High-precision dynamic calibration system and calibration method for ultrasonic partial discharge instrument

Abstract

The invention discloses a high-precision dynamic calibration system and calibration method for an ultrasonic partial discharge instrument, and relates to the technical field of dynamic calibration of the ultrasonic partial discharge instrument, and the system comprises a collection and enhancement module which is used for deploying an ultrasonic sensor in a target area to collect a partial discharge signal and carrying out the denoising and enhancement of the partial discharge signal; the feature extraction module is used for extracting features according to the denoised and enhanced signals; the matrix and calculation module is used for constructing a space-time incidence matrix based on the extracted features and dynamically calculating node weights; the error calibration and verification module is used for constructing an error model according to the node weight to carry out error calibration and verification; according to the method, the anti-noise capability is remarkably enhanced through a denoising method combining sparse decomposition and a sparse dictionary matrix, global and local features of the signals are extracted through Henkel matrix construction and singular value decomposition, and high-quality input is provided for the follow-up process through optimization reconstruction and standardized fusion of the signals; and high-precision calibration of the partial discharge signal in a complex environment is realized by combining the dynamically calculated node weight and the error calibration model.

Description

Technical Field

[0001] The invention relates to the technical field of dynamic calibration of ultrasonic partial discharge instruments, and in particular to a high-precision dynamic calibration system and a calibration method for ultrasonic partial discharge instruments. Background Art

[0002] With the increasing complexity of the operating environment of power system equipment and the increasing demand for high-precision monitoring, partial discharge (PD) signal detection technology has become an important means to ensure the reliability and safety of equipment. The detection and analysis of PD signals can effectively locate potential defects inside the equipment and prevent failures. It plays an indispensable role in power equipment status monitoring, predictive maintenance and fault diagnosis. At present, ultrasonic sensors are widely used in the detection of PD signals due to their high sensitivity, non-contact detection capabilities and adaptability to complex electromagnetic interference environments.

[0003] The existing calibration systems generally collect low-quality signals in noisy industrial environments, resulting in poor adaptability to strong noise and nonlinear interference, and difficulty in retaining the key characteristics of the signal. Secondly, signal enhancement technology mostly focuses on single optimization processing in specific frequency bands, ignoring the enhancement of full-band signal characteristics and the comprehensive utilization of multi-dimensional information. Summary of the invention

[0004] In view of the problems existing in the above-mentioned existing high-precision dynamic calibration system and calibration method of ultrasonic partial discharge instrument, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that the quality of signals collected by existing calibration systems in noisy industrial environments is generally low, resulting in poor adaptability to strong noise and nonlinear interference, and difficulty in retaining the key characteristics of the signal. Secondly, signal enhancement technology mostly focuses on single optimization processing of specific frequency bands, ignoring the enhancement of full-band signal characteristics and the comprehensive utilization of multi-dimensional information.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a high-precision dynamic calibration system for ultrasonic partial discharge instrument, comprising:

[0007] The acquisition and enhancement module deploys ultrasonic sensors in the target area to collect partial discharge signals for denoising and enhancement;

[0008] A feature extraction module extracts features based on the denoised and enhanced signals;

[0009] The matrix and calculation module constructs the spatiotemporal correlation matrix based on the extracted features and dynamically calculates the node weights;

[0010] The error calibration and verification module builds an error model based on the node weights to perform error calibration and verification;

[0011] The storage and display module stores the verification results and displays them through a visual interface.

[0012] As a preferred solution of the high-precision dynamic calibration system of ultrasonic partial discharge instrument of the present invention, the method of deploying ultrasonic sensors in the target area to collect partial discharge signals for denoising and enhancement includes:

[0013] By using GIS tools to draw the boundaries of the area and using a hierarchical grid topology to segment the regular grid, ultrasonic sensors are installed at the center of the grid according to the segmented grid, and the installation coordinates are recorded;

[0014] Based on the installed ultrasonic sensors, the partial discharge signals within the zone boundaries are captured;

[0015] The captured partial discharge signal is converted into a frequency domain signal using fast Fourier transform, and the main frequency band in the frequency domain signal is intercepted and enhanced by using frequency band interception technology;

[0016] The intercepted frequency domain signal is converted into a time domain signal by using inverse Fourier transform to obtain an enhanced signal;

[0017] Construct an M×K sparse dictionary matrix D:

[0018]

[0019] Where M represents the number of signal sampling points, K represents the total number of frequency components in the dictionary, sin(2πf e t) and cos(2πf e t) represent the sine and cosine functions respectively, f e represents the e-th frequency component, and t represents the time sampling point;

[0020] Based on the sparse dictionary matrix D, the enhanced signal is sparsely decomposed and denoised:

[0021] X'=Dε+∈,

[0022] In the formula, X' represents the enhanced signal, ε represents the sparse coefficient, and ∈ represents the residual;

[0023] The sparse decomposition is solved by the l1 norm optimization problem, and the signal is reconstructed by the solved sparse coefficient ε and the sparse dictionary matrix D to obtain the sparse reconstructed signal X1(t):

[0024] X1(t)=Dε

[0025] The sparse reconstructed signal X1(t) is converted into the Henkel matrix H using the Henkel matrix construction method:

[0026]

[0027] m+g-1=M,

[0028] In the formula, m represents the number of rows of the Henkel matrix, g represents the number of columns of the Henkel matrix, and M represents the number of sampling points of the signal;

[0029] Based on the transformed Henkel matrix, the singular value decomposition method is used to decompose the Henkel matrix to obtain the singular value vectors and singular values:

[0030] H=U∈G T ,

[0031] Where U represents the left singular vector, ∈ represents the diagonal matrix, G represents the right singular vector, T represents the transpose operation, and H represents the Henkel matrix;

[0032] By extracting singular values ​​from the diagonal matrix and using the PMI index to perform weighted screening on the singular values;

