Inter-turn short-circuit monitoring method for dry-type air-core reactors based on magnetic field difference method

Abstract

This invention relates to the field of fault testing technology, specifically to a method for monitoring inter-turn short circuits in dry-type air-core reactors based on the magnetic field difference method. This method constructs a reference electromagnetic coupling matrix by introducing regularization constraints in the initial baseline stage. Based on the recursive analysis of the three-phase current energy thermal accumulation index, it combines the magnetic field residual to fit a thermally induced magnetic field drift model. By combining the joint constraints of the reference electromagnetic coupling matrix and the thermal drift model, the normal thermally induced magnetic field drift component is dynamically removed during real-time monitoring. Under the constraints of the reference electromagnetic coupling matrix, the current vector is optimized through weighted least squares state estimation. The structural consistency deviation index is obtained by comprehensively considering the dual-dimensional deviations of the magnetic field and current for fault monitoring. This invention balances linear electromagnetic transfer laws with nonlinear thermal drift compensation for monitoring, adapting to the complex magnetic field environment and variable load operating conditions of dry-type air-core reactors, providing continuous and reliable monitoring assurance for the long-term stable operation of the equipment.

Description

Technical Field

[0001] This invention relates to the field of fault testing technology, and specifically to a method for monitoring inter-turn short circuits in dry-type air-core reactors based on the magnetic field differential method. Background Technology

[0002] Dry-type air-core reactors, as core equipment for reactive power compensation and current limiting in power systems, typically employ multi-layer parallel windings to construct a coreless magnetic circuit and are arranged in a three-phase vertical stacked structure to save substation space. In this type of structure, the intermediate phase coil not only generates a self-induced magnetic field but is also surrounded by strong alternating magnetic fields generated by the adjacent phases above and below, resulting in an extremely complex magnetic field distribution.

[0003] Conventional inter-turn short-circuit monitoring techniques often rely on electrical quantity balance or simple magnetic field differential principles, typically assuming the reactor is a linear time-invariant system with a fixed structure. However, in actual operation, fluctuations in grid load cause Joule heating from the winding current, leading to a gradual increase in coil temperature over time. This results in minute geometric deformations and changes in conductor resistivity, causing the electromagnetic transfer characteristics, which characterize the relationship between current and the spatial magnetic field, to exhibit time-varying nonlinear features as heat accumulates. This thermal effect caused by temperature rise leads to a slight drift in the spatial magnetic field distribution. When grid load fluctuations are significant, it is difficult to effectively distinguish this normal thermally induced magnetic field drift from the fault magnetic field distortion caused by early, weak short circuits, resulting in misjudgments or missed detections in monitoring. Summary of the Invention

[0004] To address the problem in existing technologies where temperature rise-induced thermal effects cause slight drifts in the spatial magnetic field distribution, making it difficult to effectively distinguish between normal thermally induced magnetic field drift and fault magnetic field distortion caused by early weak short circuits when grid load fluctuations are significant, thus leading to inaccurate monitoring, this invention aims to provide a method for monitoring inter-turn short circuits in dry-type air-core reactors based on the magnetic field differential method. The specific technical solution adopted is as follows:

[0005] The first aspect of this application provides a method for monitoring inter-turn short circuits in a dry-type air-core reactor based on the magnetic field differential method, the method comprising:

[0006] During the initial baseline phase of system startup, the phase-aligned magnetic flux density vector and three-phase current phasor vector on the reactor periphery are acquired in each analysis window;

[0007] In the initial baseline stage, a reference electromagnetic coupling matrix characterizing the current-to-magnetic field transfer relationship is constructed based on the magnetic induction intensity vector and the three-phase current phasor vector. Based on the thermal accumulation index at the moment of amplitude energy recursion analysis of the three-phase current phasor vector, the magnetic field residual vector of the analysis window is predicted through the reference electromagnetic coupling matrix. The mapping relationship between the residual vector and the thermal accumulation index is fitted and analyzed to determine the thermally induced magnetic field drift model.

[0008] In the monitoring phase following the initial baseline stage, the current thermal drift component is analyzed based on the thermally induced magnetic field drift model, and the current magnetic induction intensity vector is corrected to obtain the corrected magnetic induction vector. Under the constraint of the reference electromagnetic coupling matrix, the state is estimated based on the corrected magnetic induction vector and the three-phase current phasor vector using the weighted least squares method. The optimized current vector is fitted, and the deviation between the corrected magnetic induction intensity vector and the optimized current vector in reconstructing the magnetic field, as well as the deviation between the optimized current vector and the three-phase current phasor vector, are analyzed to obtain the structural consistency deviation index.

[0009] Monitoring inter-turn short circuits based on structural consistency deviation index.

[0010] Furthermore, the method for obtaining the reference electromagnetic coupling matrix includes:

[0011] An observation matrix is ​​constructed, consisting of the magnetic induction intensity vectors of each analysis window in the initial baseline stage, and an excitation matrix is ​​constructed, consisting of the three-phase current phasor vectors of the corresponding analysis windows. Based on the Tikhonov regularization principle, the autocorrelation matrix of the excitation matrix is ​​corrected using the collinearity suppression factor, and the reference electromagnetic coupling matrix is ​​obtained by combining the observation matrix.

[0012] Furthermore, the method for obtaining the thermal accumulation index includes:

[0013] The heat dissipation attenuation coefficient is determined based on the thermal time constant of the reactor and the sampling interval; at the initial moment, the initial heat accumulation index is determined based on the norm square of the three-phase current phasor vector.

