Insulating sleeve defect detection method and device based on electromagnetic induction

Abstract

The invention relates to the technical field of defect detection, and discloses an insulating sleeve defect detection method and device based on electromagnetic induction, and the method comprises the steps: employing a main detection coil to excite a composite insulating sleeve to generate an electromagnetic field, and employing an auxiliary compensation coil to receive a mixed induction signal; performing interface reflection component extraction on the mixed induction signal, and calculating a dielectric parameter value of the composite insulating sleeve; calculating an interface compensation parameter of the auxiliary compensation coil based on the dielectric parameter value; generating a compensation magnetic field opposite to an interface reflection phase by using an auxiliary compensation coil according to an interface compensation parameter, and receiving a compensated detection signal by using a main detection coil and performing phase gradient analysis to obtain a position coordinate and a size parameter of an internal defect of the composite insulating sleeve; according to the invention, hardware-level separation of electromagnetic excitation and interface compensation is realized, active inhibition of interface reflection is carried out from a signal source, and automation and intelligentization of defect detection of the composite insulating sleeve are further realized.

Description

Technical Field

[0001] This invention relates to the field of defect detection technology, and in particular to a method and apparatus for detecting defects in insulating bushings based on electromagnetic induction. Background Technology

[0002] Composite insulating bushings are typically composed of multiple layers of heterogeneous materials, such as epoxy resin layers, paper insulation layers, and ceramic sleeve layers. While this multi-layered heterogeneous structure provides excellent insulation performance, it also presents unprecedented technical challenges for defect detection. When electromagnetic waves propagate through this multi-layered structure, there are significant differences in dielectric constant between the layers: approximately 4.2 for the epoxy resin layer, approximately 2.8 for the paper insulation layer, and approximately 9.5 for the ceramic sleeve layer. These differences lead to strong electromagnetic reflection at the interfaces.

[0003] Traditional electromagnetic induction detection methods often encounter multi-layered interface reflections in composite insulating bushings. The intensity of the interface reflection signal is often tens of times higher than the defect scattering signal, and their spectral characteristics severely overlap, causing the defect signal to be completely masked by the strong interface reflection signal. Existing signal processing techniques mainly rely on passive methods such as frequency domain filtering or time domain processing. However, these methods have limited effectiveness in handling the spectral aliasing between interface reflections and defect signals, frequently resulting in missed or false detections, thus leading to low defect detection accuracy for composite insulating bushings. Summary of the Invention

[0004] This invention provides a method and apparatus for detecting defects in insulating bushings based on electromagnetic induction. This invention achieves hardware-level separation of electromagnetic excitation and interface compensation, and actively suppresses interface reflection from the signal source, thereby realizing the automation and intelligence of composite insulating bushing defect detection.

[0005] In a first aspect, the present invention provides a method for detecting defects in insulating bushings based on electromagnetic induction, the method comprising: The main detection coil is used to excite the composite insulating bushing to generate an electromagnetic field, while the auxiliary compensation coil is used to receive the mixed induction signal. The interface reflection component of the mixed induction signal is extracted to obtain the interface reflection signal, and the dielectric parameter value of the composite insulating sleeve is calculated based on the interface reflection signal. The interface compensation parameters of the auxiliary compensation coil are calculated based on the dielectric parameter values. The auxiliary compensation coil generates a compensation magnetic field with the opposite phase to the interface reflection based on the interface compensation parameters. At the same time, the main detection coil receives the compensated detection signal and performs phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0006] In conjunction with the first aspect, in a first implementation of the first aspect of the present invention, the step of using a main detection coil to excite a composite insulating bushing to generate an electromagnetic field, while simultaneously using an auxiliary compensation coil to receive a mixed induction signal, includes: An excitation current within a preset frequency range is input into the main detection coil to excite the axial magnetic field, thereby obtaining a main excitation magnetic field that penetrates the multi-layer structure of the composite insulating sleeve. Based on a coaxial ring structure, the auxiliary compensation coil and the main detection coil are orthogonally polarized to obtain an auxiliary receiving magnetic field with a 90-degree phase difference from the main excitation magnetic field. The main excitation magnetic field penetrates the epoxy resin layer, paper insulation layer and ceramic sleeve layer of the composite insulating sleeve to generate interface reflection and defect scattering, resulting in a composite electromagnetic response containing multi-layer interface reflection components and defect scattering components. The composite electromagnetic response is synchronously acquired using the auxiliary compensation coil to obtain a hybrid induction signal.

[0007] In conjunction with the first aspect, in a second implementation of the first aspect of the present invention, the step of extracting the interface reflection component from the hybrid induction signal to obtain an interface reflection signal, and calculating the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal, includes: The hybrid inductive signal is input into a wavelet transform algorithm for time-frequency domain decomposition to obtain a time-frequency distribution matrix; Based on the time-frequency distribution matrix, calculate the energy statistical features and the centroid position features, and generate a signal feature vector based on the energy statistical features and the centroid position features; The time-frequency energy distribution is periodically determined based on the signal feature vector to obtain the interface reflection feature components that satisfy the periodic condition; The hybrid sensing signal is filtered and reconstructed based on the interface reflection feature components to obtain the interface reflection signal. The dielectric parameters of the composite insulating sleeve are calculated based on the interface reflection signal.

[0008] In conjunction with the first aspect, in a third implementation of the first aspect of the present invention, the step of calculating the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal includes: The difference between the interface reflection signal and the preset interface reflection prediction signal is calculated to obtain the signal error value, and a parameter error function is created based on the signal error value. Based on the parameter error function, partial derivatives of dielectric constant, conductivity and layer thickness are calculated to obtain the parameter optimization gradient vector; The optimized gradient vector of the parameters is input into the least squares iterative algorithm to update the parameters, and the updated dielectric parameter combination is obtained. The dielectric parameter values ​​of the composite insulating sleeve are obtained by performing convergence discrimination based on the iteratively updated dielectric parameter combination.

[0009] In conjunction with the first aspect, in the fourth implementation of the first aspect of the present invention, the step of inputting the parameter optimization gradient vector into a least-squares iterative algorithm for parameter updating to obtain the iteratively updated dielectric parameter combination includes: The optimized gradient vector of the parameters is input into the least squares iterative algorithm, and the learning step size is set based on the optimized gradient vector of the parameters to obtain the learning step size parameter that controls the update magnitude of the parameters. The current iteration values ​​of dielectric constant, conductivity, and layer thickness are obtained based on the learning step size parameters, thus obtaining the current iteration parameter combination; The parameter optimization gradient vector is multiplied by the learning step size parameter to obtain the parameter update increment. The addition operation is performed based on the current combination of iterative parameters and the parameter update increment, and multiple different initial values ​​are calculated in parallel to obtain multiple sets of parameter update results. The convergence error is compared based on the multiple sets of parameter update results to obtain the iteratively updated dielectric parameter combination.

