A spare parts repair engineering cost intelligent management system

Abstract

This invention relates to the field of engineering cost accounting and management, and discloses an intelligent management system for spare parts repair engineering costs. The system includes: registering 3D contour, thermal imaging, and ultrasonic testing data to the same adhesive surface centerline to establish contour elongation coordinates; calculating local bending stiffness, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient based on these coordinates; and then calculating local residual strain energy and extracting the strain energy offset on the thick-walled side. Subsequently, this offset is converted into five comprehensive cost evaluation indicators: grinding, retesting, secondary heating, fixture adjustment time rate, and filling material volume. The system intelligently optimizes the repair execution sequence and cooling dwell time allocation based on a multi-parameter dispersion evaluation function, and finally integrates actual testing data to output an equivalent cost accounting value for resource consumption. This invention effectively overcomes cost drift caused by unilateral rework and achieves refined management of repair costs.

Description

Technical Field

[0001] This invention relates to the field of engineering accounting management, and more specifically, to an intelligent management system for spare parts repair engineering costs. Background Technology

[0002] In the current field of spare parts repair engineering, for composite material shell spare parts with large curvature, thickness transitions, and gradual ply changes, the industry typically uses beveling patching for repair. However, after adhesive bonding repair, the internal residual stress of such shell configurations often originates from differences in thermal expansion coefficients, chemical shrinkage, viscoelastic curing processes, and the amplification effect of internal porosity. Furthermore, damage easily begins at the adhesive interface between the patch and the base material. Related research indicates that patch thickness, beveling angle, and interface morphology directly alter the mechanical recovery state and interface quality. More seriously, when such shell spare parts with large curvature and thickness transitions undergo small-angle beveling repair, the inherent differences between the thick and thin areas in terms of heat diffusion paths, bending stiffness, and curing shrinkage constraints lead to a highly uneven equivalent stiffness field. The inequivalence of the unilateral heat dissipation path further forces residual energy to continuously shift to one side.

[0003] Traditional engineering cost management methods typically rely solely on damage projection area, patch volume, or standard repair time for ledger accounting. This extensive management model completely masks the eccentricity of residual strain energy at the interface caused by thermal expansion mismatch, chemical shrinkage, pore enlargement, and differences in unilateral heat dissipation paths. In reality, the core factor driving uncontrollable cost drift is precisely this shift of the residual energy center of gravity relative to the geometric center. This shift concentrates rework operations such as re-grinding, directional re-inspection, localized secondary heating, and repeated fixture adjustments onto one side of the patch, resulting in significant unilateral rework resource consumption. If management is based solely on average area or standard man-hours, the actual process usage and equipment wear on the thick-walled or high-curvature side will be severely underestimated, making it difficult to accurately represent and constrain unilateral rework costs. Ultimately, this leads to overall loss of control in engineering cost management and serious deviations in accounting records. Summary of the Invention

[0004] This invention provides an intelligent management system for spare parts repair engineering costs, which solves the technical problems mentioned in the background art.

[0005] This invention provides an intelligent management system for spare parts repair engineering costs, comprising: The data alignment module acquires the three-dimensional contour scan data, pulsed thermal imaging data, phased array ultrasonic data and repair process log of the spare part to be repaired, and registers them to the center line of the same adhesive surface to establish the contour elongation coordinates of each adhesive surface position. The parameter solving module obtains physical property parameters and curing parameters, and solves the local bending stiffness, curing mismatch deformation, heat dissipation delay and porosity influence coefficient at each adhesive surface position based on the contour elongation coordinates. The strain energy calculation module calculates the local residual strain energy based on the local bending stiffness, the curing mismatch deformation, the heat dissipation delay, and the porosity influence coefficient, and calculates the strain energy concentration point and the geometric center point. The thick-walled side strain energy offset is obtained from the offset distance of the strain energy concentration point relative to the geometric center point. The consumption mapping module obtains the operating parameters of the processing equipment and converts the thick-walled side strain energy offset into five comprehensive consumption evaluation indicators: grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume. The intelligent scheduling module acquires initial coded variables, encodes the repair execution order and cooldown dwell time allocation, generates a multi-parameter dispersion evaluation function and optimizes it to generate a cost management schedule and the total cost of each optimal schedule. The cost output module acquires actual testing data and enterprise accounting parameters, integrates the actual testing data for post-test correction, and outputs the equivalent accounting value of resource consumption and management accounting cost.

[0006] The beneficial effects of this invention include: by unifying and aligning multi-source detection data to the same adhesive surface to establish contour elongation coordinates, and integrating objective state parameters such as local bending stiffness, curing mismatch, heat dissipation delay, and porosity, the thick-walled side strain energy offset that dominates consumption drift is accurately extracted, and then scientifically mapped into five categories of real resource consumption rates: grinding, retesting, secondary heating, fixture adjustment, and filling consumables; at the same time, by relying on intelligent optimization scheduling and post-correction mechanisms based on multi-parameter dispersion, the traditional extensive estimation mode that only relies on damaged surface area or standard working hours is abandoned, and global optimal control from process execution sequence to cooling dwell time allocation is achieved. Finally, the equivalent accounting value and management accounting cost that accurately fit the enterprise's objective resource occupancy are output, which greatly improves the accuracy of repair project cost assessment and the scientific nature of on-site resource scheduling. Attached Figure Description

[0007] Figure 1 This is a module diagram of an intelligent management system for spare parts repair engineering costs according to the present invention. Detailed Implementation

[0008] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.

[0009] like Figure 1 As shown, an intelligent management system for spare parts repair engineering costs includes: The data alignment module acquires the three-dimensional contour scan data, pulsed thermal imaging data, phased array ultrasonic data and repair process log of the spare part to be repaired, and registers them to the center line of the same adhesive surface to establish the contour elongation coordinates of each adhesive surface position. The parameter solving module obtains physical property parameters and curing parameters, and solves the local bending stiffness, curing mismatch deformation, heat dissipation delay and porosity influence coefficient at each adhesive surface position based on the contour elongation coordinates. The strain energy calculation module calculates the local residual strain energy based on the local bending stiffness, the curing mismatch deformation, the heat dissipation delay, and the porosity influence coefficient, and calculates the strain energy concentration point and the geometric center point. The thick-walled side strain energy offset is obtained from the offset distance of the strain energy concentration point relative to the geometric center point. The consumption mapping module obtains the operating parameters of the processing equipment and converts the thick-walled side strain energy offset into five comprehensive consumption evaluation indicators: grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume. The intelligent scheduling module acquires initial coded variables, encodes the repair execution order and cooldown dwell time allocation, generates a multi-parameter dispersion evaluation function and optimizes it to generate a cost management schedule and the total cost of each optimal schedule. The cost output module acquires actual testing data and enterprise accounting parameters, integrates the actual testing data for post-test correction, and outputs the equivalent accounting value of resource consumption and management accounting cost.

[0010] Preferably, the three-dimensional contour scan data, pulsed thermal imaging data, phased array ultrasonic data, and repair process log of the spare part to be repaired are registered to the same adhesive surface centerline to establish contour elongation coordinates, including: Obtain the coordinates of the adhesive surfaces and calculate the contour elongation coordinates of each adhesive surface location. The calculation formula is as follows: ; The formula for calculating the local tangential vector is: ; The formula for calculating the curvature of a local surface is: ; The formula for calculating the width of a local beveled adhesive joint is: ; The shortest spatial distance from the coordinates of the adhesive surface to the actual heat dissipation boundary is extracted as the heat dissipation path length, and its calculation formula is as follows: ; The local porosity index is calculated using the following formula: ; In the above formula, For the first The contour elongation coordinates of each adhesive surface location. , , and These are the coordinates of the adhesive surface positions corresponding to different serial numbers. , and It is a local tangential vector. For local surface curvature, This refers to the width of the locally beveled adhesive joint. For local wall thickness, For local beveling angle, This is the length of the heat dissipation path. These are the coordinates of the boundary position on the actual heat dissipation boundary. For the actual heat dissipation boundary, The local porosity index, For pulsed thermal imaging signal-to-noise ratio, This represents the normalized thermal image defect response value. The signal-to-noise ratio of phased array ultrasound. This represents the normalized ultrasonic defect response value.