[0014] At each time after the initial time, the norm square of the three-phase current phasor vector at the current time is weighted using the heat dissipation attenuation coefficient, and the heat accumulation index at the previous time is weighted using the thermal inertia weight. The weighted sum of the two is taken as the heat accumulation index at the current time. The sum of the thermal inertia weight and the heat dissipation attenuation coefficient is 1.

[0015] Furthermore, the method for obtaining the thermally induced magnetic field drift model includes:

[0016] In the initial baseline stage, the reconstructed magnetic flux density vector at time t is obtained by multiplying the three-phase current phasor vector at time t with the reference electromagnetic coupling matrix. The predicted residual vector is obtained by the difference between the magnetic flux density vector at time t and the reconstructed magnetic flux density vector. The thermally induced magnetic field drift model is obtained by using a quadratic polynomial fitting model for each element in the magnetic flux density vector with the thermal accumulation index as the independent variable and the predicted residual vector of the reference electromagnetic coupling matrix as the dependent variable.

[0017] Furthermore, the method for obtaining the corrected magnetic induction vector includes:

[0018] Input the current thermal accumulation index into the thermally induced magnetic field drift model to obtain the current thermal drift component; use the difference between the current magnetic induction intensity vector and the thermal drift component as the corrected magnetic induction vector.

[0019] Furthermore, the method for obtaining the optimized current vector includes:

[0020] Based on the noise between the magnetic field measurement channel and the current measurement channel, the signal-to-noise ratio balance factor is obtained.

[0021] Calculate the product of the conjugate transpose of the reference electromagnetic coupling matrix and the reference electromagnetic coupling matrix itself to construct a first intermediate matrix; add the first intermediate matrix to the identity matrix weighted by the signal-to-noise ratio balance factor, and invert the sum to obtain the total weight matrix; calculate the product of the conjugate transpose of the reference electromagnetic coupling matrix and the modified magnetic induction vector, and the product of the signal-to-noise ratio balance factor and the three-phase current phasor vector, and add the products of the two to obtain the weighted observation matrix; calculate the product of the total weight matrix and the weighted observation matrix to obtain the optimized current vector.

[0022] Furthermore, the method for obtaining the structural consistency deviation index includes:

[0023] The norm square of the difference between the corrected magnetic induction vector and the reconstructed magnetic field is calculated as the magnetic induction deviation, where the reconstructed magnetic field is the product of the reference electromagnetic coupling matrix and the optimized current vector. The norm square of the difference between the optimized current vector and the three-phase current phasor vector is calculated as the product of the signal-to-noise ratio balance factor, which is used as the current deviation. The sum of the magnetic induction deviation and the current deviation is used as the structural consistency deviation index.

[0024] Furthermore, the monitoring of inter-turn short circuits based on the structural consistency deviation index includes:

[0025] Calculate the arithmetic mean of the structural consistency deviation index within the current sliding window during the monitoring phase. When the arithmetic mean is greater than the set fault threshold, output an alarm signal.

[0026] Furthermore, the method for obtaining the phase-aligned magnetic flux density vector and the three-phase current phasor vector includes:

[0027] Determine the reference phase reference in the three-phase current, calculate the real-time phase angle of the reference phase reference current, and construct a complex rotation operator using the real-time phase angle;

[0028] The magnetic flux density signal and the three-phase current signal acquired from multiple channels are multiplied by the complex rotation operator to generate phase-aligned magnetic flux density vectors and three-phase current phasor vectors.

[0029] Furthermore, the method for obtaining the signal-to-noise ratio balance factor includes:

[0030] The variances of the magnetic field measurement channel data and the current measurement channel data during the silent acquisition period of the statistical system are used to calculate the ratio of the two to obtain the signal-to-noise ratio balance factor.

[0031] Secondly, this application provides a dry-type air-core reactor inter-turn short-circuit monitoring system based on the magnetic field differential method, the system comprising:

[0032] The data monitoring module is used to acquire the phase-aligned magnetic flux density vector and three-phase current phasor vector around the reactor in each analysis window during the initial baseline phase of system startup.

[0033] The thermal state influence analysis module is used to construct a reference electromagnetic coupling matrix characterizing the current-to-magnetic field transfer relationship based on the magnetic induction intensity vector and the three-phase current phasor vector in the initial baseline stage by introducing regularization constraints. Based on the amplitude energy recursive analysis time of the three-phase current phasor vector, the thermal accumulation index is used to predict the magnetic field residual vector of the analysis window through the reference electromagnetic coupling matrix, and the mapping relationship between the residual vector and the thermal accumulation index is fitted and analyzed to determine the thermally induced magnetic field drift model.

[0034] The structural deviation analysis module is used in the monitoring phase after the initial baseline stage to analyze the current thermal drift component based on the thermally induced magnetic field drift model, correct the current magnetic induction intensity vector to obtain the corrected magnetic induction vector; under the constraint of the reference electromagnetic coupling matrix, the state is estimated based on the corrected magnetic induction vector and the three-phase current phasor vector using the weighted least squares method, the optimized current vector is fitted, and the deviation between the corrected magnetic induction intensity vector and the optimized current vector in reconstructing the magnetic field, as well as the deviation between the optimized current vector and the three-phase current phasor vector, are analyzed to obtain the structural consistency deviation index.

[0035] The monitoring module is used to monitor inter-turn short circuits based on the structural consistency deviation index.

[0036] Thirdly, this application provides a computer device including a memory and a processor. The memory is used to store computer program code, and the processor is used to call and run the computer program code from the memory to perform the method as described in the first aspect of this application or any embodiment of the first aspect.