[0010] In conjunction with the first aspect, in a fifth implementation of the first aspect of the present invention, the step of calculating the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter value includes: The dielectric parameter values ​​are input into the interface reflection calculation model to calculate the reflection coefficient, resulting in the interface reflection coefficient including the reflection amplitude and reflection phase. The ratio between the interface reflection coefficient and the incident signal amplitude is calculated, and the reflection phase is reversed to obtain the compensation weighting coefficient. The excitation amplitude and excitation phase of the auxiliary compensation coil are calculated by parameter conversion based on the compensation weight coefficient to obtain the excitation control parameters; Based on the excitation control parameters, the rate of change of compensation weight is monitored to obtain the interface compensation parameters of the auxiliary compensation coil.

[0011] In conjunction with the first aspect, in the sixth implementation of the first aspect of the present invention, the step of using the auxiliary compensation coil to generate a compensation magnetic field with the opposite phase to the interface reflection based on the interface compensation parameters, and simultaneously using the main detection coil to receive the compensated detection signal and perform phase gradient analysis to obtain the position coordinates and size parameters of the internal defect of the composite insulating sleeve, includes: The interface compensation parameters are input to a signal generator to generate an excitation signal, resulting in a compensation excitation signal with the same amplitude and opposite phase to the interface reflection signal. Based on the compensation excitation signal, the auxiliary compensation coil is driven to perform magnetic field excitation processing to obtain an anti-phase compensation magnetic field that is opposite to the direction of the interface reflected magnetic field. The anti-phase compensation magnetic field and the interface reflection magnetic field generated by the composite insulating sleeve are spatially superimposed and canceled to obtain a canceled composite magnetic field after interface reflection suppression. The main detection coil is used to receive the canceled composite magnetic field to obtain a compensated detection signal. Phase gradient analysis is performed on the compensated detection signal to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0012] In conjunction with the first aspect, in the seventh implementation of the first aspect of the present invention, the step of performing phase gradient analysis on the compensated detection signal to obtain the position coordinates and size parameters of the internal defect of the composite insulating sleeve includes: Based on the compensated detection signal, the noise standard deviation and residual interface reflection amplitude are calculated to obtain the adaptive detection threshold. The compensated detection signal is subjected to threshold discrimination based on the adaptive detection threshold to obtain a pure defect scattering signal; The pure defect scattering signal is input into the phase gradient algorithm to calculate the spatial gradient, and a phase gradient vector containing defect location information is obtained. Based on the phase gradient vector and the scattering intensity of the pure defect scattering signal, defect parameter inversion calculation is performed to obtain the position coordinates and size parameters of the internal defect of the composite insulating sleeve.

[0013] In conjunction with the first aspect, in the eighth implementation of the first aspect of the present invention, the step of inputting the pure defect scattering signal into a phase gradient algorithm to perform spatial gradient calculation to obtain a phase gradient vector includes: The pure defect scattering signal is input into a phase extraction algorithm for phase separation to obtain the defect scattering phase. Based on the defect scattering phase, phase values ​​are calculated at multiple frequency points to obtain multi-frequency phase data; The spatial phase gradient corresponding to each frequency point is obtained by calculating the partial derivative of the spatial coordinates based on the multi-frequency phase data. The phase gradient vector is obtained by weighted averaging the spatial phase gradient with the signal intensity of the pure defect scattering signal.

[0014] Secondly, the present invention provides an electromagnetic induction-based insulating sleeve defect detection device, the electromagnetic induction-based insulating sleeve defect detection device comprising: The excitation module is used to excite the composite insulating bushing to generate an electromagnetic field using the main detection coil, while simultaneously receiving the mixed induction signal using the auxiliary compensation coil. An extraction module is used to extract the interface reflection component from the mixed sensing signal to obtain the interface reflection signal, and to calculate the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal. The calculation module is used to calculate the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter values; The analysis module is used to generate a compensation magnetic field with the opposite phase to the interface reflection using the auxiliary compensation coil according to the interface compensation parameters, and at the same time use the main detection coil to receive the compensated detection signal and perform phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0015] The technical solution provided by this invention employs an orthogonal polarization configuration of the main detection coil and the auxiliary compensation coil, achieving hardware-level separation of electromagnetic excitation and interface compensation, and actively suppressing interface reflections from the signal source. Wavelet transform-based time-frequency domain decomposition technology effectively distinguishes the periodic characteristics of interface reflections from the local characteristics of defect scattering, exhibiting stronger signal recognition capabilities compared to traditional frequency domain analysis methods. Real-time dynamic estimation of dielectric parameters is achieved through a least-squares iterative algorithm, automatically adapting to changes in material properties under different operating conditions and ensuring the continuous effectiveness of the compensation effect. The auxiliary compensation coil generates an anti-phase compensation magnetic field, achieving active cancellation of interface reflections at the physical level, resulting in a higher interference suppression ratio compared to passive filtering methods. A phase gradient algorithm is used for defect localization, utilizing the high sensitivity of phase information to achieve subwavelength-level localization accuracy, significantly improving accuracy compared to traditional amplitude localization methods. Multi-frequency weighted processing and intelligent signal source separation technology further improve the stability and reliability of the detection results, realizing the automation and intelligence of composite insulating sleeve defect detection. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the steps of the insulating bushing defect detection method based on electromagnetic induction in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the insulating bushing defect detection device based on electromagnetic induction in an embodiment of the present invention. Detailed Implementation

[0018] This invention provides a method and apparatus for detecting defects in insulating sleeves based on electromagnetic induction. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the electromagnetic induction-based defect detection method for insulating bushings in this invention includes: Step S1: Use the main detection coil to excite the composite insulating sleeve to generate an electromagnetic field, and at the same time use the auxiliary compensation coil to receive the mixed induction signal; It is understood that the executing entity of this invention can be an electromagnetic induction-based insulating sleeve defect detection device, or it can be a terminal or a server; no specific limitation is made here. This embodiment of the invention will be described using a server as an example.

[0020] Specifically, the main detection coil is constructed as a ring-shaped conductor structure with an inner diameter of 80 mm and an outer diameter of 120 mm. An excitation current within a preset frequency range of 0.1 MHz to 10 MHz is input into the main detection coil to generate a high-frequency magnetic field distributed along the axial direction. This high-frequency magnetic field has sufficient penetrating power to effectively act on the internal structure of the composite insulating sleeve. An auxiliary compensation coil is coaxially configured outside the main detection coil. The auxiliary compensation coil has an inner diameter of 60 mm and an outer diameter of 100 mm and is arranged to form a 90-degree phase difference with the main detection coil according to the coil polarization direction, constructing a set of orthogonally polarized double-ring electromagnetic induction structures. This ensures that the response direction of the auxiliary compensation coil to the induced magnetic field remains perpendicular to the direction of the main excitation magnetic field. As the main excitation magnetic field penetrates the epoxy resin layer, paper insulation layer, and ceramic sleeve layer in the composite insulating sleeve structure, multi-layer interface reflections are induced between different dielectric layers due to electromagnetic impedance mismatch. Furthermore, if internal physical defects exist, additional electromagnetic scattering will be induced, thus superimposing a composite electromagnetic response containing multi-layer interface reflection components and defect scattering components throughout the transmission path. The auxiliary compensation coil is located in this orthogonal structure and can synchronously sense and receive all electromagnetic reflection and scattering signals returned from each layer of the structure through independent channels, and collect them as a mixed induction signal.

