High-precision photoacoustic spectrum monitoring method for gas in transformer oil

Abstract

The invention discloses a high-precision photoacoustic spectrum monitoring method for gas in transformer oil, and relates to the technical field of transformer oil gas monitoring, and the method comprises the following steps: constructing a photoacoustic spectrum detection system, sampling the gas in the transformer oil, and obtaining photoacoustic spectrum data of the gas, the photoacoustic spectrum detection system comprises a near-infrared tunable fiber laser, a photoacoustic cell, a lock-in amplifier and a data acquisition system; preprocessing the acquired photoacoustic spectrum data, and extracting features related to fault diagnosis from the preprocessed photoacoustic spectrum data; and the photoacoustic spectroscopic data of the mixed gas is decoupled by using a chemometrics method. According to the method, the gas photoacoustic spectroscopic data in the transformer oil are collected and analyzed in real time, abnormal conditions in the transformer are found as soon as possible, the mixed gas spectroscopic data are decoupled and separated in combination with chemometrics and a machine learning algorithm, and various gas components are accurately recognized and quantitatively analyzed.

Description

Technical Field

[0001] The invention relates to the technical field of transformer oil and gas monitoring, and in particular to a high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil. Background Art

[0002] As a key equipment in the power system, the stable operation of the transformer is crucial to ensuring the safety and reliability of the power grid. The transformer converts the voltage from one level to another through the principle of electromagnetic induction to achieve the transmission and distribution of electric energy. In power plants and substations, transformers usually need to operate at high load for a long time, and their operating status directly affects the stability and economic benefits of the entire power system. Transformer oil, as an important insulating and cooling medium for transformers, will dissolve or produce a variety of gases during the operation of the transformer. The types and contents of these gases are closely related to the operating status and fault type of the transformer. By monitoring and analyzing the gas in the transformer oil, potential faults of the transformer can be discovered in time, providing an important basis for the repair and maintenance of the transformer.

[0003] In the prior art, when detecting multiple mixed gases, there is an overlap of absorption spectra between different gases, resulting in cross-interference, which in turn affects the fault diagnosis and monitoring based on photoacoustic spectroscopy data. Therefore, how to decouple and separate the spectral data of the mixed gas and mine and analyze the photoacoustic spectroscopy data to improve the accuracy of fault diagnosis and monitoring is the problem we need to solve. To this end, a high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil is proposed. Summary of the invention

[0004] The object of the present invention is to provide a high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil to solve the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil comprises the following steps:

[0007] Step 1, constructing a photoacoustic spectroscopy detection system, sampling the gas in the transformer oil, and obtaining the photoacoustic spectroscopy data of the gas, wherein the photoacoustic spectroscopy detection system comprises a near-infrared tunable fiber laser, a photoacoustic cell, a lock-in amplifier, and a data acquisition system, wherein the near-infrared tunable fiber laser emits infrared radiation of a specific wavelength, and the frequency of the light source is modulated by a modulation disk to ensure that the light source changes periodically at a specific frequency;

[0008] Step 2, preprocessing the collected photoacoustic spectroscopy data, including filtering, denoising, and baseline correction steps, to improve data quality, and extracting features related to fault diagnosis from the preprocessed photoacoustic spectroscopy data;

[0009] Step 3: Decouple the photoacoustic spectroscopy data of the mixed gas using a chemometric method, and separate and identify the photoacoustic spectroscopy features of different gases using a machine learning algorithm to remove cross-interference between different gases;

[0010] Step 4, analyzing the decoupled and separated photoacoustic spectroscopy data, extracting characteristic information related to transformer fault diagnosis, and establishing a fault diagnosis model to identify different fault types and states;

[0011] Step 5, integrate the established fault diagnosis model into the real-time monitoring system, conduct real-time analysis of the gas in the transformer oil, combine the constructed fault diagnosis model with historical fault cases, set different fault levels, and evaluate the severity of the fault;

[0012] Step 6: Generate a transformer fault monitoring report based on the data analysis and fault assessment results, and issue an early warning signal to prompt further inspection and maintenance. The report contains information on the fault type, fault severity, development trend, and recommended maintenance measures.

[0013] A further improvement of the technical solution of the present invention is that in step 1, the construction of the photoacoustic spectroscopy detection system and the acquisition process of the gas photoacoustic spectroscopy data are as follows:

[0014] Step 101, preparing components of a photoacoustic spectroscopy detection system, including a near-infrared tunable fiber laser, a photoacoustic cell, a lock-in amplifier, and a data acquisition system, connecting and debugging the near-infrared tunable fiber laser, the modulation disk, the photoacoustic cell, the lock-in amplifier, and the data acquisition system, so that the optical path, the circuit, and the signal transmission path between the components are unobstructed;

[0015] Step 102, the gas sample in the sampling transformer oil is placed in a photoacoustic cell through vacuum degassing, the near-infrared tunable fiber laser is started, and the required output wavelength is set, and the light source is frequency modulated by a modulation disk so that the laser periodically irradiates the gas sample in the photoacoustic cell at a specific frequency;

[0016] Step 103, in the photoacoustic cell, the gas molecules generate heat energy after absorbing the laser energy, thereby generating an acoustic wave signal, and a phase-locked amplifier is used to detect the acoustic wave signal generated in the photoacoustic cell, and the acoustic wave signal is phase-locked and amplified. The amplified signal is clearer, which is convenient for subsequent data collection and analysis;

[0017] Step 104, the data acquisition system collects the photoacoustic signal output by the phase-locked amplifier in real time, and converts it into a digital signal for storage, ensuring that the data acquisition system is synchronized with the modulation frequency of the laser so as to correctly analyze the spectral data.

