A winding imaging method and system based on SAFT and multi-transducer arc scanning

Abstract

This invention discloses a winding imaging method and system based on SAFT and multi-transducer arc scanning, belonging to the field of ultrasonic signal scanning imaging technology. The method includes: mounting multiple ultrasonic transducers on a three-dimensional mechanical track; synchronously driving the transducers to emit arc-shaped ultrasonic waves and receive echo signals, generating multiple sets of original arc datasets; performing spatial intersection operations to extract the three-dimensional coordinates of the winding surface at a single moment; performing envelope extraction and SAFT delay superposition focusing processing on the echo signals to generate multi-moment enhanced signals; constructing a spatiotemporally fused three-dimensional data matrix; constructing a dynamic correction model based on the coordinate deviation of a preset reference point on the winding surface to dynamically correct the three-dimensional data matrix; and performing spatial depth mapping on the corrected three-dimensional data matrix to generate a pseudo-color three-dimensional point cloud map of the transformer winding. This invention employs a collaborative mechanism of multi-transducer arc scanning and SAFT signal focusing to achieve efficient three-dimensional imaging of winding deformation.

Description

Technical Field

[0001] This invention belongs to the field of ultrasonic signal scanning imaging technology, and more specifically, relates to a winding multidimensional imaging method and system based on SAFT and multi-transducer arc scanning. Background Technology

[0002] As a core component of power transformers, the mechanical deformation of transformer windings is one of the main causes of equipment failure. Therefore, high-precision and high-efficiency non-destructive imaging inspection of transformer windings is of great significance for ensuring the safe and stable operation of power systems.

[0003] Currently, ultrasonic imaging technology has been attempted to be applied to offline inspection of transformer windings. However, in practical applications, traditional inspection methods face significant challenges. First, the data acquisition mode of existing imaging methods is inefficient, making it difficult to quickly acquire overall spatial information of the windings, and thus failing to meet the needs of large-scale, rapid operation and maintenance of power equipment. Second, due to limitations such as the complexity of the winding structure, the attenuation of sound waves by the insulating medium, and complex electromagnetic interference in the field, the quality of the acquired echo signals is generally poor, with a low signal-to-noise ratio. This results in large errors and poor reliability in the imaging results, making it impossible to accurately identify and quantify the minute mechanical deformations of the windings.

[0004] While some existing technologies employ synthetic aperture technology to improve resolution, they have not effectively resolved the fundamental contradiction between efficiency and accuracy. Their multi-transducer collaborative scanning strategies are insufficient, data acquisition methods are redundant, and they lack adaptive correction capabilities for system errors and field interference. This results in limited practicality under complex operating conditions, preventing their application in real-world engineering. Therefore, a three-dimensional imaging method for transformer windings that can balance detection efficiency and imaging accuracy while adapting to complex operating conditions is needed. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a winding multidimensional imaging method and system based on SAFT and multi-transducer arc scanning.

[0006] The present invention adopts the following technical solution.

[0007] The first aspect of the present invention provides a winding imaging method based on SAFT and multi-transducer arc scanning, comprising the following steps:

[0008] Multiple ultrasonic transducers are mounted on a three-dimensional mechanical track, and scanning parameters are configured according to the transformer winding structure and acoustic characteristics of the insulating medium.

[0009] Synchronous drive transducer emits arc-shaped ultrasonic waves and receives echo signals to generate multiple sets of original arc datasets;

[0010] Perform a spatial intersection operation on each set of original arc datasets to extract the three-dimensional coordinates of the winding surface at a single moment.

[0011] Envelope extraction and SAFT time-delay superposition focusing processing are performed on the echo signal to generate a multi-moment enhanced signal;

[0012] The three-dimensional coordinates of the winding surface are bound to the multi-time enhancement signals according to the timestamp to construct a spatiotemporally fused three-dimensional data matrix;

[0013] Based on the coordinate deviation of a preset reference point on the winding surface, a dynamic correction model is constructed to dynamically correct the three-dimensional data matrix.

[0014] Spatial depth mapping is performed on the corrected 3D data matrix to generate a pseudo-color 3D point cloud map of the transformer winding.

[0015] Optionally, configuring the scanning parameters based on the transformer winding structure and the acoustic characteristics of the insulating medium includes:

[0016] Based on the geometry of the transformer windings, the attenuation characteristics of the laminated insulation material for ultrasonic waves, and combined with the requirements of penetration depth and signal-to-noise ratio, the center frequency of the ultrasonic transducer is set.

[0017] The scanning speed is set based on the pulse repetition frequency of the ultrasonic transducer and the motion performance of the three-dimensional mechanical track.

[0018] The sound field scanning radius is set according to the internal space dimensions of the transformer under test.

[0019] Optionally, generating multiple sets of original arc datasets includes:

[0020] The ultrasonic transducer array is synchronously driven to emit ultrasonic pulses to the transformer winding, and the three-dimensional mechanical track is controlled to move according to a preset nonlinear acceleration curve.

[0021] The system receives reflected echo signals from the surface of the transformer windings. When the intensity of the echo signal exceeds a dynamic threshold determined by the reflection coefficient and the transmitted signal voltage, it triggers a recording as a valid echo at that moment.

