Transformer full-life-cycle health diagnosis method, system and product

Abstract

The invention discloses a transformer full life cycle health diagnosis method, system and product, and the method comprises the steps: collecting electric field intensity, magnetic field intensity and acoustic emission signals in a transformer through a three-dimensional monitoring network; the electric field intensity, the magnetic field intensity and the sampling frequency of the acoustic emission signal are sequentially reduced to form a sampling time sequence of echelon triggering; in the sampling time sequence, when it is detected that the change rate of the electric field intensity is larger than a preset change rate, high-speed sampling operation of the magnetic field signals is triggered to obtain the space-time corresponding relation of electromagnetic field distortion; calculating induced current distribution in the transformer according to an electromagnetic induction law and a time-space corresponding relation; determining a mechanical stress concentration area according to the induced current distribution in combination with the physical structure characteristics of the transformer; and verifying the stress state of the mechanical stress concentration area by using the acoustic emission signal, and when the electric field intensity, the magnetic field intensity and the change trend of the acoustic emission signal form a convergence triangle, judging that a fault omen is true. The method and the device are used for improving the timeliness of diagnosing sudden faults.

Description

Technical Field

[0001] This application belongs to the field of transformer health diagnosis, and specifically relates to a transformer full life cycle health diagnosis method, system and product. Background Art

[0002] With the rapid development of power systems, transformers, as key equipment in power systems, have a direct impact on the safe and stable operation of the entire power grid due to their operating status and health. Currently, the health status diagnosis of most transformers relies mainly on regular maintenance and temporary testing. This approach not only makes it difficult to monitor the health of transformers in real time, but also makes it difficult to accurately predict potential faults, which can easily cause equipment damage and power grid accidents. In related technologies, by placing temperature, vibration, partial discharge, and other sensors at key locations on the transformer, transformer operating data can be collected in real time, and data analysis algorithms can be used to diagnose the transformer's operating status. This method can promptly detect abnormal conditions during transformer operation and provide a basis for decision-making for operation and maintenance personnel.

[0003] However, when faced with short-term, sudden transformer faults, early warning is difficult due to time delays in sensor data collection and algorithm analysis. This is particularly true when it comes to identifying the precursory features of transient faults like lightning and short circuits. Limited sampling frequency and signal processing capabilities make it difficult to capture and analyze these weak precursor signals, making it difficult to diagnose these sudden faults in a timely manner. Summary of the Invention

[0004] (1) Purpose of the invention

[0005] The purpose of the present invention is to provide a transformer full life cycle health diagnosis method, system and product, which can avoid the impact of single signal misjudgment by capturing weak precursor signals and improve the timeliness of diagnosing sudden faults.

[0006] (2) Technical solution

[0007] To address the above-mentioned issues, a first aspect of the present invention provides a transformer lifecycle health diagnosis method. The method forms a three-dimensional monitoring network by disposing corresponding sensors at different monitoring locations of the transformer. The method comprises:

[0008] The electric field strength, magnetic field strength and acoustic emission signals inside the transformer are collected through a three-dimensional monitoring network, wherein the sensors include electric field sensors, magnetic field sensors and acoustic emission sensors;

[0009] The electric field intensity, the magnetic field intensity and the acoustic emission signal are sequentially reduced according to their corresponding sampling frequencies to form a stair-trigger sampling sequence;

[0010] During the trigger sampling sequence, when it is detected that the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength, a sampling operation of the magnetic field signal is triggered to obtain a spatiotemporal correspondence of electromagnetic field distortion, wherein the preset rate of change of the electric field strength is determined based on a sampling frequency after the power plant strength is reduced;

[0011] Calculating the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion;

[0012] determining a mechanical stress concentration area of ​​the transformer according to the induced current distribution and the physical structural characteristics of the transformer;

[0013] Verifying the stress state of the mechanical stress concentration area by using the acoustic emission signal, and determining the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state;

[0014] When the change trend forms a convergent triangle, it is determined that the fault sign is established.

[0015] Furthermore, the method further comprises:

[0016] Extracting characteristic parameters of the convergent triangle, and calculating the area change rate of the convergent triangle based on the characteristic parameters;

[0017] According to the area change rate, the fault warning is divided into three levels: warning, alarm and danger;

[0018] When the danger level is reached, the fault type is output based on the position information of the mechanical stress concentration area.

[0019] Furthermore, the method further includes: obtaining an estimated development time of the fault based on the changing trend of the area change rate:

[0020] Calculating the growth rate of the area change rate at adjacent time points;

[0021] When the growth rate is greater than the first rate threshold, recording the time interval from the current moment to the first time when the area change rate exceeds the first preset threshold as a first time parameter;

[0022] When the growth rate is greater than the second rate threshold, recording the time interval from the current moment to the first time when the area change rate exceeds the second preset threshold as a second time parameter;

[0023] Calculating an estimated fault development time based on a ratio of the first time parameter to the second time parameter and a current area change rate;

[0024] The first rate threshold and the second rate threshold correspond to the judgment criteria corresponding to the growth rate of the area change rate at adjacent time points, respectively. The first preset threshold and the second preset value correspond to the judgment criteria for the absolute value of the area change rate, respectively.

[0025] Furthermore, the three-dimensional monitoring network is composed of electric field sensors, magnetic field sensors and acoustic emission sensors respectively arranged on the top, side walls and bottom of the oil tank of the transformer.

[0026] Furthermore, the obtaining of the time-space correspondence of the electromagnetic field distortion includes:

[0027] Acquiring electric field strength data corresponding to when the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength;

[0028] Obtain magnetic field strength data during acquisition operation;

[0029] The electric field intensity data and the magnetic field intensity data are paired according to sampling time to establish a time-space correspondence relationship of the electromagnetic field distortion.