[0018] A further improvement of the technical solution of the present invention is that in step 2, the process of extracting the relevant features of fault diagnosis is as follows:

[0019] Step 201, performing preprocessing operations of filtering, denoising, and baseline correction on the collected photoacoustic spectroscopy data;

[0020] Step 202, using a feature selection algorithm to extract features related to fault diagnosis from the preprocessed photoacoustic spectrum data, the features are related to the absorption characteristics, spectral shape, and intensity parameters of gas molecules, wherein the features related to fault diagnosis include the absorption peak of the gas, the intensity and area of ​​the absorption peak, the width and shape of the absorption peak, the relative intensity ratio of the spectrum, and the slope and curvature of the spectrum. By identifying the absorption peak of the gas, the type of gas present is determined, and the approximate concentration of the gas is determined by quantitative analysis in combination with the intensity and area of ​​the absorption peak. Clues of the interaction between gas molecules are provided by the width and shape of the absorption peak to assist in identifying the type of gas. The gas is distinguished and identified by the intensity ratio between specific absorption peaks of different gases, and the trend of gas concentration change over time is analyzed by the slope and curvature of the spectrum;

[0021] Step 203 , standardize the extracted fault diagnosis related feature data to ensure the comparability between different features, and integrate the feature data to form a fault feature data set for subsequent fault diagnosis model training and verification.

[0022] A further improvement of the technical solution of the present invention is that in step 3, the decoupling of the photoacoustic spectroscopy data and the separation process of the photoacoustic spectroscopy features are as follows:

[0023] Step 301, using a multivariate statistical analysis method in chemometrics to analyze the pre-processed photoacoustic spectrum data and extract characteristic information of different gas components;

[0024] Step 302: Based on the associated data of the extracted characteristic information of different gas components, a multivariate linear regression model is trained to establish a chemometric model, the photoacoustic spectrum data of each component in the mixed gas is decoupled, and the mixed photoacoustic spectrum is decomposed into the spectral contribution of each gas component. According to the decoupling result, the spectral contribution of each gas component in the mixed spectrum is separated to obtain separate gas component spectral data, and the concentration of each gas component is calculated using the chemometric model according to the decoupled photoacoustic spectrum data and the absorption characteristics of gas molecules;

[0025] Step 303, extracting features related to different gas components from the decoupled photoacoustic spectrum data, including the position, intensity, area, width and shape of the absorption peak, using the known gas component spectrum data as a training set, and using a convolutional neural network to train a gas component recognition model so that it can learn the characteristic patterns of different gas components;

[0026] In step 304, the preprocessed and decoupled mixed gas photoacoustic spectrum data is input into the trained gas component recognition model for recognition and separation. The gas component recognition model determines the components contained in the mixed gas based on the input spectral characteristics, and outputs the recognition results of each component and the predicted concentration of each gas, thereby reducing cross-interference between different gas components and improving the accuracy of the detection results.

[0027] A further improvement of the technical solution of the present invention is that the concentration of each gas component is calculated using a chemometric model, and its calculation expression is:

[0028]

[0029] Among them, C i is the concentration of the i-th gas component, i is the index of the component, A ij is the absorption value of the i-th gas at wavelength j, j is the wavelength index of the spectral data, w ij is the weight factor, b ij is an exponential parameter related to wavelength j and component i, reflecting the trend of absorption intensity with increasing concentration, indicating the saturation effect of absorption with increasing concentration, m i is the base value;

[0030] The output of the identification results of each component and the predicted concentration of each gas is calculated as follows:

[0031]

[0032] in, is the predicted gas component concentration vector, containing the predicted concentration of each gas component, X is the spectral data matrix after preprocessing and decoupling, and W k is the weight of the k-th convolution kernel, K is the number of convolution layers in the gas component recognition model, and They represent Fourier transform and inverse Fourier transform respectively. * represents the convolution operation, which is the operation used for feature extraction in convolutional neural networks. k is the activation function of the kth layer, z k is the complex coefficient of the kth layer, which is used to adjust the predicted value, and B is the reference value vector, which is used to adjust the final concentration prediction.

[0033] A further improvement of the technical solution of the present invention is that in step 4, the process of establishing the fault diagnosis model is:

[0034] Step 401, analyzing the decoupled and separated photoacoustic spectroscopy data to identify potential features associated with transformer faults;

[0035] Step 402, extracting quantitative features and qualitative features related to transformer fault diagnosis from the separated gas component photoacoustic spectrum data, including the absorption peak position, intensity, area, width and shape of the gas and the relative intensity ratio, slope and curvature of the spectrum, and analyzing the fault type and fault severity to obtain a fault feature data set;

[0036] Step 403, selecting a classification model according to data characteristics and fault types, using a fault feature data set containing known fault types and states as a training set, training the classification model to establish a fault diagnosis model to identify different fault types and states;

[0037] Step 404, input the preprocessed and decoupled mixed gas photoacoustic spectroscopy data into the trained fault diagnosis model, identify the fault type and state of the transformer according to the input spectral feature information, calculate the fault diagnosis coefficient based on the output results of the fault diagnosis model and the relevant data of the fault feature data set, and analyze the severity and urgency of the fault.