[0022] The sound field propagation distance is calculated based on the time difference between the ultrasonic pulse transmission and reception times corresponding to the effective echo.

[0023] Based on the sound field propagation distance and the real-time spatial coordinates of the transducer when the ultrasonic pulse is emitted, an arc dataset is generated by a coordinate transformation algorithm, wherein the arc represents a set of points that may exist on the surface of the transformer winding.

[0024] Optionally, the extraction of the three-dimensional coordinates of the winding surface at a single moment includes:

[0025] The spatial intersection of the arc datasets generated by multiple ultrasonic transducers in the ultrasonic transducer array at the same sampling time is calculated.

[0026] The coordinates of the geometric center point of the spatial intersection are determined as the three-dimensional coordinate points of the transformer winding surface at that moment;

[0027] Based on the adaptive circular arc generation algorithm, interpolation calculations are performed on the coordinate points caused by the lack of echo signals to supplement the three-dimensional coordinate points on the surface of the transformer winding.

[0028] Optionally, generating the multi-time enhancement signal includes:

[0029] Perform Hilbert transform on the echo signal to extract its signal envelope and filter out the carrier oscillation component;

[0030] The multi-time echo signals after envelope extraction are subjected to time-delay superposition and synthesis processing based on the time delay of ultrasound propagation to the imaging point.

[0031] During the time-delay superposition process, a superposition weighting coefficient is introduced for each echo signal, and the signal amplitude at the focal point is weighted and synthesized to generate a multi-moment enhanced signal.

[0032] Optionally, the construction of the spatiotemporally fused three-dimensional data matrix includes:

[0033] The multi-moment enhancement signal is converted from an analog signal into a digital signal of the winding position coordinate points;

[0034] Spatial coordinate timestamps are generated based on the position pulses of the grating ruler in a three-dimensional mechanical track.

[0035] The digital signal is bound to the digital signal of the winding position coordinate point to form three-dimensional winding coordinate data stored according to the timestamp;

[0036] The three-dimensional winding coordinate data is stored as a three-dimensional data matrix.

[0037] Optionally, the step of constructing a dynamic correction model to dynamically correct the three-dimensional data matrix includes:

[0038] Pre-marked reflective points on the surface of the transformer windings as spatial reference points;

[0039] Calculate the deviation vector between the actual coordinates of the spatial reference point and the corresponding coordinates in the three-dimensional data matrix;

[0040] A dynamic correction model is constructed based on the deviation vector, and the dynamic correction model dynamically adjusts the correction step size based on the gradient field of the deviation vector.

[0041] The dynamic correction model is applied to correct the winding coordinate data in the three-dimensional data matrix in real time.

[0042] Optionally, generating a pseudo-color three-dimensional point cloud map of the transformer winding includes: calculating the spatial depth value from each point in space to a preset viewpoint based on the corrected three-dimensional data matrix;

[0043] Based on the spatial depth value and the signal strength value in the three-dimensional data matrix, a three-dimensional reconstruction is performed using a volume rendering algorithm to generate a three-dimensional model of the transformer winding.

[0044] The signal intensity values ​​in the three-dimensional model are mapped to a predefined pseudo-color spectrum to generate a pseudo-color three-dimensional point cloud map of the transformer winding.

[0045] Optionally, a space-time joint confidence model is constructed based on the corrected three-dimensional data matrix, and fault points are marked, including:

[0046] For each three-dimensional coordinate point in the three-dimensional data matrix, perform the following operation:

[0047] The spatial reliability is calculated based on the spatial distance between the three-dimensional coordinate point and the geometric center of the ultrasonic transducer array, wherein the spatial reliability decreases as the spatial distance increases.

[0048] Based on the duration of continuous detection of the coordinate point, a time decay factor is introduced to correct the spatial confidence, thereby obtaining the spatial-temporal joint confidence of the coordinate point.

[0049] The calculated space-time joint confidence score is compared with the preset fault determination threshold.

[0050] If the combined spatial-temporal confidence of a certain coordinate point exceeds the fault determination threshold, then the coordinate point is marked as a potential fault point of the transformer winding.

[0051] A second aspect of the present invention provides a winding imaging system based on SAFT and multi-transducer arc scanning, for running the winding imaging method based on SAFT and multi-transducer arc scanning described in the first aspect of the present invention, comprising:

[0052] A transducer array and drive module are used to transmit arc-shaped ultrasonic waves to the transformer windings and receive the reflected echo signals; wherein the transducer array is mounted on a three-dimensional mechanical track.

[0053] The arc data acquisition and processing module is used to generate multiple sets of original arc datasets based on the echo signal, and to perform a spatial intersection operation on the multiple sets of original arc datasets to extract the three-dimensional coordinates of the winding surface.

[0054] The signal processing and data fusion module is used to perform envelope extraction and SAFT delay superposition focusing processing on the echo signal to generate a multi-time enhanced signal, and bind the three-dimensional coordinates of the winding surface with the multi-time enhanced signal according to the timestamp to construct a spatiotemporal fusion three-dimensional data matrix.