[0021] Step S2: Extract the interface reflection component from the mixed induction signal to obtain the interface reflection signal, and calculate the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal. Specifically, the mixed induced signal is input into the wavelet transform analysis module. Using a wavelet such as Morlet wavelet as the mother wavelet function, continuous wavelet transform is performed to achieve joint decomposition of the mixed signal in the time and frequency domains. This yields the time-frequency distribution matrix of the signal under different frequency and time distributions, reflecting the energy distribution of each frequency component within different time windows. Based on the time-frequency distribution matrix, statistical energy features of the signal are extracted, including parameters such as mean, variance, skewness, and kurtosis. Simultaneously, the frequency centroid and time centroid positions are calculated by combining the distribution in the frequency and time domains, forming a multi-dimensional signal feature vector that comprehensively describes the time-frequency morphology of the signal. Based on the signal feature vector, the energy distribution trend of each frequency channel is periodically judged. Utilizing the stable frequency repetition and periodic mirror structure within the time window of the interface reflection signal, a periodic identification model is constructed, and interface reflection feature components that meet the periodic conditions are selected, thus initially separating the specific frequency band signal reflecting the multi-layer interface reflection response of the composite insulating sleeve. Bandpass filtering and inverse wavelet transform are used to filter and reconstruct the mixed induced signal, retaining only the periodic components corresponding to interface reflections and suppressing non-periodic defect scattering components to obtain the interface reflection signal. By comparing the interface reflection signal with the multilayer electromagnetic model and introducing a recursive least squares estimation algorithm to minimize the error, the key parameters such as the dielectric constant, conductivity and thickness of each layer of the composite insulating sleeve are solved.

[0022] Step S3: Calculate the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter values; Specifically, the dielectric parameters of the composite insulating sleeve are input into the interface reflection calculation model. Based on the interlayer impedance abrupt change relationship defined in the interface reflection calculation model, combined with the electromagnetic propagation constant and boundary continuity conditions, the reflection coefficients between the layers in the multilayer structure are calculated, and the complete complex form of the interface reflection coefficient, including the reflection amplitude and reflection phase, is output. The interface reflection amplitude is compared with the amplitude of the incident signal applied by the main detection coil to obtain the relative response intensity between the reflection and the excitation. At the same time, a mathematical phase reversal operation is performed on the interface reflection phase, that is, a π-radian phase shift is introduced to make the reflected signal and the compensation signal cancel each other out in phase, generating a compensation weight coefficient containing both amplitude and phase components. Based on the compensation weight coefficient, the amplitude and phase transformation calculations are performed on the excitation current of the auxiliary compensation coil. The compensation amplitude is converted into a numerical control parameter of the actual excitation current, and the phase information is converted into a digital phase delay control command, forming an excitation control parameter set, which is used to drive the compensation coil to output a magnetic field signal with the same amplitude but opposite phase as the interface reflection signal in real time. A compensation weight change rate monitoring mechanism is established. The slope of the compensation weight change in each cycle is continuously calculated and compared with the set change threshold. When the change rate continuously exceeds the threshold, the parameter re-evaluation program is triggered to update the compensation control parameters and output the auxiliary compensation coil interface compensation parameters that are adaptively adjusted over time.

[0023] Step S4: Using an auxiliary compensation coil, a compensation magnetic field with the opposite phase to the interface reflection is generated according to the interface compensation parameters. At the same time, the main detection coil receives the compensated detection signal and performs phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0024] Specifically, the interface compensation parameters are input to the signal generation module. The signal generation module synthesizes a compensation excitation signal in real time according to the input amplitude and phase parameters, ensuring that the amplitude of the compensation excitation signal is completely consistent with the actual interface reflection signal. Simultaneously, a π-radian phase offset is introduced to ensure that the two signals maintain a 180-degree out-of-phase relationship on the time axis. The synthesized compensation excitation signal is used as a driving signal and applied to the auxiliary compensation coil. Under the set frequency, amplitude, and phase control, the auxiliary compensation coil generates an anti-phase compensation magnetic field with a spatial direction opposite to the interface reflection magnetic field, forming a dynamically controllable compensation source. Because the anti-phase compensation magnetic field and the original interface reflection magnetic field excited by the internal structure of the composite insulating sleeve have the characteristics of being in the same direction but opposite in phase in space, they superimpose in space and form a significant interference attenuation effect. This effectively cancels out the interface reflection magnetic field within the target area, resulting in a canceled composite magnetic field after interface reflection suppression. The main detection coil, under the background condition that the interface interference has been significantly attenuated, performs high-precision sensing and reception of the residual non-periodic energy changes in the canceled composite magnetic field, acquiring a net detection signal with the defect scattering signal as the main component. The net detection signal is input into the phase gradient analysis module, where multi-frequency phase unfolding is performed. The phase gradient change rate along the spatial coordinate axes is then calculated to deduce the relative phase perturbation position of the defect scattering source in space. Based on the known operating wavelength and phase gradient formula, the defect's position coordinates in the three-dimensional coordinate system are derived, and the positioning accuracy is optimized using a multi-frequency weighted fusion algorithm. Simultaneously, feature extraction is performed on the defect echo amplitude at different frequencies, and its time-frequency spread characteristics are analyzed to determine the defect's physical scale. Multi-dimensional quantization results, including the defect's spatial location, size parameters, and structural features, are output.

[0025] In one specific embodiment, the process of performing step S1 may specifically include the following steps: An excitation current within a preset frequency range is input into the main detection coil to excite the axial magnetic field, thereby obtaining the main excitation magnetic field that penetrates the multi-layer structure of the composite insulating sleeve. Based on the coaxial ring structure, the auxiliary compensation coil and the main detection coil are orthogonally polarized to obtain an auxiliary receiving magnetic field with a 90-degree phase difference from the main excitation magnetic field. The main excitation magnetic field penetrates the epoxy resin layer, paper insulation layer and ceramic sleeve layer of the composite insulating sleeve to generate interface reflection and defect scattering, resulting in a composite electromagnetic response containing multi-layer interface reflection components and defect scattering components. The composite electromagnetic response is synchronously acquired using an auxiliary compensation coil to obtain a mixed induction signal.