[0037] Fourthly, this application provides a computer program product comprising computer program code, which, when executed, performs the method as described in the first aspect of this application or any embodiment thereof.

[0038] Fifthly, this application provides a computer-readable storage medium that stores computer program code, which, when executed, performs the method as described in the first aspect of this application or any embodiment thereof.

[0039] The present invention has the following beneficial effects:

[0040] This invention constructs a reference electromagnetic coupling matrix by introducing regularization constraints in the initial baseline stage, avoiding instability in the reference model identification caused by the linear correlation of three-phase currents and ensuring the physical authenticity of the cold-state reference. Based on the recursive analysis of the thermal accumulation index of three-phase current energy, and combined with the fitting of the thermally induced magnetic field drift model using magnetic field residuals, the temperature state of the winding with thermal inertia can be more accurately quantified by recursively analyzing the current amplitude energy. By combining the joint constraints of the reference electromagnetic coupling matrix and the thermal drift model, the normal thermally induced magnetic field drift component can be dynamically separated in real-time monitoring, and the drift due to temperature rise can be analyzed even under conditions of large load fluctuations or frequent switching between hot and cold states. Under the constraints of the reference electromagnetic coupling matrix, the current vector is fitted and optimized through weighted least squares state estimation. The structural consistency deviation index is obtained by comprehensively considering the deviations of the magnetic field and current in both dimensions, which can specifically capture coil topology anomalies and improve the sensitivity of early weak inter-turn short-circuit fault identification. This invention takes into account both linear electromagnetic transmission laws and nonlinear thermal drift compensation for monitoring, adapting to the complex magnetic field environment and variable load operation conditions of dry-type air-core reactors, and providing continuous and reliable monitoring assurance for the long-term stable operation of equipment. Attached Figure Description

[0041] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 A flowchart of a method for monitoring inter-turn short circuits in a dry-type air-core reactor based on the magnetic field differential method, provided as an embodiment of the present invention;

[0043] Figure 2 The diagram shows a structural diagram of a dry-type air-core reactor inter-turn short-circuit monitoring system based on the magnetic field differential method, provided in one embodiment of the present invention.

[0044] Figure 3 This is a schematic diagram of a computer device structure provided in one embodiment of the present invention. Detailed Implementation

[0045] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for monitoring inter-turn short circuits in a dry-type air-core reactor based on the magnetic field differential method proposed in this invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0046] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0047] The following description, in conjunction with the accompanying drawings, details a specific scheme for a method for monitoring inter-turn short circuits in a dry-type air-core reactor based on the magnetic field differential method, provided by this invention.

[0048] Please see Figure 1 The diagram illustrates a flowchart of a method for monitoring inter-turn short circuits in a dry-type air-core reactor based on the magnetic field differential method, according to an embodiment of the present invention. The method includes the following steps:

[0049] S1: During the initial baseline phase of system startup, acquire the phase-aligned magnetic flux density vector and three-phase current phasor vector on the reactor periphery in each analysis window.

[0050] Random fluctuations in the power grid frequency can cause phase misalignment of asynchronously sampled signals, affecting the accuracy of vector analysis. Therefore, after acquiring the magnetic flux density signal around the reactor and the three-phase current signal of the circuit using a multi-channel synchronous acquisition device, phase alignment is required to eliminate the non-physical phase drift caused by power grid frequency fluctuations and convert the continuous time-domain analog signal into a complex domain vector that is aligned in time and physically corresponding in space.

[0051] In this embodiment of the invention, several magnetic field sensors are installed at monitoring positions around the reactor to detect radial or axial magnetic induction intensity. Simultaneously, the loop currents of phases A, B, and C are acquired using current transformers installed in the secondary circuit of the substation. The acquired multi-channel magnetic induction intensity signals and three-phase current signals are input into the same multi-channel synchronous data processing unit at a preset sampling frequency. The initial baseline phase is set to a preset duration after startup, which characterizes the significant temperature rise process of the reactor from a cold state to a hot state. In this embodiment, the preset duration can be set to 30 to 60 minutes, and the preset sampling frequency can be set to 5 kHz to 20 kHz. The specific acquisition settings can be adjusted by the implementer according to the specific implementation scenario and are not limited here.

[0052] Since high-frequency electromagnetic interference and harmonic components are unavoidable in the field environment, and the main magnetic field characteristics of the reactor are primarily determined by the fundamental power frequency, digital signal processing technology is used to extract the fundamental component to remove noise interference. To facilitate data extraction and alignment, windows are divided into integer multiples of the power frequency period. In this embodiment, each analysis window is 10 periods long; the specific value can be adjusted by the implementer. Within each analysis window, a Discrete Fourier Transform (DFT) is performed on the magnetic flux density signal and current signal of each channel to extract the complex phasor of the fundamental power frequency. It should be noted that the method of obtaining the fundamental phasor using Discrete Fourier Transform is a well-known technique among those skilled in the art and will not be elaborated upon here.

[0053] In this embodiment of the invention, a reference phase benchmark is determined among the three-phase currents. To ensure the stability of the phase benchmark, the current of phase B is typically selected as the reference phase benchmark. If the amplitude of phase B is too low, below the effective monitoring threshold, the system automatically switches to phase A or phase C. Specifically, the phase with the largest amplitude can be switched as the reference phase benchmark. It is understood that if the amplitudes of all three-phase currents are below the effective monitoring threshold, which is 5% of the rated current, the data is considered invalid, such as due to a shutdown, and the analysis is stopped.