[0055] An adaptive error correction module is used to construct a dynamic correction model based on the coordinate deviation of a preset reference point on the winding surface and to dynamically correct the three-dimensional data matrix.

[0056] The 3D imaging module is used to perform spatial depth mapping on the corrected 3D data matrix to generate a pseudo-color 3D point cloud map of the transformer winding.

[0057] Compared with the prior art, the beneficial effects of the present invention include at least the following:

[0058] (1) The present invention adopts multi-transducer collaborative arc scanning technology, which drives multiple transducers to move along a preset arc path synchronously through a three-dimensional mechanical track, covering a larger detection area at one time, significantly reducing the number of mechanical displacements and scanning time, greatly improving the overall detection efficiency of transformer windings, solving the problem of low efficiency caused by traditional point-by-point scanning, and meeting the rapid detection needs of large-scale operation and maintenance of power systems.

[0059] (2) This invention uses synthetic aperture focusing (SAFT) technology to perform time-delay superposition focusing on echo signals at multiple times, which effectively improves the signal-to-noise ratio and resolution of the echo signals and overcomes the problems of poor signal quality and blurred imaging caused by complex winding structure, acoustic attenuation of insulating medium and electromagnetic interference. This lays the data foundation for the accurate identification of small deformations of windings and high-precision three-dimensional imaging.

[0060] (3) The present invention introduces a dynamic adaptive correction mechanism based on reference points. The system error model is derived in reverse by measuring error, and the coordinate data is corrected in real time by using gradient-sensitive adaptive step size. This effectively compensates for system errors and multi-source interference on site, and significantly improves the accuracy of three-dimensional coordinate reconstruction and the reliability of overall imaging results.

[0061] (4) This invention achieves precise quantitative analysis of the spatial position and deformation of winding deformation through multi-transducer data fusion and three-dimensional volume rendering technology, generates intuitive pseudo-color three-dimensional point cloud map and structured diagnostic report, provides three-dimensional quantitative diagnostic capability that is lacking in traditional methods, and provides a reliable basis for transformer winding status assessment and fault early warning. Attached Figure Description

[0062] Figure 1 This is a schematic diagram of the arc-shaped data acquisition method provided in accordance with an embodiment of the present invention;

[0063] Figure 2 This is a schematic diagram illustrating the supplementary data using the midpoint circle drawing method provided in an embodiment of the present invention;

[0064] Figure 3 This is a schematic diagram of a single transducer data matrix provided according to an embodiment of the present invention;

[0065] Figure 4 This is a schematic diagram of a three-dimensional winding deformation extreme point visualization display interface provided in accordance with an embodiment of the present invention;

[0066] Figure 5 This is a schematic diagram of the transformer winding report generation interface provided according to an embodiment of the present invention;

[0067] Figure 6 This is a flowchart of a method provided according to an embodiment of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0069] Synthetic Aperture Focusing Technique (SAFT) is an advanced signal processing algorithm. Its core idea is to use a small-aperture transducer to acquire a series of signals while in motion or via an array. These signals are then coherently superimposed using an algorithm to synthesize a virtual large aperture. This method effectively overcomes the limitations of physical aperture on imaging resolution, achieving high-resolution, clear imaging of the entire detection area without mechanical focusing. It is widely used in fields such as ultrasonic non-destructive testing, radar, and sonar.

[0070] In the embodiments described in this invention, the SAFT algorithm is mainly used for time-delay superposition and focusing of echo signals. Specifically, the system acquires multiple sets of echo signals from different spatial locations, and by accurately calculating the delay of the sound wave propagation path, aligns and superimposes signals targeting the same focal point, thereby significantly enhancing the effective signal, suppressing noise, and ultimately generating a higher-resolution enhanced signal, providing a high-quality data foundation for subsequent 3D reconstruction.

[0071] In Embodiment 1, this invention provides a winding multidimensional imaging method based on SAFT and multi-transducer arc scanning, such as... Figure 6 As shown, it includes the following steps:

[0072] Step 1: Install the ultrasonic transducer array on a three-dimensional mechanical track and configure the transducer movement and scanning parameters.

[0073] Preferably, in step 1, mounting the ultrasonic transducer array on the three-dimensional mechanical track includes:

[0074] Based on the characteristics of piezoelectric transducers, such as fast response and high power density, piezoelectric ceramic ultrasonic transducers are used as the core detection unit. This invention uses three piezoelectric ceramic ultrasonic transducers and strictly installs them at the original coordinate points of the lower left corner of the three-dimensional mechanical track along the Z-axis and X / Y radial directions. The distance from the original coordinate points to the transformer oil tank is ≥10cm, so as to achieve multi-angle sound wave coverage and spatial triangulation positioning.

[0075] Preferably, in step 1, configuring the scanning parameters includes:

[0076] The scanning parameters are set based on the winding physical characteristics and system performance:

[0077] Based on the attenuation characteristics of ultrasonic waves by transformer oil temperature and laminated insulation materials, and considering factors such as penetration depth and signal-to-noise ratio, the center frequency of the ultrasonic transducer is set to 2MHz.