[0033] Set the screening threshold to 0, compare the extracted singular value with the threshold O, and when the extracted singular value is greater than or equal to the threshold O, retain the current singular value, otherwise remove the singular value;

[0034] Based on the retained singular values, calculate the energy contribution of the singular values:

[0035]

[0036] Where E(v) represents the energy contribution of the vth singular value, and They represent the vth and oth singular values ​​after Henkel matrix decomposition, respectively, and k represents the total number of singular values;

[0037] By using the matrix indexing method to extract the left singular vector U corresponding to the vth singular value from the left singular vector and the right singular vector v and the right singular vectors

[0038] The energy contribution is defined as the weight E(v) = w v ;

[0039] Combine the extracted singular vectors and weights to optimize the signal reconstruction:

[0040]

[0041] Where X2(t) represents the optimized reconstructed signal, S represents the singular value index set of the screening condition, and w v represents the weight of the vth singular value, U v and G v They represent the left singular vector and right singular vector corresponding to the vth singular value respectively;

[0042] The sparse reconstructed signal X1(t) and the optimized reconstructed signal X2(t) are normalized by using a normalization method to obtain normalized signals X'1(t) and X'2(t);

[0043] The normalized signals X'1(t) and X'2(t) are combined by a linear weighting method to obtain the final signal Z:

[0044] Z=αX′1(t)+(1-α)X′2(t),

[0045] In the formula, α represents the fusion weight of the coefficient reconstruction signal, and (1-α) represents the fusion weight of the optimized reconstructed signal.

[0046] As a preferred solution of the ultrasonic partial discharge instrument high-precision dynamic calibration system of the present invention, wherein: the feature extraction and denoising based on the denoised and enhanced signal includes:

[0047] Based on the final signal, use discrete wavelet transform to decompose the multi-layer wavelet coefficients:

[0048]

[0049] In the formula, C(a,b) represents the decomposed wavelet coefficient, φ * represents the wavelet basis function, a represents the scale parameter, b represents the translation parameter, and s represents the time variable;

[0050] Based on the obtained wavelet coefficients C(a,b), the wavelet component band energy characteristics are calculated using fast Fourier transform:

[0051]

[0052] Where Vz represents the energy characteristic of the zth frequency band, Brec(f) represents the spectrum of the signal, fz and fz +1 Indicates the start and end frequencies of the zth frequency band;

[0053] By extracting the spectrum from the energy feature and transforming the spectrum using Hilbert transform, the imaginary part is obtained and then combined with the real part to calculate the signal envelope:

[0054]

[0055] Where K(t) represents the signal envelope and β represents the Hilbert transform;

[0056] Use the normalization method to normalize the signal envelope.

[0057] As a preferred solution of the ultrasonic partial discharge instrument high-precision dynamic calibration system of the present invention, wherein: the construction of the spatiotemporal correlation matrix based on the extracted features and the dynamic calculation of the node weights include:

[0058] The normalized signal envelope was analyzed by using the Pearson product-moment correlation coefficient and the k ,t k+1 ], calculate the mean and standard deviation of the signal envelope;

[0059] Calculate covariance and variance from mean and standard deviation;

[0060] The signal envelope correlation value between ultrasonic sensor node i and ultrasonic sensor node j is further calculated by the obtained covariance and variance:

[0061]

[0062] Where M ij (t) represents the spatiotemporal correlation value of ultrasonic sensor node i and node j at time t, Cov represents covariance, Var represents variance, N i (t) and N j (t) represents the normalized signal envelope of ultrasonic sensor nodes i and j;

[0063] The distance is calculated based on the position of the ultrasonic sensor node, and the spatiotemporal correlation value is corrected using the exponential decay formula:

[0064]

[0065] Where, d ij represents the geometric distance between ultrasonic sensor node i and node j, x i ,y i , z i They represent the coordinates of ultrasonic sensor node i in three-dimensional space, x j ,y j , z j Respectively represent the coordinates of ultrasonic sensor node j in three-dimensional space, M' ij (t) represents the corrected spatiotemporal correlation value, and d0 represents the reference distance;

[0066] Based on the modified spatiotemporal correlation values, the n×n spatiotemporal correlation matrix M(t) is constructed using matrix operation methods:

[0067]

[0068] In the formula, n represents the total number of nodes;

[0069] Based on the spatiotemporal correlation matrix M(t), the weight of each ultrasonic sensor node is calculated using the weighted average method;

[0070] The weights are normalized by using a normalization method.

[0071] As a preferred solution of the high-precision dynamic calibration system of ultrasonic partial discharge instrument of the present invention, wherein: the error calibration and verification of error by constructing an error model according to node weights includes:

[0072] Based on the ultrasonic sensor receiving signal time minus the signal transmission time, the signal propagation time difference ΔT is obtained;

[0073] Multiply the obtained time difference by the signal propagation speed to obtain the measured pseudorange d mes ;

[0074] The distance between the ultrasonic sensor nodes is converted into a true pseudo-range d by using differential correction technology. tru ;

[0075] The time difference ΔT and the measured pseudorange d mes and the true pseudorange d tru As input, and construct the error model:

[0076]

[0077] In the formula, Δd i represents the pseudorange correction error of node i, represents the measured pseudorange of node i, represents the true pseudorange of node i, γ i represents the time deviation correction coefficient of node i, ΔT represents the signal propagation time difference;

[0078] Define the adaptive objective function:

[0079]

[0080] In the formula, A represents the optimization objective function, θ represents the parameter set of the model, and w' i (t) represents the dynamic weight of node i after normalization, |Δd i | represents the absolute value of the pseudorange correction error of node i, and n represents the total number of nodes;

[0081] The nonlinear features are processed by the support vector regression algorithm in combination with the radial basis kernel function, and the kernel function parameters and regularization parameters are optimized by the grid search method. The parameters are continuously updated using the gradient descent method. When the loss of the error model does not decrease significantly during the continuous iteration process, the iteration is stopped, the final error model is output, and the optimization is completed.