[0054] The real-time phase angle of the reference phase current is calculated, and a complex rotation operator is constructed using this real-time phase angle. This rotation operator is then used to perform coordinate rotation operations on all signals within the analysis window. The complex phasors extracted from the multi-channel acquired magnetic flux density signal and the three-phase current signal are multiplied by the complex rotation operator to generate phase-aligned magnetic flux density vectors and three-phase current phasor vectors. It should be noted that the complex rotation operator, also known as the phase rotation operator, describes rotation operations on the complex plane. Its acquisition method is a publicly known technique familiar to those skilled in the art and will not be elaborated upon here. In a specific embodiment of this invention, when the number of spatial magnetic field monitoring points is N, the magnetic flux density vector is an N×1-dimensional complex column vector, and the three-phase current phasor vector is a 3×1-dimensional complex column vector.

[0055] By rotating the coordinates based on the real-time phase of the selected reference phase current, the time-varying physical signal is transformed into a phase-aligned magnetic flux density vector and a three-phase current phasor vector that are aligned in time and relatively locked in phase, providing standardized data input for subsequent processing.

[0056] S2: In the initial baseline stage, based on the magnetic induction intensity vector and the three-phase current phasor vector, a reference electromagnetic coupling matrix characterizing the current-to-magnetic field transfer relationship is constructed by introducing regularization constraints; based on the thermal accumulation index at the moment of amplitude energy recursion analysis of the three-phase current phasor vector, the magnetic field residual vector of the analysis window is predicted by the reference electromagnetic coupling matrix, and the mapping relationship between the residual vector and the thermal accumulation index is fitted and analyzed to determine the thermally induced magnetic field drift model.

[0057] Because reactor windings deform and drift due to heat during operation, the magnetic field distribution exhibits nonlinear changes. Therefore, in the initial baseline stage, it is necessary to first establish the basic electromagnetic response characteristics of the reactor under conditions unaffected by significant thermal effects. Although theoretically, the magnetic field and current follow the principle of linear superposition, in a three-phase AC system, the three-phase currents are highly correlated due to Kirchhoff's laws. Directly using observational data of the magnetic field and the space magnetic field to solve for the linear transfer relationship would face serious numerical ill-conditioning problems, leading to model instability and an inability to reflect the true physical structure. Therefore, regularization constraints are introduced to address this multicollinearity problem, thereby stably identifying the inherent electromagnetic coupling matrix of the reactor in the cold state.

[0058] In one specific embodiment of the present invention, an observation matrix is ​​constructed, consisting of columnar magnetic flux density vectors from each analysis window during the initial baseline phase. Construct an excitation matrix composed of column-wise columns of the three-phase current phasor vectors corresponding to the analysis window. The number of rows in the observation matrix is ​​equal to the number of channels of the magnetic field sensor on the periphery of the reactor, and the number of columns is equal to the total number of analysis windows in the initial baseline stage. The number of rows in the excitation matrix is ​​3, corresponding to the three-phase currents A, B, and C, and the number of columns is consistent with the number of columns in the observation matrix, ensuring that the column dimensions of the two matrices are completely matched, and realizing the one-to-one correspondence of data in each analysis window.

[0059] Based on the Tikhonov regularization principle, and utilizing the collinearity suppression factor The autocorrelation matrix of the corrected excitation matrix is ​​used, and the reference electromagnetic coupling matrix is ​​obtained by combining it with the observation matrix. Specifically, the excitation matrix is ​​first calculated. autocorrelation matrix i.e., the activation matrix Its conjugate transpose The product of these factors reflects the degree of linear correlation among the three-phase currents. Since the three-phase currents exhibit strong collinearity due to Kirchhoff's laws, direct solution can easily lead to numerical divergence. Therefore, a collinearity suppression factor is used. With the identity matrix After multiplication, we obtain the regularization correction term. Autocorrelation matrix with the activation matrix Superposition and diagonal loading corrections mitigate the adverse effects of collinearity on parameter identification, yielding a regularized autocorrelation matrix. .

[0060] Inverting the regularized autocorrelation matrix yields the inverse matrix. Finally, the observation matrix is ​​calculated. The conjugate transpose of the excitation matrix The product of the products, and the result of the product Multiplying by the inverse matrix yields the reference electromagnetic coupling matrix.

[0061] The collinearity suppression factor is used to balance the fitting accuracy and numerical stability of the model. In this embodiment of the invention, the collinearity suppression factor can be determined based on the largest eigenvalue of the excitation matrix and the autocorrelation matrix, for example, a value of 10 of the largest eigenvalue. -6 Up to 10 -3 The specific value can be adjusted by the implementer based on the signal-to-noise ratio level of the acquisition system, and no restrictions are imposed here.

[0062] Ultimately, the number of rows in the reference electromagnetic coupling matrix is ​​consistent with the number of rows in the observation matrix, and the number of columns is consistent with the number of rows in the excitation matrix. Each element in the matrix represents the linear transfer coefficient of the corresponding phase current to the magnetic induction intensity of the corresponding magnetic field sensor channel. This solidifies the inherent current-magnetic field transfer relationship of the dry-type air-core reactor under healthy conditions, providing a reliable linear reference for subsequent calculation of thermally induced magnetic field residuals and magnetic field reconstruction during the monitoring period.

[0063] Joule heating generated by the current during operation causes the winding temperature to rise, which in turn leads to geometric deformation and resistivity changes, resulting in nonlinear thermal magnetic field drift. This drift is a normal operating condition disturbance. If it is not removed, it will cause the monitoring baseline to drift, which will be superimposed on the basic linear electromagnetic response, seriously interfering with fault diagnosis.