[0078] Because the minimum pulse repetition frequency of the transducer is 1 kHz and the upper limit of the mechanical track acceleration is 0.5... Considering that the transducer's own response time, including the piezoelectric crystal excitation response and acoustic wave settling time, is about 0.2-0.5ms, in order to avoid the instability of the acquired data and spatial sampling distortion caused by excessively fast scanning speed, the scanning speed in this invention is controlled at 10mm / s to ensure that the transducer has sufficient response time to complete the acoustic energy conversion and signal stabilization, while the spatial sampling density meets the Nyquist criterion.

[0079] Specifically, according to the present invention, the internal dimensions and structure of the test transformer are set with a sound field radius of 100–500 mm, which enables the three-dimensional stepping transducer to perform an all-round scan of the transformer windings.

[0080] Step 2: Synchronously drive the ultrasonic transducer array to emit arc-shaped ultrasonic waves to the transformer winding, receive multiple sets of ultrasonic transducer echo signals and generate arc data sets, and obtain the three-dimensional coordinates of the winding surface by finding the intersection of the arc data sets of multiple transducers.

[0081] Preferably, step 2 includes:

[0082] Step 2.1, synchronously driving the ultrasonic transducer array to emit arc-shaped ultrasonic waves towards the transformer winding, specifically including:

[0083] A three-dimensional mechanical track drives a transducer array to move sequentially along the corresponding dimension of the scanning direction via stepper motors. Simultaneously, the transducers are triggered to emit ultrasonic pulses and receive echo signals reflected from the transformer windings. The transducers are also controlled to emit arc-shaped ultrasonic signals with equally spaced sound field radii, and data is collected sequentially. To avoid the transducers failing to receive the corresponding ultrasonic echo signals due to excessive acceleration during sampling, this invention employs a Sigmoid nonlinear acceleration strategy to prevent motion oscillations. The formula is shown below:

[0084]

[0085] in , The minimum pulse repetition frequency is 1 kHz;

[0086]

[0087] in The minimum pulse repetition period, The transducer scanning speed is 10 mm / s.

[0088] Step 2.2: The transducer receives the echo signal. When a high level is received, arc data is generated, specifically including:

[0089] A transducer echo signal judgment and reception command is constructed. A high-level triggering mechanism is used to receive ultrasonic echo signals. Receiving a high level indicates that the ultrasonic signal of the corresponding radius has touched the winding surface.

[0090] Set the high-level threshold voltage as follows:

[0091]

[0092] k is the reflection coefficient, which is taken as 0.85-0.9. The voltage of the ultrasonic pulse transmission signal, the trigger condition is... and This ensures that the rising edge of the signal is captured.

[0093] Arc data generation principle: Each high level corresponds to an equal time delay surface, which can be used to calculate the distance between the transducer and the winding coordinate point at a certain moment.

[0094] =

[0095] in This is the time difference between the transducer pulse transmission time and the echo reception time;

[0096] The parametric equations of the arc can be obtained using algorithms for converting between polar and Cartesian coordinates. The parametric equations of the arc are as follows:

[0097]

[0098] in The coordinates of the transducer points can be calculated based on the transducer's movement time and speed, and the arc dataset can be obtained using the adaptive power-constrained circular arc generation algorithm. The original echo signal strength The magnitude of the reflection coefficient k is determined.

[0099] Step 2.3 involves finding the intersection of multiple sets of arc data received by the transducer array and extracting the three-dimensional coordinates of the transformer winding surface, specifically including:

[0100] Each time the transducer is moved, a segment of arc data can be generated in this way; the arc... Each point on the transducer position Distance between The same applies, therefore the time delay of the ultrasound signal reaching these points is... The same applies, therefore the arc The data values ​​at each point are equal and are calculated using the following formula:

[0101]

[0102] in, , Ultrasonic pulse emission signal intensity , This is the minimum straight-line distance from the transducer to the winding. This is a normalization constant for the signal strength per unit distance; The data values ​​are for the arc (dimensionless, normalized signal strength represents signal attenuation per unit distance). Normalization processing is used to make the final value... Mapping to the [0,1] interval eliminates individual differences among transducers; This is the weighting coefficient, ranging from 0.8 to 1.2. = ( (where the incident angle is used to compensate for the emission loss from the inclined surface); For time The intensity of the echo signal; Let be the radius of the sound field. The ultrasonic intensity decreases inversely with the square of the distance. The formula is obtained by dividing by . Implement parameter distance compensation, in mm.

[0103] The intersection of the arcs from the three transducers yields the ultrasonic signal intersection data:

[0104]

[0105] Where N represents the number of transducers, and in this invention, three transducers are used to scan for winding defects. For example... Figure 1 As shown, Figure 1 This is a schematic diagram of the arc-shaped data acquisition method provided according to an embodiment of the present invention. Based on known winding surface coordinate data, an adaptive power arc generation algorithm is used to supplement arc data, improving detection efficiency.

[0106] Preferably, step 2 further includes using an adaptive power arc generation algorithm to supplement the data coordinates on the arc, ensuring data integrity:

[0107] Preferably, the method for generating the arc data includes:

[0108] (1) Construct the equation of the circle

[0109] First, construct the equation of the circle based on the known coordinates.