[0038] A further improvement of the technical solution of the present invention is that the expression of the fault diagnosis coefficient is:

[0039]

[0040] Among them, FC is the fault diagnosis coefficient, which is used to evaluate the severity and urgency of the fault, n is the number of absorption peaks, α is the weight of the g-th absorption peak, which is assigned according to the importance of the peak, and p g is the position of the g-th absorption peak, s g is the intensity of the g-th absorption peak, M g is the area of ​​the g-th absorption peak, t is the number of spectral data points, r f is the weight of the relative intensity ratio of the fth data point, I f is the intensity of the fth data point, H is a set including slope and curvature, α, β, and γ are coefficients used to adjust the contribution of each part in the FC calculation. The larger the FC value, the higher the severity and urgency of the fault.

[0041] A further improvement of the technical solution of the present invention is that in step 5, different fault levels are set and the process of evaluating the severity of the fault is as follows:

[0042] Step 501, deploying the fault diagnosis model to the real-time monitoring system, combining the constructed fault diagnosis model and historical fault cases, setting different fault levels, namely, level 1 fault level, level 2 fault level, level 3 fault level and level 4 fault level, wherein the fault level increases from level 1 to level 4 as the severity of the fault increases, and matching the corresponding evaluation threshold for each fault level in combination with the fault diagnosis coefficient;

[0043] Step 502, collecting gas photoacoustic spectroscopy data in transformer oil in real time, and preprocessing the collected data, using a decoupling and separation method to decompose the mixed photoacoustic spectroscopy data into spectral contributions of each gas component;

[0044] Step 503, extracting features related to fault diagnosis from the decoupled spectral data, inputting the extracted features into the fault diagnosis model, identifying the fault type and state, calculating the fault diagnosis coefficient, determining the corresponding fault level according to the change trend of the fault diagnosis coefficient, and evaluating the severity of the fault.

[0045] A further improvement of the technical solution of the present invention is that: a plurality of the fault levels correspond to a plurality of the evaluation thresholds, wherein the evaluation thresholds include an upper threshold and a lower threshold;

[0046] The multiple fault levels and the multiple evaluation thresholds satisfy the following relationship:

[0047] Fault level 1 FC <FC L ; Indicates a minimal abnormality that may not require immediate action;

[0048] Secondary fault level FC L ≤FC <FC M ; Indicates moderate abnormality and further diagnosis or monitoring is recommended;

[0049] Three-level fault level FC H ≤FC <FC H ; Indicates a serious problem that may require repair or component replacement;

[0050] Level 4 fault level FC≥FC H ; Indicates an emergency situation requiring immediate action to avoid system failure;

[0051] Among them, FC is the fault diagnosis coefficient, FC L is the lower threshold corresponding to the second-level fault level and the upper threshold corresponding to the first-level fault level, FC M is the lower threshold corresponding to the third-level fault level and the upper threshold corresponding to the second-level fault level, FC H It is the lower threshold corresponding to the fourth fault level and the upper threshold corresponding to the third fault level.

[0052] A further improvement of the technical solution of the present invention is that in step 6, the process of generating a transformer fault monitoring report and issuing an early warning signal is:

[0053] Step 601, summarize the photoacoustic spectroscopy data collected by the real-time monitoring system and the output results of the fault diagnosis model, determine the specific type of the fault based on the analysis of the fault diagnosis model, evaluate the current severity of the fault in combination with the fault diagnosis coefficient and the fault level, and then analyze the development speed and potential impact of the fault;

[0054] Step 602, summarizing the results of fault monitoring and evaluation, and compiling a transformer fault monitoring report, including a report title, basic information, fault overview, fault severity evaluation, development trend analysis, and recommended maintenance measures;

[0055] Step 603, based on the fault level and the evaluation result, an early warning signal is automatically generated and an early warning message is issued. The early warning message includes key information such as the fault type, fault level, location of occurrence, and recommended preliminary response measures. The early warning signal and early warning message are sent to relevant personnel, including operation and maintenance personnel, technical personnel, and management, to ensure that they can respond quickly and take action.

[0056] Step 604, during the maintenance process, continuously track the maintenance progress and effect to ensure that the maintenance measures are effectively implemented. After the maintenance is completed, evaluate the maintenance effect to verify the effectiveness of the maintenance measures and whether the fault has been completely resolved, and archive the transformer fault monitoring report for future reference and analysis.

[0057] Due to the adoption of the above technical solution, the present invention has the following technical advances compared with the prior art:

[0058] 1. The present invention provides a high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil. By real-time acquisition and analysis of photoacoustic spectroscopy data of gas in transformer oil, abnormal conditions inside the transformer can be discovered early. The mixed gas spectral data are decoupled and separated in combination with chemometrics and machine learning algorithms, and various gas components are accurately identified and quantitatively analyzed. The health status of the transformer is comprehensively evaluated by extracting characteristics such as the absorption peak position, intensity, area, width and shape of the gas, as well as information such as the relative intensity ratio, slope and curvature of the spectrum, and the fault diagnosis coefficient is calculated, thereby improving the accuracy of fault detection, reducing the possibility of false alarms and missed alarms, and making fault detection more timely.