[0082] By inputting the correction error into the error model, and outputting the final error result through the error model;

[0083] Based on the final error result obtained, calculate the root mean square error;

[0084] The positioning algorithm is combined with pseudo-range to calculate the measured coordinates, and the three-dimensional positioning error is calculated by further combining the measured coordinates with the true coordinates;

[0085] The root mean square error, three-dimensional positioning error and propagation time deviation are standardized by using the standardization method, and the standardized root mean square error, three-dimensional positioning error and propagation time deviation are integrated by the weight coefficient synthesis method:

[0086]

[0087] In the formula, S represents the comprehensive deviation, ρ represents the weight coefficient, RMS' represents the normalized root mean square error, Δh' represents the normalized three-dimensional positioning error, and |ΔT'| represents the normalized propagation time deviation;

[0088] Set the error threshold to Q, and compare the comprehensive deviation S with the threshold Q. When the comprehensive deviation S is less than or equal to the threshold Q, the verification passes and the error meets the system requirements. When the comprehensive deviation S is greater than the threshold Q, the verification fails and re-optimization is performed.

[0089] As a preferred solution of the high-precision dynamic calibration system of the ultrasonic partial discharge instrument described in the present invention, the storage based on the verification results refers to storing data based on the verification results in a unified JSON data format, and adopting a hierarchical storage structure during the storage process, distinguishing data files by timestamp and task batch, and recording detailed information of all nodes in the data file of each task.

[0090] As a preferred solution of the high-precision dynamic calibration system of the ultrasonic partial discharge instrument of the present invention, the display through a visual interface refers to using the Matplotlib data visualization tool to draw a trend graph of the standardized error term and a histogram of the comprehensive deviation S, and generating a three-dimensional distribution graph by combining the node position data, and the color of each point is gradually marked according to the size of the positioning error;

[0091] The standardized error term includes a standardized root mean square error, a three-dimensional positioning error, and a propagation time deviation.

[0092] Another object of the present invention is to provide a high-precision dynamic calibration method for an ultrasonic partial discharge instrument, which comprises:

[0093] Ultrasonic sensors are deployed in the target area to collect partial discharge signals for denoising and enhancement, and feature extraction is performed based on the denoised and enhanced signals;

[0094] Based on the extracted features, a spatiotemporal correlation matrix is ​​constructed and node weights are dynamically calculated. The error model is constructed according to the node weights for error calibration and verification.

[0095] The verification results are stored and displayed through a visual interface.

[0096] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of a high-precision dynamic calibration system of an ultrasonic partial discharge instrument when executing the computer program.

[0097] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a high-precision dynamic calibration system for an ultrasonic partial discharge instrument.

[0098] The beneficial effects of the present invention are as follows: the present invention uses a denoising method combining sparse decomposition with a sparse dictionary matrix to effectively separate noise from useful signals, significantly enhance the anti-noise capability, extract global and local features of signals through Henkel matrix construction and singular value decomposition, and combine the cumulative energy curve method and the PMI index to screen key singular values, thereby realizing the comprehensive utilization of multi-dimensional information, providing high-quality input for the calibration system through optimized reconstruction and standardized fusion of signals, and combining dynamically calculated node weights and error calibration models to realize high-precision dynamic calibration of partial discharge signals in complex environments, thereby effectively improving the reliability and adaptability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0100] Figure 1 This is a schematic diagram of the structure of the high-precision dynamic calibration system of the ultrasonic partial discharge instrument.

[0101] Figure 2 This is a flow chart of a high-precision dynamic calibration method for an ultrasonic partial discharge instrument. DETAILED DESCRIPTION

[0102] 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 in conjunction with the accompanying drawings.

[0103] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and 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.

[0104] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0105] Example 1

[0106] Reference Figure 1 , which is the first embodiment of the present invention, and which provides a high-precision dynamic calibration system for ultrasonic partial discharge instrument. The high-precision dynamic calibration system for ultrasonic partial discharge instrument includes the following modules:

[0107] S1, acquisition and enhancement module, deploys ultrasonic sensors in the target area to collect partial discharge signals for denoising and enhancement;

[0108] Specifically, ultrasonic sensors are deployed in the target area to collect partial discharge signals for denoising and enhancement, including:

[0109] By using GIS tools to draw the boundaries of the area and using a hierarchical grid topology to segment the regular grid, ultrasonic sensors are installed at the center of the grid according to the segmented grid, and the installation coordinates are recorded;

[0110] Based on the installed ultrasonic sensors, the partial discharge signals within the zone boundaries are captured;

[0111] The captured partial discharge signal is converted into a frequency domain signal using fast Fourier transform, and the main frequency band in the frequency domain signal is intercepted and enhanced by using frequency band interception technology;

[0112] The intercepted frequency domain signal is converted into a time domain signal by using inverse Fourier transform to obtain an enhanced signal;

[0113] Construct an M×K sparse dictionary matrix D:

[0114]

[0115] Where M represents the number of signal sampling points, K represents the total number of frequency components in the dictionary, sin(2πf e t) and cos(2πf e t) represent the sine and cosine functions respectively, f e represents the e-th frequency component, and t represents the time sampling point;

[0116] Based on the sparse dictionary matrix D, the enhanced signal is sparsely decomposed and denoised:

[0117] X'=Dε+∈,

[0118] In the formula, X' represents the enhanced signal, ε represents the sparse coefficient, and ∈ represents the residual;

[0119] The sparse coefficient ε and residual ∈ can be obtained through the sparse representation algorithm;

[0120] The sparse decomposition is solved by the l1 norm optimization problem, and the signal is reconstructed by the solved sparse coefficient ε and the sparse dictionary matrix D to obtain the sparse reconstructed signal X1(t):