[0110]

[0111] in, These are the coordinates of the transducer's position.

[0112] (2) Calculate the discriminant

[0113]

[0114] in The radius of curvature of the arc at the corresponding point

[0115] Write an algorithm to substitute the coordinates of the available location points into the discriminant:

[0116]

[0117] (3) Coordinate point determination

[0118] when When, the next point is taken as And the discriminant is modified to:

[0119]

[0120] when When the x-coordinate of the next point is increased by one unit, the discriminant is modified to:

[0121]

[0122] Specifically, this algorithm is written into the main chip's data processing to supplement the winding coordinate point data. Based on the symmetry of the arc, only coordinate points on one-eighth of the circle are needed to obtain the coordinate data for the entire arc. The algorithm is constrained to ensure that the coordinate points are within the transducer's 45° power angle range. The power angle constraint is as follows:

[0123]

[0124] in Rated power for a single transducer This is the unit vector along the transducer axis.

[0125] like Figure 2 As shown, Figure 2 This is a schematic diagram illustrating the supplementary data using the midpoint circle drawing method provided in an embodiment of the present invention.

[0126] Preferably, the transducer emits ultrasonic pulses using the BASK radiation shift keying excitation method, generating high-energy pulse emission waveforms by controlling the pulse width of the digital baseband signal to 5μs.

[0127] Step 3: Perform envelope extraction and time-delay superposition synthetic aperture focusing processing on the echo signal in sequence, and bind the enhanced signal obtained by focusing processing with the three-dimensional coordinates of the winding surface and the timestamp to store it as a three-dimensional data matrix.

[0128] Preferably, the envelope extraction of the echo signal includes:

[0129] Envelope extraction filters out low-frequency envelope oscillations in the ultrasound radio frequency signal, eliminating carrier oscillation interference with imaging and reducing the complexity of subsequent interpolation calculations. The Hilbert transform method used in this invention is shown below:

[0130]

[0131] in This is the original ultrasonic signal. This is the Hilbert transform method.

[0132] Preferably, the time-delayed superposition synthetic aperture focusing process includes:

[0133] Echo signal processing: The transducer receives ultrasonic signal echoes at different sampling times. The signal processing chip combines the ultrasonic echo signals from multiple times through coherent sampling and superposition to create a virtual larger aperture, thereby improving imaging resolution and signal-to-noise ratio.

[0134] The intersection data of the ultrasonic echo signals from the three transducers received at different locations after signal envelope extraction is known to be: The echo data received by the transducer at different times are delayed and superimposed according to the corresponding delay time, and then convolutional operation is performed to obtain the detection data of the focal point, as shown below:

[0135]

[0136] In the formula, The time-delayed impulse function is dimensionless. This is the weighting coefficient for superimposing the echo signal, which is usually taken as 0.9~1.1. The arc coordinate data mentioned above is normalized. A larger aperture coefficient is achieved by superimposing virtual samples using SAFT delay sampling, thereby improving the ultrasonic resolution of the echo signal.

[0137] Preferably, the storage of the three-dimensional data matrix includes:

[0138] Data fusion constructs a 3D winding matrix: The SAFT-processed ultrasonic echo signal is converted from an analog echo signal to a digital signal of the winding position coordinates using a 10-bit ADC analog-to-digital converter chip. The FPGA receives the position pulses from the grating ruler of the 3D mechanical track to generate timestamps of the winding position coordinates, forming 3D winding coordinate data stored according to timestamps. This data is stored in the GPU memory of the processing layer in the form of a 3D matrix, where the matrix structure dimensions are:

[0139]

[0140] This represents the three-dimensional winding coordinate data collected by three transducers at different sampling times in a spatial coordinate system. This data can then be mapped to the fault point coordinate data of the corresponding multi-dimensional transducer using timestamps. For example... Figure 3 As shown, Figure 3 This is a schematic diagram of a single transducer data matrix provided according to an embodiment of the present invention.

[0141] Step 4: Based on the coordinate measurement error of the reference point preset on the winding surface, a dynamic correction model is constructed to perform real-time correction on the coordinates in the three-dimensional data matrix.

[0142] Preferably, step 4 includes:

[0143] Step 4.1: Pre-mark reflective points on the surface of the transformer winding as reference points;

[0144] Step 4.2, based on the deviation between the actual coordinates and the measured coordinates of the reflective marker, calculate the measurement error vector of the reference point, including:

[0145]

[0146] The received winding data in the three-dimensional coordinate system can be further adaptively corrected. This correction method specifies the coordinate data of a reference point by setting reflective markers at known locations on the winding surface, thereby deriving the system error model in reverse from the measurement error and correcting subsequent measurement data in real time. First, the error in the three-dimensional coordinates of the winding is calculated according to the error formula:

[0147]

[0148] in For measurement error , This is supplementary three-dimensional winding coordinate data based on the midpoint circle method. For the measured true three-dimensional winding coordinate data vector.