[0059] 2. The present invention provides a high-precision photoacoustic spectroscopy monitoring method for gas in transformer oil. By constructing and integrating a fault diagnosis model, different types of faults can be identified. By analyzing the decoupled spectral data, features related to fault diagnosis are extracted and input into the trained fault diagnosis model to identify the fault type and state of the transformer. In addition, by setting different fault levels and matching corresponding evaluation thresholds for each fault level, the severity of the fault is accurately evaluated, so that maintenance personnel can quickly locate the problem and take targeted maintenance measures, thereby reducing unnecessary inspection and maintenance work and improving maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0061] Figure 1 is a flow chart of the method of the present invention;

[0062] Figure 2 The present invention is a flow chart of decoupling of photoacoustic spectroscopy data and separation of photoacoustic spectroscopy features;

[0063] Figure 3 A flow chart for establishing a fault diagnosis model of the present invention;

[0064] Figure 4 A flow chart for setting different fault levels and evaluating the severity of faults for the present invention. DETAILED DESCRIPTION

[0065] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] Embodiment 1, as Figure 1 , Figure 2 As shown, the present invention provides a method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil, comprising the following steps:

[0067] Step 1, constructing a photoacoustic spectroscopy detection system, sampling the gas in the transformer oil, and obtaining the photoacoustic spectroscopy data of the gas, wherein the photoacoustic spectroscopy detection system comprises a near-infrared tunable fiber laser, a photoacoustic cell, a phase-locked amplifier, and a data acquisition system. The near-infrared tunable fiber laser emits infrared radiation of a specific wavelength, and the frequency of the light source is modulated by a modulation disk to ensure that the light source changes periodically at a specific frequency. Prepare components of the photoacoustic spectroscopy detection system, including the near-infrared tunable fiber laser, the photoacoustic cell, the phase-locked amplifier, and the data acquisition system, connect and debug the near-infrared tunable fiber laser, the modulation disk, the photoacoustic cell, the phase-locked amplifier, and the data acquisition system to ensure that the components are connected. The optical path, circuit and signal transmission path between them are unobstructed. Among them, the near-infrared tunable fiber laser is used as the light source to emit infrared radiation of a specific wavelength. The wavelength is selected according to the absorption characteristics of the gases that may exist in the transformer oil (such as H2, CO, CO2, CH4, C2H2, C2H4, C2H6, etc.). The tunability allows the laser output wavelength to be adjusted within a certain range to meet the detection needs of different gases; the photoacoustic cell is used as the place where the gas sample interacts with the laser. In the photoacoustic cell, after the gas molecules absorb the laser energy, they convert the energy into heat energy through the non-radiative relaxation process, and then generate an acoustic wave signal. The photoacoustic cell can enhance the resonance cavity of the acoustic wave signal to improve the detection sensitivity. The modulation disk is used to modulate the frequency of the light emitted by the laser, so that the light source changes periodically at a specific frequency, which helps to distinguish between background noise and photoacoustic signals caused by gas absorption. The frequency of light passing through the optical path is changed by rotation or vibration. The phase-locked amplifier is used to detect and amplify photoacoustic signals that are the same as or related to the modulation frequency. The phase-locked amplifier can suppress noise that is not related to the modulation frequency and improve the signal-to-noise ratio of the signal. The input signal is phase-locked and amplified using the feedback control principle. The data acquisition system is used to collect the photoacoustic signals output by the phase-locked amplifier and convert them into digital signals for storage and analysis. The gas sample in the sampling transformer oil is placed in the optical In the acoustic pool, a near-infrared tunable fiber laser is started and the required output wavelength is set. The light source is frequency modulated by a modulation disk so that the laser periodically irradiates the gas sample in the photoacoustic pool at a specific frequency. In the photoacoustic pool, gas molecules absorb laser energy and generate heat energy, which in turn triggers an acoustic wave signal. A phase-locked amplifier is used to detect the acoustic wave signal generated in the photoacoustic pool, and the acoustic wave signal is phase-locked and amplified. The amplified signal is clearer and is convenient for subsequent data collection and analysis. The data acquisition system collects the photoacoustic signal output by the phase-locked amplifier in real time and converts it into a digital signal for storage, ensuring that the data acquisition system is synchronized with the modulation frequency of the laser so as to correctly analyze the spectral data.