[0030] Furthermore, the calculation of the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion includes:

[0031] Calculating the induced electromotive force generated by each part of the transformer according to the law of electromagnetic induction and the time-space correspondence;

[0032] The induced electromotive force is converted into an induced current density by utilizing the resistivity distribution of the transformer winding;

[0033] In combination with the induced current density, the induced current distribution inside the transformer is calculated using a finite element method.

[0034] Furthermore, the classification of the fault warning into three levels of warning, alarm, and danger according to the area change rate includes:

[0035] Establishing a warning level determination standard, the determination standard including a first warning value and a second warning value, the first warning value being less than the second warning value;

[0036] When the area change rate is less than the first warning value, it is determined to be a warning level;

[0037] When the area change rate is greater than or equal to the first warning value and less than the second warning value, it is determined to be an alarm level;

[0038] When the area change rate is greater than or equal to the second warning value, it is determined to be a dangerous level.

[0039] Furthermore, outputting the fault type based on the position information of the mechanical stress concentration area includes:

[0040] Establishing a three-dimensional stress state model based on the position information of the mechanical stress concentration area;

[0041] Calculating the principal stress directions and shear stress directions in the stress state model;

[0042] determining a stress failure mode according to spatial distribution characteristics of the principal stress direction and the shear stress direction;

[0043] Matching the stress failure mode with a preset fault feature library and outputting the fault type;

[0044] A second aspect of the present invention provides a transformer lifecycle health diagnosis system. The system forms a three-dimensional monitoring network by disposing corresponding sensors at different monitoring locations of the transformer. The system includes:

[0045] A signal acquisition module is used to collect electric field strength, magnetic field strength and acoustic emission signals inside the transformer through a three-dimensional monitoring network. The sensors include electric field sensors, magnetic field sensors and acoustic emission sensors.

[0046] A sampling timing module is used to sequentially reduce the electric field intensity, the magnetic field intensity and the acoustic emission signal according to their corresponding sampling frequencies to form a stair-trigger sampling timing;

[0047] a time-space correspondence module, configured to trigger a sampling operation of a magnetic field signal within the trigger sampling sequence, when it is detected that the rate of change of the electric field intensity is greater than a preset rate of change of the electric field intensity, so as to obtain a time-space correspondence of the electromagnetic field distortion, wherein the preset rate of change of the electric field intensity is determined based on a sampling frequency after the power plant intensity is reduced;

[0048] An induced current distribution module, configured to calculate the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion;

[0049] A mechanical stress concentration area module, configured to determine the mechanical stress concentration area of ​​the transformer according to the induced current distribution and the physical structural characteristics of the transformer;

[0050] a change trend module, configured to verify the stress state of the mechanical stress concentration area using the acoustic emission signal, and determine the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state;

[0051] The fault judgment module is used to judge that a fault omen is established when the change trend forms a convergent triangle.

[0052] A third aspect of the present invention provides a computer-readable storage medium comprising instructions, wherein when the instructions are executed on a system, the system is caused to execute any one of the methods described above.

[0053] A fourth aspect of the present invention provides a computer program product, which, when executed on a system, enables the system to execute any one of the methods described above.

[0054] (3) Beneficial effects

[0055] The above-mentioned technical solution of the present invention has the following beneficial technical effects: The present invention provides a transformer lifecycle health diagnosis method, system, and product. By setting corresponding sensors at different monitoring locations of the transformer to form a three-dimensional monitoring network, the present invention achieves all-round monitoring of the transformer's internal electric field strength, magnetic field strength, and acoustic emission signals. Using a staggered trigger sampling sequence, the sampling frequencies of the electric field, magnetic field, and acoustic emission signals are sequentially reduced, ensuring monitoring effectiveness while reducing data storage pressure. When an abnormal electric field strength is detected, high-speed sampling of the magnetic field signal is triggered, which can accurately capture the spatiotemporal correspondence of electromagnetic field distortion, and then calculate the induced current distribution. The mechanical stress concentration area is determined by combining the physical structural characteristics of the transformer. The mechanical stress state is verified by the acoustic emission signal, and the convergent triangle formed by the change trends of the three signals is used to determine the fault signs. This completes the monitoring chain from electric field anomalies, magnetic field responses, to mechanical stress changes, avoiding the impact of single signal misjudgment and improving the accuracy and reliability of fault warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of a method for diagnosing the health of a transformer throughout its life cycle according to the present invention;

[0057] Figure 2 This is a schematic diagram of a transformer life cycle health diagnosis system according to the present invention;

[0058] Figure 3 It is a schematic diagram of the physical device structure of the transformer life cycle health diagnosis system of the present invention. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are merely exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present invention.

[0060] like Figure 1As shown, the present invention provides a transformer life cycle health diagnosis method, which forms a three-dimensional monitoring network by setting corresponding sensors at different monitoring locations of the transformer. The method includes:

[0061] S1, collecting the electric field strength, magnetic field strength, and acoustic emission signals inside the transformer through a three-dimensional monitoring network, wherein the sensors include electric field sensors, magnetic field sensors, and acoustic emission sensors. The three-dimensional monitoring network is composed of, for example, electric field sensors, magnetic field sensors, and acoustic emission sensors respectively arranged on the top, side walls, and bottom of the oil tank of the transformer, but is not limited to this composition. For example, the system can also arrange sensors at other locations inside the transformer to more comprehensively collect physical quantity information inside the transformer. In addition, in addition to electric field strength, magnetic field strength, and acoustic emission signals, the system can also collect other types of physical quantities, such as temperature, pressure, etc., to more comprehensively reflect the operating status of the transformer.