[0121] X1(t)=Dε

[0122] The sparse reconstructed signal X1(t) is converted into the Henkel matrix H using the Henkel matrix construction method:

[0123]

[0124] m+g-1=M,

[0125] In the formula, m represents the number of rows of the Henkel matrix, g represents the number of columns of the Henkel matrix, and M represents the number of sampling points of the signal;

[0126] Based on the transformed Henkel matrix, the singular value decomposition method is used to decompose the Henkel matrix to obtain the singular value vectors and singular values:

[0127] H=U∈G T ,

[0128] Where U represents the left singular vector, ∈ represents the diagonal matrix, G represents the right singular vector, T represents the transpose operation, and H represents the Henkel matrix;

[0129] By extracting singular values ​​from the diagonal matrix and using the PMI index to perform weighted screening on the singular values;

[0130] According to the analysis of the cumulative energy curve method, the screening threshold is set to 0, and the extracted singular value is compared with the threshold O. When the extracted singular value is greater than or equal to the threshold O, the current singular value is retained, otherwise the singular value is eliminated;

[0131] Based on the retained singular values, calculate the energy contribution of the singular values:

[0132]

[0133] Where E(v) represents the energy contribution of the vth singular value, and They represent the vth and oth singular values ​​after Henkel matrix decomposition, respectively, and k represents the total number of singular values;

[0134] By using the matrix indexing method to extract the left singular vector U corresponding to the vth singular value from the left singular vector and the right singular vector v and the right singular vectors

[0135] The energy contribution is defined as the weight E(v) = w v ;

[0136] Combine the extracted singular vectors and weights to optimize the signal reconstruction:

[0137]

[0138] Where X2(t) represents the optimized reconstructed signal, S represents the singular value index set of the screening condition, and w v represents the weight of the vth singular value, U v and G v They represent the left singular vector and right singular vector corresponding to the vth singular value respectively;

[0139] The sparse reconstructed signal X1(t) and the optimized reconstructed signal X2(t) are normalized by using a normalization method to obtain normalized signals X'1(t) and X'2(t);

[0140] The normalized signals X'1(t) and X'2(t) are combined by a linear weighting method to obtain the final signal Z:

[0141] Z=αX′1(t)+(1-α)X′2(t),

[0142] In the formula, α represents the fusion weight of the coefficient reconstruction signal, and (1-α) represents the fusion weight of the optimized reconstructed signal.

[0143] By deploying ultrasonic sensors in the target area to collect partial discharge signals, real-time monitoring of the equipment status can be achieved. The regional boundaries are drawn using GIS tools, and regular grids are divided through a hierarchical grid topology structure to ensure that the sensors are evenly arranged and the coverage area is complete. This sensor deployment method can effectively improve the spatial resolution of signal acquisition, reduce the blind spots of partial discharge signals, and provide high-quality original data for subsequent signal processing and analysis. The partial discharge signals are converted into frequency domain signals through fast Fourier transform and frequency band interception is performed, which can effectively filter out unnecessary frequency components and enhance the signals in the main frequency band, so that the present invention can remove irrelevant components such as environmental noise and nonlinear interference, thereby improving the signal-to-noise ratio of the signal. The enhanced signal is sparsely decomposed using a sparse dictionary matrix, and the signal can be represented as a linear combination of a few non-zero coefficients, which significantly improves the denoising effect. The sparse coefficients and residuals obtained through the sparse representation algorithm can help extract clear signal components from the noise. After the sparse reconstructed signal is converted into a Henckel matrix, the main features of the signal are extracted through singular value decomposition, which can effectively reveal the potential structure in the signal. Singular value decomposition extracts the principal components (i.e., singular values) of the signal, which helps to model and analyze the dynamic changes of the signal. Subsequently, by weighted screening of singular values ​​and calculating their energy contributions, redundant information can be removed and important signal components can be retained, thereby improving the accuracy of signal optimization reconstruction. Signal reconstruction based on the extracted singular vectors and energy contributions combined with optimized weights of the reconstructed signal can remove unnecessary noise while retaining key information in the final signal. In addition, the present invention processes the signal using a standardized method and performs linear weighted fusion to ensure the consistency and reliability of the final reconstructed signal under different parameters and environmental conditions. By optimizing the reconstruction process, the quality of the signal is greatly improved, which can provide high-precision data support for further dynamic calibration and equipment fault diagnosis.

[0144] S2, feature extraction module, extracts features based on the denoised and enhanced signals;

[0145] Specifically, feature extraction and denoising based on denoised and enhanced signals include:

[0146] Based on the final signal, use discrete wavelet transform to decompose the multi-layer wavelet coefficients:

[0147]

[0148] In the formula, C(a,b) represents the decomposed wavelet coefficient, φ * represents the wavelet basis function, a represents the scale parameter, b represents the translation parameter, and s represents the time variable;

[0149] Based on the obtained wavelet coefficients C(a,b), the wavelet component band energy characteristics are calculated using fast Fourier transform:

[0150]

[0151] Where V z represents the energy characteristics of the zth frequency band, B rec (f) represents the frequency spectrum of the signal, f z and f z+1 Indicates the start and end frequencies of the zth frequency band;

[0152] By extracting the spectrum from the energy feature and transforming the spectrum using Hilbert transform, the imaginary part is obtained and then combined with the real part to calculate the signal envelope:

[0153]

[0154] Where K(t) represents the signal envelope and β represents the Hilbert transform;

[0155] Use the normalization method to normalize the signal envelope.