[0068] Step 2, preprocessing the collected photoacoustic spectroscopy data, including filtering, denoising, and baseline correction steps to improve data quality, and extracting fault diagnosis-related features from the preprocessed photoacoustic spectroscopy data, filtering, denoising, and baseline correction preprocessing operations are performed on the collected photoacoustic spectroscopy data to improve data quality, wherein random noise, thermal noise, and pulse interference in the spectral data are eliminated by filtering to improve the signal-to-noise ratio, mainly including sliding average filtering, SG smoothing filtering, median filtering, FFT filtering, and wavelet transform filtering, selecting a suitable filtering method according to the characteristics of the spectral data and the type of noise, further reducing noise interference by denoising, and improving the clarity of the spectral data, mainly including methods based on transform domains and methods based on spatial domains, combining the three-dimensional characteristics of spectral data (spatial domain and spectral domain), selecting a suitable denoising method, eliminating baseline drift in the spectral data by baseline correction, and making the spectral curve more accurately reflect the true information of the sample, mainly including polynomial fitting, least squares method, adaptive filtering, iterative reweighted least squares Method, according to the complexity of baseline drift and the characteristics of spectral data, a suitable baseline correction method is selected, and the features related to fault diagnosis are extracted from the preprocessed photoacoustic spectral data using a feature selection algorithm. The features are related to the absorption characteristics, spectral shape and intensity parameters of gas molecules. Among them, the features related to fault diagnosis include the absorption peak of the gas, the intensity and area of ​​the absorption peak, the width and shape of the absorption peak, the relative intensity ratio of the spectrum, the slope and curvature of the spectrum. By identifying the absorption peak of the gas, the type of gas present is determined. Combined with the intensity and area of ​​the absorption peak, the approximate concentration of the gas is determined by quantitative analysis. The width and shape of the absorption peak provide clues to the interaction between gas molecules to assist in identifying the type of gas. The intensity ratio between the specific absorption peaks of different gases is used to distinguish and identify the gas. The slope and curvature of the spectrum are used to analyze the trend of gas concentration over time. The extracted fault diagnosis-related feature data are standardized to ensure the comparability between different features, and the feature data are integrated to form a fault feature data set for subsequent fault diagnosis model training and verification.

[0069] Step 3, using chemometric methods to decouple the photoacoustic spectral data of the mixed gas, and using machine learning algorithms to separate and identify the photoacoustic spectral features of different gases, remove cross-interference between different gases, and use multivariate statistical analysis methods in chemometrics to analyze the pre-processed photoacoustic spectral data to extract characteristic information of different gas components. Based on the associated data of the extracted characteristic information of different gas components, a multivariate linear regression model is trained to establish a chemometric model, and the photoacoustic spectral data of each component in the mixed gas are decoupled, and the mixed photoacoustic spectrum is decomposed into the spectral contribution of each gas component. According to the decoupling result, the spectral contribution of each gas component in the mixed spectrum is separated to obtain separate gas component spectral data, and according to the decoupled photoacoustic spectral data, Combined with the absorption characteristics of gas molecules, the concentration of each gas component is calculated using a chemometric model, and features related to different gas components are extracted from the decoupled photoacoustic spectroscopy data, including the position, intensity, area, width and shape of the absorption peak. The known gas component spectral data is used as a training set, and a convolutional neural network is used to train the gas component recognition model so that it can learn the characteristic patterns of different gas components. The preprocessed and decoupled mixed gas photoacoustic spectroscopy data is input into the trained gas component recognition model for recognition and separation. The gas component recognition model determines the components contained in the mixed gas based on the input spectral features, and outputs the recognition results of each component and the predicted concentration of each gas, thereby reducing cross-interference between different gas components and improving the accuracy of the detection results.

[0070] Furthermore, the concentration of each gas component is calculated using the chemometric model, and the calculation expression is:

[0071]

[0072] Among them, C i is the concentration of the i-th gas component, i is the index of the component, A ij is the absorption value of the i-th gas at wavelength j, j is the wavelength index of the spectral data, w ij is the weight factor, b ij is an exponential parameter related to wavelength j and component i, reflecting the trend of absorption intensity with increasing concentration, indicating the saturation effect of absorption with increasing concentration, m i is the baseline value, which is used to adjust the model to match the actual measurement value;

[0073] Output the identification results of each component and the predicted concentration of each gas. The calculation expression is:

[0074]

[0075] in, is the predicted gas component concentration vector, is a vector containing the predicted concentration of each gas component, X is the spectral data matrix after preprocessing and decoupling, W k is the weight of the k-th convolution kernel, K is the number of convolution layers in the gas component recognition model, and They represent Fourier transform and inverse Fourier transform respectively. Fourier transform is used to transform data from the spatial domain to the frequency domain, and inverse Fourier transform is used to transform data from the frequency domain back to the spatial domain. * represents the convolution operation, which is the operation used for feature extraction in convolutional neural networks. k is the activation function of the kth layer, which is a nonlinear function, such as the ReLU function, used to introduce nonlinearity, z k is the complex coefficient of the kth layer, which is used to adjust the predicted value, and B is the reference value vector, which is used to adjust the final concentration prediction;

[0076] Step 4, analyzing the decoupled and separated photoacoustic spectroscopy data, extracting characteristic information related to transformer fault diagnosis, and establishing a fault diagnosis model to identify different fault types and states;

[0077] Step 5, integrate the established fault diagnosis model into the real-time monitoring system, conduct real-time analysis of the gas in the transformer oil, combine the constructed fault diagnosis model with historical fault cases, set different fault levels, and evaluate the severity of the fault;

[0078] Step 6: Generate a transformer fault monitoring report based on the data analysis and fault assessment results, and issue an early warning signal to prompt further inspection and maintenance. The report contains information on the fault type, fault severity, development trend, and recommended maintenance measures.