[0062] In specific implementations, high-precision, high-sampling-rate sensors can be used to collect electric field strength, magnetic field strength, and acoustic emission signals. For example, to collect electric field strength, the system can use an electric field probe with a measurement range that covers the maximum electric field strength that may occur inside the transformer, and a sampling frequency that can reach the MHz level. To collect magnetic field strength, highly sensitive magnetic field sensors such as Hall sensors and magnetoresistive sensors can be used, with a measurement range and sampling frequency that also meet the characteristics of the transformer's internal magnetic field. To collect acoustic emission signals, the system can use wide-band, highly sensitive piezoelectric sensors to capture weak acoustic emission signals that may occur inside the transformer.

[0063] In step S2, based on the physical coupling mechanism between electromagnetic fields and mechanical vibrations, the electric field strength, magnetic field strength, and acoustic emission signal are sequentially reduced according to their corresponding sampling frequencies, forming a stair-trigger sampling sequence. This arrangement aims to fully utilize the physical coupling relationship between electromagnetic fields and mechanical vibrations, reduce the complexity of data acquisition and processing, and improve the real-time performance and reliability of the system.

[0064] During implementation, the sampling frequencies for electric field strength, magnetic field strength, and acoustic emission signals can be appropriately set based on the transformer's structural characteristics and fault mechanism. Generally speaking, electric field strength changes the fastest, followed by magnetic field strength, and finally acoustic emission signals. Therefore, the sampling frequency for electric field strength can be set to the highest, followed by magnetic field strength, and finally acoustic emission signals. Furthermore, the system can also be configured with trigger conditions. When the rate of change of electric field strength exceeds a certain threshold, magnetic field strength sampling is triggered; when the rate of change of magnetic field strength exceeds a certain threshold, acoustic emission signal sampling is triggered. This avoids unnecessary data collection and processing, improving system efficiency.

[0065] S3, within the trigger sampling sequence, when it is detected that the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength, triggering a sampling operation of the magnetic field signal to obtain a spatiotemporal correspondence of electromagnetic field distortion, wherein the preset rate of change of the electric field strength is determined based on the sampling frequency after the power plant strength is reduced. In this step, obtaining the spatiotemporal correspondence of electromagnetic field distortion includes:

[0066] S31, obtaining electric field strength data corresponding to when the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength;

[0067] S32, acquiring magnetic field intensity data collected by high-speed sampling operation;

[0068] S33: Pair the electric field strength data and the magnetic field strength data at sampling times to establish a spatiotemporal correspondence of the electromagnetic field distortion. During the sampling sequence, when the rate of change of the electric field strength is detected to be greater than a preset rate, the system triggers high-speed sampling of the magnetic field signal to obtain a spatiotemporal correspondence of the electromagnetic field distortion. This step aims to capture the dynamic process of electromagnetic field distortion, providing important information for subsequent fault diagnosis. In specific implementation, the system first obtains the electric field strength data corresponding to the time when the rate of change of the electric field strength is greater than a preset rate, which serves as the basis for triggering high-speed sampling of the magnetic field signal. The system then performs high-speed sampling to rapidly and continuously collect magnetic field strength data over a period of time. Finally, the system pairs the electric field strength data and magnetic field strength data at sampling times to establish a spatiotemporal correspondence of the electromagnetic field distortion. This correspondence can reflect the occurrence, development, and propagation of electromagnetic field distortion, providing important information for analyzing the location, type, and severity of the fault.

[0069] In practical applications, electromagnetic field distortion may occur on very short timescales, or multiple distortion events may occur consecutively within a short period of time, complicating data acquisition and processing. To address this issue, the system can employ multi-channel parallel sampling during high-speed sampling operations, simultaneously collecting magnetic field intensity data from multiple locations to improve both temporal and spatial resolution. Furthermore, after data acquisition, the system can perform preprocessing on the raw data, such as denoising and feature extraction, to reduce data volume and improve data quality. Furthermore, the system can utilize data compression and data fusion techniques to minimize data storage and transmission overhead while ensuring data integrity and reliability, thereby improving the system's real-time performance and scalability.

[0070] S4, calculating the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion, including:

[0071] S41, calculating the induced electromotive force generated by each part of the transformer according to the law of electromagnetic induction and the time-space correspondence relationship;

[0072] S42, converting the induced electromotive force into an induced current density using the resistivity distribution of the transformer winding;

[0073] S43: Based on the induced current density, the internal space of the transformer is discretized using the finite element method to obtain the induced current distribution within the transformer. This step can quantitatively analyze the impact of electromagnetic field distortion on the current distribution within the transformer, providing a basis for subsequent mechanical stress analysis.

[0074] In its implementation, the induced electromotive force (EMF) generated at each part of the transformer is first calculated based on the law of electromagnetic induction and the time-space correspondence. The law of electromagnetic induction describes the quantitative relationship between electromagnetic field changes and induced electromotive force, while the time-space correspondence provides temporal and spatial information about electromagnetic field changes. Combining these two, the system can calculate the distribution of induced electromotive force generated at each part of the transformer during electromagnetic field distortion. The induced electromotive force is then converted into induced current density using the resistivity distribution of the transformer windings. Finally, based on the induced current density, the finite element method is used to discretize the internal space of the transformer to obtain the induced current distribution within the transformer. The finite element method can transform complex geometric structures and boundary conditions into discrete mathematical models, facilitating numerical calculation and analysis.

[0075] S5, determining the mechanical stress concentration area of ​​the transformer based on the induced current distribution and the physical structural characteristics of the transformer. This step can identify weak points inside the transformer that are prone to mechanical stress concentration under electromagnetic field distortion, providing an important basis for determining fault signs.

[0076] During implementation, the system first needs to obtain the physical structural characteristics of the transformer, such as its geometry and material properties. This information can be obtained from transformer design parameters and test data, or from operating transformers using technologies such as digital twins. The system then overlays the induced current distribution with the physical structural characteristics for analysis, taking into account factors such as uneven current distribution and structural stress concentrations to identify areas within the transformer where mechanical stress concentrations are likely to occur. These areas are typically areas with weak insulation, structural discontinuities, and high current density, such as winding heads and core joints.

