Intelligent identification and repair method for welding defects of heavy-duty gas turbine high-temperature alloy

Abstract

The present application relates to the technical field of heavy-duty gas turbine, and particularly relates to a heavy-duty gas turbine high-temperature alloy welding defect intelligent identification and repair method. By acquiring a welding area infrared thermal image sequence to identify an abnormal temperature drop area as a candidate defect area, then performing defect verification and type determination through ultrasonic phased array scanning, calculating a repair heat input curve according to the defect type, executing pulse modulation cladding repair and adjusting parameters according to real-time temperature field and acoustic emission signals, automatic identification and accurate repair of welding defects can be realized, and the repair quality and efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of heavy-duty gas turbine technology, and in particular to a method for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines. Background Technology

[0002] Heavy-duty gas turbine high-temperature alloy components operate under high temperature and high pressure environments for extended periods. Welded joints, as structurally weak points, directly determine the safety and lifespan of the unit. Currently, the industry primarily relies on offline non-destructive testing methods for detecting welding defects, such as conventional ultrasonic, radiographic, or penetrant testing, involving point-by-point scanning of the weld seam after welding. Repair typically employs manual grinding combined with arc remelting or welding repair, depending on the operator's experience to determine defect location and repair parameters.

[0003] Isolated operations during the inspection phase struggle to capture dynamically evolving microscopic defects during welding, particularly microcracks or incomplete fusion that develop during cooling. These defects may go undetected when the weld surface is covered by oxide scale or slag. Offline inspection also faces challenges such as acoustic attenuation and signal interference caused by the coarse-grained structure of high-temperature alloys, and often results in insufficient defect location accuracy in thin-walled areas. Manual arc welding during the repair phase lacks precise control over heat input, easily leading to new thermal stress concentrations and grain coarsening in the repair area, and even inducing secondary cracks. Post-repair non-destructive testing verification is required, creating a lengthy cycle of inspection-repair-re-inspection, significantly impacting production cycle time and costs.

[0004] Furthermore, high-temperature alloys are extremely sensitive to thermal cycling, and traditional repair methods struggle to monitor the solidification process of the molten pool in real time. Operators, relying solely on visual observation of the molten pool morphology, cannot quantitatively assess residual stress release and microstructural transformation. For defects in deep or complex geometries, blindly increasing the number of repair passes often leads to overheating of the substrate or excessive dilution. These technical limitations have long made the accurate identification and high-quality repair of welding defects in high-temperature alloys for heavy-duty gas turbines a difficult task, balancing efficiency and reliability. Summary of the Invention

[0005] This invention provides a method and system for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines, which can solve the problems in the prior art.

[0006] A first aspect of this invention provides a method for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines, comprising: A continuous infrared thermal image sequence of the high-temperature alloy welding area of ​​a heavy-duty gas turbine is obtained during the welding process and the post-weld cooling stage. Temperature change curves of each spatial location of the welding area are extracted from the image. Based on the temperature change curves, abnormal temperature drop areas that deviate from the normal cooling rate are identified as candidate defect areas. The candidate defect region is subjected to ultrasonic phased array scanning to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the authenticity of the defect region and the defect type are verified and determined to identify the defect type and spatial location of the real defect region. Based on the defect type, the heat conduction anisotropy parameters of the actual defect area are determined, and the repair heat input curve is calculated by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. According to the repair heat input curve, pulse modulation cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are collected in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse cycle are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

[0007] A continuous infrared thermal image sequence of the high-temperature alloy welded area of ​​a heavy-duty gas turbine is acquired during the welding process and post-weld cooling stage. Temperature change curves at various spatial locations within the welded area are extracted from these images. Based on these temperature change curves, regions with abnormal temperature drops deviating from the normal cooling rate are identified as candidate defect regions, including: The surface of the welding area is divided into a spatial grid with a preset spatial resolution. The average pixel gray value of each frame in the infrared thermal image sequence within the corresponding grid range is extracted grid by grid. The average pixel gray value is converted into a temperature value through a radiometric calibration function to obtain the temperature-time series of each spatial grid. The radiometric calibration function is pre-calibrated based on the surface emissivity of the high-temperature alloy material and the spectral response characteristics of the infrared thermal imager. The cooling rate curves of each spatial grid are obtained by performing time differentiation on the temperature-time series of each spatial grid. The cooling rate curves of each spatial grid are compared point by point with the reference cooling rate curves of the corresponding high-temperature alloy materials under defect-free conditions. The cooling rate deviation values ​​at each time point are calculated. The duration and cumulative magnitude of the cooling rate deviation values ​​of each spatial grid exceeding the preset deviation threshold during the cooling stage are statistically analyzed. Spatial grids whose duration and cumulative magnitude exceed the corresponding thresholds are marked as abnormal temperature drop grids. Spatial connectivity analysis is performed on the spatial grid marked as abnormal temperature drop grid. Abnormal temperature drop grids that are adjacent to each other and whose spatial distance is less than the preset adjacency distance threshold are aggregated into candidate defect regions. The geometric center coordinates, area, and average cooling rate deviation of each candidate defect region are recorded.

[0008] The candidate defect region is subjected to ultrasonic phased array scanning to obtain its reflection characteristics. Based on these reflection characteristics, the authenticity and type of the candidate defect region are verified, and the defect type and spatial location of the actual defect region are determined, including: The ultrasonic phased array elements are controlled to emit ultrasonic pulses and receive echo signals according to the time-delay focusing law. By adjusting the emission delay of each element, the ultrasonic pulses emitted by each element are superimposed in phase at each scanning point of the candidate defect area. The synthesized focused beam scans the candidate defect area point by point. The focal size of the focused beam is determined according to the detection sensitivity requirements and the minimum feature size of the candidate defect area. The amplitude and phase of the reflected echo at each scanning point of the focused beam are extracted. Using the coordinates of the scanning point as an index, the amplitude and phase of the reflected echo at each scanning point are combined into a multi-dimensional vector to form the reflection feature vector of the candidate defect region. The correlation between the reflection feature vector of each scanning point and the reference reflection feature vector of the defect-free reference area is calculated. If the correlation of a certain scanning point is lower than the preset correlation threshold, the scanning point is marked as a false defect point and removed. Spatial connectivity analysis is performed on the remaining scan points after removing false defect points. The statistical mean of the reflection feature vector of each scan point in each connectivity is calculated. The statistical mean is matched with the standard reflection feature template of each defect category, and the defect category with the highest matching degree is output as the defect type of the real defect region.

[0009] Based on the defect type, the thermal conductivity anisotropy parameters of the actual defect region are determined, and the repair thermal input curve is calculated by combining the defect depth and defect width, including: According to the defect type, the thermal conductivity anisotropy feature corresponding to the defect type is queried in the thermal conductivity directionality feature library of high-temperature alloy defects. The thermal conductivity anisotropy feature includes the main thermal conductivity coefficient direction along the longitudinal direction of the weld and the secondary thermal conductivity coefficient direction perpendicular to the main thermal conductivity coefficient direction. The thermal conductivity directionality feature library of high-temperature alloy defects is constructed by offline experimental calibration of the thermal conductivity directionality data of porosity, crack and non-fusion defects under their respective typical grain orientations. An anisotropic coordinate system is constructed with the geometric center of the actual defect region as the origin, the direction of the primary thermal conductivity coefficient as the first axis, and the direction of the secondary thermal conductivity coefficient as the second axis. The first axis and the second axis are orthogonal to each other and correspond to the direction with the strongest and weakest thermal conductivity, respectively. The thermal conductivity coefficient value in the direction of the primary thermal conductivity coefficient is set as the first thermal conductivity coefficient, and the thermal conductivity coefficient value in the direction of the secondary thermal conductivity coefficient is set as the second thermal conductivity coefficient. The first thermal conductivity coefficient and the second thermal conductivity coefficient are attenuated and corrected by the material density inside the defect region corresponding to the defect type, thereby forming an anisotropic thermal conductivity parameter matrix based on the anisotropic coordinate system.

[0010] The repair thermal input curve is calculated by combining the defect depth and defect width, including: Substitute the defect depth and defect width into the two-dimensional anisotropic thermal diffusion equation with the anisotropic thermal conduction parameter matrix as coefficients, and divide the real defect region into multiple depth layers along the defect depth direction. The thickness of each depth layer is adaptively determined according to the depth change gradient of the defect depth. The layer thickness is reduced where the defect depth gradient changes sharply to improve the solution accuracy. Solve the two-dimensional anisotropic thermal diffusion equation for each depth layer, and take the recrystallization temperature of the upper surface of the depth layer as the boundary condition to obtain the minimum heat input per unit area required to raise the temperature of the material in the depth layer from the initial temperature to the recrystallization temperature. An energy margin compensation related to the depth layer position is applied to the minimum unit area heat input of each depth layer. The closer the depth layer is to the bottom of the defect, the higher the compensation coefficient is applied. The compensation coefficient increases linearly with the depth of the depth layer to obtain the target unit area heat input of each depth layer. Using the defect depth as the horizontal axis and the target unit area heat input corresponding to each depth layer as the vertical axis, the discrete points are spliced ​​to generate a stepped layered heat input curve, and the stepped layered heat input curve is smoothed by spline processing to obtain the repair heat input curve.