[0064] Therefore, by further combining historical heat dissipation data, the dynamic thermal state of the winding is quantified by recursively calculating the heat accumulation index. In this embodiment of the invention, the thermal time constant of the reactor is used as the basis. and sampling interval Determine the heat dissipation attenuation coefficient The heat dissipation decay coefficient simulates the exponential decay characteristic of heat over time, reflecting the dynamic characteristics of the natural decay of heat in the winding. It can be determined based on the exponential decay relationship. In a specific embodiment of the present invention, the ratio of the sampling interval to the thermal time constant is used. The value obtained by negative correlation mapping using a negative exponent form with the natural constant as the base. As a heat dissipation attenuation coefficient, it characterizes the proportion of heat generated by the current that remains in the winding.

[0065] At the initial moment, the equipment is assumed to be in a steady state or without historical thermal memory. The thermal state is quantified directly using the instantaneous current energy at the current moment. The initial thermal accumulation index is determined based on the norm square of the three-phase current phasor vector, which is equal to the sum of the squares of the moduli of each element in the three-phase current phasor vector at the initial moment.

[0066] At each time point after the initial time, the heat dissipation decay coefficient is used. The norm squares of the three-phase current phasor vectors at the current moment are weighted to reflect the contribution of newly generated heat, and thermal inertia weights are used. The heat accumulation index of the previous moment is weighted and retained to simulate the historical accumulation and decay inertia of heat. The weighted sum of the two is used as the heat accumulation index of the current moment. The sum of the heat inertia weight and the heat dissipation decay coefficient is 1. This recursive logic simulates the exponential decay and accumulation characteristics of heat over time and updates the real-time thermal state.

[0067] Specifically, when the heat accumulation index is too high, the initial baseline phase can be terminated early. This is because the winding has already entered a stable high-temperature operating range, and its thermally induced magnetic field drift pattern has been fully learned. Continuing to collect data will contribute little to improving model accuracy and may introduce unnecessary risks due to the equipment being in a high-temperature state for an extended period. Specifically, when a preset number of heat accumulation indices reach the effective learning threshold, the winding's thermal state is considered sufficiently activated, and baseline learning can be terminated early, initiating the monitoring phase. The preset number can be set as the length of the analysis window, and the effective learning threshold can be set as 80% of the rated heat load; these settings can be adjusted by the implementer.

[0068] Furthermore, the theoretical magnetic field strength under normal conditions is calculated using the reference electromagnetic coupling matrix, and the residual between it and the actual observed magnetic induction intensity vector is used to analyze the nonlinear deviation dominated by thermal drift. Considering the differences in thermal drift characteristics of different magnetic field sensors at corresponding spatial locations, the nonlinear mapping relationship between different thermal states and magnetic field drift is analyzed for each magnetic field sensor channel.

[0069] In this embodiment of the invention, during the initial baseline stage, the reconstructed magnetic flux density vector at time t is obtained by multiplying the three-phase current phasor vector at time t with the reference electromagnetic coupling matrix. This reconstructed magnetic flux density vector characterizes the theoretical magnetic field response under the assumption of no thermal effects and other disturbances. The predicted residual vector is obtained by the difference between the magnetic flux density vector at time t and the reconstructed magnetic flux density vector. This residual mainly includes the nonlinear magnetic field drift component caused by the winding temperature rise.

[0070] Regression analysis was performed on each magnetic field sensor channel, with the thermal accumulation index as the independent variable and the predicted residual vector of the reference electromagnetic coupling matrix as the dependent variable. A quadratic polynomial fitting model was used for each element in the magnetic induction intensity vector to obtain the thermally induced magnetic field drift model, which characterizes the nonlinear response characteristics of the magnetic field at each spatial point as the overall thermal state of the winding changes. Its output is a vector form of the predicted drift value, which can be used to compensate for the real-time magnetic field observation in the subsequent monitoring stage.

[0071] S3: In the monitoring phase after the initial baseline stage, the current thermal drift component is analyzed based on the thermally induced magnetic field drift model, and the current magnetic induction intensity vector is corrected to obtain the corrected magnetic induction vector. Under the constraint of the reference electromagnetic coupling matrix, the state is estimated based on the corrected magnetic induction vector and the three-phase current phasor vector using the weighted least squares method. The optimized current vector is fitted, and the deviation between the corrected magnetic induction intensity vector and the optimized current vector in reconstructing the magnetic field, as well as the deviation between the optimized current vector and the three-phase current phasor vector, are analyzed to obtain the structural consistency deviation index.

[0072] Because the inter-turn short-circuit fault alters the coil's topology, it distorts the current path within the coil, making the observed magnetic field uninterpretable by normal current distribution and thermal drift. Therefore, during the real-time monitoring phase, a temperature-compensated magnetic field prediction model is first used to extract the thermal drift component from the real-time observations, resulting in a corrected magnetic flux density vector.

[0073] In this embodiment of the invention, the current thermal accumulation index is input into the thermally induced magnetic field drift model to obtain the current thermal drift component, which characterizes the expected magnetic field offset component at each magnetic field monitoring point in space due to the current winding temperature rise. The difference between the current magnetic induction intensity vector and the thermal drift component is used as the corrected magnetic induction vector to dynamically remove the nonlinear drift interference caused by normal load temperature rise from the real-time observation signal. This ensures that the output corrected magnetic induction vector retains only the linear magnetic field component generated by direct current excitation and the fault magnetic field component that may be caused by coil structural abnormalities, such as inter-turn short circuits.