[0026] Specifically, a coaxial double-ring electromagnetic induction structure consisting of a main detection coil and an auxiliary compensation coil is established in the hardware system. The main detection coil is designed as a ring structure with an inner diameter of 80 mm and an outer diameter of 120 mm, and is connected to the excitation source channel. The excitation current frequency range is set from 0.1 MHz to 10 MHz, covering the characteristic response frequency range of various composite materials, ensuring that the excitation magnetic field can fully act on the internal structure of the insulating sleeve. The high-frequency current provided by the excitation source forms a stable and well-defined axial alternating magnetic field in the main detection coil. The axial alternating magnetic field is distributed along the coil axis, has good penetration ability and spatial uniformity, and can effectively couple to its internal medium when close to the surface of the insulating sleeve, and penetrate radially through multiple layers such as the epoxy resin layer, paper insulation layer, and ceramic sleeve layer. To achieve independent reception and decoupled processing of interface reflection and defect scattering responses, an auxiliary compensation coil is coaxially arranged outside the main detection coil. The auxiliary compensation coil is designed with an inner diameter of 60 mm and an outer diameter of 100 mm, and is geometrically concentric with the main detection coil, achieving a 90-degree orthogonal setting in the electromagnetic polarization direction. That is, when the main excitation magnetic field is distributed in the axial direction, the induction direction of the auxiliary coil is set in the radial direction, thus forming a typical orthogonal polarization arrangement. This orthogonal polarization configuration utilizes the principle of electromagnetic decoupling to minimize the direct interference response of the auxiliary compensation coil to the main excitation magnetic field, thereby focusing on inducing reflected signals generated by the structural interface and scattered signals caused by defect disturbances. Since the composite insulating sleeve is composed of multiple materials with significantly different dielectric constants and conductivityes, the main excitation magnetic field inevitably induces abrupt changes in electromagnetic impedance at the interfaces during penetration, resulting in obvious interface reflection signals. Simultaneously, if there are small air gaps, foreign objects, or material delamination inside the sleeve, electromagnetic wave scattering will occur at these structural anomaly locations. The composite electromagnetic field generated by the response of the entire sleeve structure contains multi-layer interface reflection components and non-periodic defect scattering components. Its time and frequency characteristics are complex and intertwined, and it is easily affected by system noise and structural coupling effects, resulting in strong interference. To obtain a high-fidelity induction record of the above composite electromagnetic field signal, the auxiliary compensation coil is configured as a synchronous induction acquisition module. By connecting to a signal conditioning channel with a high sampling rate of 50MHz and a quantization accuracy of 16 bits, it can sense and record all electromagnetic response signals returned by the insulating sleeve structure in the first instant under the action of the main excitation magnetic field. Throughout the detection cycle, the main detection coil continuously applies the excitation magnetic field, and the auxiliary compensation coil synchronously acquires the induced voltage signal in real time. The acquired waveform is framed and segmented according to the time window structure to ensure that the recorded mixed induction signal retains the original energy information and phase structure of all interface reflection and defect scattering components, forming the basic signal data source.

[0027] In one specific embodiment, the process of performing step S2 may specifically include the following steps: The hybrid inductive signal is input into a wavelet transform algorithm for time-frequency domain decomposition to obtain the time-frequency distribution matrix; The energy statistical features and centroid position features are calculated based on the time-frequency distribution matrix, and a signal feature vector is generated based on the energy statistical features and centroid position features. The time-frequency energy distribution is periodically determined based on the signal feature vector to obtain the interface reflection feature components that satisfy the periodic condition; The interface reflection signal is obtained by filtering and reconstructing the hybrid sensing signal based on the interface reflection characteristic components. The dielectric parameters of the composite insulating sleeve are calculated based on the interface reflection signal.

[0028] Specifically, the hybrid inductive signal is input into the time-frequency analysis module, and a wavelet transform algorithm constructed using the Morlet wavelet as the mother wavelet function is invoked to perform continuous wavelet transform operations. Based on the set center frequency, bandwidth factor, and time window parameters, the signal is jointly expanded in both frequency and time dimensions, thus transforming the one-dimensional time-domain signal into a two-dimensional time-frequency distribution matrix. This matrix expresses the energy distribution of the signal in different frequency bands and the local characteristic variation trends in different time windows. Based on the time-frequency distribution matrix, several sets of statistical characteristic parameters describing the spatial structure of the signal energy are extracted, including statistics such as energy mean, energy variance, energy skewness, and energy kurtosis. The time and frequency coordinates of the energy distribution centroid are calculated by combining the temporal and frequency concentrations of the energy distribution on each frequency channel. These descriptive indicators together constitute a representative time-frequency feature vector. Based on the time-frequency feature vector, the time-frequency energy distribution structure of each frequency channel is periodically identified and analyzed. Specifically, a periodic consistency discrimination model is constructed to detect whether there is an energy peak pattern that repeats with a fixed period T on the time axis, and the distribution pattern in the frequency direction is combined to determine whether the pattern conforms to the spectral mirror characteristics of interface reflection. When the system determines that a certain group of time-frequency energy blocks satisfies the periodic constraint condition in its characteristic distribution, the energy cluster is marked as an interface reflection characteristic component and separated from the original time-frequency matrix to form an interface reflection response sub-matrix. Using the inverse wavelet transform module, the identified interface reflection characteristic components are used as the retention interval. Bandpass filtering and time window constraints are applied to the mixed inductive signal to remove the non-periodic energy components corresponding to defect scattering, retaining only the interface reflection response with a periodic time-frequency structure. After restoring it to the original time axis, an interface reflection signal waveform sequence with complete phase information and waveform structure is formed. The interface reflection signal is input into the interface reflection modeling module, and the electromagnetic transmission relationships of each layer in the model are initialized based on the structural parameters of the multilayer composite insulating sleeve. The interface reflection prediction model constructed by the layered reflection theory is called, and the amplitude, phase, and time delay information of the interface reflection signal are used as constraints. The least squares error function is used to iteratively optimize parameters such as dielectric constant, conductivity, and layer thickness. By establishing a minimum error path between the theoretical reflection response and the actual reflection signal, the model gradually converges to the optimal dielectric parameter solution that meets the current measurement environment. When the error function decreases below the threshold for multiple consecutive iterations, the parameter fitting process is considered complete, and key parameters such as the dielectric constant, conductivity, and structural thickness of each layer of the composite insulating sleeve are output.

[0029] In one specific embodiment, the process of calculating the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal may specifically include the following steps: The signal error value is obtained by calculating the difference between the interface reflection signal and the preset interface reflection prediction signal, and a parameter error function is created based on the signal error value. The partial derivatives of dielectric constant, conductivity and layer thickness are calculated based on the parameter error function to obtain the parameter optimization gradient vector; The parameter optimization gradient vector is input into the least squares iterative algorithm to update the parameters, and the updated dielectric parameter combination is obtained. Convergence is determined based on the iteratively updated dielectric parameter combination to obtain the dielectric parameter values ​​of the composite insulating sleeve.

[0030] Specifically, the actual interface reflection signal obtained through time-frequency separation and filtering reconstruction of the hybrid inductive signal possesses complete amplitude envelope, phase change, and time delay information. Based on the combined parameters of dielectric constant, conductivity, and layer thickness under the current iteration step, the interface reflection prediction signal generated by the layered electromagnetic reflection model simulates the standard response behavior of the composite insulating sleeve under the current assumed electrical parameters. The interface reflection signal and the preset interface reflection prediction signal are compared point-to-point in the time and frequency domains. The complex difference between the actual interface reflection signal and the predicted reflection signal at each sampling moment is calculated, and their amplitude deviation and phase shift are recorded to construct a signal error value sequence. By accumulating and integrating the squares of the amplitudes of the signal error value sequence, an overall error measurement function describing the fit between the current parameter combination and the actual response is generated, namely the parameter error function. Its form is a nonlinear multivariable complex function to be optimized, with dielectric constant, conductivity, and layer thickness as inputs and a real error scalar as output. Within each iteration cycle, the gradient calculation module is invoked to calculate the partial derivatives of the parameter error function with respect to each parameter to be estimated. Finite difference or analytical differentiation methods are used to evaluate the first-order rate of change of the error function at the current point in the directions of dielectric constant, conductivity, and thickness, forming a parameter optimization gradient vector representing the trend of the error function. Each component of the parameter optimization gradient vector represents the trend and direction of the error function when each parameter is increased by a unit increment at the current point. The parameter optimization gradient vector is input into the least squares iterative algorithm, driving the current parameter point to update along the direction of the fastest descent of the error function. The parameters are adjusted using a gradient descent-based least squares iterative method, setting a reasonable learning rate μ as the update step size. The gradient is multiplied by the step size and used to correct the current dielectric parameter combination, forming a new iterative candidate solution. The new dielectric parameter combination is input into the interface reflection prediction model to regenerate the predicted signal, and the difference between the new signal and the actual interface reflection signal is calculated again to update the error function value, continuously executing the above iterative closed-loop process. During the iteration process, a convergence monitoring mechanism is activated simultaneously to compare the difference between the current error function value and the error value of the previous round, and a convergence judgment threshold is set. When the error change rate is lower than the convergence judgment threshold for several consecutive rounds, it is determined that the current parameter combination has reached a local optimum or global optimum solution, and the iteration is stopped. The dielectric constant, conductivity and layer thickness values ​​in the current parameter combination are locked as the equivalent electromagnetic characteristic parameters of the final composite insulating sleeve.

