Transformer winding monitoring and diagnostic data acquisition method, system, equipment and medium

Abstract

The present invention discloses a transformer winding monitoring and diagnostic data acquisition method, system, device, and medium, belonging to the technical field of power equipment status monitoring. The method comprises: constructing a multi-scale wavelet transform and an improved adaptive Kalman filter fusion architecture, wherein the multi-scale wavelet transform and the improved adaptive Kalman filter fusion architecture are used to obtain monitoring signals of transformer windings; using the multi-scale wavelet transform to perform frequency division processing on the monitoring signals to determine each scale coefficient; adjusting the noise covariance matrix based on the improved adaptive Kalman filter, filtering the each scale coefficient to determine the filtering result; reconstructing the filtering result into a reconstructed signal, and performing signal quality assessment on the reconstructed signal to determine the signal quality of the transformer windings. Through multi-scale decomposition and adaptive filtering, the present invention overcomes the limitations of traditional methods in non-stationary signal processing and strong noise interference, and improves the preservation of signal features and the signal-to-noise ratio.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment status monitoring, in particular to a transformer winding monitoring and diagnosis data acquisition method, system, equipment and medium. Background Art

[0002] In the power system, transformers are core equipment, and the winding status of their windings is directly related to the reliability of grid operation. With the continuous growth of power load and the increasing complexity of grid topology, the electromagnetic-thermal coupling effect on transformer windings has been significantly enhanced, resulting in an increasing incidence of faults such as local overheating and insulation degradation year by year.

[0003] However, the traditional periodic inspection and maintenance model has a response lag problem when dealing with sudden faults. In addition, the existing online monitoring technology generally has a low signal-to-noise ratio in the data acquisition system under complex electromagnetic interference environments, which makes it difficult to meet the needs of accurate status assessment. Therefore, there is an urgent need for a method that can improve the quality of online monitoring data acquisition for transformer windings in complex power environments to support more accurate and reliable fault diagnosis and early warning. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to propose a transformer winding online monitoring and diagnostic data acquisition method based on the fusion of multi-scale wavelet transform and improved adaptive Kalman filter. In view of the problem that the data acquisition quality of the single signal processing method in the existing technology is not high in a complex power environment, which in turn affects the accuracy and reliability of the diagnosis results, the multi-resolution analysis capability of the multi-scale wavelet transform and the dynamic filtering advantages of the improved adaptive Kalman filter are utilized to achieve high-precision noise reduction and feature extraction of the transformer winding monitoring signal.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a transformer winding monitoring and diagnosis data acquisition method, which includes the following steps:

[0007] A multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture is constructed, and the multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture is used to obtain monitoring signals of transformer windings; the monitoring signals are frequency-divided using the multi-scale wavelet transform to determine each scale coefficient; the noise covariance matrix is ​​adjusted based on the improved adaptive Kalman filter, and the each scale coefficient is filtered to determine the filtering result; the filtering result is reconstructed into a reconstructed signal, and the reconstructed signal is subjected to signal quality evaluation to determine the signal quality of the transformer winding.

[0008] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method described in the present invention, wherein: the monitoring signal is subjected to frequency division processing using the multi-scale wavelet transform, and each scale coefficient is determined by selecting a basis function that satisfies the transformer winding vibration characteristics and introducing an adaptive threshold function that considers the winding temperature field distribution characteristics. The specific formula is as follows:

[0009] ;

[0010] in, Representation scale The wavelet denoising threshold under ; Representation scale Standard deviation of wavelet coefficients; Indicates the number of wavelet coefficients at the current scale; represents the scale energy adjustment coefficient; Indicates the Layer wavelet coefficients norm; Indicates the The wavelet coefficients after layer filtering norm; represents the temperature-vibration coupling coefficient; Indicates the maximum temperature gradient of the winding; The beneficial effect of this preferred technical solution is that it introduces a dedicated basis function that meets the vibration characteristics of the transformer winding, and combines it with an adaptive threshold function that considers the distribution characteristics of the winding temperature field, thus solving the problem that traditional wavelet transform is difficult to adapt to the specific vibration mode and temperature influence of the transformer.

[0011] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method described in the present invention, the noise covariance matrix adjustment based on the improved adaptive Kalman filter includes: constructing a state prediction equation and an observation equation, the specific formulas are as follows:

[0012] ;

[0013] ;

[0014] Determine the state vector including displacement, velocity, elastic modulus and damping coefficient. The specific formula is as follows:

[0015] ;

[0016] The noise covariance matrix is ​​updated based on the prediction equation and the observation equation. The specific formula is as follows:

[0017] ;

[0018] ;

[0019] in, Indicates time The predicted value of the state; Indicates time The estimated value of the state; Indicates displacement; represents the posterior state estimate at the current time k, that is, the optimal estimate of the system state after fusing the observations; Indicates the last moment The state prediction value is used for current state prediction or error correction; Indicates speed; represents the elastic modulus; represents the damping coefficient; Represents the state transfer matrix, that is, from Time has come Dynamic mapping of moments; represents the control input matrix; represents the process noise, which is assumed to be Gaussian white noise; Represents the observation value at the current moment; Represents the observation matrix, which is used to map the state to the measurement space; represents the measurement noise; Indicates time The process noise covariance matrix of ; Indicates time The process noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; represents the forgetting factor; represents the measurement noise update forgetting factor; It represents the temperature gradient coupling coefficient, which characterizes the influence of temperature change on process noise; It represents the square norm of the temperature gradient and indicates the severity of the temperature field change; The beneficial effect of this preferred technical solution is that by adjusting the noise covariance matrix based on the improved adaptive Kalman filter and filtering each scale coefficient, accurate filtering of each frequency component in the transformer winding monitoring signal is achieved.

[0020] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method described in the present invention, wherein: filtering the scale coefficients to determine the filtering results includes: using the wavelet coefficients as observation values ​​of the improved adaptive Kalman filter; filtering the scale coefficients through filter gain matrix calculation, state update and posterior error covariance update to obtain the filtered scale coefficients; the specific formulas for the filter gain matrix calculation, state update and posterior error covariance update are as follows:

[0021] ;

[0022] in, Indicates time The Kalman gain matrix is ​​used for state update; Represents the state prediction error covariance matrix before updating; Represents the observation matrix, which is used to map the state to the observation space; represents the observation noise covariance matrix, reflecting the measurement uncertainty.

[0023] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method described in the present invention, the method of determining the signal quality of the transformer winding includes: reconstructing the scale coefficients after the filtering process to obtain a reconstructed signal; extracting the transformer winding state characteristics of the reconstructed signal, and determining the signal quality evaluation result of the reconstructed signal based on the transformer winding state characteristics; and performing feedback optimization on the feature extraction error based on the signal quality evaluation result of the reconstructed signal.

[0024] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method described in the present invention, wherein: the determination of the signal quality evaluation result of the reconstructed signal includes: constructing the fault feature vector and energy feature of the reconstructed signal based on the thermal-mechanical coupling effect model; combining the fault feature vector and energy feature, introducing a signal quality evaluation function based on the thermal-mechanical coupling model to evaluate the signal quality of the reconstructed signal, and obtaining the signal quality evaluation result of the reconstructed signal. The beneficial effect of this preferred technical solution is that it constructs the fault feature vector and energy feature based on the thermal-mechanical coupling effect model, and introduces a signal quality evaluation function that considers the thermal-mechanical coupling model, which solves the problem that traditional monitoring methods are difficult to comprehensively evaluate signal quality and difficult to adapt to complex working conditions of transformers.