[0077] In practical applications, due to the aging and wear that transformers may experience over time, their actual physical structural characteristics may deviate from the designed parameters, making it difficult to identify areas of mechanical stress concentration. To address this issue, the system can use condition monitoring and other methods to obtain real-time physical structural characteristic parameters of the transformer, such as winding deformation and core noise, to reflect the transformer's actual condition. Furthermore, the system can incorporate machine learning and other methods to analyze historical data and failure cases to establish an identification model and knowledge base for areas of mechanical stress concentration, thereby improving the accuracy and reliability of the determination. Furthermore, the system can also provide indicators such as confidence levels for the determination results to illustrate the credibility of the determination and provide a reference for subsequent decision-making.

[0078] S6, using the acoustic emission signal to verify the stress state of the mechanical stress concentration area, and determining the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state.

[0079] S7, when the change trend forms a convergent triangle, it is determined that the fault omen is established.

[0080] In steps S6 and S7, the system uses acoustic emission signals to verify the stress state of the mechanical stress concentration area and, based on the changing trends of the electric field intensity, magnetic field intensity, and acoustic emission signals, determines whether a fault premonition exists. This step is crucial for transformer fault premonition diagnosis. Multi-parameter fusion analysis can improve diagnostic accuracy and reliability. In specific implementation, the stress state of the mechanical stress concentration area determined in step S5 is first verified using acoustic emission signals. Acoustic emission signals can reflect microscopic fractures and structural changes within the material and are an effective means of assessing mechanical stress states. The system correlates characteristic parameters of the acoustic emission signals, such as energy, frequency, and duration, with the location of the mechanical stress concentration area to determine whether the area exhibits abnormal stress states, such as excessive stress or stress concentration. The system then comprehensively analyzes the changing trends of the electric field intensity, magnetic field intensity, and acoustic emission signals. When the changing trends of these three parameters exhibit a certain regularity, such as a simultaneous increase or the formation of a converging triangle, a fault premonition is determined. This determination, based on extensive historical data and expert experience, effectively reduces the risk of misjudgments and missed detections.

[0081] In actual applications, there may be multiple areas of mechanical stress concentration with abnormalities occurring simultaneously, or the changing trends of electric field strength, magnetic field strength, and acoustic emission signals may not be obvious, making it difficult to determine fault precursors. To address this issue, the system can introduce more monitoring parameters, such as partial discharge, temperature, and gas in oil, to enrich the dimensions of the judgment criteria and improve the reliability of the judgment. At the same time, the system can also combine tools such as simulation analysis and expert systems to conduct in-depth mining and analysis of monitoring data, discovering some implicit characteristics and patterns to improve the ability to identify fault precursors. In addition, the system can also establish a fault precursor level assessment system. Based on the severity and development trend of the fault precursor, it provides corresponding risk levels and disposal recommendations, such as strengthening monitoring, adjusting operating parameters, and arranging maintenance, to guide the safe operation and maintenance of the transformer.

[0082] The above steps elaborate on the basic principles and implementation process of the transformer life cycle health diagnosis method from the aspects of building a three-dimensional monitoring network, setting the sampling sequence, capturing the spatial and temporal correspondence of electromagnetic field distortion, calculating the induced current distribution, determining the area of ​​mechanical stress concentration, and determining the signs of faults. This method effectively improves the accuracy and reliability of fault warning through a comprehensive analysis of three dimensions: electric field anomalies, magnetic field responses, and mechanical stress changes. On this basis, this application further optimizes the method of fault warning and life prediction, which also includes:

[0083] S8, extract the characteristic parameters of the convergent triangle, and calculate the area change rate of the convergent triangle based on the characteristic parameters, wherein the characteristic parameters include the area, perimeter and internal angle of the convergent triangle. The system extracts the characteristic parameters of the convergent triangle and calculates the area change rate of the convergent triangle. The convergent triangle is a geometric figure formed in three-dimensional space by the changing trends of electric field intensity, magnetic field intensity and acoustic emission signals. Its characteristic parameters can reflect the severity and development trend of the fault precursor. The purpose of this step is to provide a basis for the subsequent fault warning level classification and life prediction by quantitatively analyzing the geometric characteristics of the convergent triangle.

[0084] In its implementation, the system first acquires the electric field strength, magnetic field strength, and acoustic emission signal data from the time the fault sign is determined to be present in step S7 and maps them into three-dimensional space to form a converging triangle. The system then calculates the characteristic parameters of the converging triangle, including its area, perimeter, and internal angle. The area reflects the severity of the fault sign, the perimeter its duration, and the internal angle its type and location. Finally, the system calculates the rate of change of the area of ​​the converging triangle—the rate at which its area changes per unit time—to reflect the development trend of the fault sign.

[0085] In practical applications, because the data of electric field intensity, magnetic field intensity, and acoustic emission signals may contain noise and interference, the directly formed convergent triangle may be distorted and irregular, making the extraction of characteristic parameters difficult. To address this problem, the system can preprocess the raw data by filtering and smoothing before forming the convergent triangle to eliminate the effects of noise and interference. At the same time, the system can also use efficient and stable geometric algorithms, such as triangulation and convex hull, to extract the characteristic parameters of the convergent triangle, thereby improving the accuracy and efficiency of the calculation. In addition, the system can also introduce advanced mathematical tools, such as fractal geometry and topological data analysis, to characterize the complex structure and evolution of the convergent triangle, thereby discovering more fault prediction information.

[0086] S9, classifying the fault warning into three levels: warning, alarm, and danger based on the area change rate. The purpose of this step is to provide different levels of warning information based on the severity of the fault precursor so that appropriate measures can be taken in a timely manner to avoid the occurrence of an accident. This step specifically includes:

[0087] S91, establishing a warning level determination standard, the determination standard including a first warning value and a second warning value, the first warning value being smaller than the second warning value;

[0088] S92, when the area change rate is less than the first warning value, determining it as a warning level;

[0089] S93, when the area change rate is greater than or equal to the first warning value and less than the second warning value, determining it as an alarm level;

[0090] S94: When the area change rate is greater than or equal to the second warning value, it is determined to be a dangerous level.

