31 results about "Dissolved gas analysis" patented technology

The invention discloses a transformer insulating oil chromatographic data abnormity identification method, and relates to the technical field of data identification, and the method comprises the following steps: carrying out dissolved gas analysis on an insulating oil sample through a gas chromatograph to obtain a chromatographic original response curve; performing baseline correction and de-noising processing on the chromatographic original response curve, positioning each gas peak based on a first-order derivative method and a second-order derivative method, and extracting chromatographic peak height and chromatographic peak area parameters; according to the extracted chromatographic peak height and chromatographic peak area, three Duval fault ratios are calculated, and the ratio of the total soluble gas concentration to each gas concentration threshold value is calculated; combining into a multi-dimensional diagnosis feature vector; according to the method, the chromatographic peak shape of the insulating oil sample is subjected to multi-stage baseline correction and de-noising treatment, and the gas peak position is accurately positioned based on the first-order and second-order derivative methods, so that the high precision of chromatographic peak height and chromatographic peak area extraction is ensured, the accuracy of dissolved gas concentration calculation is improved, and the misjudgment risk caused by noise or baseline drift is reduced.

The invention belongs to the technical field of power transformer monitoring, and particularly relates to a transformer state comprehensive prediction and evaluation method and system based on decision cascade fusion analysis and a storage medium. Analyzing the concentration of the characteristic gas dissolved in the oil in different operation states according to the oil sample, and calculating the generation rate of the characteristic gas; the concentration of the characteristic gas dissolved in the oil is introduced into various fault diagnosis methods based on analysis of the gas dissolved in the oil according to different decision-making means to evaluate and predict the state of the transformer; and fusing the discharge fault judgment result or the overheating fault judgment result with the concentration ratio of carbon monoxide to carbon dioxide in the oil sample, judging whether an insulation system involved in the transformer state is a pure oil insulation system or an oil paper insulation system, and completing fault classification of the transformer operation state. According to the invention, the operation state of the transformer is accurately predicted.

The invention relates to the technical field of transformer fault diagnosis, in particular to an IGWO-GBDT-based transformer fault diagnosis method and system, and the method comprises the steps: constructing an improved grey wolf optimization algorithm, and optimizing a grey wolf population position updating strategy through a nonlinear convergence factor, a dynamic weight distribution mechanism and a Levy flight disturbance mechanism; performing global optimization on the hyper-parameter combination of the gradient boosting tree by using an improved grey wolf optimization algorithm to determine an optimal hyper-parameter combination; and training the optimized gradient boosting tree by adopting a dissolved gas analysis data set, and performing transformer fault type diagnosis by using the trained gradient boosting tree. The improved grey wolf optimization algorithm (IGWO) is adopted to optimize GBDT hyper-parameters, and the diagnosis precision and efficiency are improved.

The invention discloses a transformer fault diagnosis method and device, terminal equipment and a storage medium, and belongs to the technical field of transformer fault diagnos.The method comprises the steps that image feature extraction is conducted on image data of a transformer to generate element image features; the method comprises the following steps: performing data conversion mapping on analysis data of gas dissolved in oil to generate two-dimensional image features, performing semantic alignment and structured processing on historical operation and maintenance data to generate text features, and further performing feature fusion on element image features, the two-dimensional image features and the text features to generate a multi-modal fusion feature vector; according to the invention, fault diagnosis is carried out on the transformer through the fault diagnosis model according to the multi-modal fusion feature vector, and then the large language model is adopted to output the fault diagnosis report of the transformer in the natural language format, so that the fault diagnosis efficiency and accuracy of the transformer are improved.

The invention provides a power frequency arc intelligent diagnosis method, device and equipment and a medium, and relates to the technical field of power frequency arc diagnos.The method comprises the steps that optical signal data and analysis data of gas dissolved in oil are obtained; the optical signal data and the dissolved gas analysis data are preprocessed, and a training data set is constructed based on the preprocessed data and the corresponding manual distribution labels; the training data set and a preset target loss function are adopted to train a double-flow neural network model, a target fault diagnosis model is obtained, and the double-flow neural network model comprises an optical signal network branch, an analysis data network branch and a fusion module; and inputting the real-time monitoring data into the target fault diagnosis model to obtain a target fault diagnosis result. Through the intelligent diagnosis method for the power frequency arc, early warning and accurate fault type diagnosis of the power frequency arc can be realized.