[0074] Furthermore, by using least squares state estimation, a set of optimal current vectors that best fits both the correction vector and the current measurement can be found, and the model interpretation residuals generated by the virtual current can be defined as the structural consistency deviation index.

[0075] Preferably, in this embodiment of the invention, the signal-to-noise ratio balance factor is obtained based on the noise between the magnetic field measurement channel and the current measurement channel. This factor is used to balance the weights of magnetic field measurement data and current measurement data in the optimization objective, reflecting the difference in reliability between the two measurement channels. Specifically, the variances of the magnetic field measurement channel data and the current measurement channel data during the system's silent acquisition period are statistically analyzed, and their ratio is calculated. The ratio between the variances of the magnetic field measurement channel data and the current measurement channel data yields the signal-to-noise ratio balancing factor. The system's silent acquisition period refers to the initial operating phase where the equipment is fault-free and the load is stable. The data acquired during this phase contains only measurement noise, without fault distortion or operating condition fluctuation interference. Variance statistics can characterize the noise levels of the two types of channels.

[0076] Calculate the conjugate transpose of the reference electromagnetic coupling matrix. Electromagnetic coupling matrix with reference The product of these components forms the first intermediate matrix. This matrix describes the intrinsic information strength of the current-to-magnetic field mapping under unconstrained conditions. The identity matrix is ​​obtained by weighting the first intermediate matrix with the signal-to-noise ratio balance factor. The sum is then added together, a regularization constraint is introduced using the identity matrix, and the inverse of the sum is obtained to obtain the total weight matrix. This integrates electromagnetic transmission constraints and noise calibration information. It calculates the conjugate transpose of the reference electromagnetic coupling matrix. With the corrected magnetic induction vector product Characterizes the health current-related information contained in the corrected magnetic field signal, as well as the signal-to-noise ratio balance factor. With three-phase current phasor vector product This characterizes the effective information after noise weight calibration of the current measurement signal, and the product of the two is summed to obtain the weighted observation matrix. Calculate the product of the total weight matrix and the weighted observation matrix to obtain the optimized current vector. It integrates effective information from magnetic field observation and current measurement, and reflects the theoretical current that best explains and corrects the magnetic field and actual current under healthy electromagnetic confinement and noise balance conditions, serving as the ideal healthy reference current for the current operating state.

[0077] Furthermore, when an inter-turn short circuit occurs, a short-circuit circulating current is generated inside the coil, altering the physical transmission characteristics of the device. At this point, the predictive capability of the reference electromagnetic coupling matrix fails, and adjusting the optimized current vector cannot eliminate the difference between model predictions and actual observations. Therefore, the integrity of the coil's physical structure can be assessed and the potential for faults can be reflected by combining the deviation between the corrected magnetic field and the reconstructed magnetic field (i.e., the magnetic field residual) and the deviation between the optimized current and the measured current (i.e., the current residual), and defining their weighted sum as the structural consistency deviation index.

[0078] In this embodiment of the invention, the norm square of the difference between the corrected magnetic induction vector and the reconstructed magnetic field is calculated as the magnetic induction deviation. The reconstructed magnetic field is the product of the reference electromagnetic coupling matrix and the optimized current vector. The reconstructed magnetic field characterizes the ideal magnetic field distribution corresponding to the ideal reference current. The magnetic induction deviation quantifies the degree of deviation between the current pure magnetic field signal and the ideal magnetic field reference after thermal drift removal. Furthermore, the norm square of the difference between the optimized current vector and the three-phase current phasor vector is calculated as the product of the signal-to-noise ratio (SNR) balance factor, which is used as the current deviation. The current deviation quantifies the degree of deviation between the actual current and the ideal reference current, and the influence of current noise on the deviation result is calibrated by the SNR balance factor. Finally, the sum of the magnetic induction deviation and the current deviation is used as the structural consistency deviation index. When an inter-turn short circuit occurs, the coil topology changes, causing the reference electromagnetic coupling matrix to fail. At this time, regardless of the value of the optimized current vector, the significant difference between the model and the observation cannot be eliminated, resulting in a significant increase in the value of the structural consistency deviation index.

[0079] S4: Monitoring inter-turn short circuits based on structural consistency deviation index.

[0080] Under normal operating conditions, the difference between the observed magnetic field and the reconstructed magnetic field mainly stems from the random white noise of the sensor; therefore, the structural consistency deviation index will remain at a low and stable baseline level. Conversely, when an inter-turn short-circuit fault occurs in the reactor, the physical topology inside the coil changes, such as a reduction in the effective number of turns and the generation of short-circuit circulating currents, causing a fundamental deviation between the actual electromagnetic transmission characteristics of the device and the fixed reference electromagnetic coupling matrix. This structural model mismatch cannot be eliminated by adjusting the virtual optimization current, resulting in a significant and sustained step increase in the structural consistency deviation index.

[0081] In this embodiment of the invention, the arithmetic mean of the structural consistency deviation indexes within the current sliding window during the monitoring phase is calculated. The sliding window is a preset window size before the current moment, and its length can be set to 50. The implementer can adjust this according to the specific implementation scenario, and no limitation is imposed here. The arithmetic mean of all structural consistency deviation indices is calculated. When the arithmetic mean is greater than a set fault threshold, an inter-turn short-circuit fault is determined to have occurred inside the reactor, and an alarm signal is output.