[0091] In its implementation, the system first establishes a warning level determination standard consisting of two preset thresholds. The first threshold (first warning value) is less than the second threshold (second warning value). These two thresholds can be set manually based on factors such as the transformer type, capacity, and operating conditions, combined with historical experience and expert knowledge, to balance the sensitivity and reliability of the warning. The system then compares the area change rate with the two preset thresholds. When the area change rate is less than the first preset threshold, the fault warning level is determined to be a warning level, indicating a low severity of the fault precursor and continued monitoring and observation. When the area change rate is greater than or equal to the first preset threshold and less than the second preset threshold, the fault warning level is determined to be an alarm level, indicating a high severity of the fault precursor and requiring enhanced monitoring and analysis, and, if necessary, the implementation of certain control measures. When the area change rate is greater than or equal to the second preset threshold, the fault warning level is determined to be a dangerous level, indicating a very high severity of the fault precursor and the possibility of an accident at any time, requiring immediate emergency measures such as load reduction or maintenance.

[0092] S10: When the danger level is reached, outputting a fault type based on the position information of the mechanical stress concentration area. Outputting a fault type based on the position information of the mechanical stress concentration area includes:

[0093] S101: When the fault warning level reaches a critical level, a three-dimensional stress state model is established based on the location information of the mechanical stress concentration area. When the fault warning level reaches a critical level, the system establishes a three-dimensional stress state model of the mechanical stress concentration area. This step aims to further analyze the stress distribution and failure risk of the mechanical stress concentration area in the event of a severe fault, providing a basis for fault type determination and repair decision-making.

[0094] In specific implementation, the system first needs to obtain the location and range information of the mechanical stress concentration area determined in step S5, as well as the material properties, geometric parameters, etc. of the area. Then, the system uses numerical simulation methods such as finite element analysis to discretize the mechanical stress concentration area into a series of nodes and units, and applies corresponding loads and constraints on each node, such as induced current, thermal stress, etc. Next, the system solves the stress equilibrium equation and constitutive equation in the discretized model to obtain the stress component and strain component at each node, and combines them into a complete three-dimensional stress state model. This model can intuitively display the stress distribution in the mechanical stress concentration area, such as stress magnitude, stress gradient, stress concentration coefficient, etc., providing a basis for subsequent stress analysis and failure judgment.

[0095] The three-dimensional stress state model is a numerical calculation model based on finite element analysis. The model takes the spatial coordinate data of the mechanical stress concentration area, material property parameters (elastic modulus, Poisson's ratio, density, thermal expansion coefficient), induced current distribution data, temperature distribution data and acoustic emission signal characteristics as input. The stress distribution data of normal operation, stress distribution data of different types of faults, stress failure mode data corresponding to various types of faults, material fatigue test data and stress damage data in historical maintenance records are used as training sets for model parameter calibration. The stress measurement data under actual operating conditions, stress test data from third-party laboratories and stress damage cases discovered during on-site maintenance are used as test sets for model verification. By solving the stress equilibrium equation, strain-displacement relationship and constitutive equation, the model finally outputs parameters such as the node stress tensor, principal stress and its direction, equivalent stress, stress concentration factor and strain energy density distribution, thereby achieving a comprehensive characterization of the stress state in the mechanical stress concentration area.

[0096] S102, calculating the principal stress direction and the shear stress direction in the stress state model. The purpose of this step is to further analyze the stress characteristics of the mechanical stress concentration area, determine the most dangerous stress direction, and provide a basis for determining the fault type.

[0097] In specific implementation, the system first needs to extract the stress tensor at each node based on the three-dimensional stress state model established in step S101. The stress tensor is a third-order tensor that contains complete information about the stress state at the point, such as normal stress, shear stress, etc. The system then performs eigenvalue decomposition on the stress tensor at each node to obtain three principal stress values ​​and corresponding principal stress directions. The principal stress values ​​represent the three eigenvalues ​​of the stress state at the point, and the principal stress directions represent the three eigenvectors of the stress state at the point, which together constitute the principal stress space. In the principal stress space, there is a maximum principal stress value and a corresponding maximum principal stress direction, which indicates the direction in which the material is most likely to fail in tension or compression. At the same time, the system can also calculate the angle between the maximum principal stress direction and the other two principal stress directions to obtain the shear stress direction, which indicates the direction in which the material is most likely to fail in shear.

[0098] S103: Determine a stress failure mode according to spatial distribution characteristics of the principal stress direction and the shear stress direction.

[0099] S104: Match the stress failure pattern with a preset fault signature library and output the fault type. The system matches the stress failure pattern with a preset fault signature library and outputs the fault type. This step aims to determine the possible fault type based on the stress characteristics of the mechanical stress concentration area, providing a basis for subsequent maintenance decisions.

[0100] In the specific implementation, the system first needs to establish a preset fault feature library, which contains the stress failure mode characteristics of common transformer fault types, such as insulation breakdown, inter-turn short circuit, deformation damage, etc. These stress failure mode characteristics can be obtained from historical fault cases, simulation experiments, theoretical analysis and other channels, and stored in the feature library in a certain data structure and format. Then, the system extracts the stress characteristic parameters such as the principal stress direction and shear stress direction calculated in step S102, and matches them with the stress failure mode characteristics in the preset fault feature library. The matching method can adopt pattern recognition, machine learning and other technologies, such as support vector machines, decision trees, neural networks, etc., to establish a mapping relationship between stress characteristics and fault types through training and testing. Finally, based on the matching results, the system outputs one or several most likely fault types and gives the corresponding confidence level.