The invention relates to the technical field of analysis and detection of gas dissolved in oil in an electric power system, and discloses an online monitoring system for gas dissolved in oil and a headspace elution split-flow concentration oil-gas separation method thereof, the system comprises an oil cylinder, a stepping motor, a buffer chamber, a chromatographic analysis module, a standard gas steel cylinder, a heating control module, a valve, a gas pump and a pipeline; wherein the oil cylinder and the stepping motor are horizontally arranged, a lead screw of the stepping motor is connected with a piston in the oil cylinder, and the heating control modules are arranged on the outer sides of the oil cylinder and the chromatographic analysis module; the invention also provides a headspace elution split-flow concentration oil-gas separation method of the online monitoring system for dissolved gas in oil, the headspace elution split-flow concentration oil-gas separation does not need to complete rapid headspace degassing under an ultimate vacuum condition, the degassing time is effectively shortened, and the method is similar to a conventional offline degassing method, so that the cost is reduced. The influence of the vacuum degree and the vacuum degassing rate on an analysis result does not need to be considered, and the device has the advantages of simple structure and good repeatability.

The application discloses a transformer fault diagnosis method based on a small sample unbalanced data set, comprising the following steps: S1, acquiring a transformer dissolved gas analysis data set and a comprehensive feature set; S2, constructing a transformer fault diagnosis model based on a small sample unbalanced data set; S3, according to the gas analysis data set and the comprehensive feature set, performing model training on the constructed transformer fault diagnosis model to obtain an optimal fault diagnosis model; and realizing transformer fault diagnosis based on the small sample unbalanced data set according to the optimal fault diagnosis model. The application solves the problems that the current traditional method cannot accurately diagnose transformer faults due to the technical problems such as complex fault mode recognition, specific class accurate diagnosis, insufficient model generalization ability and difficulty in fusion parameter optimization under a small sample unbalanced fault data set in the prior art.

The invention discloses a quantitative analysis method and system for overheat fault gas of an on-load tap-changer contact. The method comprises the following steps: constructing a multi-physical field coupling mathematical model of an on-load tap-changer, and defining contact resistance of a contact surface; solving the multi-physics field coupling model to obtain time domain and space domain temperature distribution of the surface of the on-load tap-changer contact, and determining the highest temperature and the volume of an overheat region; calculating the local gas production rate of each fault characteristic gas; carrying out integral calculation on local gas production rates of all fault characteristic gases in the hot area volume to obtain theoretical equilibrium concentration of each fault characteristic gas in oil, and constructing a mapping relation between an overheating parameter formed by the contact resistance and the highest temperature of the contact surface and the theoretical equilibrium concentration; the method aims at solving the problems that a traditional analysis method for the gas dissolved in the oil depends on experience criteria and is high in hysteresis quality, breaks through the bottleneck that the whole process of electricity-heat-fluidization cannot be simulated from the mechanism level, and achieves the crossing from qualitative diagnosis to quantitative prediction.

The application discloses a kind of overheat fault gas quantitative analysis method and system of on-load tap-changer contact, the method of the present application includes constructing the multi-physics field coupling mathematical model of on-load tap-changer, define contact surface contact resistance;Solving multi-physics field coupling model obtains the time domain and spatial temperature distribution of on-load tap-changer contact surface, determine the highest temperature and the volume of overheated area;Calculate the local gas production rate of each fault characteristic gas;The local gas production rate of all fault characteristic gases in the volume of hot area is integrated to obtain the theoretical equilibrium concentration of each fault characteristic gas in oil, construct the mapping relationship between the overheating parameters of contact surface contact resistance and highest temperature and theoretical equilibrium concentration.The present application aims to solve the problem that the traditional oil dissolved gas analysis method relies on empirical criterion and has high hysteresis, break through the bottleneck that cannot simulate the whole process of "electric-thermal-flow-chemical" from mechanism level, realize the leap from qualitative diagnosis to quantitative prediction.

The invention belongs to the technical field of membrane separation, and particularly relates to an oil-gas separation membrane as well as a preparation method and application thereof. The invention relates to an oil-gas separation membrane. The separation membrane comprises a PT610 resin polymer; the separation membrane is configured to separate dissolved gases in a non-polar oil phase environment and maintain the integrity of its chemical skeletal structure in the environment. The invention also provides an application of the oil-gas separation membrane in on-line oil-gas separation or dissolved gas analysis of electric power equipment.