[0079] Embodiment 2, as Figure 3 , Figure 4 As shown, based on Example 1, the present invention provides a technical solution: Preferably, in step 4, the process of establishing the fault diagnosis model is:

[0080] Analyze the decoupled and separated photoacoustic spectroscopy data, identify potential features related to transformer faults, extract quantitative and qualitative features related to transformer fault diagnosis from the separated gas component photoacoustic spectroscopy data, including the absorption peak position, intensity, area, width and shape of the gas and the relative intensity ratio, slope and curvature of the spectrum, analyze the fault type and fault severity, obtain a fault feature data set, select a classification model based on data characteristics and fault type, use a fault feature data set containing known fault types and states as a training set, train a classification model to establish a fault diagnosis model to identify different fault types and states, input the preprocessed and decoupled mixed gas photoacoustic spectroscopy data into the trained fault diagnosis model, identify the transformer fault type and state based on the input spectral feature information, calculate the fault diagnosis coefficient based on the output results of the fault diagnosis model and the relevant data of the fault feature data set, and analyze the severity and urgency of the fault;

[0081] Furthermore, the expression of the fault diagnosis coefficient is:

[0082]

[0083] Among them, FC is the fault diagnosis coefficient, which is used to evaluate the severity and urgency of the fault, n is the number of absorption peaks, α is the weight of the g-th absorption peak, which is assigned according to the importance of the peak, and p g is the position of the g-th absorption peak, s g is the intensity of the g-th absorption peak, M g is the area of ​​the g-th absorption peak, t is the number of spectral data points, r f is the weight of the relative intensity ratio of the fth data point, I f is the intensity of the fth data point, H is a set including slope and curvature, α, β, γ are coefficients used to adjust the contribution of each part in the FC calculation, ln is a natural logarithm function used to process the product of area and intensity, the larger the value of FC, the higher the severity and urgency of the fault;

[0084] In step 5, different fault levels are set and the process of evaluating the severity of the fault is as follows:

[0085] The fault diagnosis model is deployed in the real-time monitoring system. In combination with the constructed fault diagnosis model and historical fault cases, different fault levels are set, namely, level 1 fault level, level 2 fault level, level 3 fault level and level 4 fault level. The fault level increases from level 1 to level 4 as the severity of the fault increases. The corresponding evaluation threshold is matched for each fault level in combination with the fault diagnosis coefficient. The gas photoacoustic spectroscopy data in the transformer oil is collected in real time, and the collected data is preprocessed. The mixed photoacoustic spectroscopy data is decomposed into the spectral contribution of each gas component using the decoupling and separation method. The features related to fault diagnosis are extracted from the decoupled spectral data. The extracted features are input into the fault diagnosis model to identify the fault type and state, calculate the fault diagnosis coefficient, determine the corresponding fault level according to the change trend of the fault diagnosis coefficient, and evaluate the severity of the fault.

[0086] Further, the multiple fault levels correspond to multiple evaluation thresholds, wherein the evaluation threshold includes an upper threshold and a lower threshold;

[0087] Multiple fault levels and multiple evaluation thresholds satisfy the following relationship:

[0088] Fault level 1 FC <FC L ; Indicates a minimal abnormality that may not require immediate action;

[0089] Secondary fault level FC L ≤FC <FC M ; Indicates moderate abnormality and further diagnosis or monitoring is recommended;

[0090] Three-level fault level FC M ≤FC <FC H ; Indicates a serious problem that may require repair or component replacement;

[0091] Level 4 fault level FC≥FC H ; Indicates an emergency situation requiring immediate action to avoid system failure;

[0092] Among them, FC is the fault diagnosis coefficient, FC L is the lower threshold corresponding to the second-level fault level and the upper threshold corresponding to the first-level fault level, FC M is the lower threshold corresponding to the third-level fault level and the upper threshold corresponding to the second-level fault level, FC H is the lower threshold corresponding to the fourth level fault level and the upper threshold corresponding to the third level fault level;

[0093] In step 6, the process of generating a transformer fault monitoring report and issuing an early warning signal is as follows:

[0094] Summarize the photoacoustic spectroscopy data collected by the real-time monitoring system and the output results of the fault diagnosis model. According to the analysis of the fault diagnosis model, determine the specific type of fault. Combined with the fault diagnosis coefficient and fault level, evaluate the current severity of the fault, and then analyze the development speed and potential impact of the fault. Summarize the results of fault monitoring and evaluation, and write a transformer fault monitoring report, including report title, basic information, fault overview, fault severity assessment, development trend analysis and recommended maintenance measures. The report title and basic information include report title, report date, transformer number, and basic information of the monitoring point location. The fault overview briefly describes the fault type, occurrence time and preliminary cause. The fault severity assessment elaborates on the evaluation basis and results of the fault level and explains the severity of the fault. The development trend analysis is based on the data analysis results and describes the fault. Development trend, including possible expansion scope, impact on the system, etc., according to the type and degree of fault, put forward specific maintenance suggestions, including emergency treatment measures, maintenance steps, required materials and tools, etc., according to the fault level and evaluation results, automatically generate early warning signals and issue early warning information, which contains key information such as fault type, fault level, location of occurrence, and recommended initial response measures. Send early warning signals and early warning information to relevant personnel, including operation and maintenance personnel, technical personnel and management, to ensure that they can respond quickly and take action. During the maintenance process, continuously track the maintenance progress and effect to ensure that the maintenance measures are effectively implemented. After the maintenance is completed, evaluate the maintenance effect to verify the effectiveness of the maintenance measures and whether the fault has been completely resolved, and archive the transformer fault monitoring report for future reference and analysis.