[0011] According to the repair heat input curve, pulse-modulated cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are acquired in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the following is included: A reference peak power and a reference pulse width are set for pulse modulation cladding repair. Infrared thermal imager is used to collect surface infrared radiation data of the repair area in real time. The surface infrared radiation data is then converted into a two-dimensional surface temperature field through the law of thermal radiation. Acoustic emission waveform signals of the repair area during the pulse modulation cladding process are collected in real time using an acoustic emission sensor. The acoustic emission waveform signals are bandpass filtered to suppress environmental noise and electromagnetic interference. The passband range of the bandpass filter is determined based on the typical frequency band of high-temperature alloy solidification stress release. The filtered acoustic emission waveform signals are then normalized in amplitude. Short-time Fourier transform is performed on the acoustic emission waveform signal after amplitude normalization. The frequency components and energy amplitude of each time-frequency unit are extracted from the transform result. The time-frequency units that meet the preset signal-to-noise ratio condition are selected as candidate solidification events. Each candidate solidification event is located in the time domain. If the frequency component of a candidate solidification event is within the solidification stress release frequency band and its energy amplitude exceeds a preset energy threshold, then the candidate solidification event is marked as a solidification stress release event, and the frequency component and energy amplitude of the solidification stress release event are used as the solidification stress release feature.

[0012] Coordinated adjustment of peak power and pulse width for each pulse cycle, including: Calculate the average temperature deviation between the measured temperature field and the expected temperature field corresponding to the repair heat input curve within the actual defect area, and determine whether the deviation direction of the average temperature deviation is positive or negative. The peak power correction amount is generated based on the absolute value of the average temperature deviation. When the deviation direction is negative, the peak power correction amount is set to a positive value, and when the deviation direction is positive, the peak power correction amount is set to a negative value. A pulse width correction is generated based on the energy amplitude in the solidification stress release characteristics, and the pulse width correction is inversely related to the energy amplitude. The peak power correction is added to the reference peak power of the current pulse period to obtain the target peak power of the next pulse period, and the pulse width correction is added to the reference pulse width of the current pulse period to obtain the target pulse width of the next pulse period. The above adjustments are repeated until the equivalent value of residual stress in the repair area is lower than the first preset criterion and the area ratio of recrystallized grains in the metallographic structure of the repair area is higher than the second preset criterion, at which point the pulse modulation cladding repair ends.

[0013] A second aspect of the present invention provides an intelligent identification and repair system for welding defects in high-temperature alloys for heavy-duty gas turbines, comprising: The temperature anomaly unit is used to acquire a continuous infrared thermal image sequence of the high-temperature alloy welding area of ​​heavy-duty gas turbine during the welding process and the post-weld cooling stage, extract the temperature change curve of each spatial location of the welding area, and identify abnormal temperature drop areas that deviate from the normal cooling rate as candidate defect areas based on the temperature change curve. An ultrasonic defect unit is used to perform ultrasonic phased array scanning on the candidate defect region to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the candidate defect region is used to verify the authenticity of the defect and determine the defect type, thereby determining the defect type and spatial location of the real defect region. The repair parameter unit is used to determine the heat conduction anisotropy parameters of the actual defect area based on the defect type, and to calculate the repair heat input curve by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. The cladding control unit is used to perform pulse-modulated cladding repair on the actual defect area according to the repair heat input curve, and to collect the surface temperature field and acoustic emission signal of the repair area in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse period are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

[0014] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0016] By extracting temperature change curves from continuous infrared thermal image sequences, abnormal temperature drop areas deviating from the normal cooling rate are accurately identified, and candidate defect areas are screened out. Then, the authenticity and type of defects are verified by combining ultrasonic phased array scanning reflection characteristics, which effectively eliminates the interference of false defects, significantly improves the accuracy and reliability of defect detection, avoids the problems of missed detection or misjudgment in traditional single detection methods, and lays the foundation for subsequent accurate repair.

[0017] Based on the defect type, the anisotropic parameters of heat conduction are determined, and the repair heat input curve is calculated by combining the defect depth and width, thus achieving precise customization of the repair heat input. This curve limits the target heat input power and duration at each moment, ensuring that the heat distribution during the repair process is highly matched with the thermal characteristics of the defect area. This avoids overheating or insufficient cladding caused by improper heat input, significantly improving the forming quality and interfacial bonding strength of the repaired area, while reducing the size of the heat-affected zone.

[0018] During pulse-modulated cladding repair, surface temperature field and acoustic emission signals are acquired in real time. Based on the deviation between the measured and expected temperature fields and the characteristics of solidification stress release, the peak power and pulse width of each pulse cycle are adjusted in a coordinated manner. This closed-loop feedback control method dynamically compensates for heat accumulation and stress evolution during the welding process, effectively suppresses residual stress in the repair area, optimizes the metallographic structure, and continues until the preset repair qualification criteria are met. This significantly improves the repair success rate and consistency, and reduces the rework rate and subsequent processing costs. Attached Figure Description

[0019] Figure 1 A flowchart illustrating the intelligent identification and repair method for welding defects in high-temperature alloys for heavy-duty gas turbines; Figure 2 A flowchart of real-time monitoring for pulse modulation cladding repair, a method for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0022] Figure 1 This is a flowchart illustrating the intelligent identification and repair method for high-temperature alloy welding defects in heavy-duty gas turbines according to an embodiment of the present invention. The method includes: A continuous infrared thermal image sequence of the high-temperature alloy welding area of ​​a heavy-duty gas turbine is obtained during the welding process and the post-weld cooling stage. Temperature change curves of each spatial location of the welding area are extracted from the image. Based on the temperature change curves, abnormal temperature drop areas that deviate from the normal cooling rate are identified as candidate defect areas. The candidate defect region is subjected to ultrasonic phased array scanning to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the authenticity of the defect region and the defect type are verified and determined to identify the defect type and spatial location of the real defect region. Based on the defect type, the heat conduction anisotropy parameters of the actual defect area are determined, and the repair heat input curve is calculated by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. According to the repair heat input curve, pulse modulation cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are collected in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse cycle are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

[0023] A continuous infrared thermal image sequence of the high-temperature alloy welded area of ​​a heavy-duty gas turbine is acquired during the welding process and post-weld cooling stage. Temperature change curves at various spatial locations within the welded area are extracted from these images. Based on these temperature change curves, regions with abnormal temperature drops deviating from the normal cooling rate are identified as candidate defect regions, including: The surface of the welding area is divided into a spatial grid with a preset spatial resolution. The average pixel gray value of each frame in the infrared thermal image sequence within the corresponding grid range is extracted grid by grid. The average pixel gray value is converted into a temperature value through a radiometric calibration function to obtain the temperature-time series of each spatial grid. The radiometric calibration function is pre-calibrated based on the surface emissivity of the high-temperature alloy material and the spectral response characteristics of the infrared thermal imager. The cooling rate curves of each spatial grid are obtained by performing time differentiation on the temperature-time series of each spatial grid. The cooling rate curves of each spatial grid are compared point by point with the reference cooling rate curves of the corresponding high-temperature alloy materials under defect-free conditions. The cooling rate deviation values ​​at each time point are calculated. The duration and cumulative magnitude of the cooling rate deviation values ​​of each spatial grid exceeding the preset deviation threshold during the cooling stage are statistically analyzed. Spatial grids whose duration and cumulative magnitude exceed the corresponding thresholds are marked as abnormal temperature drop grids. Spatial connectivity analysis is performed on the spatial grid marked as abnormal temperature drop grid. Abnormal temperature drop grids that are adjacent to each other and whose spatial distance is less than the preset adjacency distance threshold are aggregated into candidate defect regions. The geometric center coordinates, area, and average cooling rate deviation of each candidate defect region are recorded.

[0024] Before acquiring a continuous sequence of infrared thermal images, the surface of the welded area needs to be preprocessed using spatial meshing. The visible surface of the high-temperature alloy welded area is divided into uniform rectangular spatial grids according to a preset spatial resolution. The grid size selection must balance the spatial resolution of the infrared thermal imager with the minimum defect size requirement. Typically, the grid side length is set to an order of magnitude equivalent to the actual size of a single pixel in the thermal imager to ensure that each grid contains a sufficient number of effective pixels for grayscale statistics. After the meshing is completed, for each frame in the infrared thermal image sequence, the grayscale values ​​of all pixels within the corresponding grid area are extracted grid by grid, and the mean is calculated to obtain the grayscale mean of that grid at that moment.