The invention discloses a transformer service characteristic evaluation system fused with a BiLSTM network, and the system comprises a multi-source data collection module which is used for collecting grounding resistance dynamic monitoring data, analysis data of gas dissolved in oil, vibration spectrum data, and infrared thermal imaging data of a transformer; the data preprocessing module is in communication connection with the multi-source data acquisition module and is used for performing space-time alignment, noise filtering and feature standardization processing on the acquired multi-source data; the invention relates to the technical field of power systems. According to the transformer service characteristic evaluation system fused with the BiLSTM network, grounding resistance, gas dissolved in oil, vibration spectrum and infrared thermal imaging data are integrated through the multi-source data acquisition module, and each acquisition unit has high precision, so that the one-sidedness of traditional single-source data acquisition can be avoided; more complete and accurate original data support is provided for transformer service characteristic evaluation, and evaluation deviation caused by data missing or insufficient precision is reduced.

The invention discloses a test device for simulating generation and diffusion behaviors of dissolved gas in oil and a use method. The device comprises a real transformer test device, and sampling positions are arranged at different positions of the front surface and the side surface of a transformer; an insulation discharge defect gas production simulation module; the overheating defect gas production simulation module; the three-dimensional space region oil sample multi-point collection and oil dissolved gas real-time analysis module is composed of a gas collection and separation unit, a gas detection unit and an oil dissolved gas analysis and visualization device; each sampling position is connected with one group of oil sample collection and separation unit and gas detection unit through a metal oil pipe, and finally detected diffusion information of the dissolved gas in the oil at each sampling position is transmitted to the dissolved gas in the oil analysis and visualization equipment to obtain the diffusion characteristic of the fault characteristic gas in the oil. According to the invention, different types of defects are directly set in the real transformer, so that the actual fault gas production condition of the transformer can be truly and effectively simulated.

The invention discloses a transformer fault detection method and system, and relates to the technical field of transformer fault detection, and the method comprises the steps: obtaining a plurality of training modal data, carrying out the preprocessing of each training modal data, generating a modal feature set, training a preset initial transformer detection model through employing the modal feature set, and generating a transformer detection model. The method comprises the steps of obtaining modal data of a to-be-detected transformer, performing gas feature extraction on the modal data through a time sequence feature network of a transformer detection model to obtain a corresponding gas feature vector, performing noise feature extraction on the modal data through a visual feature network of the transformer detection model to obtain a corresponding noise feature vector, and performing fault detection on the noise feature vector and the gas feature vector through a detection network of the transformer detection model to obtain a corresponding fault detection result. The technical problems that the detection accuracy of a dissolved gas analysis method is reduced and the operation reliability of the transformer is reduced under the complicated condition that multiple faults are concurrent are solved.

This invention discloses a fault diagnosis method and system based on dissolved gases in modified ester insulating oil. The method includes the following steps: S1, constructing a gas generation simulation system for overheating defects in modified ester paper insulation; S2, using the simulation system to simulate different types of overheating defects in paper insulation, performing DGA analysis, and collecting data on different types of overheating defects in modified ester paper insulation; S3, based on the DGA analysis data, finding boundary values ​​that distinguish different overheating defects, deriving a coding rule table and an overheating defect type judgment table based on the boundary values, and performing fault diagnosis based on the overheating defect type judgment table. This invention can achieve criterion correction for different types of modified esters, provide judgment intervals based on the characteristics of modified esters according to existing data, and improve the judgment accuracy of the three-ratio method for dissolved gas analysis in modified ester oil.

A system for gas analysis is provided. The system includes a gas extraction assembly having an extraction chamber having an upper end, a lower end, and an interior space. The gas extraction assembly further having one or more retainers positioned within the interior space of the extraction chamber. The gas extraction assembly further having gas permeable membrane retained within the interior space of the extraction chamber by the one or more retainers, the gas permeable membrane disposed in S-shaped curves extending from adjacent the upper end to adjacent the lower end within the interior space of the extraction chamber.