[0095] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil, characterized in that: The following steps are involved: Step 1, constructing a photoacoustic spectroscopy detection system, sampling the gas in the transformer oil, and obtaining the photoacoustic spectroscopy data of the gas, wherein the photoacoustic spectroscopy detection system comprises a near-infrared tunable fiber laser, a photoacoustic cell, a phase-locked amplifier, and a data acquisition system; Step 2, preprocessing the collected photoacoustic spectroscopy data, and extracting fault diagnosis-related features from the preprocessed photoacoustic spectroscopy data; Step 3: Decouple the photoacoustic spectroscopy data of the mixed gas using a chemometric method, and separate and identify the photoacoustic spectroscopy features of different gases using a machine learning algorithm to remove cross-interference between different gases; Step 4, analyzing the decoupled and separated photoacoustic spectroscopy data, extracting characteristic information related to transformer fault diagnosis, and establishing a fault diagnosis model to identify different fault types and states; Step 5, integrate the established fault diagnosis model into the real-time monitoring system, conduct real-time analysis of the gas in the transformer oil, combine the constructed fault diagnosis model with historical fault cases, set different fault levels, and evaluate the severity of the fault; Step 6: Generate a transformer fault monitoring report and issue an early warning signal based on the data analysis and fault assessment results.

2. The method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 1, characterized in that: In step 1, the process of constructing the photoacoustic spectroscopy detection system and acquiring the gas photoacoustic spectroscopy data is as follows: Step 101, preparing components of a photoacoustic spectroscopy detection system, including a near-infrared tunable fiber laser, a photoacoustic cell, a lock-in amplifier, and a data acquisition system, connecting and debugging the near-infrared tunable fiber laser, the modulation disk, the photoacoustic cell, the lock-in amplifier, and the data acquisition system, so that the optical path, the circuit, and the signal transmission path between the components are unobstructed; Step 102, the gas sample in the sampling transformer oil is placed in a photoacoustic cell through vacuum degassing, the near-infrared tunable fiber laser is started, and the required output wavelength is set, and the light source is frequency modulated by a modulation disk so that the laser periodically irradiates the gas sample in the photoacoustic cell; Step 103, in the photoacoustic cell, the gas molecules generate heat energy after absorbing the laser energy, thereby generating an acoustic wave signal, and the acoustic wave signal generated in the photoacoustic cell is detected by using a phase-locked amplifier, and the acoustic wave signal is phase-locked and amplified; Step 104: The data acquisition system acquires the photoacoustic signal output by the phase-locked amplifier in real time and converts it into a digital signal for storage.

3. The method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 2, characterized in that: In step 2, the process of extracting the relevant features of fault diagnosis is as follows: Step 201, performing preprocessing operations of filtering, denoising, and baseline correction on the collected photoacoustic spectroscopy data; Step 202, using a feature selection algorithm to extract features related to fault diagnosis from the preprocessed photoacoustic spectrum data, the features are related to the absorption characteristics, spectral shape, and intensity parameters of the gas molecules, wherein the features related to fault diagnosis include the absorption peak of the gas, the intensity and area of ​​the absorption peak, the width and shape of the absorption peak, the relative intensity ratio of the spectrum, and the slope and curvature of the spectrum; Step 203 , standardize the extracted fault diagnosis related feature data, and integrate each feature data to form a fault feature data set.

4. The method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 3, characterized in that: In step 3, the decoupling of photoacoustic spectroscopy data and the separation process of photoacoustic spectroscopy features are as follows: Step 301, using a multivariate statistical analysis method in chemometrics to analyze the pre-processed photoacoustic spectrum data and extract characteristic information of different gas components; Step 302: Based on the extracted correlation data of characteristic information of different gas components, a multivariate linear regression model is trained to establish a chemometric model, the photoacoustic spectrum data of each component in the mixed gas is decoupled, the mixed photoacoustic spectrum is decomposed into the spectral contribution of each gas component, and according to the decoupling result, the spectral contribution of each gas component in the mixed spectrum is separated to obtain separate gas component spectral data, and the concentration of each gas component is calculated using the chemometric model according to the decoupled photoacoustic spectrum data and the absorption characteristics of gas molecules; Step 303, extracting features related to different gas components from the decoupled photoacoustic spectrum data, including the position, intensity, area, width and shape of the absorption peak, using the known gas component spectrum data as a training set, and using a convolutional neural network to train a gas component recognition model so that it can learn the characteristic patterns of different gas components; In step 304, the preprocessed and decoupled mixed gas photoacoustic spectrum data is input into a trained gas component recognition model for recognition and separation. The gas component recognition model determines the components contained in the mixed gas based on the input spectral features, and outputs the recognition results of each component and the predicted concentration of each gas to reduce cross-interference between different gas components.