[0025] The grayscale mean itself does not directly reflect temperature; it needs to be converted into a temperature value using a radiometric calibration function. The radiometric calibration function is pre-calibrated based on the surface emissivity of the high-temperature alloy material and the spectral response characteristics of the infrared thermal imager. During calibration, multiple data acquisitions are performed on a high-temperature alloy standard sample at a known temperature under controlled conditions to establish a mapping relationship between the grayscale mean and the actual temperature. Since the surface emissivity of the high-temperature alloy fluctuates with temperature, calibration curves should be segmented within the complete temperature range covered by the welding and cooling stages. In actual use, the corresponding calibration segment is automatically selected for interpolation based on the current temperature range. After the radiometric calibration function conversion, each spatial grid corresponds to a temperature value at each acquisition time, thus forming the temperature-time series of that grid, denoted as . ,in For grid indexing, For the first Each data collection moment.

[0026] The cooling rate curves for each grid cell are obtained by performing time differentiation on the temperature-time series of each spatial grid. Time differentiation is implemented using a finite difference method. The grid in the first Cooling rate at each moment ,according to The calculations are performed using a central difference scheme to reduce the impact of noise on the differential results. During the welding process, due to the movement of heat sources and local heat accumulation, the temperature changes of each grid are quite drastic, and the cooling rate curve fluctuates significantly during this stage. After entering the post-weld cooling stage, the cooling rate in the normal region should follow the inherent cooling law of high-temperature alloy materials under this process condition, exhibiting a relatively stable attenuation trend. To suppress the interference of acquisition noise on the cooling rate curve, a slight time-domain smoothing process can be applied to the temperature-time series before differentiation. The width of the smoothing window must be selected to ensure that it does not mask the true abnormal temperature drop characteristics.

[0027] The cooling rate curves of each spatial grid were compared point-by-point with the benchmark cooling rate curves of the corresponding high-temperature alloy materials under defect-free conditions. The benchmark cooling rate curves were obtained through statistical analysis of extensive experimental data from defect-free welded specimens of similar high-temperature alloys under the same process parameters, reflecting the expected cooling rate and its statistical distribution range at each moment of the cooling stage in the normal welded region. For the first... The grid in the first Calculate the cooling rate deviation value at each time point. ,in As a reference cooling rate curve in Reference value at any given time. When defects such as porosity, cracks, and lack of fusion are present, the heat conduction path in the defective area is blocked, causing the heat dissipation rate in that area to differ from that in the normal area during the cooling phase, which is manifested as a deviation in cooling rate. The value continuously deviates from zero and exceeds a preset deviation threshold within a certain period of time. .

[0028] The analysis showed that the cooling rate deviation of each spatial grid exceeded a preset deviation threshold during the cooling phase. The duration and cumulative magnitude of the deviation. Duration is defined as the cumulative time the deviation value continuously exceeds the threshold, and cumulative magnitude is defined as the time integral of the absolute value of the deviation value during the period exceeding the threshold. The duration exceeding the threshold is considered a critical factor. And the cumulative magnitude exceeds the threshold The spatial mesh is marked as an abnormal temperature drop mesh. The dual-threshold design can effectively eliminate transient deviations caused by local transient disturbances, such as welding spatter and local peeling of surface oxide film, and improve the reliability of candidate defect identification. , and The three threshold parameters are determined comprehensively based on the type of high-temperature alloy material, welding process parameters, and thermal imager performance indicators. In practical engineering applications, they can be optimized and adjusted through verification experiments on known defect samples.

[0029] After marking the anomalous temperature drop meshes, spatial connectivity analysis is performed on all marked meshes. In actual welding defects, the response of a single defect in the surface temperature field often covers multiple adjacent meshes. If subsequent processing is performed directly on a single mesh basis, the same defect will be segmented into multiple independent targets. Spatial connectivity analysis calculates the spatial distance between each anomalous temperature drop mesh, selecting meshes with spatial distances less than a preset adjacency distance threshold. Furthermore, adjacent anomalous temperature drop meshes aggregate into the same candidate defect region. Adjacency distance threshold. The settings should refer to the typical size range of high-temperature alloy welding defects, and are usually set to be slightly larger than the side length of a single grid to ensure that adjacent grids belonging to the same defect can be correctly merged, while avoiding the mistaken merging of independent defects that are far apart in space into one area.

[0030] After aggregation, three characteristic parameters are calculated and recorded for each candidate defect region: geometric center coordinates, region area, and average cooling rate deviation within the region. The geometric center coordinates are obtained by averaging the center coordinates of all grids within the candidate region, used to locate the approximate position of the defect on the weld area surface, providing targeted positioning information for subsequent ultrasonic phased array scanning. The region area is obtained by multiplying the total number of aggregated grids by the actual area of ​​a single grid, reflecting the spatial extent of the candidate defect projected onto the surface. The average cooling rate deviation within the region is obtained by averaging the deviation values ​​of all grids within the candidate region during the threshold-exceeding period, serving as a quantitative indicator describing the severity of the thermal anomaly of the candidate defect, providing a reference for subsequent defect authenticity verification and defect type determination. These three characteristic parameters together constitute the preliminary descriptive information for each candidate defect region, supporting accurate positioning and verification in subsequent ultrasonic phased array detection.

[0031] The candidate defect region is subjected to ultrasonic phased array scanning to obtain its reflection characteristics. Based on these reflection characteristics, the authenticity and type of the candidate defect region are verified, and the defect type and spatial location of the actual defect region are determined, including: The ultrasonic phased array elements are controlled to emit ultrasonic pulses and receive echo signals according to the time-delay focusing law. By adjusting the emission delay of each element, the ultrasonic pulses emitted by each element are superimposed in phase at each scanning point of the candidate defect area. The synthesized focused beam scans the candidate defect area point by point. The focal size of the focused beam is determined according to the detection sensitivity requirements and the minimum feature size of the candidate defect area. The amplitude and phase of the reflected echo at each scanning point of the focused beam are extracted. Using the coordinates of the scanning point as an index, the amplitude and phase of the reflected echo at each scanning point are combined into a multi-dimensional vector to form the reflection feature vector of the candidate defect region. The correlation between the reflection feature vector of each scanning point and the reference reflection feature vector of the defect-free reference area is calculated. If the correlation of a certain scanning point is lower than the preset correlation threshold, the scanning point is marked as a false defect point and removed. Spatial connectivity analysis is performed on the remaining scan points after removing false defect points. The statistical mean of the reflection feature vector of each scan point in each connectivity is calculated. The statistical mean is matched with the standard reflection feature template of each defect category, and the defect category with the highest matching degree is output as the defect type of the real defect region.

[0032] Ultrasonic phased array scanning plays a crucial role in the accurate verification and classification of candidate defect regions in the detection of high-temperature alloy welding defects in heavy-duty gas turbines. The phased array probe consists of multiple independent array elements, each capable of independently controlling its transmission timing. By applying a time-delay focusing principle, a specific transmission delay is applied to each element, causing a controllable phase difference in the propagation path of the ultrasonic pulses emitted by each element. This results in in-phase superposition at the target scanning point, forming a focused, energy-concentrated synthetic beam. Specifically, let the... The acoustic path from each array element to the target scanning point is The speed of sound in high-temperature alloys is Then the first The launch delay of each array element satisfy ,in This is the maximum value of the acoustic path from all array elements to the target scan point. By updating the delay parameters of each array element point by point, the focus of the focused beam can perform point-by-point scanning coverage within the candidate defect area.

[0033] Determining the focal spot size requires considering two constraints: first, the detection sensitivity requirement, meaning the sound pressure amplitude at the focal spot must be sufficient to excite minute defects to generate detectable reflected echoes; second, the minimum feature size of the candidate defect region, meaning the focal spot size should not exceed the minimum lateral size of the defect to be detected, otherwise adjacent defect features will overlap, leading to insufficient spatial resolution. In practical engineering, high-temperature alloy welded areas may contain various defects such as microcracks, porosity, and lack of fusion. The opening width of microcracks can be as low as tens of micrometers. Therefore, the focal spot size needs to be controlled within a range comparable to the minimum feature size, which is usually achieved through coordinated optimization of parameters such as element spacing, excitation frequency, and focusing depth.

[0034] After the focused beam is excited at each scanning point, the receiver also uses a delay superposition method to perform phase alignment and synthesis of the echo signals received by each array element, thereby extracting the reflected echo information at that scanning point. For each scanning point, two key features are extracted: reflected echo amplitude and reflected echo phase. The reflected echo amplitude reflects the strength of the acoustic impedance discontinuity at that point. Volumetric defects such as pores and inclusions usually produce strong omnidirectional scattered echoes, while planar defects such as cracks exhibit direction-dependent amplitude characteristics due to orientation sensitivity. The reflected echo phase carries information about the interface type. There is a 180° difference in the reflection phase between solid-gas interfaces and solid-solid interfaces, which can be used to distinguish defects with different physical properties, such as pores and inclusions. The scanning point coordinates are used as the basis for this process. Use this as an index to determine the amplitude of the reflected echo at that point. Phase with reflected echo The combination constitutes the reflection feature vector of the scanning point. Repeat the above extraction process for all scanning points within the candidate defect region to form a complete set of reflection feature vectors for the candidate defect region.

