit, and the last valid prior value is called as a temporary reference when the temperature and humidity change exceeds the preset threshold, until the new data is replaced by the new prior value after being stabilized.

3. The method for detecting dissolved gases in transformer oil in-situ based on photoacoustic spectroscopy according to claim 1, characterized in that, In the step of performing narrowband sweep centered on the prior value of the resonance frequency, collecting the amplitude response curve and phase response curve of the detection cavity, and calibrating the characteristic fingerprint of the main resonance peak and the parasitic mode, the method comprises the following processes: The prior value of the resonance frequency is set as the center frequency, the narrowband sweep start and end range is set as the prior value minus twice the half-height width of the main peak to the prior value plus twice the half-height width of the main peak, the frequency step value is not greater than one percent of the half-height width of the main peak, and the detection cavity gas is replaced by low flow rate dry carrier gas under normal pressure; A light source with wavelength stability better than ±0.01 nanometers is used to emit a modulation light beam and collimate it into the detection cavity, and a capacitive microphone is used to synchronously collect the amplitude value and phase value of the sound wave at each frequency point to form the amplitude response curve and phase response curve data set; According to the amplitude peak position and phase mutation characteristics, a main resonance peak is identified, and the peak frequency, amplitude, half-height width, and phase mutation characteristics are recorded, and the characteristics of the parasitic modes on both sides of the main peak are identified and recorded, and a characteristic fingerprint library of the main peak and the parasitic modes is established; The current characteristic parameters are compared with the reference characteristic fingerprint, and when the main peak frequency offset is greater than ± 5 Hz or the parasitic mode amplitude is greater than 30% of the main peak amplitude, the peak site mismatch risk judgment result is output, and the frequency compensation value and the parasitic mode suppression weight are generated.

4. The method for detecting dissolved gases in transformer oil in-situ based on photoacoustic spectroscopy according to claim 3, characterized in that, After outputting the peak site mismatch risk judgment result and generating the frequency compensation value and the parasitic mode suppression weight, It includes the following processes: The frequency compensation value is applied to the frequency control of the modulated light source, so that the frequency of the modulated light source is locked to the compensated main resonance peak position under the amplitude and phase double constraint conditions; at the same time, the parasitic mode suppression weight is applied to the photoacoustic signal corresponding to the parasitic mode frequency, and the amplitude of the parasitic mode signal is dynamically reduced to prevent the parasitic mode from being misidentified as the main resonance peak by the frequency locking process, thereby ensuring that the subsequent gas concentration inversion is based on the true photoacoustic signal of the main resonance peak.

5. The method for detecting dissolved gases in transformer oil in-situ based on photoacoustic spectroscopy according to claim 1, characterized in that, In the step of calculating the frequency offset and phase offset of the main resonance peak based on the characteristic fingerprint and generating the main resonance peak credibility score and mismatch risk level, It includes the following processes: The current measured main resonance peak peak frequency is differentially calculated with the reference frequency in the reference characteristic fingerprint library to obtain the real-time frequency offset; The phase data in the current main peak center and half-height width range are compared with the reference phase curve point by point to calculate the phase offset; The frequency offset and the phase offset are normalized and combined into a main peak credibility score according to a preset weight, and a low risk, medium risk or high risk level is determined according to the score interval; In the low risk, the frequency is locked, in the medium risk, the frequency is fine-tuned and the parasitic mode suppression is enhanced, and in the high risk, the narrowband sweep is returned to reacquire and update the characteristic fingerprint library.

6. The method for detecting dissolved gases in transformer oil in-situ based on photoacoustic spectroscopy technology according to claim 5, characterized in that, Returning to narrowband sweep reacquisition and updating the characteristic fingerprint library under high risk level It includes the following processes: A narrowband sweep range is set with the latest calculated resonance frequency prior value as the center, and a low flow rate dry carrier gas is used to replace the gas in the detection cavity under normal pressure; The amplitude response curve and the phase response curve of the detection cavity are reacquired, and the peak frequency, peak amplitude, half-height width and phase mutation characteristics of the main resonance peak and all parasitic modes are identified; The new acquired characteristic parameters are used to replace the corresponding data in the reference characteristic fingerprint library, and the stability of the new characteristic fingerprint under the current environmental temperature and humidity conditions is verified; The updated characteristic fingerprint library is used as the reference data for subsequent frequency offset and phase offset calculation, ensuring that the latest acoustic characteristics of the detection cavity can be accurately reflected when the risk judgment is performed again.