[0031] In one specific embodiment, the process of inputting the parameter optimization gradient vector into the least squares iterative algorithm to update the parameters and obtain the iteratively updated dielectric parameter combination can specifically include the following steps: The parameter optimization gradient vector is input into the least squares iterative algorithm, and the learning step size is set based on the parameter optimization gradient vector to obtain the learning step size parameter that controls the magnitude of parameter update. The current iteration values ​​of dielectric constant, conductivity, and layer thickness are obtained based on the learning step size parameter, thus obtaining the current iteration parameter combination; The parameter update increment is obtained by multiplying the parameter optimization gradient vector with the learning step size parameter. The addition operation is performed based on the current combination of iterative parameters and the parameter update increment, and multiple different initial values ​​are calculated in parallel to obtain multiple sets of parameter update results. By comparing the convergence error of multiple sets of parameter update results, the iteratively updated dielectric parameter combination is obtained.

[0032] Specifically, the optimized gradient vector is input into the parameter update module, where the absolute magnitude and directional characteristics of each component of the gradient vector are analyzed to determine the rate of change of the error function and the slope of the optimal path in the current parameter space. Based on this, the learning step size parameter is adaptively set according to the gradient norm, i.e., an adjustment factor controlling the parameter update amplitude. This ensures a smaller step size when the gradient is large to avoid jumping out of the convergence region, and a larger step size when the gradient is small to improve convergence efficiency, resulting in a dynamically adjustable learning step size parameter. Based on the set step size parameter and the current gradient vector, parameter-wise channel control is performed. The baseline values ​​of dielectric constant, conductivity, and layer thickness for the current iteration are obtained from the previous iteration and used as the current iteration parameter combination, preparing for the parameter value update operation. Each component of the optimized gradient vector is numerically multiplied with the learning step size parameter to obtain the parameter update increments in the dielectric constant, conductivity, and layer thickness directions, representing the current movement magnitude in the direction of minimum error descent. Vector addition is used to element-wise accumulate the current iteration parameter combination and the parameter update increments to calculate the candidate solutions for the new round of parameter combinations. To enhance global search capabilities and overcome local optima traps, a parallel multi-starting-point optimization mechanism is introduced. During the initialization phase, multiple different initial parameter combinations are set as starting solutions, and each solution independently runs the gradient-step-increment-addition update process. This allows the system to simultaneously approach the optimal solution within multiple parameter space sub-intervals, generating multiple independent parameter update results. After each set of parameter updates, the dielectric parameters corresponding to each result are input back into the reflection prediction model to generate a predicted signal. This signal is compared with the actual interface reflection signal, and the corresponding reflection error value is calculated. This reflection error value is used as the convergence index after this iteration. The error values ​​of all candidate parameter sets are input into the convergence error comparison module for unified judgment. By comparing the magnitudes of the error functions among all groups, the group with the smallest error is selected as the optimal solution for this iteration. The learning step size and gradient direction corresponding to this parameter set are also recorded. If the errors of multiple parameter sets are close to and less than the convergence threshold, it is considered that a local convergence interval has been reached. The parameter set with the smallest fluctuation and the fastest convergence speed is selected as the final result. The resulting dielectric parameter combination includes the optimized dielectric constant, conductivity, and structural layer thickness.

[0033] In one specific embodiment, the process of performing step S3 may specifically include the following steps: The dielectric parameter values ​​are input into the interface reflection calculation model to calculate the reflection coefficient, resulting in the interface reflection coefficient including the reflection amplitude and reflection phase. The compensation weighting coefficient is obtained by calculating the ratio between the interface reflection coefficient and the incident signal amplitude, and by reversing the reflection phase. The excitation amplitude and excitation phase of the auxiliary compensation coil are converted and calculated based on the compensation weight coefficient to obtain the excitation control parameters; Based on the excitation control parameters, the rate of change of compensation weight is monitored to obtain the interface compensation parameters of the auxiliary compensation coil.

[0034] Specifically, a multi-layer electromagnetic interface reflection prediction model is established within the system. Based on the actual structure of the composite insulating bushing, the model defines the epoxy resin layer, paper insulation layer, and ceramic bushing layer as dielectric layers with different dielectric constants and conductivityes. The electromagnetic wave reflection behavior between each interface is modeled using the analytical relationship between characteristic impedance and wave propagation constant. The dielectric parameter values ​​are input into the interface reflection calculation model to calculate the reflection coefficient. Local reflection coefficients are calculated sequentially at each dielectric layer boundary based on the impedance discontinuity principle. Furthermore, multiple reflection contributions are superimposed using a matrix cascade method to obtain the overall interface reflection coefficient, which is a complex function where the real and imaginary parts represent the reflection amplitude and phase, respectively. This describes the reflection response behavior of electromagnetic waves throughout the insulating bushing structure due to electrical discontinuities. The relative reflection ratio is obtained by numerically comparing the magnitude of the interface reflection coefficient with the amplitude of the incident signal applied by the main detection coil at the current excitation frequency. For the phase information in the reflection coefficient, a π-radian numerical offset is introduced to reverse its phase, ensuring that the subsequently generated compensation magnetic field is precisely opposite in phase to the interface reflection magnetic field. This yields a complex form compensation weight coefficient containing opposite phase and proportional amplitude. The compensation weight coefficient represents the magnetic field response amplitude and phase direction that the auxiliary compensation coil should output under the current operating condition, used to achieve physical-level cancellation of the interface reflection signal. Based on the complex compensation weight coefficient, amplitude-phase control parameters are converted. The amplitude part is converted into the amplitude control command of the actual drive current, and then converted into the current output amplitude setpoint in the corresponding power driver through proportional conversion. Simultaneously, the complex angle part, i.e., the phase information, is converted into the corresponding phase modulation control command, used to set the phase adjustment amount of the signal source or phase delay circuit at the corresponding frequency point, ensuring reverse interference in the time domain. These two parts together constitute the excitation control parameters for driving the auxiliary compensation coil, including instantaneous current amplitude, current phase, frequency synchronization indicator, and delay compensation coefficient. During the continuous application of the aforementioned excitation control parameters to the compensation coil, a real-time monitoring mechanism for the rate of change of the compensation weight is simultaneously activated. By setting a certain time sampling window, the numerical difference between the current compensation weight and the weight of the previous period is statistically analyzed within each window period, and the rate of change is calculated. When the rate of change continuously exceeds a set threshold (e.g., 5%), it is considered that the dielectric response of the composite structure has shown a non-stationary trend, caused by material aging, temperature changes, or structural disturbances. At this time, the compensation parameter re-evaluation process is triggered. The optimized dielectric parameters are recalculated from the new interface reflection signal sensed in the current period, and the compensation weight is updated again to maintain the adaptive response capability of the system. A stable output interface compensation parameter is formed in the closed-loop structure of compensation weight calculation, parameter conversion, phase adjustment, and dynamic monitoring.