[0025] As a preferred solution of the transformer winding monitoring and diagnostic data acquisition method of the present invention, the specific formula of the signal quality evaluation function is as follows:

[0026] ;

[0027] in, Indicates the current time Signal quality assessment factor; Represents the observation value at the current moment; Indicates time The predicted value of the state; Represents the observation matrix, which is used to map the state to the observation space; Represents the second norm of the state prediction error, that is, the square of the residual; represents the trace of the observation noise covariance matrix, i.e., the sum of the diagonal elements; represents the stress gradient tensor obtained based on the thermal-mechanical coupling model; Represents the stress gradient weight coefficient, which is used to measure the impact of thermal-mechanical coupling on signal quality evaluation.

[0028] Another object of the present invention is to provide a transformer winding monitoring and diagnostic data acquisition system.

[0029] To solve the above technical problems, the present invention provides the following technical solutions: a transformer winding monitoring and diagnosis data acquisition system, comprising: a signal acquisition module, used to construct a multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture, the multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture is used to obtain the monitoring signal of the transformer winding; a multi-scale wavelet transform module, used to use the multi-scale wavelet transform to perform frequency division processing on the monitoring signal and determine each scale coefficient; an improved adaptive Kalman filter module, used to adjust the noise covariance matrix based on the improved adaptive Kalman filter, and filter each scale coefficient to determine the filtering result; a feature fusion diagnosis module, used to reconstruct the filtering result into a reconstructed signal, and perform signal quality evaluation on the reconstructed signal to determine the signal quality of the transformer winding.

[0030] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the processor executes the computer program, the steps of the transformer winding monitoring and diagnostic data acquisition method are implemented.

[0031] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, the steps of the transformer winding monitoring and diagnostic data acquisition method are implemented.

[0032] The beneficial effects of the present invention are as follows: the data acquisition method based on the fusion of multi-scale wavelet transform and improved adaptive Kalman filtering proposed in the present invention effectively combines the advantages of multi-scale wavelet transform in time-frequency analysis and the ability of improved adaptive Kalman filtering in noise suppression, and can improve the accuracy and reliability of transformer winding monitoring data acquisition in the power environment; through multi-scale decomposition and adaptive filtering, it effectively solves the limitations of traditional methods in non-stationary signal processing and strong noise interference, and improves the retention of signal characteristics and signal-to-noise ratio; it can provide high-quality data support for transformer winding status monitoring, provide a more accurate basis for subsequent fault diagnosis and early warning, help improve the reliability of transformer operation, reduce maintenance costs, and is of great significance to the construction of smart grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0034] Figure 1 A schematic flow chart of a method for online monitoring and diagnosis of transformer windings provided by one embodiment of the present invention.

[0035] Figure 2 A schematic diagram of the structure of a computer device provided in one embodiment of the present invention.

[0036] Figure 3 This is a comparison chart of the average accuracy of various methods provided by one embodiment of the present invention under the same test data set.

[0037] Figure 4 This is a graph showing the accuracy variation trends of three methods under different noise levels provided by one embodiment of the present invention.

[0038] Figure 5 A schematic diagram of the noise reduction effects on noisy signals using different methods provided in one embodiment of the present invention.

[0039] Figure 6 A schematic diagram of residual scatter distribution provided by one embodiment of the present invention.

[0040] Figure 7 A comparison chart of the three methods provided in one embodiment of the present invention in terms of characteristic peak retention.

[0041] Figure 8 A schematic diagram of the performance of each method provided in one embodiment of the present invention on five key performance indicators.

[0042] Figure 9 A distribution diagram of vibration characteristic parameters of a transformer in normal and fault states provided by one embodiment of the present invention.

[0043] Figure 10 A temperature characteristic parameter distribution diagram of a transformer in normal and fault states provided by one embodiment of the present invention.

[0044] Figure 11 This is a distribution diagram of electrical characteristic parameters of a transformer in normal and fault states provided by one embodiment of the present invention.

[0045] Figure 12 A comparison chart of the diagnostic performance of three methods in vibration, temperature, and electrical characteristics provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0047] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0048] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0049] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a transformer winding monitoring and diagnostic data acquisition method, comprising:

[0050] S100: Construct a multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture, which is used to obtain monitoring signals of transformer windings.

[0051] S200: Perform frequency division processing on the monitoring signal using multi-scale wavelet transform to determine each scale coefficient.

[0052] S300: adjusting the noise covariance matrix based on the improved adaptive Kalman filter, and performing filtering processing on each scale coefficient to determine a filtering result.

[0053] S400: Reconstruct the filtering result into a reconstructed signal, and perform signal quality evaluation on the reconstructed signal to determine the signal quality of the transformer winding.

[0054] It should be noted that transformers are subject to complex electromagnetic-thermal coupling effects during operation, and winding status monitoring faces multiple technical challenges. For example, there is serious aliasing between high-frequency discharge pulses and low-frequency mechanical vibration characteristics in the time-frequency domain, making it difficult for traditional filtering methods to effectively separate them; changes in temperature field gradients will significantly affect material parameters and signal propagation characteristics, resulting in the failure of fixed threshold denoising methods; the noise covariance matrix of the existing Kalman filtering algorithm lacks an adjustment mechanism and cannot adapt to the time-varying characteristics of winding status parameters; these problems result in insufficient monitoring signal quality, limited sensitivity of partial discharge detection, and significant errors in mechanical deformation feature extraction.

[0055] Therefore, in order to address the above-mentioned problems of multi-physical field coupling interference, time-varying noise suppression, and insufficient feature extraction accuracy, a composite processing framework integrating wavelet packet decomposition and improved Kalman filtering is established through steps S100-S400 to achieve multi-level frequency band separation; an adaptive threshold function of the temperature field is introduced to achieve accurate extraction of high-frequency discharge pulses and low-frequency vibration features under a specific wavelet basis; the noise suppression capability is enhanced by coupling the dynamic covariance update algorithm of the forgetting factor and the temperature gradient coefficient; the closed-loop optimization mechanism of the bidirectional long short-term memory network diagnostic network is combined to reduce the feature extraction error; the synergistic effect of multi-scale decomposition and adaptive filtering is used to effectively improve the signal-to-noise ratio and feature retention of the monitoring signal in complex electromagnetic environments, provide more reliable data support for power equipment status assessment, and improve the accuracy of fault warning.

[0056] Example 2, reference Figure 1 , which is the second embodiment of the present invention, provides a transformer winding monitoring and diagnostic data acquisition method based on the above embodiment.

[0057] In the embodiment of the present invention, step S100 constructs a fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter, including:

[0058] The monitoring signal is frequency-divided using a five-layer wavelet packet decomposition framework. The Daubechies 8th-order wavelet basis function is used to extract the 0.1-5MHz discharge pulse in the high-frequency band, and the symmetrical 6th-order wavelet function is used in the low-frequency band to retain the 10-500Hz mechanical vibration characteristics.

[0059] A 16-dimensional feature matrix including temperature, vibration and electrical parameters is established, and the observation weights of the improved adaptive Kalman filter are dynamically adjusted based on the sliding window mutual information entropy to adapt to the dynamic changes of the signal.