[0011] Three-dimensional contour scan data is spatial point cloud or mesh data that describes the true geometric shape of the outer surface and adhesive area of ​​the spare part to be repaired. It can be obtained by performing a full-area scan of the spare part to be repaired using a structured light scanner, laser scanner, articulated arm coordinate measuring machine, or contact coordinate measuring machine.

[0012] Pulse thermal imaging data is a sequence of images that continuously records the temperature field changes over time on the surface of a spare part after a short period of thermal excitation. It can be obtained by applying a controlled thermal pulse to the spare part using a pulse thermal imaging system and simultaneously acquiring an infrared thermal image sequence.

[0013] Phased array ultrasound data consists of echo amplitude, flight time, and imaging results obtained by multi-element ultrasound probes at different incident angles and focusing depths. It can be obtained by performing line or area scans along the repair area using a phased array ultrasound testing system following a pre-defined scanning trajectory.

[0014] The repair process log is a data set that records process information such as beveling, heat curing, fixture clamping, dwell time, process sequence, and equipment status. It can be obtained through manufacturing execution systems, records exported from equipment controllers, records filled in manually on process cards, and summaries of sensor historical records.

[0015] The coordinates of the adhesive surface are three-dimensional spatial points obtained by discretizing along the centerline of the adhesive surface or the path representing the adhesive surface. This can be obtained by extracting the centerline of the adhesive surface from the three-dimensional contour scan data and then sampling at a predetermined discrete distance.

[0016] Local wall thickness is the total thickness of the base material or patch laminate measured along the local thickness direction at a discrete location on the adhesive surface. It can be obtained through ultrasonic thickness measurement, 3D scanning and comparison with the design model, or cross-sectional measurement.

[0017] The local beveling angle is the angle between the beveling surface of the adhesive-bonded repair area at a discrete location and the reference plane or local tangential plane. It can be obtained by reading repair process design values, three-dimensional profile fitting angle measurement, or contact profile measurement.

[0018] The actual heat dissipation boundary is the set of edges, contact surfaces, or free surfaces that can actually transfer heat to the outside during the repair process. It can be determined by combining three-dimensional geometric models, tooling contact relationships, insulation coverage, and heating area layout.

[0019] The coordinates of the boundary positions on the actual heat dissipation boundary are the coordinates of the discrete boundary points that make up the actual heat dissipation boundary. These coordinates can be obtained by discretizing the actual heat dissipation boundary into a set of boundary points or a boundary mesh and then extracting the node coordinates.

[0020] Normalized thermal image defect response value is a dimensionless characterization quantity that compresses the thermal image anomaly response corresponding to a certain adhesive surface location into a comparable range according to a uniform scale. It is used to eliminate amplitude differences between different thermal excitation intensities and different imaging batches.

[0021] The pulsed thermal imaging signal-to-noise ratio (SNR) is the ratio of the thermal image defect response to the background noise intensity, used to characterize the reliability and usability of the thermal imaging anomaly results.

[0022] Normalized ultrasonic defect response value is a dimensionless characterization quantity that converts the local abnormal echo or imaging response detected by phased array ultrasound to a unified comparison scale, and is used for same-scale fusion with thermal imaging abnormal response.

[0023] The signal-to-noise ratio (SNR) of phased array ultrasound is the ratio of the ultrasonic defect echo to the intensity of background noise or structural clutter, used to characterize the reliability of the ultrasonic anomaly result.

[0024] The spatial straight-line distance is the shortest straight-line distance between the position coordinates of two adjacent adhesive surfaces in a unified three-dimensional coordinate system. It is used to convert discrete coordinate points into contour elongation coordinates accumulated along the real geometric path.

[0025] The contour elongation coordinate value is the actual spatial length accumulated along the centerline of the adhesive surface from the starting position to the current discrete position. It is used to map the position on the complex three-dimensional curved surface into a single sequence of coordinates, which facilitates subsequent statistics and scheduling.

[0026] The local tangential vector is the local forward direction obtained by normalizing the direction of the difference in position coordinates between adjacent adhesive surfaces.

[0027] Local surface curvature is a local characterization of the degree of turning of the adhesive surface path between adjacent positions, used to reflect the bending strength of a complex curved shell at the current position.

[0028] The effective bonding width of the beveled transition zone is determined by both the local wall thickness and the local bevel angle.

[0029] The heat dissipation path length is the shortest spatial distance between a certain position on the adhesive surface and the actual heat dissipation boundary. It is used to approximately characterize the length of the heat transfer path that needs to be traversed for heat to be released outward from that position.

[0030] The local porosity index is a measure of the porosity of local pores or interfaces obtained by weighting and fusing the thermal imaging anomaly response and the ultrasonic anomaly response according to their respective signal-to-noise ratios. It is used to describe the amplifying effect of pore defects on curing quality and subsequent rework costs.

[0031] In detail, because 3D contour scanning data reflects external geometry, pulsed thermal imaging data reflects transient surface thermal response, phased array ultrasonic data reflects internal echo characteristics, and the repair process log reflects process status, these four types of data originally exist under different coordinate systems, different time bases, and different sampling densities. Therefore, they must first be uniformly mapped to the same adhesive surface centerline to ensure that geometric states, thermal anomalies, and ultrasonic anomalies at the same location correspond one-to-one. For example, the same thick-walled transition zone may appear as a cold spot on thermal imaging and as a high-echo area on ultrasonic imaging. If it is not first registered to the same adhesive surface centerline, it is impossible to confirm whether these two anomalies originate from the same rework location.

[0032] In detail, because the adhesive surfaces of highly curved shells often simultaneously exhibit spatial transitions, thickness shifts, and ply variations, directly comparing the thermal and cost states at various locations in three-dimensional space would lead to sequencing difficulties. Therefore, accumulating the coordinates of the adhesive surface locations along the actual path into profile elongation coordinates allows complex surfaces to be mapped into a monotonically unfolding one-dimensional sequence. For example, even after the same patch edge bypasses a surface corner in space, the profile elongation coordinates can still continuously represent the preceding and following relationships, thus facilitating the calculation of the geometric center point, strain energy concentration point, and scheduling sequence.

[0033] In detail, because thermal imaging is exceptionally sensitive to near-surface heat diffusion, while ultrasound is sensitive to echoes from internal interfaces and pores, the two detection methods have different advantages in depth and noise resistance. Therefore, by weighting and fusing the normalized thermal imaging defect response value and the normalized ultrasonic defect response value according to their respective signal-to-noise ratios, misjudgment by a single method can be avoided. For example, when thermal imaging shows severe surface reflection, noise increases, and the ultrasonic signal-to-noise ratio is higher in this case, so the fusion result will automatically increase the contribution of the ultrasonic result.

[0034] In detail, although the thick and thin areas may be located on the same patch, the path length for heat to reach the actual heat dissipation boundary is not the same. Therefore, the shortest spatial distance from the adhesive surface to the actual heat dissipation boundary is used to represent the heat dissipation path length, which can explicitly incorporate the unequal heat dissipation on one side into the model. For example, although the geometric distance on the side blocked by the fixture is similar, if the heat dissipation boundary is restricted, its effective heat dissipation path will be longer, and the risk of subsequent curing mismatch and rework will be more concentrated.

[0035] In detail, because the actual bonding transition range of bevel repair is not determined solely by the planar projection, but is simultaneously affected by both the local wall thickness and the local bevel angle, using both the local wall thickness and the local bevel angle to calculate the local bevel bonding width can more accurately reflect the area affected by the bonding. For example, under the same wall thickness conditions, a smaller bevel angle will result in a longer bevel transition zone, and the grinding and filling area will also expand accordingly.

[0036] In detail, the method for registering multi-source data to the same adhesive surface centerline is as follows: First, a unique three-dimensional coordinate reference is established using three-dimensional contour scanning data. Then, the adhesive surface centerline is extracted based on the patch boundary line, the base material transition line, and the step positions in the repair process log. Subsequently, the pixel positions in the pulsed thermal imaging data are projected onto the three-dimensional surface through a calibration plate and reference points. The scanning positions in the phased array ultrasonic data are mapped onto the same three-dimensional surface through the probe encoder coordinates and tooling positioning holes. Finally, the thermal imaging and ultrasonic results are projected onto the discrete positions corresponding to the adhesive surface centerline according to the minimum distance principle. For example, more than three common reference points can be set at both ends of the patch to complete the geometric calibration first, and then the abnormal responses of the thermal imaging and ultrasound are interpolated onto the centerline.