[0156] By using discrete wavelet transform to perform multi-level time-frequency decomposition of the signal, the high-frequency noise in the signal can be effectively separated from the low-frequency signal. By using fast Fourier transform to calculate the frequency band energy characteristics of the wavelet component, the spectral distribution of the signal can be further analyzed, and the energy concentration area of ​​the main frequency band can be identified. Then, by extracting the frequency band energy characteristics, the present invention can accurately locate the key frequency range of the partial discharge signal, improve the analysis efficiency of the signal, and provide more accurate signal characteristics for subsequent calibration and positioning. The imaginary part of the spectrum is extracted using Hilbert transform, and combined with the real part to calculate the envelope of the signal. This step reveals the amplitude change trend of the signal, can effectively remove the influence of noise on the signal, and make the main characteristics of the signal more prominent. In the final stage of signal processing, the signal envelope is standardized by the normalization method to ensure that the various characteristics of the signal are compared at the same scale, eliminating the influence of signal amplitude differences, and making the comparison between different signals more unified and reliable.

[0157] S3, the matrix and calculation module, builds the spatiotemporal correlation matrix based on the extracted features and dynamically calculates the node weights;

[0158] Specifically, the spatiotemporal correlation matrix is ​​constructed based on the extracted features and the node weights are dynamically calculated, including:

[0159] The normalized signal envelope was analyzed by using the Pearson product-moment correlation coefficient and the k ,t k+1 ], calculate the mean and standard deviation of the signal envelope:

[0160]

[0161] In the formula, μ represents the mean value of the signal envelope, T represents the total duration of the time window, and t k and t k+1 represents the start and end time of the kth time window, σ represents the standard deviation of the signal envelope, and N(t) represents the normalized signal envelope;

[0162] Calculate the covariance and variance from the mean and standard deviation:

[0163]

[0164] In the formula, Cov[N i (t),N j (t)] represents the signal envelope N i (t) and N j (t) Covariance within the time window, μ i and μ j Represents the signal envelope N i (t) and N j (t) is the mean value in the time window, N i (t) and N j (t) represent the normalized signal envelopes of ultrasonic sensor nodes i and j, Var[N i (t)] represents the signal envelope N i The variance of (t);

[0165] The signal envelope correlation value between ultrasonic sensor node i and ultrasonic sensor node j is further calculated by the obtained covariance and variance:

[0166]

[0167] Where M ij (t) represents the spatiotemporal correlation value of ultrasonic sensor node i and node j at time t, Cov represents covariance, Var represents variance, N i (t) and N j (t) represents the normalized signal envelope of ultrasonic sensor nodes i and j;

[0168] The distance is calculated based on the position of the ultrasonic sensor node, and the spatiotemporal correlation value is corrected using the exponential decay formula:

[0169]

[0170] Where, d ij represents the geometric distance between ultrasonic sensor node i and node j, x i ,y i , z iThey represent the coordinates of ultrasonic sensor node i in three-dimensional space, x j ,y j , z j Respectively represent the coordinates of ultrasonic sensor node j in three-dimensional space, M' ij (t) represents the corrected spatiotemporal correlation value, and d0 represents the reference distance;

[0171] The reference distance d0 of the corresponding frequency signal can be obtained by consulting relevant literature;

[0172] Based on the modified spatiotemporal correlation values, the n×n spatiotemporal correlation matrix M(t) is constructed using matrix operation methods:

[0173]

[0174] In the formula, n represents the total number of nodes;

[0175] Based on the spatiotemporal correlation matrix M(t), the weight of each ultrasonic sensor node is calculated using the weighted average method:

[0176]

[0177] In the formula, p i (t) represents the dynamic weight of ultrasonic sensor node i at time t, and n represents the total number of nodes;

[0178] The weights are normalized by using a normalization method.

[0179] By using the Pearson product-moment correlation coefficient to analyze the normalized signal envelope, the present invention can efficiently quantify the correlation of signal envelopes between different ultrasonic sensor nodes, which provides a key input for the construction of the spatiotemporal correlation matrix, enables the signal correlation between nodes to be quantitatively described, and improves the accuracy and reliability of signal analysis. By calculating the mean, standard deviation and covariance of the signal envelope, it provides necessary data support for further analysis of the spatiotemporal relationship and dynamic weight between nodes, and calculating the covariance and variance of the signal envelope helps to quantify the signal similarity and distribution characteristics between different sensors. The covariance measures the linear relationship between two signals, while the variance reflects the fluctuation amplitude of the signal. In the construction of the spatiotemporal correlation matrix, the calculation of covariance and variance helps capture the correlation between nodes and their changes within the time window, further improving the characterization of signal characteristics. When constructing the spatiotemporal correlation matrix, based on the geometric distance between nodes, the covariance and variance are used. The exponential decay formula is used to correct the spatiotemporal correlation value, which can effectively take into account the attenuation effect in the signal propagation process. Through the matrix operation method, combined with the correlation analysis of covariance, variance, signal envelope and distance attenuation, the spatiotemporal correlation matrix is ​​finally constructed, so that the matrix of the present invention not only captures the spatiotemporal relationship between sensor nodes, but also takes into account the physical position and signal propagation characteristics between nodes, laying the foundation for subsequent dynamic calibration and weight calculation. Based on the spatiotemporal correlation matrix, the weighted average method is used to calculate the dynamic weight of each ultrasonic sensor node, ensuring that the key nodes of the signal can get more weight in the entire system, thereby improving the key focus area in the signal processing process, and the dynamic weight calculation not only makes the system more refined, but also enhances the sensitivity to specific areas or nodes. In the calculation process, the normalization method is used to standardize the signal envelope and node weights, ensuring the comparability of data between different signals and nodes.