[0060] By forming a closed-loop optimization with the bidirectional long short-term memory network diagnostic network and the signal processing module, real-time correction of the decomposition layer number and forgetting factor can be achieved, further improving the adaptability and accuracy of data acquisition.

[0061] It should be noted that, in view of the inherent limitations of traditional methods between time-frequency analysis accuracy and dynamic tracking capabilities, the present invention proposes a collaborative processing architecture of multi-scale wavelet transform and improved adaptive Kalman filter. The specific formula of the fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter is as follows:

[0062] ;

[0063] in, Indicates the The wavelet coefficients after layer decomposition, i.e., the feature quantity; Indicates the Layer-adaptive wavelet decomposition kernel function, reflecting the feature extraction weight of the scale; Indicates the The input signal or wavelet coefficients of the layer; Indicates the time domain sampling point index; represents a set of integers, representing all possible convolution kernel indices; Indicates the After downsampling the layer signal, The sampling value of each position; Represents the index of the convolution kernel, which is used to indicate the shift index range corresponding to the filter when calculating the current wavelet coefficient.

[0064] Furthermore, the design criteria of the fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter are determined by the following invariant optimization model:

[0065] ;

[0066] ;

[0067] in, Represents the signal in the frequency domain The spectrum below; Indicates the The expression of layer kernel function in frequency domain; Indicates the Representation of layer wavelet coefficients in frequency domain; represents the regularization weight factor in variational optimization; represents the variation regularization term, that is, the total norm of the rate of change of the frequency domain kernel function; Represents the gradient of the kernel function in the frequency domain; represents the complex conjugate form of the j-th layer frequency domain kernel function; J represents the total number of layers of multi-scale wavelet transform; Represents the expression of the j-th layer wavelet kernel function in the frequency domain, which determines the filtering characteristics of the corresponding frequency band; ablah j (ω) represents the gradient of the j-th layer wavelet kernel function in the frequency domain.

[0068] Furthermore, in the dynamic filtering stage, a state space equation with scale adaptability is constructed. The specific formula is as follows:

[0069] ;

[0070] in, represents the unupdated predicted state vector; Representation scale The state transition matrix under ; Representation scale Status input; Represents the process noise term, which satisfies Gaussian distribution; Representation scale The noise variance of the lower process.

[0071] Furthermore, the process noise covariance matrix The time-varying parameters of are dynamically calculated as follows:

[0072] ;

[0073] in, Representation scale The process noise covariance estimate of ; Indicates the The total length of the layer wavelet coefficients; Indicates the Tier wavelet coefficients.

[0074] Furthermore, the iterative update formula of the Kalman gain matrix is ​​as follows:

[0075] ;

[0076] in, Representation scale The Kalman gain under represents the prediction error covariance matrix; Represents the observation matrix (mapping state to observation space); represents the observation noise covariance matrix, along with the new information term Dynamic updates; Indicates the The innovation term of the layer is the residual between the actual observation value and the predicted observation value; Represents the observation matrix The transpose of is used to map the error information in the observation space back to the state space to correct the state estimate.

[0077] It should be noted that the present invention realizes the efficient acquisition and preprocessing of transformer winding monitoring signals by constructing a fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter. By combining the advantages of multi-scale wavelet transform in signal decomposition and the expertise of improved adaptive Kalman filter in noise suppression, the problem that traditional single signal processing methods are difficult to cope with multi-source interference in the working environment of transformer windings is solved; it can effectively filter out external interference and measurement noise while retaining key feature information, thereby improving the reliability of subsequent analysis; compared with the common single filtering method in the prior art, the present invention captures the feature changes in different frequency domains through multi-scale analysis, and combines the adjustment capability of adaptive Kalman filter to improve the perception accuracy of transformer winding status in the power grid environment, providing a more solid data foundation for early warning of faults, and ultimately achieving the beneficial effects of improving power grid operation safety and reducing maintenance costs.

[0078] In an embodiment of the present invention, in step S200, the monitoring signal is frequency-divided using multi-scale wavelet transform, and determining each scale coefficient includes selecting a basis function that meets the vibration characteristics of the transformer winding and introducing an adaptive threshold function that considers the distribution characteristics of the winding temperature field.

[0079] In the embodiment of the present invention, a basis function that satisfies the transformer winding vibration characteristics is selected, and a specific formula for introducing an adaptive threshold function that considers the winding temperature field distribution characteristics is as follows:

[0080] ;

[0081] in, Representation scale The wavelet denoising threshold under is the set of positive real numbers; Representation scale Standard deviation of wavelet coefficients; Indicates the number of wavelet coefficients at the current scale; represents the scale energy adjustment coefficient; Indicates the Layer wavelet coefficients norm; Indicates the The wavelet coefficients after layer filtering norm; represents the temperature-vibration coupling coefficient; Indicates the maximum temperature gradient of the winding; represents the temperature adjustment exponential factor, is the set of real numbers.

[0082] It should be noted that by using multi-scale wavelet transform to perform frequency division processing on the monitoring signal and determine the coefficients of each scale, accurate decomposition and analysis of the complex monitoring signal of the transformer winding can be achieved; the present invention introduces a special basis function that meets the vibration characteristics of the transformer winding, and combines it with an adaptive threshold function that considers the temperature field distribution characteristics of the winding, which solves the technical problem that the traditional wavelet transform is difficult to adapt to the specific vibration mode and temperature influence of the transformer; it can adaptively extract key features at different frequency scales based on the unique electromagnetic-thermal-mechanical coupling characteristics of the transformer winding, so that the signal decomposition is more in line with the physical characteristics of the transformer winding; traditional methods often only focus on the vibration signal itself and ignore the influence of the temperature field on the vibration characteristics. The present invention introduces the temperature-vibration coupling coefficient and the temperature gradient parameter to construct a coupling relationship model between temperature and vibration, so that it can more accurately identify the real fault characteristics and changes caused by environmental factors, improve the accuracy of feature extraction and anti-interference ability, and ultimately provide a more reliable feature basis for transformer winding status assessment.

[0083] It should be noted that the analysis of transformer winding vibration characteristics includes:

[0084] The vibration response of transformer windings under electromagnetic-mechanical coupling constitutes a key physical characteristic for fault diagnosis. The axial helical symmetry of the winding structure causes its dynamic characteristics to be subject to dual constraints of geometric and material parameters. Its natural frequency distribution can be derived from the three-dimensional elastic body wave equation, as shown in the following formula:

[0085] ;

[0086] in, represents the divergence of the stress tensor; represents the displacement field vector, that is, the displacement change of each point in the winding over time; Indicates the density of the material, that is, the mass per unit volume; Represents Young's modulus, an index of the elastic properties of a material; Indicates time; Represents the volume force term, that is, the external excitation force per unit volume, such as the electromagnetic Lorentz force; Represents the time second derivative of displacement, corresponding to the acceleration distribution; Indicates the introduction of geometric nonlinear terms to consider the influence of structural deformation.