The invention discloses a power transformer health index calculation method based on principal component analysis and a mass-spring-damping model. The method comprises the following steps: collecting multi-stage data of analysis parameters and oil quality parameters of dissolved gas in transformer oil; determining a weight coefficient of each parameter by using a principal component analysis method; calculating a DGA index and an oil quality index in combination with a mode coefficient in a CIGRE technical manual 761; enabling the transformer to be equivalent to a second-order mass-spring-damping system, and mapping a DGA index and an oil quality index into a damping coefficient and an inherent frequency parameter of the system respectively; and constructing a system state space matrix, and calculating a system matrix infinite norm ratio of the new transformer to the transformer in operation to obtain a transformer health index. The method improves the accuracy and reliability of health index calculation, and is suitable for transformer state evaluation and life prediction.

The utility model discloses a device for analyzing gas dissolved in oil by a spectrometry method, and relates to the technical field of analysis of gas dissolved in oil. The device comprises a box body, wherein a dissolved gas analyzer is fixed on one side of the box body; a plurality of interfaces are fixed at the top of the dissolved gas analyzer; a plurality of supporting devices are fixed in the box body, each supporting device comprises a positioning rod, and one end of each positioning rod is fixedly connected with the box body; the outer side of the positioning rod is movably connected with a bracket; a fastening frame is fixed at one end of the bracket; a plurality of positioning grooves are formed in two sides of the fastening frame; and a clamping plate is movably connected into the fastening frame. Under the action of the supporting device, the device can be used for stably mounting a gas conveying pipeline when gas dissolved in oil is detected, so that looseness between the conveying pipeline and a connector caused by vibration generated during gas conveying is prevented, the air tightness is prevented from being reduced, external gas permeates into the detection gas, and the detection reliability is improved. Therefore, errors of the detection result are avoided.

The invention discloses a transformer fault diagnosis method based on a small sample unbalanced data set. The method comprises the following steps: S1, acquiring an analysis data set and a comprehensive feature set of gas dissolved in a transformer; s2, constructing a transformer fault diagnosis model based on the small sample unbalanced data set, and S3, performing model training on the constructed transformer fault diagnosis model according to the gas analysis data set and the comprehensive feature set to obtain an optimal fault diagnosis model; and transformer fault diagnosis based on the small sample unbalanced data set is realized according to the optimal fault diagnosis model. According to the method, the problem that the transformer fault cannot be accurately diagnosed due to the technical problems of complex fault mode recognition, specific category accurate diagnosis, insufficient model generalization ability, difficulty in fusion parameter optimization and the like in the condition of adapting to a small sample unbalanced fault data set in the prior art is solved.

This invention relates to the technical field of transformer fault diagnosis, and in particular to a transformer fault diagnosis method and system based on the IGWO-GBDT. This method constructs an improved gray wolf optimization algorithm (IGWO) and optimizes the position update strategy of the gray wolf population through a nonlinear convergence factor, a dynamic weight allocation mechanism, and a Levy flight perturbation mechanism. The improved gray wolf optimization algorithm is used to globally optimize the hyperparameter combination of a gradient boosting tree to determine the optimal hyperparameter combination. The optimized gradient boosting tree is trained using a dissolved gas analysis dataset and then used to diagnose transformer fault types. This method uses the improved gray wolf optimization algorithm (IGWO) to optimize GBDT hyperparameters, improving diagnostic accuracy and efficiency.

The invention discloses a transformer state signal analysis method, and relates to the technical field of transformers, and the method comprises the steps: converting dissolved gas analysis data into physical characteristic data based on a preset prior knowledge base; judging a current sample mode of the target transformer according to the physical characteristic data; if the current sample mode is a known mode, inputting the physical characteristic data into a pre-trained multivariate classifier to obtain a basic fault physical mode label of the target transformer; if the current sample mode is an unknown mode, inputting the physical feature data and the equipment parameters into a pre-trained semantic mapping model to obtain distribution representation data in a semantic space, the distribution representation data being a probability distribution parameter set; determining the negative logarithm likelihood of the distribution representation data and the plurality of composite fault prototypes; and determining a composite fault physical mode label of the target transformer according to the composite fault prototype corresponding to the minimum one of the plurality of negative logarithm likelihoods. According to the invention, the accuracy of transformer oil state monitoring is improved.