[0180] S4, error calibration and verification module, constructs an error model according to the node weights to perform error calibration and verification;

[0181] Specifically, error calibration and verification are performed by constructing an error model based on node weights, including:

[0182] Based on the ultrasonic sensor receiving signal time minus the signal transmission time, the signal propagation time difference ΔT is obtained;

[0183] Multiply the obtained time difference by the signal propagation speed to obtain the measured pseudorange d mes ;

[0184] The distance between the ultrasonic sensor nodes is converted into a true pseudo-range d by using differential correction technology. tru ;

[0185] The time difference ΔT and the measured pseudorange d mesand the true pseudorange d tru As input, and construct the error model:

[0186]

[0187] In the formula, Δd i represents the pseudorange correction error of node i, represents the measured pseudorange of node i, represents the true pseudorange of node i, γ i represents the time deviation correction coefficient of node i, ΔT represents the signal propagation time difference;

[0188] The time deviation correction coefficient γ of node i can be obtained by the signal-to-noise ratio measurement method. i ;

[0189] Define the adaptive objective function:

[0190]

[0191] In the formula, A represents the optimization objective function, θ represents the parameter set of the model, and w' i (t) represents the dynamic weight of node i after normalization, |Δd i | represents the absolute value of the pseudorange correction error of node i, and n represents the total number of nodes;

[0192] The nonlinear features are processed by the support vector regression algorithm in combination with the radial basis kernel function, and the kernel function parameters and regularization parameters are optimized by the grid search method. The parameters are continuously updated using the gradient descent method. When the loss of the error model does not decrease significantly during the continuous iteration process, the iteration is stopped, the final error model is output, and the optimization is completed.

[0193] By inputting the correction error into the error model, and outputting the final error result through the error model;

[0194] Based on the final error result obtained, calculate the root mean square error:

[0195]

[0196] In the formula, RMS represents the root mean square error, n represents the total number of nodes, (Δd i ) 2 The square of the corrected error of node i;

[0197] Use the positioning algorithm combined with pseudorange to calculate the measured coordinates, and further combine the measured coordinates with the true coordinates to calculate the three-dimensional positioning error:

[0198]

[0199] In the formula, Δh represents the three-dimensional positioning error, xrue ,y rue , z rue represents the real coordinates, x eas ,y eas , z eas Indicates the measurement coordinates;

[0200] The root mean square error, three-dimensional positioning error and propagation time deviation are standardized by using the standardization method, and the standardized root mean square error, three-dimensional positioning error and propagation time deviation are integrated by the weight coefficient synthesis method:

[0201]

[0202] In the formula, S represents the comprehensive deviation, ρ represents the weight coefficient, RMS' represents the normalized root mean square error, Δh' represents the normalized three-dimensional positioning error, and |ΔT'| represents the normalized propagation time deviation;

[0203] The comprehensive deviation S can be fitted to the formula by using multiple linear regression to minimize the fitting error and determine the weight coefficient ρ;

[0204] According to the system performance requirements, the error threshold is set to Q, and the comprehensive deviation S is compared with the threshold Q. When the comprehensive deviation S is less than or equal to the threshold Q, the verification passes and the error meets the system requirements. When the comprehensive deviation S is greater than the threshold Q, the verification fails and re-optimization is performed.

[0205] By calculating the time difference of the ultrasonic sensor receiving the signal, the preliminary pseudo-range is obtained, thereby providing a basic distance estimation for the subsequent calibration, and the position difference between multiple ultrasonic sensor nodes is corrected by the differential correction technology, which can effectively reduce the pseudo-range error caused by environmental factors or the distance between sensors. In the error model, the Pearson product-moment correlation coefficient is used to analyze the normalized signal envelope, calculate the covariance and variance, and then obtain the spatiotemporal correlation value, so that the present invention can construct the relationship matrix between each sensor node through this step, and accurately quantify the signal correlation between nodes, and use the exponential decay formula to correct the spatiotemporal correlation value to ensure the spatiotemporal The correlation matrix can truly reflect the signal strength and correlation between sensors, and through the support vector regression algorithm, combined with the radial basis kernel function, the nonlinear error characteristics in the pseudorange measurement are effectively processed. Under the use of root mean square error and three-dimensional positioning error indicators, combined with normalization and weight coefficient calculation, the positioning accuracy of the system is comprehensively evaluated. Then, the comprehensive deviation is fitted through multivariate linear regression, the error is minimized and the optimal weight coefficient is determined, thereby optimizing the system performance. By setting the error threshold and comparing it with the actual error result, it is determined whether the system meets the accuracy requirements. When the error exceeds the set threshold, the system will automatically optimize to ensure that the positioning accuracy meets the expected requirements.

[0206] S5, storage and display module, stores the verification results and displays them through a visual interface;

[0207] Specifically, storage based on verification results means storing data in a unified JSON data format based on the verification results, and adopting a hierarchical storage structure during the storage process, distinguishing data files by timestamp and task batch, and recording detailed information of all nodes in the data file of each task.

[0208] By adopting a unified JSON data format for storage, it not only helps to ensure the consistency and standardization of the data format, but also enhances the cross-platform compatibility of the system. By adopting a hierarchical storage structure and distinguishing data by timestamp and task batch, a large number of data files can be effectively managed. Using timestamps as the basis for distinguishing data can ensure that data is stored in chronological order. The introduction of task batches makes data storage and management more orderly and systematic. Due to the use of a hierarchical storage structure and JSON format, the data storage system of the present invention has good scalability and flexibility. As the amount of tasks increases, new task batches can be easily added to the storage system without affecting the existing data structure and storage method.

[0209] Furthermore, displaying through a visual interface means using the Matplotlib data visualization tool to draw a trend graph of the standardized error term and a bar graph of the comprehensive deviation S, and generating a three-dimensional distribution graph by combining the node position data, and the color of each point is gradually marked according to the size of the positioning error;

[0210] The standardized error term includes a standardized root mean square error, a three-dimensional positioning error, and a propagation time deviation.