[0035] During the defect verification stage, the reflection feature vectors of each scanning point are... The reference reflection feature vector of the defect-free reference region Correlation calculations were performed. The reference reflection feature vector was derived from a defect-free area confirmed by metallographic cutting in the same batch of high-temperature alloy welded joints, acquired under the same ultrasonic parameters, and represents the reflection feature background of the normal welded area. Correlation Defined as and The normalized inner product between them, i.e. When the correlation of a certain scan point Higher than the preset relevance threshold When the reflection characteristics of a point are highly similar to those of a normal area, it is considered a false defect point resulting from misjudgment during the thermal imaging stage, and is marked and removed from the candidate set; when Below If a point is found to have a genuine acoustic impedance anomaly, it is retained as a valid detection point. A preset correlation threshold is used. The calibration needs to be performed based on the microstructure uniformity and ultrasonic background noise level of the specific high-temperature alloy material. It is usually determined by optimizing the receiver operating characteristic curve on known defect samples to achieve a reasonable balance between false alarm rate and false negative rate.

[0036] Spatial connectivity analysis is performed on the remaining valid scan points after removing spurious defect points. This analysis is based on the adjacency relationships of the scan points in 3D space: if the Euclidean distance between two valid scan points in 3D space does not exceed a certain multiple of the scan step distance, they are considered spatially connected and belong to the same defect connectivity region. By progressively merging points that satisfy the connectivity condition, all valid scan points are ultimately divided into several independent connectivity regions, each corresponding to an independent real defect region. The spatial extent of the connectivity region directly provides the 3D position coordinates and spatial extension range of the defect, i.e., the defect depth and width information. These parameters will be directly used in the subsequent calculation of the repair thermal input curve.

[0037] Calculate the statistical mean of the reflection feature vectors of all scan points within each connected component to obtain the representative feature vector of that connected component. The statistical mean is calculated by taking the arithmetic mean of the reflection feature vectors of all scan points within the connected domain, component by component. This suppresses single-point measurement noise while preserving the overall reflection feature information of the entire defect region. Subsequently, Matching is performed against a pre-established standard reflection feature template library for each defect category. The standard reflection feature template library contains standard feature templates for typical defect categories such as porosity, microcracks, lack of fusion, and inclusions. Each template is derived from statistical analysis of ultrasonic phased array data of a large number of known defect samples, exhibiting strong intra-class representativeness. The matching degree is calculated using... With various templates Weighted Euclidean distance between ,in Represented by the weight matrix Weighted norm, weight matrix The weighting is determined based on the contribution of each feature component to the defect category differentiation ability, with higher weights assigned to feature components that have strong differentiation ability. The defect category with the smallest output distance, representing the highest matching degree, is then determined. This serves as the result of determining the defect type of the actual defect region corresponding to the connected domain.

[0038] After determining the defect type for all connected regions within the candidate defect area, a complete set of real defect areas is obtained. Each real defect area has a clear defect type label and three-dimensional spatial location information, providing accurate input parameters for subsequent calculation of the repair thermal input curve based on the defect type. In actual testing, the combined use of ultrasonic phased array scanning and infrared thermal imaging can effectively overcome the limitations of single detection methods: infrared thermal imaging has a high detection rate for surface and near-surface defects, but is susceptible to false alarms due to background interference from thermal radiation during welding; ultrasonic phased array scanning has the ability to penetrate internal defects, and through a correlation verification mechanism, it can effectively eliminate false defects in the thermal imaging stage. The two complement each other to form a reliable dual-modal defect identification link, ensuring that repair decisions are based on accurate defect information.

[0039] Based on the defect type, the thermal conductivity anisotropy parameters of the actual defect region are determined, and the repair thermal input curve is calculated by combining the defect depth and defect width, including: According to the defect type, the thermal conductivity anisotropy feature corresponding to the defect type is queried in the thermal conductivity directionality feature library of high-temperature alloy defects. The thermal conductivity anisotropy feature includes the main thermal conductivity coefficient direction along the longitudinal direction of the weld and the secondary thermal conductivity coefficient direction perpendicular to the main thermal conductivity coefficient direction. The thermal conductivity directionality feature library of high-temperature alloy defects is constructed by offline experimental calibration of the thermal conductivity directionality data of porosity, crack and non-fusion defects under their respective typical grain orientations. An anisotropic coordinate system is constructed with the geometric center of the actual defect region as the origin, the direction of the primary thermal conductivity coefficient as the first axis, and the direction of the secondary thermal conductivity coefficient as the second axis. The first axis and the second axis are orthogonal to each other and correspond to the direction with the strongest and weakest thermal conductivity, respectively. The thermal conductivity coefficient value in the direction of the primary thermal conductivity coefficient is set as the first thermal conductivity coefficient, and the thermal conductivity coefficient value in the direction of the secondary thermal conductivity coefficient is set as the second thermal conductivity coefficient. The first thermal conductivity coefficient and the second thermal conductivity coefficient are attenuated and corrected by the material density inside the defect region corresponding to the defect type, thereby forming an anisotropic thermal conductivity parameter matrix based on the anisotropic coordinate system.

[0040] After obtaining the defect type and spatial location of the actual defect region, it is necessary to establish an accurate thermal conduction description for each actual defect region to provide a physical basis for the subsequent calculation of the heat input curve for repair. Due to differences in grain orientation, phase composition, and defect morphology, the thermal conduction behavior of the welded area of ​​high-temperature alloys exhibits a significant direction dependence, with different defect types showing marked differences in heat transfer paths. Porous defects contain enclosed gas with extremely low thermal conductivity, and heat is mainly conducted along the base material, bypassing the pores. Crack defects, due to discontinuous contact on both sides of the crack surface, experience strong heat flow blockage in the direction perpendicular to the crack surface, while thermal conduction along the crack extension direction is less affected. Incomplete fusion defects, due to the presence of incomplete metallurgical bonding interfaces in local areas, exhibit attenuation characteristics in the interface normal direction that are intermediate between those of pores and cracks. To quantitatively describe these differences, offline experiments were conducted to systematically calibrate the thermal conduction directionality data of three typical defects under their respective common grain orientations, constructing a thermal conduction directionality feature library for high-temperature alloy defects.

[0041] Offline calibration experiments used samples of the same grade as those used in actual heavy-duty gas turbine high-temperature alloys. Under controlled conditions, standard defects such as porosity, cracks, and lack of fusion were artificially prepared. The thermal conductivity coefficients of each defect along different grain orientations were measured using laser scintillation and steady-state heat flow methods, and the correspondence between each measurement direction and the defect geometry was recorded. For each defect type, at least three typical grain orientations were covered to ensure sufficient representativeness of the feature library. After calibration, the thermal conductivity directionality data corresponding to each defect type were stored in a structured format, including the primary thermal conductivity direction, the secondary thermal conductivity direction, and the corresponding measured values ​​of thermal conductivity coefficients in each direction, forming an online queryable feature library of thermal conductivity directionality for high-temperature alloy defects.

[0042] During the actual repair calculation phase, the defect type index determined by ultrasonic phased array detection is used. The anisotropic thermal conductivity characteristics corresponding to this defect type were retrieved from a database of thermal conductivity directionality features for high-temperature alloy defects. The search results included descriptions of two key directions: the primary thermal conductivity direction extending longitudinally along the weld, and the secondary thermal conductivity direction perpendicular to the primary thermal conductivity direction. The primary thermal conductivity direction corresponds to the path of least heat transfer resistance and easiest heat flow, typically coinciding with the main extension direction of the defect or the columnar growth direction of the grains. The secondary thermal conductivity direction corresponds to the path of greatest heat transfer resistance, where heat propagation is significantly hindered by defect morphology or grain boundary distribution. These two directions are orthogonal to each other in three-dimensional space, jointly defining the main framework of thermal conduction behavior in the defect region.

[0043] An anisotropic coordinate system is constructed with the geometric center of the actual defect region as the origin, the direction of the primary thermal conductivity coefficient as the first axis, and the direction of the secondary thermal conductivity coefficient as the second axis. The geometric center of the defect is obtained by calculating the centroid of the three-dimensional defect volume reconstructed by ultrasonic phased array scanning, ensuring that the origin of the coordinate system corresponds precisely to the actual location of the defect. The first and second axes are both unit vectors and are strictly orthogonal. The third axis is determined by the cross product of the first and second axes, and the three axes together form a right-handed orthogonal coordinate system. Under this anisotropic coordinate system, the thermal conduction behavior can be characterized by thermal conductivity coefficients that are independently described along the three axes, thus transforming the anisotropic problem, which was originally coupled in the global coordinate system, into a decoupled form in the anisotropic coordinate system, facilitating the numerical processing of subsequent heat input calculations.