The invention discloses an in-operation transformer fault diagnosis method based on analysis of gas dissolved in oil, and the method comprises the steps: collecting a fault sample of the content of gas dissolved in transformer oil with a clear fault type, and carrying out the classification and coding of the fault sample; aDASYN-MK is adopted to carry out adaptive oversampling on a small number of fault samples, and equalization processing of the fault samples is realized; performing normalization processing on the characteristic gas of the fault sample by using a normalization formula; adopting an improved frost ice optimization algorithm to automatically optimize hyper-parameters of the random forest model; dividing the equalized fault samples into a training set and a test set, and establishing a transformer fault intelligent diagnosis model based on the random forest based on hyper-parameters obtained through optimization; and inputting the data of the dissolved gas in the in-operation transformer oil, and carrying out fault diagnosis by using the method provided by the invention. According to the method, the fault diagnosis accuracy of the in-operation transformer is effectively improved.

The method for analyzing the dissolved gas in the transformer oil comprises a dissolved gas content measuring method and a dissolved gas concentration measuring method, and the gas content measuring method comprises the steps that pressure is pumped to be vacuum, and the initial pressure before oil injection is obtained after the pressure is balanced; heating after oil injection to control the temperature to be constant at a first degassing temperature, carrying out degassing treatment, obtaining the balance pressure after degassing after the pressure is balanced, obtaining the pressure difference according to the initial pressure and the balance pressure, and calculating and obtaining the gas content of the oil sample to be detected in combination with a gas content correction model; according to the gas concentration measurement method, the in-oil concentration of the gas is obtained by obtaining the gas phase concentration, the gas content and the gas-liquid conversion coefficient. Compared with the prior art, through the core concept of designing the gas content correction model, the non-linear precision problem of the vacuum degassing method is solved, the differential pressure measurement process is standardized, and the measurement precision and reliability of the whole process are effectively improved.

The invention relates to a microfluidic device for analyzing dissolved gas in insulating oil and a control method. The micro-fluidic device comprises a micro-fluidic chip and a heating device, the micro-fluidic chip is provided with an oil sample inlet, a micro-channel area and a gas-liquid separation area, the oil sample inlet is communicated with the micro-channel area, the micro-channel area is communicated with the gas-liquid separation area, a pump cavity used for containing fluid is formed in the micro-channel area, and the heating device is arranged in the pump cavity. The pump cavity is respectively communicated with the oil sample inlet and the gas-liquid separation area, a piezoelectric brake is arranged in the pump cavity, the piezoelectric brake is connected with an elastic pump membrane of the pump cavity, and the heating device is used for heating the micro-channel area; the piezoelectric brake and the heating device are integrated on the micro-fluidic chip, and the micro-fluidic chip is small in size and high in extraction efficiency.

The present invention provides a helium ionization gas chromatograph for analyzing dissolved gases in transformer oil, comprising: a gas cleaning and purging system, a medium and heavy hydrocarbon cutting and backflushing system, an air interference cutting system, a sampling system, a helium ion detector, and a control system. The gas cleaning and purging system is connected to the medium and heavy hydrocarbon cutting and backflushing system; the medium and heavy hydrocarbon cutting and backflushing system is respectively connected to the air interference cutting system and the sampling system; the air interference cutting system is connected to the sampling system; and the sampling system is connected to the helium ion detector. The present invention avoids the interference of high molecular hydrocarbons on low molecular hydrocarbons, improves the efficiency of continuous sample analysis, and improves the reliability and repeatability of sample analysis results. At the same time, the helium ion detector has a wide analysis range and high detection limit sensitivity, which is conducive to timely discovering internal faults of detection equipment, enabling equipment with potential faults to be repaired plannedly and economically, and avoiding equipment damage and unplanned power outages.

The invention discloses an analysis sample generation method and device for dissolved gas in transformer oil, a medium and a product, and the method comprises the steps: classifying transformer fault types corresponding to time sequence DGA data, and determining the classified data; adding noise to the classified data according to a forward Markov chain, and determining noise data obeying Laplacian distribution; based on a reverse Markov chain, denoising the noise data obeying Laplacian distribution, and determining denoised data; carrying out trend decomposition on the de-noised data based on an encoder and a decoder, and determining a long-term trend characteristic and a time-frequency characteristic of the de-noised data; and according to the long-term trend characteristic and the time-frequency characteristic of the de-noised data, determining reconstructed time sequence DGA data, and taking the reconstructed time sequence DGA data as an analysis sample of the dissolved gas in the transformer oil. According to the method, the sample is analyzed based on the high-quality dissolved gas in the transformer oil, the accuracy and efficiency of transformer fault diagnosis are improved, and the performance of a power transformation system is improved.