[0149] Step 4.3 involves constructing and using a dynamic correction model to correct the coordinates in the 3D data matrix in real time, including:

[0150] The adaptive correction formula is shown below:

[0151]

[0152] The adaptive step size is:

[0153]

[0154]

[0155] Its function is to control the correction intensity and avoid overfitting or oscillation. Its physical meaning is to approximate the measured value towards the reference point and compensate for systematic deviations. The error gradient vector is a differential representation of the spatial error, and its range is adjusted according to the system state: low gradient region. <0.5 indicates a gradual change in systematic error, suitable for wall correction; 0.5 < <2.0, significant error variation requires appropriate correction; high gradient region >2.0 indicates a local anomaly triggering protection mechanism. This correction method can resolve certain errors caused by noise interference when data is received through the transducer. The model improves the coordinate reconstruction accuracy to 0.03mm through adaptive step size while ensuring system stability.

[0156] Step 5: Perform spatial depth calculation based on the corrected 3D data matrix to generate a 3D point cloud map.

[0157] Preferably, step 5 includes:

[0158] Perform spatial depth calculations on the corrected 3D data matrix to generate a pseudo-color 3D point cloud map, including:

[0159] This paper describes how to generate pseudo-color 3D winding point cloud maps using Python code written in PyCharm. First, a 3D winding imaging GUI interface is built to import the 3D winding data in Cartesian coordinates into the volume rendering algorithm. The core formula is shown below:

[0160]

[0161] in These are the coordinates of the three-dimensional winding points. The coordinates of the center point of the scene. Let be the unit vector in the direction of the viewpoint. The depth value is used. A clear, accurate, and noise-free 3D winding structure model is constructed using a volume rendering algorithm. Based on the generated 3D winding model, a high-precision pseudo-color 3D point cloud map is rendered and output. This image can intuitively display the overall shape, spatial arrangement, details of key parts, and smooth surface after noise reduction of the winding. Simultaneously, built-in software is used to analyze the deformation trend and extreme points of winding deformation. The extreme points of winding deformation are obtained from the multi-transducer confidence calculation model described above. Figure 4 As shown, Figure 4 This is a schematic diagram of a three-dimensional winding deformation extreme point visualization display interface provided in accordance with an embodiment of the present invention.

[0162] Preferably, step 5 further includes:

[0163] When the combined confidence level of spatial location and time decay reaches a preset threshold, the winding fault point is marked, and a diagnostic report containing the fault location coordinates and radial deformation is output.

[0164] Specifically, a spatial-temporal joint confidence model is constructed based on a three-dimensional data matrix, and fault points are marked, including:

[0165] Fault point coordinates extracted from multi-transducer data: The winding data coordinates with timestamps are jointly calculated using a confidence calculation model of the three-dimensional coordinates of multiple transducers. The formula is as follows:

[0166] First, build based on Confidence model:

[0167]

[0168] The formula for spatial credibility assessment is:

[0169]

[0170] in, This represents the spatial reliability attenuation coefficient, controlling the decrease in reliability with distance. In high-frequency precision detection, strong attenuation results in a small reliability range. The value is typically between 0.8 and 1.2.

[0171] The coordinates of the fault point are:

[0172]

[0173] The coordinates of the transducer's geometric center are:

[0174]

[0175] Where N is the number of ultrasonic transducers;

[0176] The time factor formula is:

[0177]

[0178] in, The initial confidence level is set to 0.3. The time constant is taken as 10s. This is the time of the first fault detection.

[0179] When the three-dimensional winding coordinate data of the multi-transducer with timestamps are input into the confidence calculation interval model, the confidence value containing spatial and temporal factors is... When the accuracy rate is 95%, it indicates that the three-dimensional winding fault coordinate point can be marked as the actual fault coordinate point.

[0180] A quantitative diagnostic report is generated based on the fault point marking results, including:

[0181] The reconstructed 3D model of the winding is automatically analyzed to detect potential winding deformations and automatically generate a structured winding fault diagnosis report based on the analysis results. The report analysis includes: the detected fault type and the location coordinates of the fault point in 3D space, and assesses the overall condition of the winding based on the severity of the winding or quantitative indicators. Figure 5 As shown, Figure 5 This is a schematic diagram of the transformer winding report generation interface provided in accordance with an embodiment of the present invention.

[0182] The winding multidimensional imaging method based on SAFT and multi-transducer arc scanning provided by this invention has the following technical advantages compared with the prior art:

[0183] 1. In view of the problem of low efficiency in transformer winding detection in the prior art, the present invention provides a multi-transducer collaborative arc scanning technology, which synchronously drives multiple transducers to move along a preset arc path through a three-dimensional mechanical track, reducing the number of mechanical displacements and significantly improving the scanning detection efficiency.

[0184] 2. To address the problem of large positioning errors in existing technologies, this invention provides an adaptive dynamic correction mechanism. Based on the reference point measurement error, the system error model is derived in reverse, and the coordinate offset is corrected in real time through a gradient-sensitive adaptive step size, which greatly improves the positioning accuracy.