[0037] In detail, the discrete sampling method for the coordinates of the adhesive surface is as follows: taking the starting point of the centerline of the adhesive surface as the zero point, sampling is performed sequentially along the actual contour direction at a fixed contour extension distance. When encountering areas of abrupt curvature changes, abrupt thickness changes, or areas with dense anomalies, sampling can be locally densified. The endpoint is the end position of the centerline of the adhesive surface. All sampling points are uniformly numbered in a single direction from the starting point to the endpoint. For example, a step distance of 1 mm can be used for standard sections, and 0.5 mm step distance can be used near corners to ensure that local curvature and anomaly responses are not lost due to sparse sampling.

[0038] In detail, the methods for obtaining local wall thickness and local beveling angle are as follows: local wall thickness is preferentially obtained through ultrasonic thickness measurement or the thickness field of the design model; local beveling angle is preferentially obtained through fitting the angle between the beveling surface and the local reference surface using a 3D profile. When there is a difference between the measured value and the design value, the measured value shall prevail, and the deviation shall be recorded in the process log. For example, if the actual beveling angle at a certain position after patch processing is smaller than the design value, the local beveling adhesive width should be recalculated using the measured angle.

[0039] In detail, the method for determining the actual heat dissipation boundary is as follows: the free edges, exposed surfaces, and thermally conductive contact surfaces in the repair area that can effectively exchange heat with ambient air, heat dissipation fixtures, or thermally conductive clamps are defined as the actual heat dissipation boundary. Areas covered by insulation blankets, sealed by thermal insulation tape, or blocked by non-thermally conductive clamps that hardly participate in heat dissipation in the short term are excluded. For example, when one side of the edge of the same patch is in contact with the metal pressure plate, that contact edge can be considered as the actual heat dissipation boundary, while the other side covered by a thick insulation layer is not considered as an effective heat dissipation boundary.

[0040] In detail, the calculation methods for normalized thermal image defect response value, normalized ultrasonic defect response value, pulsed thermal image signal-to-noise ratio, and phased array ultrasonic signal-to-noise ratio are as follows: First, a fixed-length analysis window is established around each adhesive surface location. Then, the peak temperature difference, thermal attenuation slope anomaly, ultrasonic echo amplitude anomaly, or imaging brightness anomaly are extracted separately and normalized to the 0-1 range according to the maximum and minimum values ​​of all locations in this batch. The signal-to-noise ratio is calculated using the ratio of the target response amplitude to the root mean square value of the noise in the adjacent defect-free background area. For example, the maximum abnormal temperature difference in the thermal image can be taken as the target response within the 3rd to 8th second after thermal excitation, and the background noise is taken as the fluctuation value of the defect-free neighborhood in the same frame.

[0041] In detail, the sign distinction between local tangential vectors and local wall thicknesses is as follows: local tangential vectors are uniformly denoted as path direction quantities, and local wall thicknesses are uniformly denoted as thickness quantities; the two must not use the same sign. When the coordinates of adjacent adhesive surfaces coincide, the local tangential vectors should be reconstructed using two adjacent non-coincident points. When the sum of the pulsed thermal imaging signal-to-noise ratio and the phased array ultrasonic signal-to-noise ratio is zero, the local porosity index should be set to 0 and marked as a low-confidence location. For example, when both thermal imaging and ultrasonic waves are distorted in a certain area, the original formula should not be used for direct calculation to avoid undefined results.

[0042] Preferably, the calculation of local bending stiffness, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient at each adhesive surface location based on the contour elongation coordinates includes: The formula for calculating the single-layer thickness weight and the local equivalent elastic modulus is as follows: ; ; The local bending stiffness is calculated using the following formula: ; The amount of deformation due to curing mismatch is calculated using the following formula: ; The formula for calculating the local heat dissipation delay time and the heat dissipation delay is as follows: ; ; The porosity influence coefficient is calculated using the following formula: ; In the above formula, As a single-layer thickness weight, For single-layer thickness, For the single-layer thickness in the summation term, This represents the total number of plies. For the local equivalent elastic modulus, For the fiber layup angle, The elastic modulus of the first principal axis, The elastic modulus of the second principal axis. It is the in-plane shear modulus. For local bending stiffness, For local wall thickness, For the locally equivalent Poisson's ratio, This represents the amount of deformation due to curing mismatch. Due to the difference in the coefficients of thermal expansion between the patch and the base material, For localized curing temperature difference, The coefficient of chemical shrinkage during full curing. This refers to the degree of localized curing. and This is the local heat dissipation delay time. This is the length of the heat dissipation path. For local thermal diffusivity, To delay heat dissipation, This represents the total number of adhesive bonding locations. The porosity influence coefficient is... This represents the local porosity index.

[0043] Single-layer thickness refers to the actual thickness of a single fiber reinforcement layer or single patch layer in a localized plywood structure after curing. It is used to calculate the single-layer thickness weight and participate in the solution of the local equivalent elastic modulus. It can be obtained from plywood design documents, material specifications, microscopic section measurements, or ultrasonic layer-by-layer thickness measurements.

[0044] The total number of ply layers refers to the total number of ply layers involved in stress and heat transfer analysis at a specific location. It is used to determine the summation range for calculating the weight of single-layer thickness. This can be obtained from patch ply design documents, base material structural drawings, or actual cross-sectional verification results.

[0045] The primary modulus of elasticity is the elastic modulus of a single-layer composite material along the principal fiber direction, reflecting the level of tensile or compressive stiffness along the fiber direction. It can be obtained from material supplier datasheets, tensile tests on single-layer specimens, or the company's material database.

[0046] The second principal modulus of elasticity is the modulus of elasticity perpendicular to the principal fiber direction of a single-layer composite material, used to reflect the level of lateral stiffness under stress. It can be obtained from material supplier datasheets, transverse tensile tests of single-layer specimens, or the company's material database.

[0047] In-plane shear modulus is a parameter representing the deformation resistance of a single-layer composite material under in-layer shear, used to characterize the shear response of a ply under angular stress transition. It can be obtained from material data sheets, laminate shear tests, or the company's existing physical property database.

[0048] The fiber layup angle is the angle between the principal direction of a single fiber layer and the reference direction of the local analysis coordinate system. It is used to convert the principal property of a single layer to the local stress direction. It can be obtained from layup design documents, tape laying program records, or image recognition results of the actual layup direction.

[0049] The local equivalent Poisson's ratio is a comprehensive parameter characterizing the degree of vertical contraction of a local laminate when it is subjected to tension or compression in one direction. It is used to correct for the lateral coupling effect in the calculation of local bending stiffness. It can be obtained from a laminate equivalent performance database, numerical simulation results, or local specimen mechanical test results.

[0050] The difference in the coefficients of thermal expansion between the patch and the base material represents the difference in their ability to freely expand and contract under the same temperature change. It characterizes the source of thermal mismatch during curing and cooling. This can be obtained through thermomechanical analysis of the patch and base materials or by comparing data from a material database.

[0051] Local curing temperature difference is the temperature difference at a local location relative to the reference temperature or target curing temperature during the curing process. It reflects the difference in thermal strain caused by uneven curing thermal history. It can be obtained through thermocouple arrays, infrared thermographic calibration results, or curing process records.

[0052] The full-curing chemical shrinkage coefficient is a characterizing parameter of the volume or length shrinkage of a resin system due to chemical reactions when it reaches full curing. It is used to estimate the deformation contribution caused by purely chemical curing. It can be obtained through resin system curing tests, material handbooks, or company process databases.

[0053] Local curing degree is the proportion of the reaction that has been completed at a local location during the current curing stage, used to indicate how much chemical shrinkage has been released. It can be obtained through differential scanning calorimetry, dielectric analysis sensors, curing kinetic model inversion, or process log estimation.

[0054] Local thermal diffusivity is a parameter that describes how quickly heat diffuses along the dominant heat transfer direction within a localized material. It determines the timescale for heat propagation and temperature equalization. It can be obtained through flash thermophysical property testing, material databases, or by combining a thermal analysis model with fiber orientation corrections.

[0055] The single-layer thickness weight is the proportion of a certain single-layer thickness to the current local total thickness, used to quantify the contribution of different single layers to the local equivalent elastic modulus.

[0056] The local equivalent elastic modulus is a comprehensive stiffness parameter obtained by converting the principal direction elastic modulus and shear modulus of each single layer under different ply angles to the current analysis direction. It is used to characterize the equivalent stress capacity of the local laminate as a whole.