[0087] Furthermore, when the transformer winding is subjected to the Lorentz force excitation generated by the leakage magnetic field, the axial vibration acceleration obeys a nonlinear second-order differential equation, and the specific formula is as follows:

[0088] ;

[0089] in, represents the equivalent mass, i.e. the mass of the winding segment being analyzed; It represents the damping coefficient, that is, the system's ability to attenuate vibration; Indicates axial vibration displacement; represents velocity, i.e. the first derivative of displacement with respect to time; represents the second derivative of displacement with respect to time; It represents the nonlinear stiffness factor, which characterizes the change of system stiffness when the amplitude increases; Indicates the total number of modal or harmonic components; Indicates the The magnetic induction intensity amplitude of the harmonic component; Indicates the Harmonic current components; It represents the magnetic-mechanical coupling coefficient, which is used to represent the effect of harmonic frequency on system response; Indicates the modal or harmonic frequencies; Indicates the The initial phase of the mode; Represents a time variable.

[0090] It should be noted that the analysis of the winding temperature field distribution characteristics includes:

[0091] The transformer winding temperature field exhibits a three-dimensional non-steady-state distribution characteristic, and its temporal and spatial evolution law is described by the modified Fourier-Kirchhoff equation as follows:

[0092] ;

[0093] in, represents temperature as a function of position r and time t; It represents the rate of change of temperature with time, that is, the partial derivative of the temperature field with respect to time, and its unit is K / s; It is the time partial derivative operator, which means the operation of finding the partial derivative of the time variable t; α is the thermal diffusion coefficient, the unit is m² / s; represents the Laplace operator, which is used to calculate the spatial second-order derivative of the temperature field; r represents the position, usually a three-dimensional coordinate (x, y, z), which is used to describe the position dependence of the field variable; Indicates the volumetric heat source intensity per unit volume, in W / m³, such as iron core eddy current loss; Indicates the specific heat capacity of the material, with the unit of J / (kg·K), that is, the amount of heat required to raise the temperature of the unit mass of the material by 1K; Indicates the material density in kg / m³; Represents the current density vector, with the unit of A / m², that is, the current intensity per unit area; Indicates the electrical conductivity of the material in S / m; It represents the square of the modulus of the current density vector, which is the core term of the Joule heat power density.

[0094] Under rated operating conditions, the axial and radial temperature gradients of the winding follow a dual-scale distribution law, and the specific formula is as follows:

[0095] ;

[0096] in, and represent the axial and radial temperature gradients, respectively; Indicates the winding working current, the unit is A; Indicates AC resistance, unit is Ω; Indicates the thermal conductivity of insulating oil, the unit is W / (m·K); Indicates the equivalent cross-sectional area of ​​the cooling channel, in m²; Indicates the eddy current loss power density, in W / m³; Indicates the thermal conductivity of the turn insulation material, the unit is W / (m·K); Indicates the characteristic length of the winding structure, in meters; and represent the axial and radial normalized coordinates respectively; represents the error function.

[0097] When a turn-to-turn short circuit fault occurs, the heat source intensity at the fault point shows an exponential growth characteristic. The specific formula is as follows:

[0098] ;

[0099] in, Indicates time Fault heat source intensity; Indicates the initial value of normal heat source intensity; Indicates the current time; Indicates the starting moment of fault development; Represents the heat source growth time constant.

[0100] The aging process of insulation materials causes nonlinear degradation of the equivalent thermal conductivity. The specific formula is as follows:

[0101] ;

[0102] in, Indicates the insulation material at time Thermal conductivity under ; represents the initial thermal conductivity; Indicates the material aging rate coefficient; Indicates the current time; Indicates the aging time threshold.

[0103] The mechanical stress distribution caused by the thermal-electric coupling effect can be expressed as:

[0104] ;

[0105] in, represents the axial stress; represents Young's modulus; 、 and denote axial, radial and hoop strains, respectively; represents the coefficient of thermal expansion; Indicates temperature rise change; represents the Poisson-related coupling factor.

[0106] Temperature gradient variance matrix The dynamic relationship equation with the Kalman filter parameters is as follows:

[0107] ;

[0108] ;

[0109] in, Indicates the corrected time The process noise covariance matrix of ; Indicates the sampling period; represents the thermal relaxation time constant; Represents the diagonal matrix form of the temperature gradient variance matrix; represents the temperature gradient coupling coefficient; represents the square norm of the temperature gradient; represents the identity matrix; It represents the characteristic length of the heat diffusion path in the z-axis direction (or specified direction) (in meters); α represents the thermal diffusivity (in meters per second).

[0110] It should be noted that when the transformer winding online monitoring system is coupled with power frequency harmonics, random pulses and broadband noise, the traditional threshold denoising method is prone to cause the loss of characteristic waveform edge information; it represents a breakthrough in the contradictory relationship between time-frequency characteristics and dynamic noise, and constructs a hybrid architecture of multi-scale wavelet transform and improved adaptive Kalman filter. Its mathematical representation is defined by the following key equations, including:

[0111] Discrete wavelet transform realizes signal feature extraction through multi-scale orthogonal decomposition, and the mathematical representation is as follows:

[0112] ;

[0113] Where x represents the original input signal in the form of a discrete time series; x[n] represents the value of the original signal x at the time domain sampling point n, usually the amplitude value of the nth sampling point in a one-dimensional discrete sequence; n represents the discrete sampling index in the time domain; Indicates the In the first-order scale space Wavelet coefficients of displacement positions; Represented by the scale factor and displacement factor Generated orthogonal wavelet basis functions; Represents the mother wavelet function that satisfies the admissibility condition, satisfying .

[0114] In the design of the improved Kalman filter, the dynamic estimation mechanism of the noise covariance matrix and its state prediction and measurement update process satisfy:

[0115] ;

[0116] in, express The prior state estimation vector at the moment; Represents the linearized state transfer matrix of the nonlinear system; represents the prior error covariance matrix, which characterizes the statistical characteristics of the prediction error; represents the dynamic weight matrix; Represents the prior state estimation vector at the current moment k, that is, the predicted value; Indicates the last moment The predicted value of the state; Represents the control input matrix, which is used to transform the control input vector Introducing status updates; represents the control input vector, which can be external excitation, disturbance or command input; Represents the state estimation error covariance matrix at the previous moment, which measures the statistical uncertainty of the prediction error; Represents the linearized state transfer matrix of the nonlinear system The transpose of is used for the propagation of covariance; Indicates the corrected time The process noise covariance matrix is ​​.

[0117] In the embodiment of the present invention, the specific formula for constructing the time-varying covariance scaling factor is as follows:

[0118] ;

[0119] in, represents the dynamic weight matrix; W jrepresents the j-th layer wavelet coefficient; k represents the time domain sampling point index; L represents the half-width of the sliding window; ∥⋅∥ 2 Represents the L2 norm square, used to calculate energy; SNR k Represents the signal-to-noise ratio at time k, which is used to dynamically adjust the covariance weighting coefficient; the numerator The local wavelet energy is calculated by sliding the window to reflect the transient characteristics of the signal; the denominator Establish a parameterized signal-to-noise ratio adjustment channel, when the signal-to-noise ratio at time k The covariance constraint is automatically enhanced during descent; the diagonal matrix construction ensures independent adjustment of each state component, achieving refined control of noise suppression.

[0120] Furthermore, the sliding window length is adaptively determined by the following formula:

[0121] ;

[0122] in, Indicates time Adaptive sliding window length; Represents the 2-norm of the current wavelet coefficient, that is, the energy intensity; Indicates the time interval between adjacent sample points; The time constant that represents the dynamic response of the system.

[0123] In the embodiment of the present invention, the step S300 of adjusting the noise covariance matrix based on the improved adaptive Kalman filter includes:

[0124] Construct state prediction equations and observation equations.