[0185] 3. To address the lack of three-dimensional quantization capabilities in existing technologies, this invention provides a multi-transducer data fusion and three-dimensional volume rendering technology. By integrating the arc scanning data of multiple transducers through clock synchronization, performing SAFT focusing processing, and generating a pseudo-color three-dimensional point cloud map, it achieves accurate quantitative analysis of the spatial position and deformation of the winding.

[0186] In Embodiment 2, this invention provides a winding multidimensional imaging system based on SAFT and multi-transducer arc scanning, which is based on the winding multidimensional imaging method based on SAFT and multi-transducer arc scanning described in Embodiment 1. The system includes:

[0187] A transducer array and drive module are used to emit arc-shaped ultrasonic waves to the transformer windings and receive the reflected echo signals; wherein the transducer array is mounted on a three-dimensional mechanical track and is driven by a stepper motor to achieve multi-angle scanning.

[0188] The arc data acquisition and processing module is used to generate multiple sets of original arc datasets based on the echo signal, and to perform spatial intersection operation on the multiple sets of original arc datasets to extract the three-dimensional coordinates of the winding surface; at the same time, it supplements the missing coordinate points based on the midpoint circle method.

[0189] The signal processing and data fusion module is used to perform envelope extraction and SAFT delay superposition focusing processing on the echo signal to generate a multi-time enhanced signal, and bind the three-dimensional coordinates of the winding surface with the multi-time enhanced signal according to the timestamp to construct a spatiotemporal fusion three-dimensional data matrix.

[0190] An adaptive error correction module is used to construct a dynamic correction model based on the coordinate deviation of a preset reference point on the winding surface and to dynamically correct the three-dimensional data matrix.

[0191] The 3D imaging module is used to perform spatial depth mapping on the corrected 3D data matrix to generate a pseudo-color 3D point cloud map of the transformer winding.

[0192] The fault diagnosis module is used to construct a spatial-temporal joint confidence model and mark fault points based on the three-dimensional data matrix or the pseudo-color three-dimensional point cloud map, and output a diagnostic report containing the fault location and deformation.

[0193] This invention provides a three-dimensional multi-transducer arc scanning algorithm based on synthetic aperture technology. The three-dimensional multi-transducer arc scanning strategy involves placing multiple transducers in a three-dimensional spatial coordinate system (XYZ) to emit arc-shaped ultrasonic waves to detect transformer windings. A three-dimensional mechanical track device, constructed using CAD, simultaneously moves the transducers along the scanning direction at the same mechanical speed to traverse the transformer windings for ultrasonic detection. A single transducer acquires echo signal data in an arc shape with a constant time period using the arc scanning algorithm. An echo signal data receiving device is programmed using FPGA, and an adaptive correction algorithm is introduced to eliminate noise interference during reception.

[0194] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method of winding imaging based on SAFT and multi-transducer arcuate scanning, characterized in that, The method comprises the following steps: installing multiple ultrasonic transducers on a three-dimensional mechanical track to form an ultrasonic transducer array, configuring scanning parameters according to the transformer winding structure and the acoustic characteristics of the insulating medium; synchronously driving the transducers to emit arc-shaped ultrasonic waves and receive echo signals, and generating multiple groups of original arc line data sets; performing a space intersection operation on each group of original arc line data sets to extract three-dimensional coordinates of the winding surface at a single time; performing envelope extraction and SAFT delay-and-sum focusing processing on the echo signals to generate multiple-time enhanced signals; binding the three-dimensional coordinates of the winding surface and the multiple-time enhanced signals according to timestamps to construct a three-dimensional data matrix with space-time fusion; constructing a dynamic correction model based on the coordinate deviation of a preset reference point on the winding surface to dynamically correct the three-dimensional data matrix; the dynamic correction of the three-dimensional data matrix by the dynamic correction model comprises: taking a preset light reflection marker point on the winding surface of the transformer as a spatial reference point; calculating the deviation vector of the actual coordinates of the spatial reference point and the corresponding coordinates in the three-dimensional data matrix; constructing a dynamic correction model according to the deviation vector, and the dynamic correction model dynamically adjusts the correction step length based on the gradient field of the deviation vector; applying the dynamic correction model to real-time correct the winding coordinate data in the three-dimensional data matrix; performing spatial depth mapping on the corrected three-dimensional data matrix to generate a pseudo-color three-dimensional point cloud map of the transformer winding; the generation of the pseudo-color three-dimensional point cloud map of the transformer winding comprises: calculating the spatial depth value of each point in space to a preset viewpoint based on the corrected three-dimensional data matrix; performing three-dimensional reconstruction through volume rendering algorithm according to the spatial depth value and the signal intensity value in the three-dimensional data matrix to generate a three-dimensional model of the transformer winding; mapping the signal intensity value in the three-dimensional model to a predefined pseudo-color spectrum to generate a pseudo-color three-dimensional point cloud map of the transformer winding; constructing a space-time joint confidence model based on the corrected three-dimensional data matrix and marking fault points, comprising: for each three-dimensional coordinate point in the three-dimensional data matrix, the following operations are performed: based on the spatial distance between the three-dimensional coordinate point and the geometric center of the ultrasonic transducer array, the spatial confidence of the three-dimensional coordinate point is calculated, wherein the spatial confidence decreases with the increase of the spatial distance; based on the length of time during which the coordinate point is continuously detected, a time attenuation factor is introduced to correct the spatial confidence to obtain the space-time joint confidence of the coordinate point; comparing the calculated space-time joint confidence with a preset fault judgment threshold; if the space-time joint confidence of a coordinate point exceeds the fault judgment threshold, the coordinate point is marked as a potential fault point of the transformer winding.