[0044] The thermal conductivity value corresponding to the direction of the principal thermal conductivity (first axis) is set as the first thermal conductivity. The thermal conductivity value corresponding to the direction of the secondary thermal conductivity (second axis) is set as the second thermal conductivity. Thermal conductivity in the third axis direction Values ​​are assigned based on the 3D calibration data corresponding to the defect type in the feature library. Among the three typical defect types, porosity defects have extremely low internal gas thermal conductivity. , , All three values ​​were significantly lower than the baseline value of the base material, and the differences among them were relatively small, indicating a weak degree of anisotropy; crack-type defects along the in-plane direction (i.e., the first axis) Approaching the horizontal plane of the parent material, but perpendicular to the crack plane (i.e., the second axis). Significant attenuation, with the most pronounced anisotropy; defects of the non-fusion type and The attenuation rate falls between that of porosity and cracks, and is closely related to the contact area ratio of the unfused interface and the degree of interface oxidation.

[0045] After assigning the initial thermal conductivity value, it is also necessary to adjust the value based on the material density within the defect region. and Attenuation correction is applied. Material density. Defined as the ratio of the actual solid material volume within the defect region to the total volume of the defect region, with a value range of [value missing]. The three-dimensional echo amplitude distribution reconstructed from an ultrasonic phased array was statistically calculated after threshold segmentation. Lower density indicates a larger proportion of pores or discontinuous interfaces within the defect region, resulting in a stronger blocking effect on heat conduction. The corrected first thermal conductivity... With the second thermal conductivity By respectively , The correction factor is obtained by multiplying it by the density correction factor. The specific functional form of the correction factor is determined based on the experimental fitting results of different defect types. Linear correction is used for porosity defects, and power function correction is used for crack and non-fusion defects to more accurately reflect the nonlinear influence of the internal microstructure of various defects on heat conduction attenuation.

[0046] The revised , and The combination forms a diagonal matrix, where the thermal conductivity along the third axis remains unchanged when unaffected by density correction, or is corrected according to the same rule, thus constituting an anisotropic thermal conductivity parameter matrix based on an anisotropic coordinate system. ,Right now The matrix is ​​diagonal in an anisotropic coordinate system, and its form in the global coordinate system is determined by a coordinate transformation matrix. (Constituted by unit vectors of the three axes of an anisotropic coordinate system arranged in columns) is converted into a global anisotropic heat conduction tensor. This allows for direct use of the thermal field simulation model during subsequent calculations of the repair heat input curve. Through this process, a complete mapping from defect type to anisotropic heat conduction parameter matrix is ​​achieved, providing a physically accurate and numerically operable foundation for calculating the repair heat input curve.

[0047] The repair thermal input curve is calculated by combining the defect depth and defect width, including: Substitute the defect depth and defect width into the two-dimensional anisotropic thermal diffusion equation with the anisotropic thermal conduction parameter matrix as coefficients, and divide the real defect region into multiple depth layers along the defect depth direction. The thickness of each depth layer is adaptively determined according to the depth change gradient of the defect depth. The layer thickness is reduced where the defect depth gradient changes sharply to improve the solution accuracy. Solve the two-dimensional anisotropic thermal diffusion equation for each depth layer, and take the recrystallization temperature of the upper surface of the depth layer as the boundary condition to obtain the minimum heat input per unit area required to raise the temperature of the material in the depth layer from the initial temperature to the recrystallization temperature. An energy margin compensation related to the depth layer position is applied to the minimum unit area heat input of each depth layer. The closer the depth layer is to the bottom of the defect, the higher the compensation coefficient is applied. The compensation coefficient increases linearly with the depth of the depth layer to obtain the target unit area heat input of each depth layer. Using the defect depth as the horizontal axis and the target unit area heat input corresponding to each depth layer as the vertical axis, the discrete points are spliced ​​to generate a stepped layered heat input curve, and the stepped layered heat input curve is smoothed by spline processing to obtain the repair heat input curve.

[0048] Obtain the anisotropic heat conduction parameter matrix Next, the geometric dimensions of the defect need to be incorporated into the thermal diffusion equation to calculate the heat input curve that can guide the repair process. Defect depth With defect width All images are from 3D images reconstructed by ultrasonic phased array scanning, including defect depth. The defect width is defined as the maximum distance the defect extends in the direction perpendicular to the weld surface. Defined as the maximum lateral dimension of the defect in the direction parallel to the weld surface. and Substitute it as a geometric boundary parameter The two-dimensional anisotropic thermal diffusion equation, which uses the diffusion coefficient tensor, describes the propagation of heat in anisotropic media within the plane of the defect cross-section. Its form is: ,in Density of high-temperature alloy materials For specific heat capacity, For temperature field distribution, It is a time variable. Because... Different defect types have different off-diagonal components. Solving the equation directly in the global coordinate system will introduce a large numerical coupling error. Therefore, before substituting the geometric parameters, the coordinate system of the defect region is aligned to the anisotropic principal axis direction to simplify the numerical solution of the equation.

[0049] The actual defect region is discretized into layers along the defect depth direction, covering the entire depth range. Divided into There are several depth layers. The thickness of each depth layer is not uniformly distributed, but adaptively determined based on the depth gradient along the defect depth direction. Specifically, the curvature distribution of the defect boundary along the depth direction is extracted from the ultrasonic phased array reconstructed image and quantized into a depth gradient sequence. ,in Indicates the first The rate of change of the local slope of the defect boundary at the depth of the layer. Larger depth locations (i.e., locations with abrupt boundary changes) correspond to layer thicknesses. A smaller value is chosen to improve the accuracy of the solution to the heat diffusion equation in this region; At smaller depths (i.e., with gently sloping boundaries), The layer thickness can be appropriately increased to reduce the amount of calculation. With gradient They satisfy an inverse proportional relationship ,in Based on the reference value for the base layer thickness, This is the gradient sensitivity coefficient, used to control how sensitive the layer thickness is to changes in gradient. The sum of the layer thicknesses of all depth layers is strictly equal to... ,Right now This ensures that the layers cover the entire range of defect depth.

[0050] Solve the two-dimensional anisotropic thermal diffusion equation for each depth layer separately, so that the surface temperature of the upper surface of that depth layer reaches the recrystallization temperature of the high-temperature alloy. For thermal boundary conditions, the lower surface and sides adopt insulating boundary conditions or continuity conditions with adjacent layers. The initial condition is the measured temperature of the defect area before repair. This value is provided by the final frame reading from the infrared thermal imager before repair. Under the above boundary and initial conditions, the solution for each depth layer is obtained to determine the temperature of the material in that layer from [previous value]. Rise to Minimum required heat input per unit area . The physical meaning is: under ideal conditions where heat loss is negligible, the first... The minimum heat energy density applied per unit cross-sectional area of ​​the depth layer allows the material in that layer to just complete the temperature rise required for recrystallization. This is due to the anisotropic thermal conductivity parameter matrix. The presence of this element results in different heat diffusion paths at different depths, leading to variations in the heat diffusion paths between layers. There are differences; the deeper layers near the bottom of the defect require a longer path for heat to be conducted to the upper surface. Typically, it is above the depth layer near the top of the defect.

[0051] Relying solely on The insufficient heat input for repair is due to heat loss via radiation, convection cooling, and thermal diffusion losses caused by the high thermal conductivity of the high-temperature alloy itself during the actual repair process. These factors all result in the effective heat reaching the target depth layer being lower than the theoretically calculated value. Therefore, for each depth layer... Apply energy margin compensation related to depth layer location, and introduce a compensation coefficient. . Numbering by depth layer The value increases linearly (i.e., the closer to the bottom of the defect), and the specific expression is: ,in The initial compensation coefficients are those corresponding to the top depth layer. This represents the maximum compensation coefficient corresponding to the lowest depth layer. and All were pre-calibrated based on the thermophysical properties of high-temperature alloy materials and experience data from repair processes. After compensation was applied, the... Target heat input per unit area in depth layer for The physical basis of this linear incremental compensation strategy is that the bottom of the defect is farthest from the heat source for repair, and the energy dissipation in the heat transfer path is the greatest. Therefore, the highest compensation ratio is required to ensure that the bottom material can be fully recrystallized, thereby eliminating residual porosity or cracks.