[0101] S11: Based on the changing trend of the area change rate, an estimated development time of the fault is obtained. This step specifically includes:

[0102] S111, calculating the growth rate of the area change rate at adjacent time points. The purpose of this step is to further analyze the development trend of fault signs and provide a basis for life prediction.

[0103] In specific implementation, the system first needs to obtain a series of area change rate data calculated in step S8 and arrange them into a time series in chronological order. Then, the system differentiates the area change rate at two adjacent time points to obtain the increment of the area change rate, and then divides it by the time interval between the two adjacent time points to obtain the growth rate of the area change rate. The growth rate can reflect the speed and acceleration of the area change rate and is an important indicator for evaluating the development trend of fault precursors. In order to improve the calculation accuracy and stability of the growth rate, the system can use some numerical difference methods, such as forward difference, central difference, least square difference, etc., to fit and smooth the area change rate data to reduce the influence of noise and interference.

[0104] In practical applications, since area change rate data may contain jitter and anomalies, directly calculating the growth rate of adjacent time points may result in large fluctuations, which brings difficulties to trend analysis and life prediction. To solve this problem, the system can use some data preprocessing techniques, such as moving average, exponential smoothing, wavelet denoising, etc., to filter out high-frequency fluctuations and anomalies in the area change rate data and extract the main trend characteristics. At the same time, the system can also introduce some time series analysis models, such as autoregressive models and ARIMA models, to fit and predict the changing trend of area change rate and improve the long-term stability of life prediction. In addition, the system can also combine machine learning algorithms, such as support vector regression and long short-term memory networks, to establish a nonlinear mapping relationship between area change rate and fault development time, thereby improving the accuracy and adaptability of life prediction.

[0105] S112: When the growth rate exceeds the first rate threshold, the time interval between the current moment and the first time the area change rate exceeded the first preset threshold is recorded as the first time parameter. The first rate threshold and the second rate threshold refer to the criteria for determining the growth rate of the area change rate at adjacent time points and are manually set. The first and second preset values ​​refer to the criteria for determining the absolute value of the area change rate and are also manually set. The purpose of this step is to obtain early characteristic information about fault development and provide basic data for subsequent lifespan prediction.

[0106] In its specific implementation, the system first needs to monitor the growth rate of the area change rate of the converging triangle in real time. When the growth rate exceeds a pre-set first rate threshold, the recording operation of the first time parameter is triggered. The first time parameter is defined as the time interval from the current moment to the first time the area change rate exceeds the first preset threshold, reflecting the time scale from the initial development of the fault to reaching a certain level of severity. The system can obtain the value of the first time parameter by recording the moment when the area change rate exceeds the first preset threshold and calculating the difference with the current moment. This value can be stored in the system's database and used together with other characteristic parameters as historical data on the development of the fault for subsequent analysis and prediction.

[0107] At step S113, when the growth rate exceeds the second rate threshold, the time interval between the current moment and the first time the area change rate exceeds the second preset threshold is recorded as a second time parameter. This step aims to obtain late-stage characteristic information about the fault development, providing more comprehensive data support for life prediction.

[0108] In specific implementation, the system needs to set a second rate threshold and a second preset threshold, which are used to trigger the recording of the second time parameter and mark the moment when the area change rate reaches the dangerous level. When the growth rate of the area change rate exceeds the second rate threshold, and the absolute value of the area change rate exceeds the second preset threshold, the system records the time interval from the current moment to the first time the area change rate exceeds the second preset threshold as the second time parameter. Similar to the first time parameter, the second time parameter reflects the time scale for the fault to develop from severity to dangerous level, and is an important indicator for assessing the threat level of the fault. The system can calculate the value of the second time parameter through time difference and store it in the database for subsequent analysis and prediction.

[0109] S114: Based on the ratio of the first time parameter to the second time parameter and the current area change rate, the estimated fault development time is calculated. This step is the ultimate goal of transformer fault early warning and life prediction. By comprehensively analyzing the characteristic parameters obtained in the early stages, the time course of the fault development is determined, providing a basis for subsequent decision-making and actions.

[0110] In specific implementations, the system first calculates the ratio of the first time parameter to the second time parameter. This ratio reflects the relationship between the time scale from the initial development of the fault to the severity level and the time scale from the severity level to the dangerous level. Generally speaking, the larger the ratio, the faster the fault develops and the shorter the time until the danger arrives. The system then combines the current area change rate with a pre-established life prediction model to calculate the expected development time of the fault. This model can be an analytical model based on physical mechanisms or a data-driven machine learning model. By inputting characteristic quantities such as the time parameter ratio and the area change rate, it outputs the remaining time for the fault to develop. This time can serve as an important reference for the remaining life of the transformer, providing a quantitative basis for operation and maintenance decisions.

[0111] In practical applications, accurately predicting the development time of transformer failures is challenging due to the complex mechanisms and diverse influencing factors. To improve the reliability and practicality of predictions, the system can take the following measures: First, the system can establish targeted life prediction models for different types and models of transformers, fully considering the influence of factors such as structural characteristics, material properties, and operating conditions, thereby improving the model's applicability and accuracy. Second, the system can incorporate a confidence assessment module to quantify the credibility of prediction results, providing risk warnings and reference advice to decision makers. Third, the system can visualize prediction results, using charts and animations to intuitively present fault development trends and key nodes, facilitating understanding and judgment by maintenance personnel. Finally, the system provides interfaces for manual intervention and correction, allowing expert users to adjust and optimize prediction results based on their experience and on-site conditions. This enables human-machine collaboration and complementary advantages, continuously improving the quality and effectiveness of predictions.