2. The winding imaging method based on SAFT and multi-transducer arc-shaped scanning according to claim 1, wherein: the configuration of the scanning parameters according to the transformer winding structure and the acoustic characteristics of the insulating medium comprises: setting the center frequency of the ultrasonic transducer according to the geometric shape of the transformer winding, the attenuation characteristics of the ultrasonic wave to the laminated insulating material, and the requirements of the penetration depth and signal-to-noise ratio. The scanning speed is set based on the pulse repetition frequency of the ultrasonic transducer and the motion performance of the three-dimensional mechanical track; The acoustic field scanning radius is set according to the internal space size of the transformer to be measured.

3. The winding imaging method based on SAFT and multi-transducer arc scanning according to claim 1, characterized in that: the generating of the multiple sets of original arc data sets comprises: synchronously driving the ultrasonic transducer array to emit ultrasonic pulses to the transformer winding and controlling the three-dimensional mechanical track to move according to the preset nonlinear acceleration curve; receiving the reflected echo signals of the transformer winding surface, and triggering the recording of the effective echo at the moment when the echo signal intensity exceeds the dynamic threshold value determined by the reflection coefficient and the transmission signal voltage; calculating the acoustic field propagation distance according to the time difference between the transmission and reception time instants of the ultrasonic pulse corresponding to the effective echo; generating the arc data set by a coordinate conversion algorithm based on the acoustic field propagation distance and the real-time spatial coordinates of the transducer when the ultrasonic pulse is transmitted, wherein the arc represents the point set that may exist on the surface of the transformer winding.

4. The winding imaging method based on SAFT and multi-transducer arc scanning according to claim 3, characterized in that: the extracting of the three-dimensional coordinates of the winding surface at a single time instant comprises: calculating the spatial intersection of the arc data sets generated by multiple ultrasonic transducers in the ultrasonic transducer array at the same sampling time instant; determining the geometric center point coordinates of the spatial intersection as the three-dimensional coordinate point of the transformer winding surface at the moment; based on the adaptive arc generation algorithm, interpolating the coordinate points caused by the absence of echo signals to supplement the three-dimensional coordinate points of the transformer winding surface.

5. The winding imaging method based on SAFT and multi-transducer arc scanning according to claim 1, characterized in that: the generating of the multiple time instant enhanced signals comprises: performing Hilbert transform on the echo signals to extract the signal envelope and filter out the carrier oscillation components; delay and add processing the multiple time instant echo signals after envelope extraction according to the delay time of ultrasonic propagation to the imaging point; in the delay and add process, introducing a weighting coefficient for each echo signal to weight and synthesize the signal amplitude at the focusing point, generating the multiple time instant enhanced signals.

6. The winding imaging method based on SAFT and multi-transducer arc scanning according to claim 4 or 5, characterized in that: the constructing of the three-dimensional data matrix with time and space fusion comprises: converting the multiple time instant enhanced signals from analog signals to digital signals of winding position coordinate points; generating a space coordinate timestamp based on the position pulse of the grating ruler of the three-dimensional mechanical track; binding the digital signals with the digital signals of the winding position coordinate points to form three-dimensional winding coordinate data stored according to the timestamp; storing the three-dimensional winding coordinate data as a three-dimensional data matrix.

7. A system for performing a method of SAFT-based multi-transducer arcuate scan imaging of windings according to any one of claims 1 to 6, wherein the system comprises: a plurality of transducers arranged in an arcuate configuration; a signal generator for generating a signal to be transmitted by the plurality of transducers; a signal receiver for receiving signals transmitted by the plurality of transducers; and a processor for processing the received signals to generate an image of the windings. It comprises: a transducer array and a driving module for emitting arc-shaped ultrasonic waves to the transformer winding and receiving reflected echo signals; wherein the transducer array is installed on a three-dimensional mechanical track; An arc data acquisition and processing module is configured to generate a plurality of sets of original arc data sets based on the echo signals and perform a spatial intersection operation on the plurality of sets of original arc data sets to extract three-dimensional coordinates of a winding surface; A signal processing and data fusion module is configured to perform envelope extraction and SAFT delay-and-sum focusing processing on the echo signals to generate a plurality of time-point enhanced signals, and bind the three-dimensional coordinates of the winding surface and the plurality of time-point enhanced signals according to time stamps to construct a three-dimensional data matrix with spatio-temporal fusion; An adaptive error correction module is configured to construct a dynamic correction model based on coordinate deviations of preset reference points of the winding surface and perform dynamic correction on the three-dimensional data matrix; A three-dimensional imaging module is configured to perform spatial depth mapping on the corrected three-dimensional data matrix to generate a pseudo-color three-dimensional point cloud map of the transformer winding.

Images