[0052] Obtain the target heat input per unit area at each depth layer Then, with the defect depth as the horizontal axis, the corresponding depth layers are plotted. Using the vertical axis as the coordinate, the heat input data points of each discrete depth layer are arranged sequentially to form a stepped layered heat input curve. This stepped curve intuitively reflects the overall increase in heat input with increasing defect depth, as well as the local fluctuation characteristics caused by layer thickness changes at the abrupt gradient change at the defect boundary. However, the stepped curve has abrupt jumps between layers. If directly used to control pulse-modulated cladding repair equipment, it will cause instantaneous impacts in power output during interlayer switching, leading to local overheating or thermal stress concentration. Therefore, the stepped layered heat input curve is subjected to cubic spline smoothing, with the center depth of each depth layer as the spline node. As the function values ​​at the nodes, a smooth transition curve is constructed between the nodes using cubic spline interpolation. Simultaneously, natural boundary conditions (second derivative equal to zero) are applied at both ends of the curve to ensure no artificial warping occurs at the endpoints. After spline smoothing, a continuous and smooth repair heat input curve is obtained. ,in For the defect depth coordinates, The target heat input power density per unit area is given at any depth location.

[0053] Repairing the heat input curve The final output is presented to the pulse-modulated cladding repair control unit in the form of a power-time relationship. During actual repair execution, as the repair heat source advances layer by layer towards the bottom of the defect, the control unit adjusts its output based on the current repair depth. Query And combined with the repair of the light spot area Converting heat input per unit area into actual output power At the same time, based on the duration of each pulse cycle, it ensures that the cumulative heat input within each pulse cycle is consistent with... Consistent. This method of generating heat input curves based on the joint calculation of defect geometric parameters and anisotropic thermal conductivity can provide precisely matched thermal energy supply to defect materials at different depths during the repair process, avoiding incomplete repair due to insufficient heat input or thermal damage to the parent material due to excessive heat input. This lays a reliable energy input foundation for the coordinated control of subsequent pulse-modulated cladding repair.

[0054] Figure 2 A flowchart of real-time monitoring for pulse modulation cladding repair, a method for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines.

[0055] According to the repair heat input curve, pulse-modulated cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are acquired in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the following is included: A reference peak power and a reference pulse width are set for pulse modulation cladding repair. Infrared thermal imager is used to collect surface infrared radiation data of the repair area in real time. The surface infrared radiation data is then converted into a two-dimensional surface temperature field through the law of thermal radiation. Acoustic emission waveform signals of the repair area during the pulse modulation cladding process are collected in real time using an acoustic emission sensor. The acoustic emission waveform signals are bandpass filtered to suppress environmental noise and electromagnetic interference. The passband range of the bandpass filter is determined based on the typical frequency band of high-temperature alloy solidification stress release. The filtered acoustic emission waveform signals are then normalized in amplitude. Short-time Fourier transform is performed on the acoustic emission waveform signal after amplitude normalization. The frequency components and energy amplitude of each time-frequency unit are extracted from the transform result. The time-frequency units that meet the preset signal-to-noise ratio condition are selected as candidate solidification events. Each candidate solidification event is located in the time domain. If the frequency component of a candidate solidification event is within the solidification stress release frequency band and its energy amplitude exceeds a preset energy threshold, then the candidate solidification event is marked as a solidification stress release event, and the frequency component and energy amplitude of the solidification stress release event are used as the solidification stress release feature.

[0056] Before performing pulse-modulated cladding repair on the actual defect area, a reference peak power and a reference pulse width need to be pre-set based on the repair heat input curve. The reference peak power is determined based on the anisotropic parameters of heat conduction in the defect area and the target heat input, ensuring that the output power is within a safe and controllable range at the beginning of the repair. The reference pulse width is estimated based on the thermal diffusion characteristics of the high-temperature alloy material and the response time of the molten pool, avoiding excessive concentration of energy in a single pulse that could lead to local overheating or microcrack propagation. The setting of the reference parameters provides an initial operating point for subsequent real-time closed-loop regulation, ensuring that the pulse-modulated cladding repair process has a reasonable energy output basis before entering dynamic feedback control.

[0057] During the restoration process, continuous infrared radiation data of the repaired area is acquired using an infrared thermal imager. The frame rate must meet the time resolution requirements for temperature transients within a single pulse cycle. The infrared imager's detection band is selected from the mid-wave or long-wave infrared bands, where the high-temperature alloy exhibits high radiation intensity within the restoration temperature range, to ensure the signal-to-noise ratio meets the temperature measurement accuracy requirements. The acquired surface infrared radiation data is converted into a two-dimensional surface temperature field according to the law of thermal radiation. Specifically, the emissivity of the high-temperature alloy surface in the repaired area is incorporated into the conversion. The emissivity varies with the surface oxidation state and temperature. Therefore, before repair, the emissivity of the target material within a typical repair temperature range needs to be calibrated, and the radiance data is corrected based on the calibration results to obtain a physically meaningful two-dimensional surface temperature field distribution. The obtained two-dimensional surface temperature field is discretized and stored in spatial pixels, with each pixel corresponding to a spatial location on the surface of the repair area. The pixel temperature value serves as the measured basis for subsequent deviation calculations.

[0058] A sensor array is deployed at an appropriate location in the repair area using acoustic emission sensors to acquire acoustic emission waveform signals generated during pulse-modulated cladding in real time. The coupling method and installation position of the sensors must comprehensively consider the geometry of the repair area and the workpiece temperature to ensure that the attenuation and mode conversion of the acoustic emission waveform along the propagation path do not lead to distortion of the effective signal. The acquired raw acoustic emission waveform signals contain broadband noise from electromagnetic interference from welding equipment, cooling airflow vibration, and mechanical structure conduction, which needs to be bandpass filtered to suppress these interferences. The passband range of the bandpass filter is determined based on the typical frequency band of solidification stress release in high-temperature alloys. This frequency band is usually obtained by statistical analysis of acoustic emission signals generated by similar high-temperature alloy materials in controlled solidification experiments, covering the characteristic frequency range corresponding to microscopic physical processes such as solidification crack initiation and interdendritic stress concentration release. The bandpass filter adopts a linear-phase finite-impulse response structure to avoid introducing nonlinear phase distortion within the passband, thereby preserving the temporal waveform integrity of the acoustic emission waveform. After filtering, the acoustic emission waveform signal is normalized to map the waveform amplitude to a uniform dimensionless range, eliminating amplitude baseline drift caused by differences in sensor sensitivity or changes in coupling conditions, and making the subsequent feature extraction results comparable across channels.

[0059] A short-time Fourier transform (SFT) is performed on the amplitude-normalized acoustic emission waveform signal to obtain the energy distribution of the signal in the joint time-frequency domain. The type and length of the SFT window function need to be selected comprehensively based on the duration of the solidification stress release event and the frequency resolution requirements: a window length that is too short will result in insufficient frequency resolution, making it impossible to distinguish different physical mechanisms in adjacent frequency bands; a window length that is too long will result in a decrease in time resolution, making it difficult to accurately locate fast-changing solidification events in the time domain. A Hanning window or a flat-top window is selected as the window function to suppress the interference of spectral leakage on the energy estimation of adjacent time-frequency units. The transform result forms a time-frequency matrix, where each time-frequency unit corresponds to the energy amplitude at a specific time and a specific frequency component. This will satisfy the preset signal-to-noise ratio condition. The time-frequency unit is used as a candidate solidification event, where This serves as a baseline for estimating background noise energy in the corresponding frequency band during the non-repair silence period. A preset signal-to-noise ratio threshold is used to distinguish between real physical events and residual noise.

[0060] Each candidate solidification event was time-domain localized to determine its occurrence time within the repair pulse sequence, and its frequency components and energy amplitude were extracted as feature quantities. The criteria for determining solidification stress release events involve two aspects: first, the dominant frequency component of the candidate solidification event. It must fall within the pre-defined solidification stress relief frequency band. Within this frequency band, determined by the solidification physics mechanism of high-temperature alloys, the elastic wave radiation characteristics during the release of interdendritic micro-stress and the advancement of the solidification front must be reflected; secondly, the energy amplitude of the candidate solidification event must exceed a preset energy threshold. This threshold is determined statistically based on the typical energy levels of known solidification stress release events, and is used to exclude the influence of low-energy background disturbances on the judgment results. Candidate solidification events that simultaneously meet both of the above conditions are marked as solidification stress release events, and their corresponding dominant frequency components... With energy amplitude Together, they constitute the solidification stress release characteristics, which are used for subsequent coordinated adjustment of pulse peak power and pulse width.