[0125] Determine the state vector consisting of displacement, velocity, elastic modulus, and damping coefficient.

[0126] Update the noise covariance matrix based on the prediction equation and the observation equation.

[0127] Specifically, in the embodiment of the present invention, the specific formula of the state prediction equation is as follows:

[0128] ;

[0129] The specific formula of the observation equation is as follows:

[0130] ;

[0131] The specific formula of the state vector is as follows:

[0132] ;

[0133] Furthermore, the specific formula for updating the noise covariance matrix is ​​as follows:

[0134] ;

[0135] ;

[0136] in, Indicates time The predicted value of the state; Indicates time The estimated value of the state; Indicates displacement; represents the posterior state estimate at the current time k, that is, the optimal estimate of the system state after fusing the observations; Indicates the last moment The state prediction value is used for current state prediction or error correction; Indicates speed; represents the elastic modulus; represents the damping coefficient; Represents the state transfer matrix, that is, from Time has come Dynamic mapping of moments; represents the control input matrix; represents the process noise, which is assumed to be Gaussian white noise; Represents the observation value at the current moment; Represents the observation matrix, which is used to map the state to the measurement space; represents the measurement noise; Indicates time The process noise covariance matrix of ; Indicates time The process noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; represents the forgetting factor; represents the measurement noise update forgetting factor; It represents the temperature gradient coupling coefficient, which characterizes the influence of temperature change on process noise; It represents the square norm of the temperature gradient and indicates the severity of the temperature field change; Represents the identity matrix.

[0137] It should be noted that in the winding mechanical vibration propagation model, the axial vibration propagation characteristics of the transformer winding constitute the theoretical basis for fault diagnosis; based on the elastic dynamics theory of inhomogeneous media, the vibration propagation process of the winding under electromagnetic excitation is described by the modified D'Alembert principle, and the axial coordinate system is established. , vibration displacement field Satisfies the following formula:

[0138] ;

[0139] in, represents the winding cross-sectional area function; and They represent the spatial distribution of elastic modulus and density caused by insulation aging; represents the viscous damping coefficient; represents the Lorentz force density term, which reveals the effect of material degradation on the vibration wave propagation speed through the variable coefficient form Dynamic impact mechanism of Indicates the current time.

[0140] Furthermore, a state space model is constructed to adapt to the improved adaptive Kalman filter framework, where the specific formula for defining the state vector is as follows:

[0141] ;

[0142] The nonlinear observation equation is derived, and the specific formula is as follows:

[0143] ;

[0144] in, Indicates time The state vector of ,speed , elastic modulus and damping coefficient ; Indicates time The state vector of represents the nonlinear state transfer function; represents the process noise impact matrix; represents process noise; represents the observation matrix; represents the observed quantity; represents the observation noise.

[0145] Furthermore, process noise and observation noise The covariance matrix of is updated in real time in the improved adaptive Kalman filter. The specific formula is as follows:

[0146] ;

[0147] in, represents the corrected process noise covariance matrix; represents the forgetting factor; represents the Kalman gain matrix, represents the transpose of the Kalman gain matrix; represents the innovation vector, i.e. the residual term; Represents the outer product of the new information, reflecting the update of the covariance estimate.

[0148] In the embodiment of the present invention, filtering is performed on each scale coefficient in step S300 to determine the filtering result, including:

[0149] The wavelet coefficients are used as observation values ​​of the improved adaptive Kalman filter.

[0150] Through filtering gain matrix calculation, state update and posterior error covariance update, each scale coefficient is filtered to obtain each scale coefficient after filtering.

[0151] In the embodiment of the present invention, the specific formulas for filter gain matrix calculation, state update, and posterior error covariance update are as follows:

[0152] ;

[0153] in, Indicates time The Kalman gain matrix is ​​used for state update; Represents the state prediction error covariance matrix before updating; Represents the observation matrix, which is used to map the state to the observation space; represents the observation noise covariance matrix, reflecting the measurement uncertainty.

[0154] It should be noted that the present invention adjusts the noise covariance matrix based on the improved adaptive Kalman filter and filters each scale coefficient to achieve precise filtering of each frequency component in the transformer winding monitoring signal; wherein, by constructing a complete state vector including displacement, velocity, elastic modulus and damping coefficient, and introducing the temperature gradient coupling coefficient, the technical problems of fixed parameters and insufficient adaptability of traditional Kalman filtering when facing the complex working environment of the transformer winding are solved; a noise covariance matrix update mechanism based on temperature gradient is proposed, which can automatically adjust the filtering parameters according to the real-time temperature field changes of the transformer winding through the forgetting factor and the temperature gradient coupling coefficient, thereby adapting to the changes in winding characteristics under different load conditions; compared with the filtering method with static parameter setting in the prior art, the present invention can maintain the stability of filtering performance under extreme working conditions such as severe temperature fluctuations, effectively suppress various non-stationary noise interferences, improve the signal-to-noise ratio of fault characteristics, and provide clearer signal characteristics for subsequent winding status diagnosis, ultimately achieving high precision and high reliability of transformer winding monitoring and diagnosis.

[0155] It should be noted that the multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture constructs a feature-level fusion model, and its dynamic weight allocation mechanism can be expressed as a nonlinear superposition of multi-source signals. The specific formula is as follows:

[0156] ;

[0157] in, Indicates the fusion feature vector at time The output value of Indicates the number of fusion source signals; Indicates the The signal source at time The weight factor of Indicates that the signal is on scale The wavelet decomposition coefficients under , namely the wavelet characteristics; Indicates the Improved filter response kernel of the channel, dynamically adjusted with time and measurement variables; It represents the Hadamard product, which is used to represent the element-by-element product of elements in corresponding positions, that is, the matrix or vector is multiplied one by one; Represents tensor product, which is used to fuse features of different modalities.

[0158] Furthermore, the process noise covariance matrix is ​​updated using exponentially weighted recursion. The specific formula is as follows:

[0159] ;

[0160] in, Indicates the The noise covariance matrix at time t; Indicates the The noise covariance matrix at time t; represents the forgetting factor, ; represents the new information, i.e. the observation residual; Indicates the current observation value; represents the observation matrix; Indicates the prediction status of the previous step.

[0161] Furthermore, the final decision fusion layer adopts the modified DS evidence synthesis rule, and the specific formula is as follows:

[0162] ;

[0163] in, Indicates the target state Confidence value of Indicates the Source signal pair The support probability of represents the fusion weight; Indicates the Source signal pair The support probability of represents the confidence correction factor, based on the Kullback-Leibler divergence distance Calculate the source confidence level, P k Indicates the The support probability of the source signal to the target state, P ref represents the probability distribution under the standard state; K represents the total number of fused signal sources; Θ j Represents the jth candidate target state; this formula is used to calculate the normalized confidence value of each state based on multi-source support and confidence correction factor .

[0164] In the embodiment of the present invention, determining the signal quality of the transformer winding in step S400 includes:

[0165] The scale coefficients after filtering are reconstructed to obtain a reconstructed signal.

[0166] The transformer winding state characteristics of the reconstructed signal are extracted, and the signal quality evaluation result of the reconstructed signal is determined according to the transformer winding state characteristics.

[0167] The feature extraction error is fed back and optimized based on the signal quality evaluation result of the reconstructed signal.