5. The method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 4, characterized in that: The concentration of each gas component is calculated using the chemometric model, and the calculation expression is: Among them, C i is the concentration of the i-th gas component, i is the index of the component, A ij is the absorption value of the i-th gas at wavelength j, j is the wavelength index of the spectral data, w ij is the weight factor, b ij is an exponential parameter related to wavelength j and component i, reflecting the trend of absorption intensity with increasing concentration, indicating the saturation effect of absorption with increasing concentration, m i is the base value; The output of the identification results of each component and the predicted concentration of each gas is calculated as follows: in, is the predicted gas component concentration vector, X is the spectral data matrix after preprocessing and decoupling, and W k is the weight of the k-th convolution kernel, K is the number of convolution layers in the gas component recognition model, and They represent Fourier transform and inverse Fourier transform respectively. * represents the convolution operation, which is the operation used for feature extraction in convolutional neural networks. k is the activation function of the kth layer, z k is the complex coefficient of the kth layer, and B is the reference value vector used to adjust the final concentration prediction.

6. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 5, characterized in that: In step 4, the process of establishing the fault diagnosis model is as follows: Step 401, analyzing the decoupled and separated photoacoustic spectroscopy data to identify potential features associated with transformer faults; Step 402, extracting quantitative features and qualitative features related to transformer fault diagnosis from the separated gas component photoacoustic spectrum data, including the absorption peak position, intensity, area, width and shape of the gas and the relative intensity ratio, slope and curvature of the spectrum, and analyzing the fault type and fault severity to obtain a fault feature data set; Step 403, selecting a classification model according to data characteristics and fault types, using a fault feature data set containing known fault types and states as a training set, training the classification model to establish a fault diagnosis model to identify different fault types and states; Step 404, input the preprocessed and decoupled mixed gas photoacoustic spectroscopy data into the trained fault diagnosis model, identify the fault type and state of the transformer according to the input spectral feature information, calculate the fault diagnosis coefficient based on the output results of the fault diagnosis model and the relevant data of the fault feature data set, and analyze the severity and urgency of the fault.

7. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 6, characterized in that: The expression of the fault diagnosis coefficient is: Among them, FC is the fault diagnosis coefficient, which is used to evaluate the severity and urgency of the fault, n is the number of absorption peaks, α is the weight of the g-th absorption peak, and p g is the position of the g-th absorption peak, s g is the intensity of the g-th absorption peak, M g is the area of ​​the gth absorption peak, t is the number of spectral data points, r f is the weight of the relative intensity ratio of the fth data point, I f is the intensity of the fth data point, H is a set containing slope and curvature, and α, β, and γ are coefficients used to adjust the contribution of each part in the FC calculation.

8. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 7, characterized in that: In step 5, different fault levels are set and the process of evaluating the severity of the fault is as follows: Step 501, deploying the fault diagnosis model to the real-time monitoring system, combining the constructed fault diagnosis model and historical fault cases, setting different fault levels, namely, level 1 fault level, level 2 fault level, level 3 fault level and level 4 fault level, wherein the fault level increases from level 1 to level 4 as the severity of the fault increases, and matching the corresponding evaluation threshold for each fault level in combination with the fault diagnosis coefficient; Step 502, collecting gas photoacoustic spectroscopy data in transformer oil in real time, and preprocessing the collected data, using a decoupling and separation method to decompose the mixed photoacoustic spectroscopy data into spectral contributions of each gas component; Step 503, extracting features related to fault diagnosis from the decoupled spectral data, inputting the extracted features into the fault diagnosis model, identifying the fault type and state, calculating the fault diagnosis coefficient, determining the corresponding fault level according to the change trend of the fault diagnosis coefficient, and evaluating the severity of the fault.

9. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 8, characterized in that: A plurality of the fault levels correspond to a plurality of the evaluation thresholds, wherein the evaluation thresholds include an upper threshold and a lower threshold; The multiple fault levels and the multiple evaluation thresholds satisfy the following relationship: Fault level 1 FC <FC L ; Secondary fault level FC L ≤FC<FC M ; Three-level fault level FC M ≤FC<FC H ; Level 4 fault level FC≥FC H ; Among them, FC is the fault diagnosis coefficient, FC L is the lower threshold corresponding to the second-level fault level and the upper threshold corresponding to the first-level fault level, FC M is the lower threshold corresponding to the third-level fault level and the upper threshold corresponding to the second-level fault level, FC H It is the lower threshold corresponding to the fourth fault level and the upper threshold corresponding to the third fault level.

10. A method for high-precision photoacoustic spectroscopy monitoring of gas in transformer oil according to claim 9, characterized in that: In step 6, the process of generating a transformer fault monitoring report and issuing an early warning signal is as follows: Step 601, summarize the photoacoustic spectroscopy data collected by the real-time monitoring system and the output results of the fault diagnosis model, determine the specific type of the fault based on the analysis of the fault diagnosis model, evaluate the current severity of the fault in combination with the fault diagnosis coefficient and the fault level, and then analyze the development speed and potential impact of the fault; Step 602, summarizing the results of fault monitoring and evaluation, and compiling a transformer fault monitoring report, including a report title, basic information, fault overview, fault severity evaluation, development trend analysis, and recommended maintenance measures; Step 603, automatically generate a warning signal and issue a warning message based on the fault level and the evaluation result. The warning message includes key information such as the fault type, fault level, location, and recommended initial response measures. The warning signal and warning message are sent to relevant personnel, including operation and maintenance personnel, technical personnel, and management. Step 604, during the maintenance process, the maintenance progress and effect are continuously tracked. After the maintenance is completed, the maintenance effect is evaluated to verify the effectiveness of the maintenance measures and whether the fault has been completely resolved, and the transformer fault monitoring report is archived for future reference.

Images