[0112] like Figure 2 As shown, the second aspect of the present invention provides a transformer life cycle health diagnosis system, wherein corresponding sensors are arranged at different monitoring locations of the transformer to form a three-dimensional monitoring network, and the system includes:

[0113] The signal acquisition module 21 is used to collect the electric field strength, magnetic field strength and acoustic emission signals inside the transformer through a three-dimensional monitoring network. The sensors include electric field sensors, magnetic field sensors and acoustic emission sensors.

[0114] The sampling timing module 22 is used to sequentially reduce the electric field intensity, the magnetic field intensity and the acoustic emission signal according to their corresponding sampling frequencies to form a stair-trigger sampling timing;

[0115] a time-space correspondence module 23 configured to trigger a sampling operation of a magnetic field signal within the trigger sampling sequence when it is detected that the rate of change of the electric field intensity is greater than a preset rate of change of the electric field intensity, so as to obtain a time-space correspondence of the electromagnetic field distortion, wherein the preset rate of change of the electric field intensity is determined based on a sampling frequency after the power plant intensity is reduced;

[0116] An induced current distribution module 24 is configured to calculate the induced current distribution inside the transformer according to the electromagnetic induction law and the time-space correspondence of the electromagnetic field distortion;

[0117] A mechanical stress concentration area module 25 is configured to determine the mechanical stress concentration area of ​​the transformer according to the induced current distribution and the physical structural characteristics of the transformer;

[0118] a change trend module 26, configured to verify the stress state of the mechanical stress concentration area using the acoustic emission signal, and determine the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state;

[0119] The fault determination module 27 is configured to determine that a fault omen is established when the change trend forms a convergent triangle.

[0120] A third aspect of the present invention provides a computer-readable storage medium comprising instructions, which, when executed on a system, causes the system to execute any one of the methods described above.

[0121] A fourth aspect of the present invention provides a computer program product, which, when executed on a system, enables the system to execute any one of the methods described above.

[0122] The following describes the system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of the physical device structure of a transformer full life cycle health diagnosis system provided in an embodiment of the present application.

[0123] It should be noted that Figure 3 The structure of the system shown is only an example and should not limit the functions and scope of use of the embodiments of the present invention.

[0124] like Figure 3 As shown, the system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 into the random access memory (RAM) 303, such as executing the method in the above embodiment. Various programs and data required for system operation are also stored in the RAM 303. The CPU 301, ROM 302 and RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0125] The following components are connected to the I / O interface 305: an input section 306 including a camera, an infrared sensor, etc.; an output section 307 including a liquid crystal display (LCD) and a speaker; a storage section 308 including a hard disk; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed in the drive 310 as needed so that computer programs read therefrom can be installed into the storage section 308 as needed.

[0126] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product that includes a computer program carried on a computer-readable medium, the computer program including a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, the various functions defined in the present invention are performed.

[0127] This application provides a transformer life cycle health diagnosis method, system, and product, which have the following advantages:

[0128] 1. This application realizes all-round monitoring of the electric field strength, magnetic field strength and acoustic emission signals inside the transformer by setting up a three-dimensional monitoring network on the top, side walls and bottom of the transformer oil tank. By using the sampling timing setting of the cascade trigger, the sampling frequency of the electric field, magnetic field and acoustic emission signals is reduced in sequence, which reduces the pressure of data storage while ensuring the monitoring effect. When the electric field strength anomaly is detected, the high-speed sampling of the magnetic field signal is triggered, which can accurately capture the time and space correspondence of the electromagnetic field distortion, and then calculate the induced current distribution, and judge the mechanical stress concentration area in combination with the physical structure characteristics of the transformer. The mechanical stress state is verified by the acoustic emission signal, and the convergent triangle formed by the change trend of the three signals is used to determine the fault signs, realizing a complete monitoring chain from electric field anomaly, magnetic field response to mechanical stress change, avoiding the impact of single signal misjudgment, and improving the accuracy and reliability of fault warning.

[0129] 2. This application extracts characteristic parameters of a converging triangle, such as its area, perimeter, and interior angle, to calculate the area change rate. Based on this, the fault warning level is classified into three levels: warning, alarm, and danger. When the fault warning level reaches danger, the fault type and expected development time are output. This hierarchical warning mechanism, based on the characteristic parameters of the converging triangle, makes the fault warning assessment process more quantitative and standardized. By monitoring the area change rate, the dynamic trend of fault development can be reflected. Combined with the location information of mechanical stress concentration areas, this not only enables timely issuance of graded warnings, but also predicts the fault type and development time, facilitating the timely implementation of targeted maintenance measures. The principle behind this graded warning is that the area change rate reflects the speed of fault development: a small rate of change indicates slow fault development, corresponding to the warning level; a medium rate of change indicates accelerated fault development, corresponding to the alarm level; a large rate of change indicates rapid fault deterioration, corresponding to the danger level. The fault type prediction principle combines the location of the stress concentration area to determine the fault type: the top area may indicate a casing failure or gasket damage; the coil area may indicate winding deformation or a short circuit; the core area may indicate core grounding or local overheating; and the bottom area may indicate oil deterioration or sediment accumulation. The principle of predicting fault development time is based on trend analysis of the area change rate. This method uses the area change rate to establish a fault development curve, predicting the fault evolution rate through curve fitting. Combined with the material properties and service life of the stress concentration location, a comprehensive assessment is conducted to determine the critical time for fault development. This invention utilizes a multifunctional early warning synergy: the area change rate provides dynamic information on fault development, while the stress concentration location provides information on the spatial distribution of the fault. This combination of the two provides higher reliability, avoids misjudgment of a single parameter, improves prediction accuracy, and enables comprehensive prediction of fault type and timing.

[0130] 3. This application calculates the growth rate of the area change rate at adjacent time points, records the time intervals when the area change rate exceeds different preset thresholds, and calculates the expected fault development time based on the ratio of these time parameters. By analyzing the growth law of the area change rate, the acceleration characteristics of the fault development are captured, and the ratio relationship of the time parameters is used to reflect the nonlinear characteristics of the fault development, thereby obtaining a fault development time prediction result that is more in line with the actual situation and improving the accuracy of the fault development time prediction.