[0168] In the embodiment of the present invention, determining the signal quality evaluation result of the reconstructed signal in step S400 includes:

[0169] Based on the thermal-mechanical coupling effect model, the fault feature vector and energy feature of the reconstructed signal are constructed.

[0170] Combining the fault feature vector and energy feature, a signal quality evaluation function based on the thermal-mechanical coupling model is introduced to evaluate the signal quality of the reconstructed signal, and the signal quality evaluation result of the reconstructed signal is obtained.

[0171] In the embodiment of the present invention, the specific formula of the signal quality evaluation function is as follows:

[0172] ;

[0173] in, Indicates the current time Signal quality assessment factor; Represents the observation value at the current moment; Indicates time The predicted value of the state; Represents the observation matrix, which is used to map the state to the observation space; Represents the second norm of the state prediction error, that is, the square of the residual; represents the trace of the observation noise covariance matrix, i.e., the sum of the diagonal elements; represents the stress gradient tensor obtained based on the thermal-mechanical coupling model; Represents the stress gradient weight coefficient, which is used to measure the impact of thermal-mechanical coupling on signal quality evaluation.

[0174] It should be noted that for the thermal-mechanical coupling effect model, the temperature gradient and mechanical stress generated by the transformer winding during operation form a dynamic coupling system. The interaction mechanism can be mathematically described through the theoretical framework of thermoelasticity, and a generalized heat conduction equation including the energy conversion mechanism is established. The specific formula is as follows:

[0175] ;

[0176] in, Indicates the material density; Indicates the specific heat capacity of the material; Indicates temperature; Indicates time; represents thermal conductivity; represents the plastic stress tensor; represents the plastic strain rate tensor; represents the current density; Indicates the resistivity of the material; Represents the heat conduction term, reflecting the temperature gradient diffusion line representation.

[0177] Furthermore, considering the basic structural characteristics of anisotropic materials, the strain tensor induced by thermal expansion can be decomposed to represent elastic strain and thermal strain. The specific formula is as follows:

[0178] ;

[0179] in, represents the total strain tensor component; represents the material flexibility tensor; represents the stress tensor component; represents the thermal expansion coefficient tensor; Indicates the current temperature; Indicates the reference temperature; represents the thermo-plastic coupling coefficient; represents the bulk modulus; represents the temperature excitation function; Represents the time-accumulated thermal excitation term.

[0180] In the embodiment of the present invention, the dynamic response of the winding structure satisfies the differential form of the law of conservation of momentum, and the specific formula is as follows:

[0181] ;

[0182] in, represents the stress tensor component; represents the elastic stiffness tensor; represents the total strain tensor; represents the thermal expansion coefficient tensor; Indicates temperature changes; represents the coupling correction coefficient; It represents the unit displacement rate and reflects the deformation dynamics of the structure.

[0183] In the embodiment of the present invention, step S400 performs feedback optimization on the feature extraction error based on the signal quality evaluation result of the reconstructed signal, including:

[0184] The bidirectional long short-term memory network diagnostic network and the signal processing unit are used to form a closed-loop optimization to correct the number of decomposition layers and the forgetting factor.

[0185] It should be noted that the present invention reconstructs the filtering results into a reconstructed signal and performs signal quality evaluation on the reconstructed signal to achieve comprehensive evaluation and feedback optimization of the transformer winding state. The fault feature vector and energy feature are constructed based on the thermal-mechanical coupling effect model, and a signal quality evaluation function considering the thermal-mechanical coupling model is introduced, which solves the technical difficulties of traditional monitoring methods in comprehensively evaluating signal quality and adapting to complex working conditions of transformers. Secondly, a closed-loop mechanism from signal reconstruction to quality evaluation to feedback optimization is established. The influence of thermal-mechanical coupling on signal quality is quantified by stress gradient tensor and stress gradient weight coefficient, which can automatically identify and correct signal distortion caused by temperature change. Compared with the common one-way signal processing process in the prior art, the present invention can evaluate the reliability of the reconstructed signal in real time and adjust the pre-processing parameters in a targeted manner, thereby improving monitoring accuracy. At the same time, this quality assessment method based on the physical model can also effectively distinguish between real fault characteristics and normal operating fluctuations, reduce the false alarm rate, enhance the credibility of the diagnosis results, and provide more scientific and reliable data support for transformer operation and maintenance decisions.

[0186] Exemplarily, performing feedback optimization on the feature extraction error according to the signal quality evaluation result of the reconstructed signal includes:

[0187] Based on the fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter, a joint parameter optimization model with time-varying characteristics is established. Assume that the monitoring signal at each moment is , then the multi-scale decomposition coefficient matrix It can be expressed as:

[0188] ;

[0189] in, Indicates time The multi-scale decomposition coefficient matrix of ; Indicates the Layer wavelet decomposition coefficients; Indicates the Layer band basis matrix, corresponding to the frequency domain wavelet basis; represents the total number of wavelet decomposition scales; It means to perform weighted superposition on the output results of all wavelet decomposition layers k=1,2,…,N to construct the feature fusion matrix at multiple scales, i.e. The multi-scale decomposition coefficient matrix of .

[0190] Furthermore, a time-frequency sensitivity factor is constructed to quantify the response intensity of each scale component to line group faults. The specific formula is as follows:

[0191] ;

[0192] in, Representation scale Response sensitivity factor; Indicates the layer wavelet coefficient vector; Represents a typical failure mode template; represents the Hadamard product (element-wise product); Represents the 2-norm, used for energy normalization.

[0193] Furthermore, scale Response sensitivity factor Driving noise covariance matrix The dynamic correction of , the specific formula is as follows:

[0194] ;

[0195] in, Representation scale time The observation noise covariance matrix of ; Indicates the frequency band noise energy baseline value; Representation scale Response sensitivity factor; represents the degradation time constant of the time-frequency response; represents the identity matrix; represents the adjustment coefficient, which controls the degree of spectral correction of the covariance estimate; Represented by extracting the wavelet coefficient autocorrelation matrix Construct a diagonal matrix from the main diagonal elements of .

[0196] Furthermore, scale Response sensitivity factor Drives the dynamic adjustment of the subsequent covariance matrix and adjusts with the scale Response sensitivity factor The change in the forecast covariance Perform adaptive correction, where the predicted state covariance matrix in the Kalman filter process is The specific formula for adaptive correction is as follows:

[0197] ;

[0198] in, Representation scale The next predicted state covariance matrix; represents the state transition matrix; Representation scale The next predicted state covariance matrix; represents the derivative of the covariance with respect to sensitivity, that is, the rate of change; Indicates the sensitivity change, which is used to adaptively correct the prediction error.

[0199] In summary, the data acquisition method based on the fusion of multi-scale wavelet transform and improved adaptive Kalman filtering proposed in the present invention effectively combines the advantages of multi-scale wavelet transform in time-frequency analysis and the ability of improved adaptive Kalman filtering in noise suppression, and can improve the accuracy and reliability of transformer winding monitoring data acquisition in the power environment; through multi-scale decomposition and adaptive filtering, it effectively solves the limitations of traditional methods in non-stationary signal processing and strong noise interference, and improves the retention of signal characteristics and signal-to-noise ratio; it can provide high-quality data support for transformer winding status monitoring, provide a more accurate basis for subsequent fault diagnosis and early warning, help improve the reliability of transformer operation, reduce maintenance costs, and is of great significance to the construction of smart grids.