[0057] Local bending stiffness is a measure of local bending capacity obtained by combining local equivalent elastic modulus, local wall thickness, and local equivalent Poisson's ratio. It is used to describe the ability of a local structure to resist bending deformation.

[0058] Curing mismatch deformation is a characterization of local mismatch deformation determined by the difference in thermal expansion coefficients between the patch and the base material, the local curing temperature difference, the total curing chemical shrinkage coefficient, and the local degree of curing. It is used to indicate the strength of the source of residual deformation after curing.

[0059] Local heat dissipation delay time is the characteristic cooling lag time exhibited at a certain adhesive surface location due to a longer heat dissipation path and lower thermal diffusivity. It is used to measure the degree of hysteresis in temperature decay at that location.

[0060] The total number of adhesive surface locations is the total number of location points obtained after discrete sampling along the centerline of the adhesive surface, which is used for averaging and determining the range of discrete summation.

[0061] Heat dissipation delay is a dimensionless hysteresis coefficient obtained by normalizing the local heat dissipation delay time relative to the average heat dissipation delay time at all locations. It is used to compare the speed of heat dissipation at different locations.

[0062] The porosity influence coefficient is an amplification factor obtained by adding one to the local porosity index.

[0063] In detail, because the stiffness of composite laminates has a significant directionality, different monolayers contribute differently to the local stress direction under different fiber layup angles. Therefore, it is necessary to synthesize the first principal axial elastic modulus, the second principal axial elastic modulus, and the in-plane shear modulus according to the fiber layup angle and monolayer thickness to obtain the local equivalent elastic modulus. For example, the zero-degree layer contributes highly to axial stiffness, while the forty-five-degree layer is more sensitive to shear coupling. Simply averaging these values ​​would mask the true differences in local stiffness.

[0064] In detail, because the patch and the base material will experience thermal mismatch due to their different thermal expansion capacities during heating, heat preservation, and cooling, and the resin curing reaction itself will also bring about chemical shrinkage, the difference in thermal expansion coefficients between the patch and the base material, the local curing temperature difference, the total curing chemical shrinkage coefficient, and the local degree of curing are combined to form the curing mismatch deformation, which can more completely characterize the source of residual deformation after curing. For example, even with a small temperature difference but a high degree of curing, the same type of patch may still show significant mismatch due to the contribution of chemical shrinkage.

[0065] In detail, because the rate of local temperature dissipation depends on the length of the heat diffusion path and the material's heat transfer capacity, and the characteristic time of the heat diffusion process typically increases with the square of the path length, defining the local heat dissipation delay time by dividing the square of the heat dissipation path length by the local thermal diffusivity captures the core characteristic that thick-walled and shielded sides cool more slowly. For example, when the path length is doubled, the cooling lag does not increase linearly but is significantly amplified.

[0066] In detail, because porosity doesn't just represent a single detection anomaly, but also amplifies the impact of local thermal resistance, interface discontinuities, and curing defects on rework resources, converting the local porosity index into a porosity influence coefficient allows the porosity level to be directly incorporated into subsequent energy and cost calculations. For example, two locations with similar curing temperature differences, but with more porosity, often require more frequent retesting and localized repairs.

[0067] In detail, because the absolute heat dissipation time for different spare parts and patch lengths is not directly comparable, dividing the local heat dissipation delay time by the overall average to form the heat dissipation delay can convert the cooling lag under different tasks into a relative comparative quantity. For example, although the absolute cooling time of a certain area is not long, if it is significantly higher than the average for the same task, it can still be identified as a high-risk location for subsequent rework.

[0068] In detail, the mapping of the first principal axis elastic modulus, the second principal axis elastic modulus, the in-plane shear modulus, the fiber layup angle, and the single-layer thickness to each adhesive surface position is as follows: First, establish the correspondence between layer numbers and spatial positions based on the layup design documents of the patch and the base material. Then, extract the corresponding single-layer parameters according to the actual ply combinations passed through at each adhesive surface position. Finally, calculate the single-layer thickness weight and the local equivalent elastic modulus according to the effective stacking order at that position. For example, in the thickness transition zone, the outer 4 layers may belong to the newly added layers of the patch, while the inner original layers belong to the base material layers. They should be included in the calculation according to the actual spatial coverage area.

[0069] The detailed instructions for using the local equivalent modulus of elasticity are as follows: This method is for local laminate equivalent analysis under small deformation conditions. It assumes that the bonding between individual layers is complete, the local layup is considered uniform within the analysis step, and the local reference direction is established tangentially along the adhesive surface path. For example, this equivalent treatment can be used when the analysis step is 1 mm and the layup direction does not change significantly within this range; if there is obvious delamination or discontinuous layup, the local equivalent parameters should be reconstructed segment by segment.

[0070] In detail, the methods for determining the difference in thermal expansion coefficients between the patch and the base material, the total chemical shrinkage coefficient, and the degree of localized curing are as follows: the difference in thermal expansion coefficients between the patch and the base material is obtained through thermomechanical analysis tests on both materials; the total chemical shrinkage coefficient is obtained through resin curing volume change tests; and the degree of localized curing is obtained through differential scanning calorimetry or inversion recording by a dielectric sensor. For example, the rate of increase in localized curing degree of the same resin system varies under different heating rates, and a heating history consistent with the actual process should be used for calibration.

[0071] In detail, the convention for signifying curing mismatch deformation is as follows: a positive value is defined when the patch exhibits greater shrinkage relative to the base material or tends to bend inwards towards the base material after being constrained; a negative value is defined when the patch exhibits predominant elongation relative to the base material or tends to bulge outwards. The reference temperature is uniformly taken as the local temperature peak before natural cooling after curing. For example, if outward bulging occurs at the same location after cooling, a negative sign should be assigned according to the pre-defined orientation rule to ensure consistent interpretation of subsequent warping direction.

[0072] In detail, the selection method for local thermal diffusivity is as follows: when the main heat transfer direction is along the thickness direction, the equivalent thermal diffusivity in the thickness direction should be used; when there is a clear dominance of in-plane heat conduction, the local equivalent thermal diffusivity can be established by combining the main fiber direction and boundary conditions, but the same selection rule must be maintained throughout the entire bonded surface. For example, when the thick-walled area is compressed by the clamp and mainly dissipates heat to the outer surface, the thermal diffusivity in the thickness direction can be used preferentially.

[0073] In detail, the average local heat dissipation delay time is calculated as follows: The arithmetic mean of the local heat dissipation delay times at all effective adhesive bonding locations is calculated, excluding locations with missing thermal images, ultrasonic distortion, or invalid geometric sampling. If a location's value far exceeds the overall distribution upper limit due to abnormal local thermal diffusivity, the average value of adjacent effective locations can be used as a substitute, and the substitute record should be retained. For example, if a point's parameters are abnormal due to probe decoupling, it should not be directly included in the average value calculation.

[0074] In detail, the calibration method for the porosity influence coefficient is as follows: First, based on historical repair samples, statistically analyze the correspondence between the local porosity index and the rework frequency, repair area, and number of retests. Then, determine the applicable range of adding 1 to the local porosity index as the basic amplification factor. When the local porosity index exceeds the preset high-risk threshold, an upper limit or segmented amplification rule can be set for the porosity influence coefficient. For example, when the local porosity index is less than 0.3, the linear addition rule can be used, while when it is higher than 0.8, an upper limit constraint can be used to avoid excessive amplification.

[0075] Preferably, the local residual strain energy is calculated based on the local bending stiffness, the curing mismatch deformation, the heat dissipation delay, and the porosity influence coefficient. The strain energy offset on the thick-walled side is obtained from the offset distance of the strain energy concentration point relative to the geometric center point, including:

[0076] The local residual strain energy is calculated using the following formula: ; The geometric center point is calculated using the following formula: ; The strain energy concentration point is calculated using the following formula: ; The absolute value of the difference between the strain energy concentration point and the geometric center point is obtained as the offset distance, and the thick-walled side strain energy offset is calculated using the following formula: ; In the above formula, This is local residual strain energy. For local bending stiffness, This refers to the width of the locally beveled adhesive joint. This represents the amount of deformation due to curing mismatch. To delay heat dissipation, The porosity influence coefficient is... The geometric center point, These are the coordinates of the outline's elongation. This represents the total number of adhesive bonding locations. As the point of strain energy concentration, This represents the shift in strain energy on the thick-walled side. The maximum value of the contour elongation coordinates. This represents the minimum value of the contour elongation coordinates.