[0061] Frequency components of solidification stress release events This reflects the evolution of the solidification microstructure within the repaired area: when When the frequency band is biased towards the lower end, it usually corresponds to larger-scale stress release behavior, indicating a higher risk of residual stress accumulation within the repair area; when When distributed at the higher end of the frequency band, it corresponds to smaller microscopic stress release events, indicating a relatively uniform solidification process. Energy amplitude This directly reflects the intensity of a single solidification stress release event. A persistently high level of solidification stress in the repair area indicates that the solidification rate exceeds expectations. This necessitates reducing the peak power of subsequent pulses or extending the pulse interval to slow the solidification rate and reduce residual stress accumulation. By integrating the aforementioned solidification stress release characteristics with the deviation of the measured temperature field from the expected temperature field into the repair control logic, closed-loop adjustment of the peak power and pulse width for each pulse cycle is achieved. This ensures that the final residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

[0062] Coordinated adjustment of peak power and pulse width for each pulse cycle, including: Calculate the average temperature deviation between the measured temperature field and the expected temperature field corresponding to the repair heat input curve within the actual defect area, and determine whether the deviation direction of the average temperature deviation is positive or negative. The peak power correction amount is generated based on the absolute value of the average temperature deviation. When the deviation direction is negative, the peak power correction amount is set to a positive value, and when the deviation direction is positive, the peak power correction amount is set to a negative value. A pulse width correction is generated based on the energy amplitude in the solidification stress release characteristics, and the pulse width correction is inversely related to the energy amplitude. The peak power correction is added to the reference peak power of the current pulse period to obtain the target peak power of the next pulse period, and the pulse width correction is added to the reference pulse width of the current pulse period to obtain the target pulse width of the next pulse period. The above adjustments are repeated until the equivalent value of residual stress in the repair area is lower than the first preset criterion and the area ratio of recrystallized grains in the metallographic structure of the repair area is higher than the second preset criterion, at which point the pulse modulation cladding repair ends.

[0063] During pulse-modulated cladding repair, there is often a dynamic deviation between the real-time acquired surface temperature field of the repair area and the expected temperature field. This deviation directly reflects the degree of matching between the current heat input and the actual thermal response of the defect area. To achieve coordinated adjustment of the peak power and pulse width of each pulse cycle, it is first necessary to quantitatively compare the measured temperature field with the expected temperature field, extract the average temperature deviation value, and determine the direction of the deviation.

[0064] Within the actual defect area, the difference between the measured temperature field and the expected temperature field is calculated point by point in space. The arithmetic mean of the temperature differences at all spatial sampling points is then obtained to obtain the average temperature deviation value within the current pulse cycle. Specifically, suppose there are a total of [number] defects within the actual defect area. The temperature sampling point, the first The measured temperature at each sampling point was The target temperature at this point corresponds to the expected temperature field. The average temperature deviation value is .when When the measured temperature is higher than expected, it indicates a positive deviation, meaning the current heat input is too high, the molten pool temperature exceeds the design range, and there is a risk of overheating; when When the measured temperature is lower than expected, it indicates that the actual temperature is lower than expected, which is defined as a negative deviation. This means that the current heat input is insufficient and the cladding area has not reached the temperature level required for full fusion.

[0065] Peak power correction in accordance with The size is generated proportionally, and the calculation relationship is as follows: ,in This is the peak power temperature response coefficient, which is pre-calibrated based on the thermophysical properties of the high-temperature alloy material and the area of ​​the repaired spot. When the deviation direction is negative... A positive value indicates that the peak power of the next pulse cycle needs to be increased to compensate for the heat; when the deviation is in the positive direction, A negative value indicates that the peak power of the next pulse cycle needs to be reduced to prevent overheating. This notation ensures that the direction of peak power adjustment is always opposite to the direction of temperature deviation, forming a negative feedback control mechanism. The magnitude of the peak power adjustment is constrained by upper and lower limits to avoid instability in the repair process due to excessive adjustment in a single instance.

[0066] Pulse width correction The generation is based on the energy amplitude in the solidification stress release characteristics. The two are inversely related. When A larger pulse size indicates more intense stress release during solidification and a faster solidification rate. It is necessary to appropriately extend the pulse interval or shorten the pulse width to allow the molten pool more time to complete orderly solidification, thereby reducing the solidification stress level. Conversely, when... When the pulse width is small, the solidification process is relatively smooth, and the pulse width can be appropriately increased to improve the repair efficiency. Specifically, and The reverse relationship is expressed as ,in The pulse width acoustic emission response coefficient is denoted as . This serves as a reference value for the solidification stress release energy, corresponding to the maximum allowable solidification stress release energy level in the repair process design. When measured... Exceed hour, A negative value indicates a shortened pulse width; when measured... Below hour, A positive value indicates an extended pulse width.

[0067] After obtaining the peak power correction and pulse width correction, they are respectively superimposed on the reference parameters of the current pulse period to obtain the target parameters for the next pulse period. Let the reference peak power of the current pulse period be... Then the target peak power of the next pulse cycle is Let the reference pulse width of the current pulse period be... Then the target pulse width for the next pulse period is Both the target peak power and the target pulse width must be saturated and limited within the safe range allowed by the repair process to ensure that the adjusted parameters do not exceed the rated capacity of the equipment or fall below the minimum energy density required to maintain cladding. The adjusted target parameters will serve as the execution command for the next pulse cycle and will be sent in real time to the power control unit and timing control unit of the pulsed laser or plasma cladding equipment.

[0068] The above adjustment process is repeated cyclically after each pulse cycle, forming a closed-loop control. In each cycle, the surface temperature field and acoustic emission signal of the repair area are reacquired, and the calculations are recalculated. and And update accordingly. and The parameters are then superimposed on the actual baseline parameters from the previous cycle, gradually approaching the optimal repair state. This iterative adjustment method can effectively track the dynamic changes in the thermal response of the defect region. Especially when there is material inhomogeneity in the defect depth direction, the anisotropy of thermal conduction in each depth layer will cause a systematic shift between the expected temperature field and the measured temperature field. The closed-loop adjustment mechanism can adaptively compensate for this shift.

[0069] The criterion for terminating repair is composed of two independent indicators: Equivalent value of residual stress. The measurements were obtained by X-ray diffraction or blind hole method of the repaired area. Below the first preset criterion At that time, the residual stress was determined to meet the requirements; the area ratio of recrystallized grains in the metallographic structure was... Obtained through metallographic cross-section analysis or electron backscatter diffraction (EBSD) detection, when Higher than the second preset criterion At that time, the metallographic structure is determined to meet the requirements. Both indicators must be met simultaneously for the pulse-modulated cladding repair to end. If only one indicator is met, the pulse-modulated repair continues, and the control strategy is adjusted accordingly based on the type of unmet indicator: if the residual stress is too high, the peak power is reduced first and the pulse width is appropriately shortened to slow down the solidification rate; if the recrystallized grain ratio is insufficient, the heat input is appropriately increased to promote sufficient grain recrystallization. This dual-criteria termination mechanism ensures that the repair results meet the quality requirements of high-temperature alloy welding repair in both mechanical properties and microstructure, avoiding insufficient or over-repair problems caused by judging based on a single indicator.

[0070] A second aspect of the present invention provides an intelligent identification and repair system for welding defects in high-temperature alloys for heavy-duty gas turbines, comprising: The temperature anomaly unit is used to acquire a continuous infrared thermal image sequence of the high-temperature alloy welding area of ​​heavy-duty gas turbine during the welding process and the post-weld cooling stage, extract the temperature change curve of each spatial location of the welding area, and identify abnormal temperature drop areas that deviate from the normal cooling rate as candidate defect areas based on the temperature change curve. An ultrasonic defect unit is used to perform ultrasonic phased array scanning on the candidate defect region to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the candidate defect region is used to verify the authenticity of the defect and determine the defect type, thereby determining the defect type and spatial location of the real defect region. The repair parameter unit is used to determine the heat conduction anisotropy parameters of the actual defect area based on the defect type, and to calculate the repair heat input curve by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. The cladding control unit is used to perform pulse-modulated cladding repair on the actual defect area according to the repair heat input curve, and to collect the surface temperature field and acoustic emission signal of the repair area in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse period are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