[0082] In one specific embodiment of the present invention, the mean and standard deviation of all structural consistency deviation indices during the initial baseline phase can be calculated to obtain a set fault threshold. As an example, the expression for setting the fault threshold is: In the formula, This indicates that a fault threshold has been set. This represents the mean of all structural consistency deviation indices during the initial baseline phase. This is expressed as the standard deviation of all structural consistency deviation indices during the initial baseline phase. The confidence coefficient is used to control the false alarm rate. Its value can be 6, and the implementer can adjust it as needed.

[0083] In summary, this invention avoids the instability in benchmark model identification caused by the linear correlation of three-phase currents by introducing regularization constraints to construct a benchmark electromagnetic coupling matrix in the initial baseline stage, thus ensuring the physical authenticity of the cold-state benchmark. Based on the recursive analysis of the thermal accumulation index of three-phase current energy, combined with the fitting of the thermally induced magnetic field drift model using magnetic field residuals, and through recursive analysis of current amplitude energy, the temperature state of windings with thermal inertia can be more accurately quantified. By combining the joint constraints of the benchmark electromagnetic coupling matrix and the thermal drift model, normal thermally induced magnetic field drift components can be dynamically stripped away in real-time monitoring, and the drift due to temperature rise can be analyzed even under conditions of large load fluctuations or frequent switching between hot and cold states. Under the constraints of the benchmark electromagnetic coupling matrix, the current vector is optimized by fitting weighted least squares state estimation, and the structural consistency deviation index is obtained by comprehensively considering the deviations of the magnetic field and current in both dimensions. This index can specifically capture coil topology anomalies and improve the sensitivity of early weak inter-turn short-circuit fault identification. This invention combines linear electromagnetic transfer law with nonlinear thermal drift compensation for monitoring, adapting to the complex magnetic field environment and variable load operation conditions of dry-type air-core reactors, and providing continuous and reliable monitoring guarantee for the long-term stable operation of the equipment.

[0084] This application also provides a dry-type air-core reactor inter-turn short-circuit monitoring system based on the magnetic field differential method. Please refer to [link to relevant documentation]. Figure 2 The diagram shows a structural diagram of a dry-type air-core reactor inter-turn short-circuit monitoring system based on the magnetic field difference method according to an embodiment of the present invention. The system includes: a data monitoring module 201, a thermal state influence analysis module 202, a structural deviation analysis module 203, and a monitoring module 204.

[0085] The data monitoring module 201 is used to acquire the phase-aligned magnetic flux density vector and three-phase current phasor vector on the periphery of the reactor in each analysis window during the initial baseline phase of system startup.

[0086] The thermal state influence analysis module 202 is used to construct a reference electromagnetic coupling matrix characterizing the current-to-magnetic field transfer relationship based on the magnetic induction intensity vector and the three-phase current phasor vector during the initial baseline stage by introducing regularization constraints; based on the thermal accumulation index at the moment of amplitude energy recursion analysis of the three-phase current phasor vector, the magnetic field residual vector of the analysis window is predicted through the reference electromagnetic coupling matrix, the mapping relationship between the residual vector and the thermal accumulation index is fitted and analyzed, and the thermally induced magnetic field drift model is determined.

[0087] The structural deviation analysis module 203 is used in the monitoring stage after the initial baseline stage to analyze the current thermal drift component based on the thermally induced magnetic field drift model, correct the current magnetic induction intensity vector to obtain the corrected magnetic induction vector; under the constraint of the reference electromagnetic coupling matrix, the state is estimated based on the corrected magnetic induction vector and the three-phase current phasor vector by the weighted least squares method, the optimized current vector is fitted, and the deviation between the corrected magnetic induction intensity vector and the optimized current vector reconstructing the magnetic field, as well as the deviation between the optimized current vector and the three-phase current phasor vector, are analyzed to obtain the structural consistency deviation index.

[0088] Monitoring module 204 is used to monitor inter-turn short circuits based on structural consistency deviation index.

[0089] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the dry-type air-core reactor inter-turn short circuit monitoring system based on the magnetic field differential method and the dry-type air-core reactor inter-turn short circuit monitoring method based on the magnetic field differential method provided in the above embodiments belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0090] This application also provides a computer device; please refer to [link / reference]. Figure 3 The illustration shows a schematic diagram of a computer device structure according to an embodiment of the present invention. The computer device includes a memory 301, a processor 302, and a computer program 303 stored in the memory 301 and running on the processor 302. When the processor 302 executes the computer program 303, the computer device can execute any of the aforementioned methods for monitoring inter-turn short circuits of dry air-core reactors based on the magnetic field differential method.

[0091] This application also provides a computer program product that, when run on a computer device, enables the computer device to execute any of the aforementioned methods for monitoring inter-turn short circuits in dry-type air-core reactors based on the magnetic field differential method.

[0092] This application also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer device, the computer device can execute any of the aforementioned methods for monitoring inter-turn short circuits in dry-type air-core reactors based on the magnetic field difference method.

[0093] In the embodiments provided in this application, it should be understood that the computer device, computer program product and computer-readable storage medium provided are all used to perform the corresponding methods provided above, and therefore the beneficial effects they can achieve can be referred to the beneficial effects of the methods provided above, which will not be repeated here.