The invention provides a chromatographic detection system and detection method for analyzing dissolved gas in transformer oil, and the system comprises an integrated solenoid valve type valve terminal module which is used for carrying out flow path switching and controlling a sample to enter an integrated constant flow separation module; the integrated constant-current separation module comprises an input module for a constant-current sample, a first chromatographic column channel for separating a first gas in the sample and a second chromatographic column channel for separating a second gas in the sample; the composite detection module comprises a first catalytic combustion sensor arranged at the outlet of the first chromatographic column channel, and an infrared optical sensor and a second catalytic combustion sensor which are arranged at the outlet of the second chromatographic column channel and are connected in series; and the back-blowing stable base line module is connected with a back-blowing gas circuit controlled by the integrated electromagnetic valve type valve terminal module, and is used for back-blowing the double chromatographic column to remove residual components. On the basis of integration of electromagnetic valve terminal control, double-column parallel separation and multi-sensor detection, the high-standard requirement for online monitoring of the state of power equipment is met.

The present invention discloses a method and system for evaluating the insulation state of a transformer based on dissolved gas analysis in oil. The method includes: obtaining the normal historical data of the transformer, preprocessing it, and constructing a corresponding set of grid spaces; calculating the relative entropy thresholds of all grids in each dimension of the set of grid spaces; comparing the relative entropy of each grid with the corresponding relative entropy threshold, performing grid adaptive partitioning, and then comparing the number of normal historical data in each grid with a threshold to determine normal data clusters; calculating the centroids of all normal data clusters as cluster centers; determining the normal state distance range based on the distances between all normal historical data and the cluster centers; obtaining the real-time data of the transformer to be evaluated, preprocessing it, and determining the insulation state according to the relationship between the distance between the real-time data and the cluster centers and the normal state distance range. The present invention is not affected by manually set thresholds, has high adaptability, and does not require fault sample data.

The present invention provides a transformer malfunction diagnostic device and a malfunction diagnosis method using same, wherein a rule-based learning method is combined with a deep-learning-based learning method based on artificial intelligence. The transformer malfunction diagnosis device according to an embodiment of the present invention comprises: a data scaling unit for scaling dissolved gas analysis data acquired from a transformer, a provisionally labeled data acquisition unit for converting unlabeled data, among the scaled dissolved gas analysis data, to provisionally labeled data and acquiring same; a prelearning unit for performing prelearning for the unlabeled data, among the scaled dissolved gas analysis data, and the provisionally labeled data; and a relearning unit for performing relearning for a labeled data through the transformer of parameters optimized by performing the prelearning.

The invention discloses a transformer monitoring method and system based on analysis of gas dissolved in oil, terminal equipment and a storage medium. According to the method, preliminary fault type identification is carried out through a trained decision tree model according to the gas concentration of each dissolved gas, the characteristic ratio of a plurality of gas concentrations and the gas concentration change rate of each dissolved gas obtained through calculation of chromatographic data of each dissolved gas, so that a preliminary fault prediction identification result of the transformer is obtained; and carrying out secondary fault identification on the transformer through the trained support vector machine classifier to obtain a final fault identification result of the transformer to be detected. According to the invention, preliminary fault identification and secondary fault identification are carried out through the decision tree model and the support vector classifier, and refined fault identification is carried out on the transformer, so that the accuracy of fault identification is improved; the problem that the accuracy of fault identification is low due to the fact that the fault type cannot be finely identified by single fault identification operation in the existing transformer monitoring technology is solved.

The invention discloses a transformer fault diagnosis method, and relates to the field of fault diagnosis, and the method comprises the steps: obtaining a dissolved gas analysis data set of a transformer, carrying out the optimization of the dissolved gas analysis data set through employing a factor analysis method, and obtaining a dissolved gas analysis optimization data set; inputting the dissolved gas analysis optimization data set into the trained fault diagnosis model to obtain a transformer fault diagnosis result; the fault diagnosis model comprises a feature precoding module, a TransformerFat module, an attention module and a trans-attention fusion module; the feature precoding module is used for obtaining a time sequence feature vector; the TransformerFeat module is used for carrying out feature extraction on the time sequence feature vector to obtain a global feature; the attention module is used for performing feature extraction on the global features to obtain local features; and the trans-attention fusion module is used for fusing the global features and the local features to obtain a transformer fault diagnosis result. According to the invention, the accuracy and robustness of transformer fault diagnosis are improved.