[0200] Example 3 is the third embodiment of the present invention. This embodiment provides a transformer winding monitoring and diagnostic data acquisition system, including: a signal acquisition module, used to construct a multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture, the multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture is used to obtain the monitoring signal of the transformer winding; a multi-scale wavelet transform module, used to use the multi-scale wavelet transform to perform frequency division processing on the monitoring signal and determine each scale coefficient; an improved adaptive Kalman filter module, used to adjust the noise covariance matrix based on the improved adaptive Kalman filter, and filter each scale coefficient to determine the filtering result; a feature fusion diagnosis module, used to reconstruct the filtering result into a reconstructed signal, and perform signal quality evaluation on the reconstructed signal to determine the signal quality of the transformer winding.

[0201] Example 4 is the fourth embodiment of the present invention, which is different from the first three embodiments in that:

[0202] like Figure 2 As shown, if the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or the portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media that can store program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0203] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0204] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, and then editing, interpreting, or processing in another suitable manner as necessary, and then storing it in a computer memory.

[0205] It should be understood that various components of the present invention may be implemented using hardware, software, firmware, or a combination thereof. In the aforementioned embodiments, multiple steps or methods may be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or combination of the following technologies known in the art may be used: a discrete logic circuit having logic gate circuits for implementing logic functions on data signals, an application-specific integrated circuit having suitable combinational logic gate circuits, a programmable gate array (PGA), a field-programmable gate array (FPGA), etc.

[0206] Example 5, with reference to Figures 3 to 12 , which is the fifth embodiment of the present invention, provides a transformer winding monitoring and diagnostic data acquisition method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0207] In order to verify the application effect of the present invention in the transformer winding online monitoring system, this embodiment carried out specific testing and verification. The test environment includes hardware configuration and software tools to ensure the comprehensiveness and accuracy of the test. Table 1 shows the specific parameter settings of the simulation model and algorithm. These parameter settings ensure that the simulation environment can comprehensively test and verify the performance and effects of the present invention in different scenarios, providing reliable data support and theoretical basis for practical applications.

[0208] Table 1 Simulation model and algorithm parameter settings

[0209] ,

[0210] According to the data in Table 1, the test is carried out. Figure 3 and Figure 4 The performance comparison and analysis results of three fault diagnosis methods are presented; specifically, Figure 3 The average accuracy of each method under the same test data set is presented. Among them, the accuracy of the multi-scale wavelet transform-improved adaptive Kalman filter fusion method reaches 92.36%, which is 13.84% and 8.57% higher than the traditional multi-scale wavelet transform method and the traditional improved adaptive Kalman filter method respectively; at the same time, the standard deviation of the multi-scale wavelet transform-improved adaptive Kalman filter fusion method is 1.93, which is lower than the 3.21 of the traditional multi-scale wavelet transform method and the 2.87 of the traditional improved adaptive Kalman filter method, indicating that its diagnostic results are more stable and reliable. Figure 4The accuracy trends of the three methods under different noise levels are shown. As the noise level increases from 0.1 to 0.5, the accuracy of the multiscale wavelet transform-improved adaptive Kalman filter fusion method decreases from 91.68% to 80.36%, a decrease of 11.32%. In comparison, the accuracy of the traditional multiscale wavelet transform method and the traditional improved adaptive Kalman filter method decreases from 76.83% to 56.78% and from 82.17% to 64.51%, respectively, with decreases of 20.05% and 17.66%, respectively. Data analysis shows that the multiscale wavelet transform-improved adaptive Kalman filter fusion method outperforms the traditional single method in all indicators. In a low noise environment (0.1), the accuracy of the multiscale wavelet transform-improved adaptive Kalman filter fusion method is 14.85% higher than that of the traditional multiscale wavelet transform method and 9.51% higher than that of the traditional improved adaptive Kalman filter method. This advantage is further enhanced in a high-noise environment (0.5), with the accuracy of the multiscale wavelet transform-improved adaptive Kalman filter fusion method exceeding that of the traditional multiscale wavelet transform method and the traditional improved adaptive Kalman filter method by 23.58% and 15.85%, respectively. This result fully demonstrates the superiority of the multiscale wavelet transform-improved adaptive Kalman filter fusion method in transformer winding fault diagnosis, especially in complex noise environments, where it exhibits strong robustness and accuracy. This method effectively combines the multi-resolution analysis capabilities of the multiscale wavelet transform with the dynamic filtering advantages of the improved adaptive Kalman filter, providing a highly efficient and reliable technical means for online monitoring and fault diagnosis of transformer windings.

[0211] like Figure 5 The denoising effects of different methods on noisy signals are demonstrated. Among them, the multi-scale wavelet transform-improved adaptive Kalman filter method effectively suppresses noise interference while maintaining the overall trend of the signal, especially in the three characteristic peak areas of 2.3-2.7s, 5.1-5.5s and 8.2-8.6s. Compared with the single methods of multi-scale wavelet transform and improved adaptive Kalman filter, the multi-scale wavelet transform-improved adaptive Kalman filter combines the multi-resolution analysis capability of multi-scale wavelet transform and the dynamic filtering advantage of improved adaptive Kalman filter, achieving a better balance between feature preservation and noise suppression. Figure 6 This is further verified by the residual scatter distribution of the multi-scale wavelet transform-improved adaptive Kalman filter. The residual distribution of the multi-scale wavelet transform-improved adaptive Kalman filter is more concentrated and has a smaller amplitude, indicating that its noise reduction effect is more stable. Quantitative analysis shows that the average residual amplitude of the multi-scale wavelet transform-improved adaptive Kalman filter is 0.09 and 0.05 lower than that of the multi-scale wavelet transform and the improved adaptive Kalman filter, respectively, and the noise reduction effect is significantly improved.

[0212] like Figure 7The performance of the three methods in retaining characteristic peaks was compared. Among them, the peak error mean of the multi-scale wavelet transform-improved adaptive Kalman filter method was the lowest, which was 0.037, which was 28.85% and 17.78% lower than that of the multi-scale wavelet transform (0.052) and the improved adaptive Kalman filter (0.045), respectively. At the same time, the error standard deviation of the multi-scale wavelet transform-improved adaptive Kalman filter was also the smallest, only 0.008, indicating that it was more stable in processing different characteristic peaks. Figure 8 A comprehensive comparison of the performance of various methods in five key performance indicators was conducted. The multi-scale wavelet transform-improved adaptive Kalman filter outperformed the single method in terms of signal-to-noise ratio improvement (32.15%), feature retention (94.27%) and residual stability (92.35%), demonstrating its superiority in transformer winding signal processing; although it has slight disadvantages in algorithm complexity (68.24%) and computational time (79.15%), considering its significantly improved signal quality, the multi-scale wavelet transform-improved adaptive Kalman filter method is still an efficient and reliable signal processing solution, providing more accurate data support for transformer winding fault diagnosis.

[0213] like Figure 9 、 Figure 10 and Figure 11 The distribution of vibration, temperature, and electrical characteristic parameters of the transformer under normal and fault conditions is shown. Regarding vibration characteristics, the average vibration intensity in the fault state (3.72 m / s²) is significantly higher than that in the normal state (2.35 m / s²), an increase of 58.30%. Similarly, temperature characteristics also show significant differences, with the average temperature in the fault state (89.25°C) increasing by 13.63°C compared to the normal state (75.62°C). Among electrical characteristics, the average harmonic distortion rate in the fault state (3.42%) far exceeds that in the normal state (1.25%), an increase of 173.60%. These significant changes in characteristic parameters provide a reliable data foundation for fault diagnosis and highlight the advantages of the multi-scale wavelet transform and improved adaptive Kalman filter fusion framework in feature extraction.