[0035] In one specific embodiment, the process of performing step S4 may specifically include the following steps: The interface compensation parameters are input to the signal generator to generate an excitation signal, resulting in a compensation excitation signal with the same amplitude and opposite phase as the interface reflection signal. The auxiliary compensation coil is driven by the compensation excitation signal to perform magnetic field excitation processing, and an anti-phase compensation magnetic field opposite to the direction of the interface reflected magnetic field is obtained. The anti-phase compensation magnetic field and the interface reflection magnetic field generated by the composite insulating bushing are spatially superimposed and canceled to obtain the canceled composite magnetic field after interface reflection suppression. The main detection coil is used to receive the canceled composite magnetic field to obtain the compensated detection signal. Phase gradient analysis was performed on the compensated detection signal to obtain the location coordinates and size parameters of the internal defects in the composite insulating sleeve.

[0036] Specifically, the interface compensation parameters are input into the signal generator, which has amplitude control, phase modulation, and frequency synchronization functions. Its input parameters include the target compensation amplitude, phase reversal angle (typically π radians), excitation frequency, and time synchronization delay. Upon receiving the interface compensation parameters, the system sets the peak level of the output current based on the amplitude parameters and adjusts the phase of the output signal to be 180 degrees out of phase with the interface reflection signal in the current main excitation magnetic field, while maintaining frequency consistency to fully satisfy the time-domain phase interference condition. This generates a compensation excitation signal with the same amplitude and completely opposite phase to the interface reflection signal. The compensation excitation signal is then applied in real-time to the auxiliary compensation coil, driving it to output a compensation magnetic field with precise amplitude and phase. In the compensation coil, the excitation signal forms an alternating magnetic field with a specific waveform and direction. The direction and phase of this alternating magnetic field are designed to be completely opposite to the magnetic field generated by reflection from the multi-layer dielectric interface in the composite insulating sleeve, resulting in destructive interference when spatially superimposed. At the structural level, the electromagnetic field excited by the main detection coil forms a multi-reflection interface with the epoxy resin layer, paper insulation layer, and ceramic sleeve layer of the composite insulating sleeve. The interface reflected magnetic field returns from the boundaries of each dielectric layer in the composite structure, superimposing to form a macroscopic periodic reflection response. Meanwhile, the auxiliary compensation coil, by controlling the formation of a reverse magnetic field, will spatially superimpose the compensation magnetic field with a high coupling rate to the interface reflected magnetic field. Since the two are completely identical in amplitude and completely opposite in phase, physical interference cancellation occurs in the superposition region, resulting in a canceled composite magnetic field where interface reflection is effectively suppressed. The main detection coil is used to receive the canceled composite magnetic field, recording the composite magnetic field response under the state where interface reflection interference is filtered out. The electromagnetic scattering signal generated by real defects in the insulation structure will no longer be submerged by the strong reflection background, significantly improving the signal-to-noise ratio and yielding a compensated detection signal with prominent scattering characteristics in the structure. Phase gradient analysis is performed on the compensated detection signal. The detection signal is expanded using multi-frequency phase expansion to extract the phase change curve corresponding to each frequency component. The gradient of the phase distribution is calculated in the spatial domain, i.e., the phase of the detection signal is differentiated along different spatial directions to form a spatial gradient vector field. The relative position of the source causing the phase disturbance in the three-dimensional coordinate system is deduced from the phase gradient. Combining the operating wavelength, magnetic field propagation model, and weighted averaging algorithm, the position coordinates (x, y, z) of the defect inside the composite insulating sleeve are obtained. Simultaneously, the frequency bandwidth, time-domain duration, and amplitude envelope of the defect signal are extracted. The geometrical parameters of the defect, such as radius, length, or expanded area, are calculated using a scale feature inversion algorithm.

[0037] In one specific embodiment, the process of performing phase gradient analysis on the compensated detection signal to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve can specifically include the following steps: Based on the compensated detection signal, the noise standard deviation and residual interface reflection amplitude are calculated to obtain the adaptive detection threshold. The compensated detection signal is threshold-discriminated based on the adaptive detection threshold to obtain the pure defect scattering signal; The pure defect scattering signal is input into the phase gradient algorithm to calculate the spatial gradient, and a phase gradient vector containing defect location information is obtained. The defect parameters are inverted based on the phase gradient vector and the scattering intensity of the pure defect scattering signal to obtain the position coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0038] Specifically, the noise distribution in the compensated detection signal is statistically modeled, and a zero-signal reference window is established by extracting the lowest energy segment. Within this zero-signal reference window, the amplitude sequences at multiple frequency points are statistically processed to obtain the standard deviation of the background noise. Simultaneously, the residual interface reflection amplitude in the compensated signal is calculated by performing energy-weighted integration on the low-frequency continuous structure region in the time-frequency distribution. Based on the noise standard deviation and the residual interface reflection amplitude, an adaptive detection threshold for signal judgment is calculated. This threshold is dynamically adjusted according to the noise level of the current detection environment and the residual interface condition after compensation. Based on the adaptive detection threshold, the compensated detection signal is sequentially subjected to amplitude discrimination. Each sampling point is analyzed to determine whether the signal amplitude exceeds the currently calculated adaptive detection threshold, and whether its time-frequency distribution exhibits non-periodic and spatially localized characteristics. When a signal segment meets the above conditions, it is marked as a valid defect scattering response. Continuous segments are then extracted to form a pure defect scattering signal sequence, eliminating all background noise interference and the influence of residual interface signals to obtain a high signal-to-noise ratio defect information carrier. The pure defect scattering signal is input into the phase gradient analysis module. Phase demodulation is performed on the signal at multiple excitation frequency points, and a phase distribution mapping is established along the spatial dimensions. Gradient calculations are performed on the phase change rate of the signal along the x, y, and z directions in space, forming a phase gradient vector field describing the location characteristics of the electromagnetic disturbance source. Regions with significantly increased phase change amplitude in the phase gradient vector field are identified as inversion results corresponding to the defect location. These spatially abrupt regions are mapped onto the three-dimensional coordinate grid of the insulating sleeve, and the relative position coordinates of the defect in the three-dimensional coordinate system are calculated by combining the operating wavelength corresponding to the current excitation frequency, thus completing the defect location. Key parameters such as the energy intensity, spectral width, and duration of the defect signal are analyzed. Combined with the distribution density of the phase gradient in each direction, and matched with the established defect feature template, the type and geometric characteristics of the defect are determined. For example, point defects exhibit a phase gradient concentrated at a single spatial point with a wide corresponding spectral distribution; linear defects cause the phase gradient to be continuously distributed along a spatial direction, forming a highly directional energy ridge structure; while planar defects exhibit a larger-scale gradually varying region and multi-band coupling characteristics in their phase distribution. By jointly analyzing these distribution characteristics and scattering intensity, the size parameters of the defects are inverted, including the lateral size, longitudinal extension range, and morphological classification index, and multidimensional parameter results including defect spatial coordinates, size range, and structural type are output.