[0071] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0072] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0073] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0074] 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 them; 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 or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent identification and repair of welding defects in high-temperature alloys for heavy-duty gas turbines, characterized in that, include: A continuous infrared thermal image sequence of the high-temperature alloy welded area of ​​a heavy-duty gas turbine is acquired during the welding process and post-weld cooling stage. Temperature change curves at various spatial locations within the welded area are extracted from these images. Based on these temperature change curves, regions with abnormal temperature drops deviating from the normal cooling rate are identified as candidate defect areas. The surface of the welding area is divided into a spatial grid with a preset spatial resolution. The average pixel gray value of each frame in the infrared thermal image sequence within the corresponding grid range is extracted grid by grid. The average pixel gray value is converted into a temperature value through a radiometric calibration function to obtain the temperature-time series of each spatial grid. The radiometric calibration function is pre-calibrated based on the surface emissivity of the high-temperature alloy material and the spectral response characteristics of the infrared thermal imager. The cooling rate curves of each spatial grid are obtained by performing time differentiation on the temperature-time series of each spatial grid. The cooling rate curves of each spatial grid are compared point by point with the reference cooling rate curves of the corresponding high-temperature alloy materials under defect-free conditions. The cooling rate deviation values ​​at each time point are calculated. The duration and cumulative magnitude of the cooling rate deviation values ​​of each spatial grid exceeding the preset deviation threshold during the cooling stage are statistically analyzed. Spatial grids whose duration and cumulative magnitude exceed the corresponding thresholds are marked as abnormal temperature drop grids. Spatial connectivity analysis is performed on the spatial grid marked as abnormal temperature drop grid. Abnormal temperature drop grids that are adjacent to each other and whose spatial distance is less than the preset adjacency distance threshold are aggregated into candidate defect regions. The geometric center coordinates, area, and average cooling rate deviation of each candidate defect region are recorded. The candidate defect region is scanned by an ultrasonic phased array to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the authenticity of the defect and the type of defect are verified and determined to identify the defect type and spatial location of the real defect region. In this process, each element of the ultrasonic phased array is controlled to emit ultrasonic pulses and receive echo signals according to the time-delay focusing rule. By adjusting the emission delay of each element, the ultrasonic pulses emitted by each element are superimposed in phase at each scanning point of the candidate defect region to synthesize a focused beam to scan the candidate defect region point by point. The focal size of the focused beam is determined based on the detection sensitivity requirements and the minimum feature size of the candidate defect region. The amplitude and phase of the reflected echo at each scanning point of the focused beam are extracted. Using the coordinates of the scanning point as an index, the amplitude and phase of the reflected echo at each scanning point are combined into a multi-dimensional vector to form the reflection feature vector of the candidate defect region. The correlation between the reflection feature vector of each scanning point and the reference reflection feature vector of the defect-free reference area is calculated. If the correlation of a certain scanning point is lower than the preset correlation threshold, the scanning point is marked as a false defect point and removed. Spatial connectivity analysis is performed on the remaining scan points after removing false defect points. The statistical mean of the reflection feature vector of each scan point in each connectivity is calculated. The statistical mean is matched with the standard reflection feature template of each defect category, and the defect category with the highest matching degree is output as the defect type of the real defect region. Based on the defect type, the heat conduction anisotropy parameters of the actual defect area are determined, and the repair heat input curve is calculated by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. According to the repair heat input curve, pulse modulation cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are collected in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse cycle are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

2. The method according to claim 1, characterized in that, Based on the defect type, the thermal conductivity anisotropy parameters of the actual defect region are determined, and the repair thermal input curve is calculated by combining the defect depth and defect width, including: According to the defect type, the thermal conductivity anisotropy feature corresponding to the defect type is queried in the thermal conductivity directionality feature library of high-temperature alloy defects. The thermal conductivity anisotropy feature includes the main thermal conductivity coefficient direction along the longitudinal direction of the weld and the secondary thermal conductivity coefficient direction perpendicular to the main thermal conductivity coefficient direction. The thermal conductivity directionality feature library of high-temperature alloy defects is constructed by offline experimental calibration of the thermal conductivity directionality data of porosity, crack and non-fusion defects under their respective typical grain orientations. An anisotropic coordinate system is constructed with the geometric center of the actual defect region as the origin, the direction of the primary thermal conductivity coefficient as the first axis, and the direction of the secondary thermal conductivity coefficient as the second axis. The first axis and the second axis are orthogonal to each other and correspond to the direction with the strongest and weakest thermal conductivity, respectively. The thermal conductivity coefficient value in the direction of the primary thermal conductivity coefficient is set as the first thermal conductivity coefficient, and the thermal conductivity coefficient value in the direction of the secondary thermal conductivity coefficient is set as the second thermal conductivity coefficient. The first thermal conductivity coefficient and the second thermal conductivity coefficient are attenuated and corrected by the material density inside the defect region corresponding to the defect type, thereby forming an anisotropic thermal conductivity parameter matrix based on the anisotropic coordinate system.

3. The method according to claim 2, characterized in that, The repair thermal input curve is calculated by combining the defect depth and defect width, including: Substitute the defect depth and defect width into the two-dimensional anisotropic thermal diffusion equation with the anisotropic thermal conduction parameter matrix as coefficients, and divide the real defect region into multiple depth layers along the defect depth direction. The thickness of each depth layer is adaptively determined according to the depth change gradient of the defect depth. The layer thickness is reduced where the defect depth gradient changes sharply to improve the solution accuracy. Solve the two-dimensional anisotropic thermal diffusion equation for each depth layer, and take the recrystallization temperature of the upper surface of the depth layer as the boundary condition to obtain the minimum heat input per unit area required to raise the temperature of the material in the depth layer from the initial temperature to the recrystallization temperature. An energy margin compensation related to the depth layer position is applied to the minimum unit area heat input of each depth layer. The closer the depth layer is to the bottom of the defect, the higher the compensation coefficient is applied. The compensation coefficient increases linearly with the depth of the depth layer to obtain the target unit area heat input of each depth layer. Using the defect depth as the horizontal axis and the target unit area heat input corresponding to each depth layer as the vertical axis, the discrete points are spliced ​​to generate a stepped layered heat input curve, and the stepped layered heat input curve is smoothed by spline processing to obtain the repair heat input curve.

4. The method according to claim 1, characterized in that, According to the repair heat input curve, pulse-modulated cladding repair is performed on the actual defect area. The surface temperature field and acoustic emission signal of the repair area are acquired in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the following is included: A reference peak power and a reference pulse width are set for pulse modulation cladding repair. Infrared thermal imager is used to collect surface infrared radiation data of the repair area in real time. The surface infrared radiation data is then converted into a two-dimensional surface temperature field through the law of thermal radiation. Acoustic emission waveform signals of the repair area during the pulse modulation cladding process are collected in real time using an acoustic emission sensor. The acoustic emission waveform signals are bandpass filtered to suppress environmental noise and electromagnetic interference. The passband range of the bandpass filter is determined based on the typical frequency band of high-temperature alloy solidification stress release. The filtered acoustic emission waveform signals are then normalized in amplitude. Short-time Fourier transform is performed on the acoustic emission waveform signal after amplitude normalization. The frequency components and energy amplitude of each time-frequency unit are extracted from the transform result. The time-frequency units that meet the preset signal-to-noise ratio condition are selected as candidate solidification events. Each candidate solidification event is located in the time domain. If the frequency component of a candidate solidification event is within the solidification stress release frequency band and its energy amplitude exceeds a preset energy threshold, then the candidate solidification event is marked as a solidification stress release event, and the frequency component and energy amplitude of the solidification stress release event are used as the solidification stress release feature.

5. The method according to claim 4, characterized in that, Coordinated adjustment of peak power and pulse width for each pulse cycle, including: Calculate the average temperature deviation between the measured temperature field and the expected temperature field corresponding to the repair heat input curve within the actual defect area, and determine whether the deviation direction of the average temperature deviation is positive or negative. The peak power correction amount is generated based on the absolute value of the average temperature deviation. When the deviation direction is negative, the peak power correction amount is set to a positive value, and when the deviation direction is positive, the peak power correction amount is set to a negative value. A pulse width correction is generated based on the energy amplitude in the solidification stress release characteristics, and the pulse width correction is inversely related to the energy amplitude. The peak power correction is added to the reference peak power of the current pulse period to obtain the target peak power of the next pulse period, and the pulse width correction is added to the reference pulse width of the current pulse period to obtain the target pulse width of the next pulse period. The above adjustments are repeated until the equivalent value of residual stress in the repair area is lower than the first preset criterion and the area ratio of recrystallized grains in the metallographic structure of the repair area is higher than the second preset criterion, at which point the pulse modulation cladding repair ends.

6. A smart identification and repair system for welding defects in high-temperature alloys of heavy-duty gas turbines, used to implement the method as described in any one of claims 1-5, characterized in that, include: The temperature anomaly unit is used to acquire a continuous infrared thermal image sequence of the high-temperature alloy welding area of ​​heavy-duty gas turbine during the welding process and the post-weld cooling stage, extract the temperature change curve of each spatial location of the welding area, and identify abnormal temperature drop areas that deviate from the normal cooling rate as candidate defect areas based on the temperature change curve. An ultrasonic defect unit is used to perform ultrasonic phased array scanning on the candidate defect region to obtain the reflection characteristics of the candidate defect region. Based on the reflection characteristics, the candidate defect region is used to verify the authenticity of the defect and determine the defect type, thereby determining the defect type and spatial location of the real defect region. The repair parameter unit is used to determine the heat conduction anisotropy parameters of the actual defect area based on the defect type, and to calculate the repair heat input curve by combining the defect depth and defect width. The repair heat input curve limits the target heat input power and heat input duration at each moment during the repair process. The cladding control unit is used to perform pulse-modulated cladding repair on the actual defect area according to the repair heat input curve, and to collect the surface temperature field and acoustic emission signal of the repair area in real time. Based on the deviation between the measured temperature field and the expected temperature field and the solidification stress release characteristics in the acoustic emission signal, the peak power and pulse width of each pulse period are adjusted in a coordinated manner until the residual stress and metallographic structure of the repair area meet the preset repair qualification criteria.

7. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 5.

Images