[0077] Local residual strain energy is a local energy concentration proxy formed by comprehensively considering local bending stiffness, local oblique bonding width, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient. It is used to measure the strength of the intrinsic state at a certain location that may drive rework after curing.

[0078] The geometric center point is the position of the geometric center obtained by weighting the contour elongation coordinates with the width of the local beveled adhesive joint as the weight. It is used to characterize the centroid position of the adhesive joint area in the contour elongation direction.

[0079] The strain energy concentration point is the location of the energy centroid obtained by weighting the profile elongation coordinate values ​​with local residual strain energy as the weight. It is used to characterize the concentration location of the rework driving force in the profile elongation direction.

[0080] Offset distance is the absolute distance between the strain energy concentration point and the geometric center point, used to quantify the degree of eccentricity of the energy center of gravity relative to the geometric centroid.

[0081] The maximum value of the profile extension coordinate is the end coordinate value of the discrete point column of the adhesive surface in the profile extension direction, which is used to represent the upper limit of the total extension of the segment involved in the analysis.

[0082] The minimum value of the profile extension coordinate is the starting coordinate value of the discrete point list of the adhesive surface in the profile extension direction, which is used to represent the lower limit of the total extension of the segment involved in the analysis.

[0083] The thick-walled side strain energy offset is an offset index obtained by normalizing the offset distance relative to the total span of the profile extension. It is used to characterize the degree to which local residual strain energy is concentrated on the thick-walled side or the highly constrained side.

[0084] In detail, the real cause of concentrated rework on one side is not the damaged area itself, but rather the localized inherent release tendency formed by uneven stiffness, curing mismatch, delayed heat dissipation, and enlarged porosity. Therefore, by combining local bending stiffness, local beveling width, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient to construct local residual strain energy, multiple structural states can be uniformly compressed into a single cost-driven proxy. For example, in two repair areas of the same area, if one has a stronger thermal mismatch and more porosity, it will exhibit higher resource consumption during subsequent rework.

[0085] In detail, because the effective bonding participation range varies at different locations on the patch, with wider areas contributing more to the overall geometric distribution, the geometric center point is obtained by weighting the contour elongation coordinate value with the local beveling bonding width. This method can more accurately represent the geometric centroid of the bonding area. For example, on a patch with a longer beveling on one side, using a simple average position would underestimate the pulling effect of the wide transition zone on the overall geometric center.

[0086] In detail, because the driving force for rework is mainly concentrated in areas with high local residual strain energy, rather than being evenly distributed across the entire adhesive surface, the strain energy concentration point is obtained by weighting the profile elongation coordinates with local residual strain energy. This allows us to directly locate the center of the position that truly dominates subsequent rework. For example, when there are minor anomalies in multiple locations, the energy weighting method can highlight the high-risk sections that require priority treatment.

[0087] In detail, because the more significant the offset of the energy center of gravity relative to the geometric center, the more likely the rework workload is to be concentrated on one side, the offset distance between the strain energy concentration point and the geometric center point is normalized according to the total span to become the strain energy offset on the thick-walled side. This allows us to use a unified index to characterize the degree of unilateral rework resource eccentricity. For example, tasks with large offsets often exhibit the phenomenon of localized repeated grinding and directional retesting concentrated on the same side.

[0088] In detail, the dimensionless method for local residual strain energy is as follows: this quantity should be clearly defined as a cost-driven proxy quantity, rather than an absolute energy in a strictly thermodynamic sense; during implementation, historical rework samples should be used to dimensionlessly or proportionally calibrate the results of combining local bending stiffness, local beveling width, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient, so that it monotonically corresponds to the actual rework intensity. For example, the initial combination values ​​at all locations can be divided by the average value of the same task to form the relative local residual strain energy, which can be used for subsequent offset and scheduling analysis.

[0089] In detail, the discrete summation method for the geometric center point and strain energy concentration point is as follows: the summation range is uniformly taken as all effective adhesive surface positions from the start point to the end point of the profile elongation, and missing measurement points are filled by interpolation between adjacent points; if the local beveling adhesive width or local residual strain energy is zero, the point is retained in the coordinate sequence but its weight can be zero. For example, when there is an invalid sampling point at the end of the patch, linear interpolation of the preceding and following adjacent points can be used before participating in the overall summation.

[0090] In detail, the method for determining the thick-walled side is as follows: First, establish a uniform positive direction from the starting point to the ending point along the contour extension direction. Then, compare the average local wall thickness of the adjacent regions on both sides of each discrete position, and define the side with the larger wall thickness as the thick-walled side. Subsequently, associate the offset direction of the strain energy concentration point relative to the geometric center point with the thick-walled side direction, and output the strain energy offset of the thick-walled side with a directional sign. For example, if the average wall thickness is larger to the right of the geometric center point, and the strain energy concentration point is also located on the right, then the offset is recorded as an offset towards the thick-walled side.

[0091] In detail, the boundary treatment method for offset distance and thick-walled side strain energy offset is as follows: when all local residual strain energies are zero, the strain energy concentration point is directly taken as the geometric center point; when the maximum and minimum values ​​of the profile elongation coordinates are equal, the thick-walled side strain energy offset is defined as zero and the segment is marked as not having elongation distribution significance. For example, when there is an extremely short patch or only a single effective sampling point, the offset should not be calculated as for a normal segment.

[0092] In detail, the method for correcting the direction after taking the absolute value of the offset distance is as follows: Before calculating the absolute value, retain the original signed difference and record both the offset distance and the offset direction in the output. If the management side only focuses on the degree of concentration, the absolute value can be used; if the process side needs to determine whether the offset is towards the thick-walled side or the thin-walled side, the direction indicator must be read. For example, with the same offset of 0.2, one towards the thick-walled side and the other towards the thin-walled side, their subsequent rework strategies are different.

[0093] Preferably, converting the thick-walled side strain energy offset into grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume includes: The local warping curvature and local contour trimming compensation amount are calculated using the following formula: ; ; The grinding time rate and filling material volume are calculated using the following formula: ; ; The retest scan repetition frequency and retest time consumption rate are calculated using the following formulas: ; ; The secondary heating time consumption rate is calculated using the following formula: ; The fixture attitude deflection angle and fixture adjustment time rate are calculated using the following formula: ; ; In the above formula, For local warping deformation curvature, This represents the amount of deformation due to curing mismatch. To delay heat dissipation, The porosity influence coefficient is... This refers to the width of the locally beveled adhesive joint. This is the amount of compensation for local contour trimming. To reduce the time spent polishing, To extend the walking distance, For equipment volume removal rate, To fill the volume of consumables, To measure the frequency of repeated scans, and This is local residual strain energy. This represents the total number of adhesive bonding locations. For the retest time consumption rate, The scanning speed of the probe movement. This refers to the time consumption for secondary heating. Density of the parent material Specific heat capacity of the base material For local wall thickness, For localized curing temperature difference, This is the power for localized heating. For the clamp attitude deflection angle, It is the local surface normal vector. For adjacent local surface normal vectors, Adjusting the fixture time rate This represents the angular velocity of the clamp's movement.

[0094] The profile spread distance is the spacing between adjacent positions when discretizing along the profile spread direction of the adhesive surface. It is used to control the discretization scale of analysis accuracy, computational scale, and polishing volume and scan length. It is preferably between 0.5 mm and 2 mm. This range can balance the geometric resolution of complex curvature areas, the projection accuracy of thermal and ultrasonic data, and the efficiency of scheduling calculations.

[0095] The equipment volumetric removal rate is the volume of material that a grinding or finishing machine can reliably remove per unit time. It is used to convert the amount of local contour finishing compensation into the grinding time. It can be obtained through equipment no-load and load calibration tests, standard test block trial cutting and grinding, or equipment process database.

[0096] The probe movement scanning speed is the linear velocity of the ultrasonic or other testing probe as it steadily moves along the scanning trajectory during retesting. It is used to convert the scan length and repetition frequency into the retesting time. It can be obtained through scanning equipment control parameters, encoder records, or standard scan calibration.