7. The method for on-site detection of dissolved gases in transformer oil based on photoacoustic spectroscopy according to claim 1, characterized in that, The adaptive tracking control of the modulation frequency according to the mismatch risk level judgment result includes the following processes: When the risk level is low risk, the current modulation light source frequency remains unchanged; when the risk level is medium risk, fine scanning is performed within ± 1 Hz with a step of 0.01 Hz to find the frequency point with the maximum amplitude and consistent phase transition direction as the locking point; when the risk level is high risk, the characteristic fingerprint is reacquired by returning to narrowband sweep; After the lock-in frequency is established, according to the parasitic mode peak position and half-width interval in the characteristic fingerprint library, a suppression frequency band is constructed within a range of 5 Hz around the lock-in frequency, and the gain compression ratio is determined according to the amplitude ratio of the parasitic mode, and a 1 Hz gradual transition zone is set at both ends of the suppression frequency band to ensure smooth connection of the response.

8. The method for on-site detection of dissolved gases in transformer oil based on photoacoustic spectroscopy according to claim 1, characterized in that, The steps for stabilizing the gas sound speed and fixing the position of the main resonance peak after frequency locking include: Collect the outer wall temperature, multiple-point temperature and humidity in the cavity and perform a ±5Hz, 1Hz step micro-sweep frequency judgment to determine the drift source; Paste a resistance heating sheet on the photoacoustic action area, 10mm above and below, the target temperature is the first lock-in 5min average, the temperature rise is ≤0.1℃ / min, and the steady-state fluctuation is ≤±0.05℃; Pass nitrogen with a dew point of ≤-40℃ into the back of the water-resistant gas permeable membrane, the flow rate is 1-5sccm, and the back pressure is 20-50Pa; Two rounds of ±5Hz re-sweeping are performed at intervals of 5min, the center frequency difference is ≤3Hz, the bandwidth difference is ≤10%, and the amplitude difference is ≤3%, then the parameters are fixed, if not up to standard, the temperature is adjusted by 0.1℃ or the flow rate is adjusted by 1sccm and re-verified.

9. The method for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy according to claim 1, characterized in that, Under the condition that the gas sound speed and the position of the main resonance peak are stable, the following steps are included to ensure detection accuracy and achieve closed-loop control: Under the condition that the gas sound speed and the position of the main resonance peak are stable, a reference gas path filled with standard gas is introduced through a bypass of the light path, the temperature, pressure and flow rate of the reference gas are controlled, the reference gas enters the micro absorption cavity for photoacoustic signal acquisition, the converted reference gas concentration is compared with the real-time inverted sample gas concentration, when the deviation exceeds the preset threshold, narrowband sweep measurement is performed, the position of the main resonance peak is recalibrated and the modulation frequency lock-in parameters are updated, until the concentration deviation is within the threshold.

10. A device for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy, for implementing a method for in-situ detection of dissolved gases in transformer oil based on photoacoustic spectroscopy according to any one of claims 1 to 9, characterized in that, The system includes a temperature and humidity driven resonance frequency prior value calculation module, a narrowband sweep and characteristic peak calibration module, a resonance peak shift analysis and mismatch risk evaluation module, an adaptive frequency lock-in and parasitic mode suppression module, a sound speed stabilization and resonance peak fixing control module, and a reference absorption line closed-loop calibration and precision maintenance module; The temperature and humidity driven resonance frequency prior value calculation module acquires the temperature and humidity data in the detection cavity, calculates the current gas sound speed based on the analytical model of temperature and humidity and gas sound speed variation, and calculates the resonance frequency prior value from the gas sound speed; The narrowband sweep and characteristic peak calibration module performs narrowband sweep with the resonance frequency prior value as the center, acquires the amplitude response curve and phase response curve of the detection cavity, identifies and calibrates the main resonance peak and the characteristic fingerprint of the adjacent parasitic mode; The resonance peak shift analysis and mismatch risk evaluation module calculates the frequency shift and phase shift of the main resonance peak based on the characteristic fingerprint, generates the main resonance peak credibility score and mismatch risk level; The adaptive frequency lock-in and parasitic mode suppression module performs adaptive tracking control of the modulation frequency according to the mismatch risk level, locks the modulation light source frequency to the position of the main resonance peak under the dual constraints of amplitude and phase, and dynamically suppresses the gain of parasitic modes. The sound speed stabilizing and resonance peak fixing control module, after frequency locking, implements local heating control of the detection cavity aiming at sound speed change caused by temperature change, and implements membrane side dry gas input adjustment aiming at sound speed change caused by humidity change, so as to stabilize the sound speed of the gas and fix the position of the main resonance peak. The reference absorption line closed loop calibration and precision maintenance module, under the condition of sound speed and main resonance peak position stabilization, introduces a known concentration reference absorption line for online verification, when the deviation between the real-time inversion gas concentration and the reference concentration exceeds the preset threshold, returns to re-scan and corrects the frequency locking parameter, and completes the closed loop control of the gas detection precision.

Images