[0214] like Figure 12The figure shows a comparison of the diagnostic performance of three methods: SVM, CNN, and multi-scale wavelet transform-improved adaptive Kalman filter for vibration, temperature, and electrical characteristics. The results show that the multi-scale wavelet transform-improved adaptive Kalman filter method performs best on all feature types, with average diagnostic accuracies of 97.23%, 94.68%, and 92.35%, respectively, which are 11.91% and 7.51% higher than the SVM and CNN methods, respectively. The fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter is particularly advantageous in vibration feature diagnosis, with an accuracy of up to 97.23%. This is attributed to the fact that this method combines the multi-resolution analysis capability of multi-scale wavelet transform with the dynamic filtering advantages of improved adaptive Kalman filter, which can more effectively extract and process the complex vibration signals of transformer windings. In addition, the fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter also maintains a high level of diagnostic accuracy for temperature and electrical characteristics, at 94.68% and 92.35%, respectively, demonstrating the stability and adaptability of the present invention in multi-type feature processing.

[0215] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for collecting data for transformer winding monitoring and diagnosis, characterized in that: include, Constructing a multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture, wherein the multi-scale wavelet transform and improved adaptive Kalman filter fusion architecture is used to obtain monitoring signals of transformer windings; Performing frequency division processing on the monitoring signal by using the multi-scale wavelet transform to determine each scale coefficient; Adjusting the noise covariance matrix based on the improved adaptive Kalman filter, and filtering the scale coefficients to determine the filtering result; Reconstructing the filtering result into a reconstructed signal, and performing signal quality evaluation on the reconstructed signal to determine the signal quality of the transformer winding; The monitoring signal is subjected to frequency division processing using the multi-scale wavelet transform. Determining each scale coefficient includes selecting a basis function that satisfies the vibration characteristics of the transformer winding and introducing an adaptive threshold function that considers the distribution characteristics of the winding temperature field. The specific formula is as follows: ； in, Representation scale The wavelet denoising threshold under ; Representation scale Standard deviation of wavelet coefficients; Indicates the number of wavelet coefficients at the current scale; represents the scale energy adjustment coefficient; Indicates the Layer wavelet coefficients norm; Indicates the The wavelet coefficients after layer filtering norm; represents the temperature-vibration coupling coefficient; Indicates the maximum temperature gradient of the winding; represents the temperature regulation index factor; Performing filtering on the scale coefficients to determine filtering results includes: The wavelet coefficients are used as observation values ​​of the improved adaptive Kalman filter; Performing filtering processing on the scale coefficients through filtering gain matrix calculation, state update and posterior error covariance update to obtain filtered scale coefficients; The specific formulas for the filter gain matrix calculation, state update, and posterior error covariance update are as follows: ； in, Indicates time The Kalman gain matrix of Represents the state prediction error covariance matrix before updating; represents the observation matrix; represents the observation noise covariance matrix.

2. The transformer winding monitoring and diagnostic data acquisition method according to claim 1, wherein: The noise covariance matrix is ​​adjusted based on the improved adaptive Kalman filter, comprising: Construct the state prediction equation and observation equation. The specific formulas are as follows: ； ； Determine the state vector including displacement, velocity, elastic modulus and damping coefficient. The specific formula is as follows: ； The noise covariance matrix is ​​updated based on the prediction equation and the observation equation. The specific formula is as follows: ； ； in, Indicates time The predicted value of the state; Indicates time The estimated value of the state; Indicates displacement; represents the estimated value of the state at the current moment k; Indicates the last moment The predicted value of the state; Indicates speed; represents the elastic modulus; represents the damping coefficient; represents the state transition matrix; represents the control input matrix; represents process noise; Represents the observation value at the current moment; represents the observation matrix; represents the measurement noise; Indicates time The process noise covariance matrix of ; Indicates time The process noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; Indicates time The measurement noise covariance matrix of ; represents the forgetting factor; represents the measurement noise update forgetting factor; represents the temperature gradient coupling coefficient; represents the square norm of the temperature gradient; Represents the identity matrix.

3. The transformer winding monitoring and diagnostic data acquisition method according to claim 2, wherein: The determining of the signal quality of the transformer winding includes: Reconstructing the scale coefficients after the filtering process to obtain a reconstructed signal; extracting transformer winding state characteristics of the reconstructed signal, and determining a signal quality evaluation result of the reconstructed signal according to the transformer winding state characteristics; Feedback optimization is performed on the feature extraction error according to the signal quality evaluation result of the reconstructed signal.

4. The transformer winding monitoring and diagnostic data acquisition method according to claim 3, wherein: Determining a signal quality evaluation result of the reconstructed signal includes: Based on the thermal-mechanical coupling effect model, the fault feature vector and energy feature of the reconstructed signal are constructed; In combination with the fault feature vector and energy feature, a signal quality evaluation function based on a thermal-mechanical coupling model is introduced to evaluate the signal quality of the reconstructed signal, thereby obtaining a signal quality evaluation result of the reconstructed signal.

5. The transformer winding monitoring and diagnostic data acquisition method according to claim 4, characterized in that: The specific formula of the signal quality evaluation function is as follows: ； in, Indicates the current time Signal quality assessment factor; Represents the observation value at the current moment; Indicates time The predicted value of the state; represents the observation matrix; represents the second norm of the state prediction error; represents the trace of the observation noise covariance matrix; represents the stress gradient tensor obtained based on the thermal-mechanical coupling model; Represents the stress gradient weight coefficient.

6. A transformer winding monitoring and diagnostic data acquisition system, applying the transformer winding monitoring and diagnostic data acquisition method according to any one of claims 1 to 5, characterized in that: include, A signal acquisition module is used to construct a fusion architecture of multi-scale wavelet transform and improved adaptive Kalman filter, which is used to obtain monitoring signals of transformer windings; A multi-scale wavelet transform module is used to perform frequency division processing on the monitoring signal using a multi-scale wavelet transform and determine each scale coefficient. The frequency division processing on the monitoring signal using a multi-scale wavelet transform and the determination of each scale coefficient include selecting a basis function that satisfies the vibration characteristics of the transformer winding and introducing an adaptive threshold function that considers the distribution characteristics of the winding temperature field; An improved adaptive Kalman filter module is used to adjust the noise covariance matrix based on the improved adaptive Kalman filter, and perform filtering on each scale coefficient to determine the filtering result; Perform filtering on each scale coefficient to determine the filtering result, including: The wavelet coefficients are used as observation values ​​of the improved adaptive Kalman filter; Performing filtering processing on the scale coefficients through filtering gain matrix calculation, state update and posterior error covariance update to obtain filtered scale coefficients; The feature fusion diagnosis module is used to reconstruct the filtering result into a reconstructed signal, and perform signal quality evaluation on the reconstructed signal to determine the signal quality of the transformer winding.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the transformer winding monitoring and diagnostic data acquisition method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the transformer winding monitoring and diagnostic data acquisition method according to any one of claims 1 to 5 are implemented.

Images