[0097] The density of the base material is the mass per unit volume of the base material of the spare part to be repaired, used to estimate the heat required for localized heating. It can be obtained from material handbooks, weighing and volume measurements, or the company's material database.

[0098] The specific heat capacity of a base material is the amount of heat required to raise its temperature by one unit mass, and it is used to calculate the theoretical heat required for localized secondary heating. It can be obtained from material thermophysical property databases, differential scanning calorimetry (DSC) tests, or enterprise process databases.

[0099] Local heating power is the effective power input to the repair area by the local secondary heating device under actual operating conditions, used to convert the required heat into time. It can be obtained through the nameplate parameters of the heating equipment, real-time measurement by a power meter, or recording by the process control system.

[0100] The local surface normal vector is a vector representation of the direction of the surface normal at a discrete location, used to describe the orientation of the local surface that the fixture needs to align.

[0101] The adjacent local surface normal vector is a vector representation of the surface normal direction at adjacent discrete positions. It is used to compare with the current local surface normal vector and calculate the fixture attitude deflection angle.

[0102] The angular velocity of a fixture is the amount of angular change that a fixture can complete per unit time during attitude adjustment. It is used to convert the fixture's attitude deflection angle into the time required for fixture adjustment. It can be obtained through fixture servo system parameters, no-load and load calibration tests, or equipment motion control records.

[0103] Local warpage curvature is a curvature index that is amplified by the curvature amount of curing mismatch deformation, heat dissipation delay and porosity influence coefficient and normalized relative to the width of the local beveled adhesive joint. It is used to describe the warpage trend that may occur on the local repair surface.

[0104] The local contour trimming compensation amount is the local trimming thickness or trimming amount calculated from the local beveling and bonding width and the local warping deformation curvature.

[0105] Grinding time rate is a time consumption index that corresponds to the removal rate of the volume to be removed and the equipment volume, which is formed by the local bevel bonding width, the contour extension distance, and the local contour trimming compensation amount. It is used to characterize the time burden of local grinding rework.

[0106] The comprehensive cost assessment index is a set of indicators that uniformly express the consumption of five types of resources: grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume.

[0107] The volume of filler material is determined by the local beveling width, the profile elongation distance, and the local profile trimming compensation amount.

[0108] The retest frequency is an index of the number of times a local residual strain energy is repeatedly detected, determined by the degree of amplification of the local residual strain energy relative to the overall average value. It is used to reflect the degree to which high-risk areas need to be repeatedly confirmed.

[0109] The retesting time consumption rate is a local retesting time consumption index determined by the retesting scan repetition frequency, the contour expansion distance, and the probe moving scan speed. It is used to characterize the workload of directional retesting.

[0110] The secondary heating time consumption rate is a local secondary heating time consumption index determined by the base material density, base material specific heat capacity, local beveling width, local wall thickness, profile extension distance, local curing temperature difference, heat dissipation delay, and local heating power. It is used to characterize the thermal rework time.

[0111] The fixture attitude deflection angle is the angle between the current local surface normal vector and the adjacent local surface normal vector, used to characterize the magnitude of attitude change that the fixture needs to adjust.

[0112] The fixture adjustment time consumption rate is a local fixture adjustment time consumption index determined by the fixture attitude deflection angle and the fixture movement angular velocity. It is used to characterize the workload of clamping and repeated attitude correction.

[0113] In detail, because the direct visual manifestations requiring rework after repair are often localized unevenness and warping, and this warping is not a single geometric error, but rather the result of the combined effects of curing mismatch deformation, heat dissipation delay, and porosity, it is more accurate to first construct the localized warping curvature and then map it to the localized contour trimming compensation amount to better reflect the actual grinding and filling needs. For example, for the same thickness difference, areas with more severe cooling unevenness often produce a larger amount of warping trimming.

[0114] In detail, since the amount of material to be removed and reworked during rework is essentially governed by the amount of local contour trimming compensation, using the same compensation amount to simultaneously drive the sanding time rate and the filling material volume can maintain logical consistency between material removal and material replenishment. For example, if the trimming amount of a certain patch edge increases, it will not only increase the sanding workload but also increase the consumption of subsequent glue and leveling materials.

[0115] In detail, because areas with high local residual strain energy are more likely to experience re-inspection disputes, repeated confirmations, and additional scans, converting the amplification of local residual strain energy relative to the average value into the frequency of retesting can reflect the process rule that higher risks require more frequent retesting. For example, high-curvature thick-walled sides often require multiple ultrasonic verifications to confirm whether the interface quality is stable.

[0116] In detail, because localized secondary heating needs to overcome not only geometric volume issues, but also the density and specific heat capacity of the base material, the local curing temperature difference, and the heat loss caused by delayed heat dissipation, using these parameters together to estimate the secondary heating time rate can more accurately reflect the burden of thermal rework than empirical time. For example, for two locations with similar volumes, if one location dissipates heat more slowly and has a higher requirement for temperature recovery, its secondary heating time will increase significantly.

[0117] In detail, because the surface orientation of curved spare parts changes constantly at adjacent positions, whether the fixture needs repeated orientation adjustments depends primarily on the change in the local surface normal vector, rather than solely on the distance. Therefore, the fixture attitude deflection angle is defined by the angle between the normal vectors of adjacent local surfaces. Combined with the fixture's angular velocity, the fixture adjustment time rate is obtained, which can capture the clamping burden on large curvature shells. For example, near corners and thickness transition coupling areas, the fixture often requires more frequent small-angle corrections.

[0118] In detail, because on-site cost management doesn't just look at a single work hour, but is simultaneously affected by five categories of resources: grinding, retesting, secondary heating, fixture adjustment, and replenishment consumables, unifying these five quantities into a comprehensive cost assessment index provides a unified input for subsequent scheduling optimization. For example, a certain location may not have a high grinding time, but if the frequency of retesting and fixture adjustments is significantly high, it will still become a key area for cost management.

[0119] In detail, the application of local warpage curvature is as follows: it should be clearly defined as applicable to local small deformation repair scenarios, and the combined result of curing mismatch deformation, heat dissipation delay, and porosity influence coefficient should be regarded as an approximate representation of the warpage trend; when there are obvious debonding bulges or local instability, they should be handled according to the defect repair rules before entering this formula. For example, when the local bulge height at the edge exceeds the range of conventional grinding, the ordinary warpage curvature model should not be directly applied.

[0120] In detail, the method for determining the value of the local contour trimming compensation is as follows: when the curvature of a local warp corresponds to the need to remove high points, the local contour trimming compensation is taken as a positive value and used in the grinding amount calculation; when the curvature of a local warp corresponds to the need to fill depressions, the absolute value of the local contour trimming compensation is used in the filling material volume calculation and treated as zero in the grinding time rate. For example, if a convex high point is measured at a certain location, grinding is required, while if there is an inward depression, the main increase is in the filling material rather than the grinding time.

[0121] In detail, the measurement methods for grinding time rate, retesting time rate, secondary heating time rate, and fixture adjustment time rate are as follows: The calculation result for each discrete position is uniformly defined as the time consumption of the corresponding process step, with units of seconds or minutes. When a rate value along the contour extension needs to be expressed, it can be divided by the contour extension distance to obtain the time consumption rate per unit length. For example, minutes can be used when outputting documents to financial accounting, while the average time per millimeter can be included when outputting to process planning.

[0122] In detail, the rounding method for the repetition frequency of the retest scan is as follows: first, obtain the continuous value according to the formula, and then convert it into the actual number of executions using the rounding up rule. The minimum value is 1, and the maximum value can be set to 3 to 5 times according to the detection resource limit. For example, if the calculated value for a certain position is 1.2 times, it will actually be executed 2 times. If the calculated value is 3.8 times, it can be executed 4 times or truncated to 4 times according to the preset upper limit.

[0123] In detail, the correction method for the secondary heating time rate is as follows: after dividing the theoretical heat by the local heating power, it should also be multiplied by a heating efficiency correction factor to account for heat loss, equipment start-up delay, and the effects of poor local thermal contact; the heating efficiency correction factor can be calibrated through a standard test block heating test. For example, equipment with the same nominal power may have significantly different actual effective heating capacities under different tooling clamping conditions.

[0124] In detail, the calculation method for the fixture attitude deflection angle is as follows: First, the local surface normal vector and the normal vectors of adjacent local surfaces should be normalized to unit length vectors. Then, the dot product should be calculated, and the result should be limited to the range of -1 to +1. Finally, the included angle should be calculated to avoid undefined results due to numerical errors. For example, if the normal vector is obtained by mesh fitting, if it is not normalized first, the dot product may be greater than 1, making it impossible to calculate the angle.