[0039] In one specific embodiment, the process of inputting the pure defect scattering signal into the phase gradient algorithm to calculate the spatial gradient and obtain the phase gradient vector can specifically include the following steps: The pure defect scattering signal is input into a phase extraction algorithm for phase separation to obtain the defect scattering phase; Phase values ​​are calculated at multiple frequency points based on the defect scattering phase to obtain multi-frequency phase data; The spatial phase gradient corresponding to each frequency point is obtained by calculating the partial derivative of the spatial coordinates based on the multi-frequency phase data. The phase gradient vector is obtained by weighted averaging of the spatial phase gradient and the signal intensity of the pure defect scattering signal.

[0040] Specifically, by inputting the pure defect scattering signal into the phase extraction algorithm, the real and imaginary parts of the signal in the complex domain are transformed into complex exponential form, and the arctangent function is performed on its complex amplitude signal to separate the phase information carried in the defect scattering signal, obtaining defect scattering phase data at a single frequency point, reflecting the relative propagation state of the scattered wave in the composite structure and the structural disturbance response characteristics. Based on the defect scattering phase, the signal is demodulated at multiple preset frequency points, that is, the phase extraction algorithm is repeatedly executed under different excitation frequency conditions, demodulating the defect scattering signal corresponding to each frequency point into a set of phase response datasets, forming multi-frequency phase data covering the target frequency band. The multi-frequency phase data constitutes a phase distribution set spanning the frequency domain, used to reflect the response consistency of wave phase changes caused by defects under different wavelength conditions, exhibiting strong frequency stability and structural sensitivity. The multi-frequency phase data is input into the spatial mapping model, and based on the three-dimensional spatial grid established by the sleeve detection structure, the partial derivatives of the phase values ​​are calculated along the x, y, and z directions respectively, forming a directional derivative matrix describing the gradient distribution of phase changes in space. By calculating the local derivative of the phase at each frequency point in three-dimensional space using the finite difference method or fitting algorithm, the spatial phase gradient corresponding to the current frequency point is obtained. This gradient represents the rate of phase change per unit distance in each direction, measuring the direction and intensity of the phase disturbance source's influence in space. Since the amplitude and direction of the phase gradient of the defect signal at different frequency points are affected by factors such as scattering effects, propagation paths, and excitation wavelengths, there are certain offsets and inconsistencies. Signal intensity is introduced as a weighting factor, and the spatial phase gradient corresponding to each frequency point is calculated by weighting the corresponding defect scattering signal amplitude. The spatial phase gradient at each frequency point is matched with its scattering intensity, assigning higher weighting coefficients to frequencies with greater intensity and lower influence weights to those with lower intensity. The spatial phase gradient data of all frequency points are then weighted, accumulated, and normalized to form a comprehensive phase gradient vector.

[0041] The above describes the method for detecting defects in insulating bushings based on electromagnetic induction in the embodiments of the present invention. The following describes the device for detecting defects in insulating bushings based on electromagnetic induction in the embodiments of the present invention. Please refer to [link / reference]. Figure 2 One embodiment of the insulating bushing defect detection device based on electromagnetic induction in this invention includes: Excitation module 201 is used to excite the composite insulating bushing to generate an electromagnetic field using the main detection coil, and at the same time use the auxiliary compensation coil to receive the mixed induction signal; Extraction module 202 is used to extract the interface reflection component from the mixed induction signal, obtain the interface reflection signal, and calculate the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal. Calculation module 203 is used to calculate the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter value; Analysis module 204 is used to generate a compensation magnetic field with the opposite phase to the interface reflection based on the interface compensation parameters using an auxiliary compensation coil. At the same time, it uses the main detection coil to receive the compensated detection signal and perform phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

[0042] Through the coordinated operation of the aforementioned components, and by employing an orthogonal polarization configuration of the main detection coil and the auxiliary compensation coil, hardware-level separation of electromagnetic excitation and interface compensation is achieved. Compared to traditional single-coil systems, this system can actively suppress interface reflections at the signal source, fundamentally solving the problem of aliasing between interface interference and defect signals. The wavelet transform-based time-frequency domain decomposition technique can simultaneously preserve the time and frequency information of the signal, effectively distinguishing the periodic mirror characteristics of interface reflections from the non-periodic local characteristics of defect scattering. Compared to traditional single-frequency domain analysis methods, it possesses stronger signal recognition capabilities and anti-interference performance. Real-time dynamic estimation of dielectric parameters is achieved through a least-squares iterative algorithm, automatically adapting to changes in the composite insulating sleeve material properties under different operating conditions. Compared to fixed-parameter methods, it offers stronger environmental adaptability and detection accuracy, ensuring the continuous effectiveness of the compensation effect. The auxiliary compensation coil generates a compensation magnetic field with the opposite phase to the interface reflection, achieving active cancellation of interface reflections at the physical level. Compared to traditional passive filtering methods, this achieves a higher interference suppression ratio, significantly improving the detection conditions for weak defect signals. A phase gradient algorithm is employed for defect spatial localization. Leveraging the high sensitivity of phase information, subwavelength-level localization accuracy is achieved, significantly improving accuracy compared to traditional amplitude-based methods. This meets the precise diagnostic requirements for minute defects inside composite insulating sleeves. By calculating the phase gradient at multiple frequency points and performing weighted averaging based on signal strength, random errors from single-frequency measurements are effectively suppressed, improving the stability and reliability of defect parameter inversion and ensuring the consistency and reproducibility of detection results. This invention can automatically identify and separate interface reflection and defect scattering components, avoiding the subjectivity and complexity of manual parameter adjustment and achieving automation and intelligence in the detection process.

[0043] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0044] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0045] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting defects in insulating bushings based on electromagnetic induction, characterized in that, include: The main detection coil is used to excite the composite insulating bushing to generate an electromagnetic field, while the auxiliary compensation coil is used to receive the mixed induction signal. The interface reflection component of the mixed induction signal is extracted to obtain the interface reflection signal, and the dielectric parameter value of the composite insulating sleeve is calculated based on the interface reflection signal. The interface compensation parameters of the auxiliary compensation coil are calculated based on the dielectric parameter values. The auxiliary compensation coil generates a compensation magnetic field with the opposite phase to the interface reflection based on the interface compensation parameters. At the same time, the main detection coil receives the compensated detection signal and performs phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

2. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 1, characterized in that, The method of using a main detection coil to excite a composite insulating bushing to generate an electromagnetic field, and simultaneously using an auxiliary compensation coil to receive the mixed induced signal, includes: An excitation current within a preset frequency range is input into the main detection coil to excite the axial magnetic field, thereby obtaining a main excitation magnetic field that penetrates the multi-layer structure of the composite insulating sleeve. Based on a coaxial ring structure, the auxiliary compensation coil and the main detection coil are orthogonally polarized to obtain an auxiliary receiving magnetic field with a 90-degree phase difference from the main excitation magnetic field. The main excitation magnetic field penetrates the epoxy resin layer, paper insulation layer and ceramic sleeve layer of the composite insulating sleeve to generate interface reflection and defect scattering, resulting in a composite electromagnetic response containing multi-layer interface reflection components and defect scattering components. The composite electromagnetic response is synchronously acquired using the auxiliary compensation coil to obtain a hybrid induction signal.

3. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 1, characterized in that, The step of extracting the interface reflection component from the mixed induction signal to obtain the interface reflection signal, and calculating the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal, includes: The hybrid inductive signal is input into a wavelet transform algorithm for time-frequency domain decomposition to obtain a time-frequency distribution matrix; Based on the time-frequency distribution matrix, calculate the energy statistical features and the centroid position features, and generate a signal feature vector based on the energy statistical features and the centroid position features; The time-frequency energy distribution is periodically determined based on the signal feature vector to obtain the interface reflection feature components that satisfy the periodic condition; The hybrid sensing signal is filtered and reconstructed based on the interface reflection feature components to obtain the interface reflection signal. The dielectric parameters of the composite insulating sleeve are calculated based on the interface reflection signal.

4. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 3, characterized in that, The step of calculating the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal includes: The difference between the interface reflection signal and the preset interface reflection prediction signal is calculated to obtain the signal error value, and a parameter error function is created based on the signal error value. Based on the parameter error function, partial derivatives of dielectric constant, conductivity and layer thickness are calculated to obtain the parameter optimization gradient vector; The optimized gradient vector of the parameters is input into the least squares iterative algorithm to update the parameters, and the updated dielectric parameter combination is obtained. The dielectric parameter values ​​of the composite insulating sleeve are obtained by performing convergence discrimination based on the iteratively updated dielectric parameter combination.

5. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 4, characterized in that, The step of inputting the optimized gradient vector of the parameters into a least-squares iterative algorithm for parameter updating to obtain the iteratively updated dielectric parameter combination includes: The optimized gradient vector of the parameters is input into the least squares iterative algorithm, and the learning step size is set based on the optimized gradient vector of the parameters to obtain the learning step size parameter that controls the update magnitude of the parameters. The current iteration values ​​of dielectric constant, conductivity, and layer thickness are obtained based on the learning step size parameters, thus obtaining the current iteration parameter combination; The parameter optimization gradient vector is multiplied by the learning step size parameter to obtain the parameter update increment. The addition operation is performed based on the current combination of iterative parameters and the parameter update increment, and multiple different initial values ​​are calculated in parallel to obtain multiple sets of parameter update results. The convergence error is compared based on the multiple sets of parameter update results to obtain the iteratively updated dielectric parameter combination.

6. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 1, characterized in that, The calculation of the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter value includes: The dielectric parameter values ​​are input into the interface reflection calculation model to calculate the reflection coefficient, resulting in the interface reflection coefficient including the reflection amplitude and reflection phase. The ratio between the interface reflection coefficient and the incident signal amplitude is calculated, and the reflection phase is reversed to obtain the compensation weighting coefficient. The excitation amplitude and excitation phase of the auxiliary compensation coil are calculated by parameter conversion based on the compensation weight coefficient to obtain the excitation control parameters; Based on the excitation control parameters, the rate of change of compensation weight is monitored to obtain the interface compensation parameters of the auxiliary compensation coil.

7. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 1, characterized in that, The auxiliary compensation coil generates a compensation magnetic field with an opposite phase to the interface reflection based on the interface compensation parameters. Simultaneously, the main detection coil receives the compensated detection signal and performs phase gradient analysis to obtain the location coordinates and dimensional parameters of the internal defects in the composite insulating sleeve. This includes: The interface compensation parameters are input to a signal generator to generate an excitation signal, resulting in a compensation excitation signal with the same amplitude and opposite phase to the interface reflection signal. Based on the compensation excitation signal, the auxiliary compensation coil is driven to perform magnetic field excitation processing to obtain an anti-phase compensation magnetic field that is opposite to the direction of the interface reflected magnetic field. The anti-phase compensation magnetic field and the interface reflection magnetic field generated by the composite insulating sleeve are spatially superimposed and canceled to obtain a canceled composite magnetic field after interface reflection suppression. The main detection coil is used to receive the canceled composite magnetic field to obtain a compensated detection signal. Phase gradient analysis is performed on the compensated detection signal to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

8. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 7, characterized in that, The step of performing phase gradient analysis on the compensated detection signal to obtain the location coordinates and size parameters of the internal defect of the composite insulating sleeve includes: Based on the compensated detection signal, the noise standard deviation and residual interface reflection amplitude are calculated to obtain the adaptive detection threshold. The compensated detection signal is subjected to threshold discrimination based on the adaptive detection threshold to obtain a pure defect scattering signal; The pure defect scattering signal is input into the phase gradient algorithm to calculate the spatial gradient, and a phase gradient vector containing defect location information is obtained. Based on the phase gradient vector and the scattering intensity of the pure defect scattering signal, defect parameter inversion calculation is performed to obtain the position coordinates and size parameters of the internal defect of the composite insulating sleeve.

9. The method for detecting defects in insulating bushings based on electromagnetic induction according to claim 8, characterized in that, The step of inputting the pure defect scattering signal into a phase gradient algorithm for spatial gradient calculation to obtain a phase gradient vector includes: The pure defect scattering signal is input into a phase extraction algorithm for phase separation to obtain the defect scattering phase. Based on the defect scattering phase, phase values ​​are calculated at multiple frequency points to obtain multi-frequency phase data; The spatial phase gradient corresponding to each frequency point is obtained by calculating the partial derivative of the spatial coordinates based on the multi-frequency phase data. The phase gradient vector is obtained by weighted averaging the spatial phase gradient with the signal intensity of the pure defect scattering signal.

10. A defect detection device for insulating bushings based on electromagnetic induction, characterized in that, For performing the electromagnetic induction-based insulating sleeve defect detection method as described in any one of claims 1-9, the electromagnetic induction-based insulating sleeve defect detection device comprises: The excitation module is used to excite the composite insulating bushing to generate an electromagnetic field using the main detection coil, while simultaneously receiving the mixed induction signal using the auxiliary compensation coil. An extraction module is used to extract the interface reflection component from the mixed sensing signal to obtain the interface reflection signal, and to calculate the dielectric parameter value of the composite insulating sleeve based on the interface reflection signal. The calculation module is used to calculate the interface compensation parameters of the auxiliary compensation coil based on the dielectric parameter values; The analysis module is used to generate a compensation magnetic field with the opposite phase to the interface reflection using the auxiliary compensation coil according to the interface compensation parameters, and at the same time use the main detection coil to receive the compensated detection signal and perform phase gradient analysis to obtain the location coordinates and size parameters of the internal defects of the composite insulating sleeve.

Images