Claims

1. A method for diagnosing the health of a transformer throughout its life cycle, characterized in that: The method forms a three-dimensional monitoring network by arranging corresponding sensors at different monitoring locations of the transformer. The method includes: The electric field strength, magnetic field strength and acoustic emission signals inside the transformer are collected through a three-dimensional monitoring network, wherein the sensors include electric field sensors, magnetic field sensors and acoustic emission sensors; The electric field intensity, the magnetic field intensity and the acoustic emission signal are sequentially reduced according to their corresponding sampling frequencies to form a stair-trigger sampling sequence; During the trigger sampling sequence, when it is detected that the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength, a sampling operation of the magnetic field signal is triggered to obtain a spatiotemporal correspondence of electromagnetic field distortion, wherein the preset rate of change of the electric field strength is determined based on a sampling frequency after the power plant strength is reduced; Calculating the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion; determining a mechanical stress concentration area of ​​the transformer according to the induced current distribution and the physical structural characteristics of the transformer; Verifying the stress state of the mechanical stress concentration area by using the acoustic emission signal, and determining the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state; When the change trend forms a convergent triangle, it is determined that the fault sign is established.

2. The transformer life cycle health diagnosis method according to claim 1, characterized in that: The method further comprises: Extracting characteristic parameters of the convergent triangle, and calculating the area change rate of the convergent triangle based on the characteristic parameters; According to the area change rate, the fault warning is divided into three levels: warning, alarm and danger; When the danger level is reached, the fault type is output based on the position information of the mechanical stress concentration area.

3. The transformer life cycle health diagnosis method according to claim 2, characterized in that: The method further includes: obtaining an estimated development time of the fault based on a change trend of the area change rate: Calculating the growth rate of the area change rate at adjacent time points; When the growth rate is greater than the first rate threshold, recording the time interval from the current moment to the first time when the area change rate exceeds the first preset threshold as a first time parameter; When the growth rate is greater than the second rate threshold, recording the time interval from the current moment to the first time when the area change rate exceeds the second preset threshold as a second time parameter; Calculating an estimated fault development time based on a ratio of the first time parameter to the second time parameter and a current area change rate; The first rate threshold and the second rate threshold correspond to the judgment criteria corresponding to the growth rate of the area change rate at adjacent time points, respectively. The first preset threshold and the second preset value correspond to the judgment criteria for the absolute value of the area change rate, respectively.

4. The transformer life cycle health diagnosis method according to claim 1, characterized in that: The three-dimensional monitoring network is composed of electric field sensors, magnetic field sensors and acoustic emission sensors respectively arranged on the top, side walls and bottom of the oil tank of the transformer.

5. The transformer life cycle health diagnosis method according to claim 1, characterized in that: The obtaining of the time-space correspondence of the electromagnetic field distortion includes: Acquiring electric field strength data corresponding to when the rate of change of the electric field strength is greater than a preset rate of change of the electric field strength; Obtain magnetic field strength data during acquisition operation; The electric field intensity data and the magnetic field intensity data are paired according to sampling time to establish a time-space correspondence relationship of the electromagnetic field distortion.

6. The transformer life cycle health diagnosis method according to claim 1, characterized in that: The calculating of the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion includes: Calculating the induced electromotive force generated by each part of the transformer according to the law of electromagnetic induction and the time-space correspondence; The induced electromotive force is converted into an induced current density by utilizing the resistivity distribution of the transformer winding; In combination with the induced current density, the induced current distribution inside the transformer is calculated using a finite element method.

7. The transformer life cycle health diagnosis method according to claim 2, characterized in that: The classification of the fault warning into three levels of warning, alarm and danger according to the area change rate includes: Establishing a warning level determination standard, the determination standard including a first warning value and a second warning value, the first warning value being less than the second warning value; When the area change rate is less than the first warning value, it is determined to be a warning level; When the area change rate is greater than or equal to the first warning value and less than the second warning value, it is determined to be an alarm level; When the area change rate is greater than or equal to the second warning value, it is determined to be a dangerous level.

8. A transformer life cycle health diagnosis system, characterized in that: The system forms a three-dimensional monitoring network by setting corresponding sensors at different monitoring locations of the transformer. The system includes: A signal acquisition module is used to collect electric field strength, magnetic field strength and acoustic emission signals inside the transformer through a three-dimensional monitoring network. The sensors include electric field sensors, magnetic field sensors and acoustic emission sensors. A sampling timing module is used to sequentially reduce the electric field intensity, the magnetic field intensity and the acoustic emission signal according to their corresponding sampling frequencies to form a stair-trigger sampling timing; a time-space correspondence module, configured to trigger a sampling operation of a magnetic field signal within the trigger sampling sequence, when it is detected that the rate of change of the electric field intensity is greater than a preset rate of change of the electric field intensity, so as to obtain a time-space correspondence of the electromagnetic field distortion, wherein the preset rate of change of the electric field intensity is determined based on a sampling frequency after the power plant intensity is reduced; An induced current distribution module, configured to calculate the induced current distribution inside the transformer according to the law of electromagnetic induction and the time-space correspondence of the electromagnetic field distortion; A mechanical stress concentration area module, configured to determine the mechanical stress concentration area of ​​the transformer according to the induced current distribution and the physical structural characteristics of the transformer; a change trend module, configured to verify the stress state of the mechanical stress concentration area using the acoustic emission signal, and determine the change trend of the electric field intensity, the magnetic field intensity, and the acoustic emission signal according to the stress state; The fault judgment module is used to judge that a fault omen is established when the change trend forms a convergent triangle.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a system, the system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product is run on a system, the system is caused to perform the method according to any one of claims 1 to 7.

Images