[0125] In detail, the method for converting the volume of replenishing consumables into the actual material usage is as follows: the theoretical geometric volume should be multiplied by the material loss coefficient and the curing shrinkage compensation coefficient to obtain the actual usage. The material loss coefficient can be determined based on statistics of scraping, overflow, and trimming losses. For example, if the theoretical volume is 10 cubic centimeters, and the historical average loss is 15 and the curing shrinkage compensation is 5, then the actual usage volume should be increased accordingly.

[0126] Preferably, the repair execution order and cooldown dwell time allocation are encoded, and optimization is performed based on a multi-parameter dispersion evaluation function to generate a cost management schedule, including: The calculation formulas for the distribution ratio of cooling dwell time and the cooling dwell time duration are as follows: ; ; The cross-segment strain energy conduction and the corrected local residual strain energy are calculated using the following formulas: ; ; The community dispersion weight is calculated using the following formula: ; ; The multi-parameter dispersion evaluation function is calculated, and its calculation formula is as follows: ; In the above formula, Assign a percentage to the cooling dwell time. and Weighted by dwell time, This represents the total number of adhesive bonding locations. For cooling dwell time, This refers to the time consumption for secondary heating. This represents the strain energy transfer across different sections. This refers to the local residual strain energy at the preceding position. This refers to the cooling dwell time at the front position. This refers to the localized heat dissipation delay time at the front position. To correct local residual strain energy, This is local residual strain energy. and For the discrete entropy of the group probability, The total population. The group probability of the comprehensive cost assessment index. The parameter is the community dispersion weight. For multi-parameter dispersion evaluation function, and For various comprehensive cost evaluation indicators, and These are the minimum and maximum values ​​of the comprehensive cost evaluation index, respectively.

[0127] Repair execution sequence encoding is a discrete sequence used to represent the processing order of each adhesive surface position in a candidate scheduling scheme, driving cross-segment strain energy transfer and multi-scheme optimization. Preferably, it is a non-repeating integer sequence consisting of the numbers of all adhesive surface positions from one to all others. This representation facilitates combined optimization operations such as crossover, mutation, exchange, and reversal, and can be directly mapped to the sequence of on-site work steps.

[0128] The dwell time weight is a preset weight used to describe the relative priority of each location in the cooling dwell time allocation, and is used to generate the cooling dwell time allocation ratio through exponential mapping. The preferred weight is 0.2 to 3. This range reflects the differences in priority between locations without causing extreme concentration after exponential mapping.

[0129] The cooling dwell time allocation ratio is the proportion of cooling time allocation for each location obtained by exponential mapping and normalization of dwell time weights. It is used to determine how the total cooling dwell time is allocated among different locations.

[0130] Cooling dwell time is the actual cooling waiting time at a certain location, calculated based on the proportion of cooling dwell time allocation and the total global secondary heating time rate. It is used to reflect the cooling dwell time arrangement of that location in the schedule.

[0131] Cross-section strain energy transfer is the equivalent contribution of local residual strain energy at the preceding position to subsequent positions after propagation under a predetermined repair execution sequence and cooling dwell arrangement. It is used to describe the transmission effect of work steps and cooling wait on the risk of rework in adjacent sections.

[0132] The corrected local residual strain energy is a corrected value obtained by superimposing the original local residual strain energy at the current location with the strain energy transmission amount across sections. It is used to reflect the change in the local rework driving force after the actual execution of the scheduling scheme.

[0133] Population probability is the proportion of a certain comprehensive cost assessment index in the entire candidate scheduling population, used to characterize the dispersion and information content of the index in the population.

[0134] Population probability discrete entropy is a discrete evaluation quantity calculated based on population probability, used to measure the uniformity of a certain comprehensive cost assessment index in the population.

[0135] The community dispersion weight is an index weight obtained by further converting the population probability discrete entropy. It is used to reflect the ability of different comprehensive cost evaluation indicators to distinguish the optimal results in the current population.

[0136] The minimum value of the comprehensive cost evaluation index is the minimum observed value of a certain comprehensive cost evaluation index among all candidate scheduling schemes, and is used for range normalization processing.

[0137] The maximum value of the comprehensive cost assessment index is the maximum observed value of a certain comprehensive cost assessment index among all candidate scheduling schemes, and is used for range normalization processing.

[0138] The multi-parameter dispersion evaluation function is an overall evaluation value obtained by weighting and summing various comprehensive cost evaluation indicators according to the dispersion weight of parameter clusters after range normalization. It is used to compare the advantages and disadvantages of different scheduling schemes.

[0139] The total population size represents the number of candidate scheduling schemes simultaneously retained and evaluated during the intelligent scheduling optimization process. It controls the search scope, stability, and computational cost. A value of 30 to 100 is preferred. This range typically allows for maintaining scheme diversity while keeping the computational burden within an acceptable range for engineering systems.

[0140] The difference in the median of a population is a statistic used to compare the degree of change in the population center level across different iterations or schedules, and is used to determine whether the optimization process continues or has reached a stable state.

[0141] Cost management scheduling is the optimal or near-optimal execution plan obtained by comprehensively considering the order of repair execution, the allocation of cooling time, and five comprehensive cost assessment indicators. It is used to guide on-site resource allocation and work step arrangement.

[0142] The total cost of each optimal schedule is the total consumption of the five categories of indicators (grinding, retesting, secondary heating, fixture adjustment, and replenishment materials) under the optimal cost management schedule. It is used for subsequent resource equivalence accounting and cost accounting.

[0143] In detail, because the allocation of cooldown dwell time needs to reflect the priority differences between different locations while ensuring that the total time is not unallocable, we first use dwell time weights to represent relative importance, and then obtain the cooldown dwell time allocation ratio through exponential mapping and normalization. This ensures that the sum of the ratios for each location remains stable while preserving the differences. For example, if the dwell time weight of a certain location is slightly higher, the cooldown time allocated to it will increase, but it will not directly squeeze out other locations to zero.

[0144] In detail, because the impact of rework at the initial stage on subsequent stages gradually diminishes with intermediate cooling pauses and localized heat dissipation delays, converting the local residual strain energy at the initial stage into cross-section strain energy transfer through negative exponential decay can reflect the transmission pattern of risks to subsequent stages based on scheduling order. For example, if a high-risk stage is scheduled earlier and allowed sufficient cooling time, its amplified impact on subsequent stages will be weakened.

[0145] In detail, because the distinguishing ability of different comprehensive cost evaluation indicators is not constant at different population stages, some indicators vary greatly in the current population while others vary very little. Therefore, automatically generating parameter community dispersion weights based on the population probability discrete entropy allows indicators that are better able to distinguish the merits of solutions to receive higher weights. For example, in a batch of candidate scheduling, the time spent on fixture adjustment varies greatly, while the volume of filling consumables varies very little; in this case, the former should contribute more to the evaluation function.

[0146] In detail, because the five comprehensive cost evaluation indicators have different dimensions and orders of magnitude, they cannot be directly added. Therefore, range normalization is performed first, and then the sum is weighted according to the dispersion weight of the parameter clusters to form a unified evaluation function that can be compared across indicators. For example, the grinding time is a time quantity, and the filling material volume is a volume quantity. Only by unifying the scale can the comprehensive advantages and disadvantages of the scheduling schemes be directly ranked.

[0147] In detail, because a few extreme solutions can easily distort the average, and on-site scheduling is more concerned with whether the overall population tends to stabilize and improve, using the difference in the population median as the optimization criterion can improve the robustness of the optimization process to abnormal solutions. For example, if an abnormally low-cost but obviously unfeasible solution appears in a certain iteration, the median criterion will not be overly influenced by it like the mean.

[0148] In detail, the definition of the repair execution order encoding is as follows: For each candidate scheduling scheme, a non-repeating position number sequence is established, and the order from left to right in the sequence is the actual repair execution order of the scheme; the encoding initialization can be generated by a mixture of random permutation and heuristic permutation. For example, for a task containing 20 adhesive surface positions, its repair execution order encoding can be a non-repeating sort from 1 to 20.