[0094] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A dry-type air-core reactor turn-to-turn short circuit monitoring method based on magnetic field differential method, characterized in that, The method comprises: In the initial baseline stage of system startup, the magnetic induction intensity vector and the three-phase current phasor vector of the reactor are obtained in each analysis window; In the initial baseline stage, based on the magnetic induction intensity vector and the three-phase current phasor vector, a regularization constraint is introduced to construct a benchmark electromagnetic coupling matrix representing the current-to-magnetic field transmission relationship; based on the thermal accumulation index of the amplitude energy recursive analysis moment of the three-phase current phasor vector, the magnetic field residual vector of the analysis window is predicted by the benchmark electromagnetic coupling matrix, the mapping relationship between the analysis residual vector and the thermal accumulation index is fitted, and the thermal magnetic field drift model is determined; In the monitoring stage after the initial baseline stage, the current thermal drift component is analyzed based on the thermal magnetic field drift model, and the corrected magnetic induction vector is obtained by correcting the current magnetic induction intensity vector; under the constraint of the benchmark electromagnetic coupling matrix, the state estimation is performed based on the corrected magnetic induction vector and the three-phase current phasor vector by the weighted least square method, the current vector is fitted and optimized, and the deviation of the corrected magnetic induction intensity vector and the optimized current vector from the reconstructed magnetic field is analyzed, and the deviation of the optimized current vector from the three-phase current phasor vector is analyzed, to obtain the structural consistency deviation index; The turn-to-turn short circuit is monitored based on the structural consistency deviation index.

2. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the benchmark electromagnetic coupling matrix comprises: An observation matrix composed of the magnetic induction intensity vectors of each analysis window in the initial baseline stage is constructed, and an excitation matrix composed of the three-phase current phasor vectors of the corresponding analysis window is constructed; based on the Tikhonov regularization principle, the autocorrelation matrix of the excitation matrix is corrected by using a collinearity suppression factor, and the benchmark electromagnetic coupling matrix is obtained by combining the observation matrix.

3. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the thermal accumulation index comprises: The thermal dissipation attenuation coefficient is determined according to the thermal time constant of the reactor and the sampling interval; at the initial moment, the initial thermal accumulation index is determined based on the norm square of the three-phase current phasor vector; At each moment after the initial moment, the norm square of the three-phase current phasor vector at the current moment is weighted by using the thermal dissipation attenuation coefficient, and the thermal accumulation index at the previous moment is weighted and retained by using the thermal inertia weight, and the weighted sum of the two is taken as the thermal accumulation index at the current moment; wherein the sum of the thermal inertia weight and the thermal dissipation attenuation coefficient is 1.

4. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the thermal magnetic field drift model comprises: In the initial baseline stage, the reconstructed magnetic induction intensity vector at the moment is obtained by multiplying the three-phase current phasor vector at the moment and the benchmark electromagnetic coupling matrix; the prediction residual vector is obtained by the difference between the magnetic induction intensity vector at the moment and the reconstructed magnetic induction intensity vector; the thermal magnetic field drift model is obtained by using a quadratic polynomial fitting model for each element in the magnetic induction intensity vector, taking the thermal accumulation index as the independent variable and the prediction residual vector of the benchmark electromagnetic coupling matrix as the dependent variable.

5. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the corrected magnetic induction vector comprises: The current thermal drift component is obtained by inputting the current thermal accumulation index into the thermal magnetic field drift model; the difference between the current magnetic induction intensity vector and the thermal drift component is taken as the corrected magnetic induction vector.

6. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the optimized current vector comprises: Obtaining a signal-to-noise ratio balance factor based on noise between the magnetic field measurement channel and the current measurement channel; Calculating a product of a conjugate transpose matrix of the reference electromagnetic coupling matrix and the reference electromagnetic coupling matrix itself to construct a first intermediate matrix; adding the first intermediate matrix and a unit matrix weighted by the signal-to-noise ratio balance factor, and inverting the addition result to obtain a total weight matrix; calculating a product of the conjugate transpose matrix of the reference electromagnetic coupling matrix and the modified magnetic induction vector, and a product of the signal-to-noise ratio balance factor and the three-phase current phasor vector, and adding the two product results to obtain a weighted observation matrix; and calculating a product of the total weight matrix and the weighted observation matrix to obtain the optimized current vector.

7. The method according to claim 6, wherein the method is characterized in that, The method for obtaining the structural consistency deviation index comprises: Calculating a norm square of a difference between the modified magnetic induction vector and a reconstructed magnetic field as a magnetic induction deviation, the reconstructed magnetic field being a product of the reference electromagnetic coupling matrix and the optimized current vector; calculating a product of a norm square of a difference between the optimized current vector and the three-phase current phasor vector and the signal-to-noise ratio balance factor as a current deviation; and taking a sum of the magnetic induction deviation and the current deviation as the structural consistency deviation index.

8. The method according to claim 1, wherein the method is characterized in that, The method for monitoring the inter-turn short circuit based on the structural consistency deviation index comprises: Calculating an arithmetic mean of the structural consistency deviation index in a current sliding window in a monitoring stage, and outputting an alarm signal when the arithmetic mean is greater than a set fault threshold.

9. The method according to claim 1, wherein the method is characterized in that, The method for obtaining the phase-aligned magnetic induction intensity vector and the three-phase current phasor vector comprises: Determining a reference phase reference in the three-phase current, and calculating a real-time phase angle of the reference phase reference current; constructing a complex rotation operator by using the real-time phase angle; Multiplying the multi-channel collected magnetic induction intensity signals and the three-phase current signals by the complex rotation operator respectively to generate the phase-aligned magnetic induction intensity vector and the three-phase current phasor vector.

10. The method according to claim 6, wherein the method is characterized in that, The method for obtaining the signal-to-noise ratio balance factor comprises: Calculating a signal-to-noise ratio balance factor by calculating a ratio of a variance of the magnetic field measurement channel data to a variance of the current measurement channel data during system quiet acquisition.

Images