[0211] By using Matplotlib to draw a trend graph of the standardized error term and a histogram of the comprehensive deviation S, these graphs can show the changing trends of the positioning error and the deviation. Through the visual display of the standardized error term, the present invention can eliminate the dimensional differences between different error sources (such as root mean square error, three-dimensional positioning error and propagation time deviation), so that various error terms can be compared at the same scale, and the three-dimensional visualization display method can intuitively reveal which areas or nodes in the spatial distribution have larger positioning errors, thereby providing optimization directions. Moreover, through the visualization interface, users can intuitively and conveniently operate and view the operating status of the system, so that users do not need to deeply understand complex mathematical models and data analysis processes, and can monitor system performance and positioning errors through a graphical interface.

[0212] Example 2

[0213] Reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides a high-precision dynamic calibration method for an ultrasonic partial discharge instrument, which includes:

[0214] Ultrasonic sensors are deployed in the target area to collect partial discharge signals for denoising and enhancement, and feature extraction is performed based on the denoised and enhanced signals;

[0215] Based on the extracted features, a spatiotemporal correlation matrix is ​​constructed and node weights are dynamically calculated. The error model is constructed according to the node weights for error calibration and verification.

[0216] The verification results are stored and displayed through a visual interface.

[0217] 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 part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0218] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0219] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), 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 disk read-only memory (CDROM). In addition, the computer-readable medium may even be a 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, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0220] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0221] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. 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 high-precision dynamic calibration system for ultrasonic partial discharge instrument, characterized by: include, The acquisition and enhancement module deploys ultrasonic sensors in the target area to collect partial discharge signals for denoising and enhancement; A feature extraction module extracts features based on the denoised and enhanced signals; The matrix and calculation module constructs the spatiotemporal correlation matrix based on the extracted features and dynamically calculates the node weights; The error calibration and verification module builds an error model based on the node weights to perform error calibration and verification; The storage and display module stores the verification results and displays them through a visual interface.

2. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 1, characterized in that: The deployment of ultrasonic sensors in the target area to collect partial discharge signals for denoising and enhancement includes: By using GIS tools to draw the regional boundaries and adopting a hierarchical grid topology structure to segment the regular grid, ultrasonic sensors are installed at the center of the grid according to the segmented grid, and the installation coordinates are recorded; Based on the installed ultrasonic sensors, the partial discharge signals within the zone boundaries are captured; The captured partial discharge signal is converted into a frequency domain signal using fast Fourier transform, and the main frequency band in the frequency domain signal is intercepted and enhanced by using frequency band interception technology; The intercepted frequency domain signal is converted into a time domain signal by using inverse Fourier transform to obtain an enhanced signal; Construct an M×K sparse dictionary matrix D: Where M represents the number of signal sampling points, K represents the total number of frequency components in the dictionary, sin(2πf e t) and cos(2πf e t) represent the sine and cosine functions respectively, f e represents the e-th frequency component, and t represents the time sampling point; Based on the sparse dictionary matrix D, the enhanced signal is sparsely decomposed and denoised: X'=Dε+∈, In the formula, X' represents the enhanced signal, ε represents the sparse coefficient, and ∈ represents the residual; The sparse decomposition is solved by the l1 norm optimization problem, and the signal is reconstructed by the solved sparse coefficient ε and the sparse dictionary matrix D to obtain the sparse reconstructed signal X1(t): X1(t)=Dε The sparse reconstructed signal X1(t) is converted into the Henkel matrix H using the Henkel matrix construction method: m+g-1=M, In the formula, m represents the number of rows of the Henkel matrix, g represents the number of columns of the Henkel matrix, and M represents the number of sampling points of the signal; Based on the transformed Henkel matrix, the singular value decomposition method is used to decompose the Henkel matrix to obtain the singular value vectors and singular values: H=U∈G T , Where U represents the left singular vector, ∈ represents the diagonal matrix, G represents the right singular vector, T represents the transpose operation, and H represents the Henkel matrix; By extracting singular values ​​from the diagonal matrix and using the PMI index to perform weighted screening on the singular values; Set the screening threshold to 0, compare the extracted singular value with the threshold O, and when the extracted singular value is greater than or equal to the threshold O, retain the current singular value, otherwise remove the singular value; Based on the retained singular values, calculate the energy contribution of the singular values: Where E(v) represents the energy contribution of the vth singular value, θ v and θ o They represent the vth and oth singular values ​​after Henkel matrix decomposition, respectively, and k represents the total number of singular values; By using the matrix indexing method to extract the left singular vector U corresponding to the vth singular value from the left singular vector and the right singular vector v and the right singular vectors The energy contribution is defined as the weight E(v) = w v ; Combine the extracted singular vectors and weights to optimize the signal reconstruction: Where X2(t) represents the optimized reconstructed signal, S represents the singular value index set of the screening condition, and w v represents the weight of the vth singular value, U v and G v They represent the left singular vector and right singular vector corresponding to the vth singular value respectively; The sparse reconstructed signal X1(t) and the optimized reconstructed signal X2(t) are normalized by using a normalization method to obtain normalized signals X'1(t) and X'2(t); The normalized signals X'1(t) and X'2(t) are combined by a linear weighting method to obtain the final signal Z: Z = αX'1(t)+(1-α)X'2(t), In the formula, α represents the fusion weight of the coefficient reconstruction signal, and (1-α) represents the fusion weight of the optimized reconstructed signal.

3. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 2, characterized in that: The feature extraction and denoising according to the denoised and enhanced signal comprises: Based on the final signal, use discrete wavelet transform to decompose the multi-layer wavelet coefficients: In the formula, C(a,b) represents the decomposed wavelet coefficient, φ * represents the wavelet basis function, a represents the scale parameter, b represents the translation parameter, and s represents the time variable; Based on the obtained wavelet coefficients C(a,b), the wavelet component band energy characteristics are calculated using fast Fourier transform: Where V z represents the energy characteristics of the zth frequency band, B rec (f) represents the frequency spectrum of the signal, f z and f z+1 Indicates the start and end frequencies of the zth frequency band; By extracting the spectrum from the energy feature and transforming the spectrum using Hilbert transform, the imaginary part is obtained and then combined with the real part to calculate the signal envelope: Where K(t) represents the signal envelope, and β represents the Hilbert transform; Use the normalization method to normalize the signal envelope.

4. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 3, characterized in that: The step of constructing a spatiotemporal association matrix based on the extracted features and dynamically calculating node weights includes: The normalized signal envelope was analyzed by using the Pearson product-moment correlation coefficient and the k ,t k+1 ], calculate the mean and standard deviation of the signal envelope; Calculate covariance and variance from mean and standard deviation; The signal envelope correlation value between ultrasonic sensor node i and ultrasonic sensor node j is further calculated by the obtained covariance and variance: Where M ij (t) represents the spatiotemporal correlation value of ultrasonic sensor node i and node j at time t, Cov represents covariance, Var represents variance, N i (t) and N j (t) represents the normalized signal envelope of ultrasonic sensor nodes i and j; The distance is calculated based on the position of the ultrasonic sensor node, and the spatiotemporal correlation value is corrected using the exponential decay formula: Where, d ij represents the geometric distance between ultrasonic sensor node i and node j, x i ,y i , z i They represent the coordinates of ultrasonic sensor node i in three-dimensional space, x j ,y j , z j Respectively represent the coordinates of ultrasonic sensor node j in three-dimensional space, M' ij (t) represents the corrected spatiotemporal correlation value, and d0 represents the reference distance; Based on the modified spatiotemporal correlation values, the n×n spatiotemporal correlation matrix M(t) is constructed using matrix operation methods: In the formula, n represents the total number of nodes; Based on the spatiotemporal correlation matrix M(t), the weight of each ultrasonic sensor node is calculated using the weighted average method; The weights are normalized by using a normalization method.

5. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 4, characterized in that: The error calibration and verification of the error model constructed according to the node weights includes: Based on the ultrasonic sensor receiving signal time minus the signal transmission time, the signal propagation time difference ΔT is obtained; Multiply the obtained time difference by the signal propagation speed to obtain the measured pseudorange d mes ; The distance between the ultrasonic sensor nodes is converted into a true pseudo-range d by using differential correction technology. tru ; The time difference ΔT and the measured pseudorange d mes and the true pseudorange d tru As input, and construct the error model: In the formula, Δd i represents the pseudorange correction error of node i, represents the measured pseudorange of node i, represents the true pseudorange of node i, γ i represents the time deviation correction coefficient of node i, ΔT represents the signal propagation time difference; Define the adaptive objective function: In the formula, A represents the optimization objective function, θ represents the parameter set of the model, and w' i (t) represents the dynamic weight of node i after normalization, |Δd i | represents the absolute value of the pseudorange correction error of node i, and n represents the total number of nodes; The nonlinear features are processed by the support vector regression algorithm in combination with the radial basis kernel function, and the kernel function parameters and regularization parameters are optimized by the grid search method. The parameters are continuously updated using the gradient descent method. When the loss of the error model does not decrease significantly during the continuous iteration process, the iteration is stopped, the final error model is output, and the optimization is completed. By inputting the correction error into the error model, and outputting the final error result through the error model; Based on the final error result obtained, the root mean square error is calculated; Use the positioning algorithm combined with pseudo-range to calculate the measured coordinates, and further combine the measured coordinates with the true coordinates to calculate the three-dimensional positioning error; The root mean square error, three-dimensional positioning error and propagation time deviation are standardized by using the standardization method, and the standardized root mean square error, three-dimensional positioning error and propagation time deviation are integrated by the weight coefficient synthesis method: In the formula, S represents the comprehensive deviation, ρ represents the weight coefficient, RMS' represents the normalized root mean square error, Δh' represents the normalized three-dimensional positioning error, and |ΔT'| represents the normalized propagation time deviation; Set the error threshold to Q, and compare the comprehensive deviation S with the threshold Q. When the comprehensive deviation S is less than or equal to the threshold Q, the verification passes and the error meets the system requirements. When the comprehensive deviation S is greater than the threshold Q, the verification fails and re-optimization is performed.

6. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 5, characterized in that: The storage based on the verification result refers to storing data in a unified JSON data format based on the verification result, adopting a hierarchical storage structure during the storage process, distinguishing data files by timestamp and task batch, and recording detailed information of all nodes in the data file of each task.

7. The high-precision dynamic calibration system for ultrasonic partial discharge instrument according to claim 6, characterized in that: The display through the visual interface refers to using the Matplotlib data visualization tool to draw a trend graph of the standardized error term and a histogram of the comprehensive deviation S, and generating a three-dimensional distribution graph by combining the node position data, and the color of each point is gradually marked according to the size of the positioning error; The standardized error term includes a standardized root mean square error, a three-dimensional positioning error, and a propagation time deviation.

8. A high-precision dynamic calibration method for an ultrasonic partial discharge instrument based on the high-precision dynamic calibration system for an ultrasonic partial discharge instrument according to any one of claims 1 to 7, characterized in that: include, Ultrasonic sensors are deployed in the target area to collect partial discharge signals for denoising and enhancement, and feature extraction is performed based on the denoised and enhanced signals; Based on the extracted features, a spatiotemporal correlation matrix is ​​constructed and node weights are dynamically calculated. The error model is constructed according to the node weights for error calibration and verification. The verification results are stored and displayed through a visual interface.

9. A computer device comprising: A memory and a processor; the memory stores a computer program, wherein the processor implements the steps of the high-precision dynamic calibration system of an ultrasonic partial discharge instrument according to any one of claims 1 to 7 when executing the computer program.

10. 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 high-precision dynamic calibration system for ultrasonic partial discharge instrument according to any one of claims 1 to 7 are implemented.

Images