[0149] In detail, the setting method for dwell time weights is as follows: the initial dwell time weights for all positions should be limited to a uniform range, preferably a closed interval of 0.2 to 3, and continuous adjustment in small steps should be allowed during the optimization process; at the same time, total cooling resource constraints and minimum dwell time constraints for a single position should be set. For example, the cooling dwell time at any position should not be less than 30 seconds to ensure that the actual process is executable.

[0150] In detail, the method for converting the cooling dwell time allocation ratio into cooling dwell duration is as follows: First, sum the total secondary heating time rates to obtain the total cooling resource benchmark for this round of tasks, and then allocate it according to the cooling dwell time allocation ratio of each location; if there is a total cycle time limit on site, the total cooling dwell duration should be compressed proportionally. For example, if the total cooling resource benchmark for a certain round is 60 minutes, and the ratio of a certain location is 0.1, then its initial cooling dwell duration is 6 minutes.

[0151] In detail, the calculation method for cross-segment strain energy transfer is as follows: The cumulative cooling dwell time of all intermediate positions between the previous and current positions should be read according to the repair execution sequence. Then, the local heat dissipation delay time of the previous position is used as the decay time scale to calculate the attenuation contribution of the local residual strain energy of the previous position to the current position. For example, if the heat dissipation of the previous position is slow and the cumulative cooling time of the intermediate positions is short, a larger proportion of the cross-segment strain energy transfer will be retained.

[0152] In detail, the method for correcting the local residual strain energy and updating the calculation of grinding time, retesting time, secondary heating time, fixture adjustment time, and filling material volume is as follows: the retest scanning frequency is directly recalculated using the corrected local residual strain energy instead of the original local residual strain energy; the heat dissipation delay in the curvature of local warping deformation can be corrected in conjunction with the risk coefficient corresponding to the corrected local residual strain energy; the secondary heating time, fixture adjustment time, and filling material volume are recalculated using the updated local contour trimming compensation amount. For example, if the corrected local residual strain energy increases, the number of retests can increase, and the risk of local warping can also be amplified simultaneously.

[0153] In detail, the population probability is calculated as follows: First, divide each comprehensive cost assessment indicator into intervals within the current population using a fixed number of bins. Then, count the frequency of each interval and divide it by the total population size to obtain the population probability. The number of bins can be 5 to 10 intervals depending on the amount of data. For example, when the total population size is 50, a certain cost indicator can be divided into 6 intervals, and then the discrete entropy can be calculated accordingly.

[0154] In detail, the implementation method for optimizing the median difference in the population is as follows: A new population is generated in each iteration, the multi-parameter dispersion evaluation function for all schemes is calculated, the median is taken as the central level for that round, and compared with the median of the previous round; iteration stops when the median difference is lower than a preset threshold for three consecutive rounds; otherwise, selection, exchange, mutation, and local rearrangement are continued. For example, the threshold can be set to 0.001 to prevent the algorithm from running ineffectively during small fluctuations in later stages.

[0155] In detail, the method for generating the cost management schedule and the total cost of each optimal schedule at the optimal optimization node is as follows: After the iteration terminates, the scheme with the smallest multi-parameter dispersion evaluation function is selected as the optimal scheme, and the corresponding repair execution order encoding, cooling dwell time sequence, and cumulative values ​​of five comprehensive cost evaluation indicators are output together. For example, the output results should simultaneously show which position is processed in step 1, which position is processed in step 2, and how long each step requires to stay.

[0156] Preferably, the actual test data is integrated for post-hoc correction, and the equivalent accounting value of resource consumption and management accounting cost are output, including: The a posteriori local residual strain energy is calculated using the following formula: ; The equivalent accounting value of resource consumption is calculated using the following formula: ; The management and accounting costs are calculated using the following formula: ; In the above formula, For the a posteriori local residual strain energy, This is the predicted value of local residual strain energy. To predict the variance of uncertainty, This represents the measured value of local residual strain energy. The variance of the measured uncertainty. This is the equivalent accounting value for resource consumption. For the category of consumable items, For collection elements of the consumption item category, The optimal parameter is the community dispersion weight. This represents the total cost of the optimal scheduling for each item. For a sample set of historically related resources of the same type, For the set of location values ​​of similar historically related resource samples, To manage and account for costs, The unit cost rate for enterprise system consumption.

[0157] Actual inspection data is a collection of data formed by verifying the actual state after the repair is completed. It typically includes re-measured thermal images, re-measured ultrasonic data, geometric verification, and relevant quality records, used for post-testing correction. It can be obtained by summarizing the re-measured thermal image data, phased array ultrasonic data, geometric re-inspection data, and quality records after rework.

[0158] The measured values ​​of local residual strain energy are obtained by inversion or conversion based on actual detection data, and are used for posterior fusion with model predictions. These values ​​can be obtained through state inversion from re-measured thermographic images, re-measured ultrasonic images, deformation measurement results, and geometric verification results.

[0159] The measured uncertainty variance is a measure of the uncertainty of the measured local residual strain energy due to instrument error, repeated detection fluctuations and inversion error. It is used to determine the confidence weight of the measured value in the posterior fusion.

[0160] The predicted value of local residual strain energy is the result of local residual strain energy calculated based on the aforementioned geometric, material, curing, thermal boundary and porosity related parameter model, and is used for posterior fusion with the measured value.

[0161] The prediction uncertainty variance is a measure of the uncertainty of the predicted local residual strain energy due to model simplification, parameter fluctuations and boundary condition uncertainties. It is used to determine the confidence weight of the predicted values ​​in the posterior fusion.

[0162] The historical resource sample set is a collection of historical repair resource consumption records filtered based on similar spare parts structure, similar damage type, similar process path, and similar resource configuration. It is used to construct a benchmark for equivalent resource consumption accounting. It can be obtained through conditional searches using the enterprise's historical work order database, maintenance execution records, equipment ledgers, and cost systems.

[0163] The median value of a historically related resource sample set is the median statistical value of a certain consumption item category in the historically related resource sample set. It is used to mitigate the impact of extreme values ​​and serve as a benchmark for resource equivalence comparison.

[0164] The unit cost rate in an enterprise system refers to the unit resource price or unit time price set by the enterprise cost system for different cost item categories. It is used to convert the total cost of optimal scheduling into monetary cost. It can be obtained through the enterprise resource planning system, cost center rate table, or financial accounting benchmark table.

[0165] The consumable item category is a classification identifier used to distinguish five types of resource consumption: grinding, retesting, secondary heating, fixture adjustment, and replenishment consumables. It is used for classification, summarization, and accounting. The preferred categories are grinding, retesting, secondary heating, fixture adjustment, and replenishment consumables. This classification corresponds one-to-one with the aforementioned comprehensive consumable evaluation indicators, facilitating historical sample matching and enterprise cost mapping.

[0166] The numerator is a composite term obtained by weighting the predicted value of local residual strain energy according to the prediction uncertainty and adding it to the measured value of local residual strain energy according to the measured uncertainty. It is used for a posteriori fusion calculation.

[0167] The denominator is the total weight term obtained by summing the reciprocals of the predicted uncertainty variance and the measured uncertainty variance. It is used to normalize the numerator and form the posterior estimate.

[0168] The posterior local residual strain energy is a correction result obtained by integrating the predicted and measured values ​​of local residual strain energy, and is used to reflect the local rework driving force more closely to the actual state.

[0169] The equivalent resource consumption value is a dimensionless or quasi-dimensional resource occupancy intensity index obtained by combining the optimal parameter community dispersion weight, the total cost of each optimal schedule, and the central position value of similar historical related resource samples. It is used to compare the resource occupancy levels of different repair tasks on enterprises.

[0170] Management accounting cost is the cost result obtained by multiplying the unit consumption rate of the enterprise system by the total consumption of each optimal schedule item and summing them up. It is used to form the cost accounting value under the internal management scope of the enterprise.

[0171] In detail, because relying solely on model predictions is affected by parameter simplification and boundary errors, while relying solely on measured values ​​is affected by detection noise and inversion errors, posteriorly fusing the predicted and measured values ​​of local residual strain energy according to their respective uncertainties yields a more robust correction result. For example, when the measured value at a certain location fluctuates greatly, the system will automatically reduce the influence of the measured value on the final estimate.

[0172] In detail, because the total original cost of different tasks is affected by the size of spare parts and differences in processes, direct comparison is not fair. Therefore, the median value of historically related resource samples of the same type is introduced as a normalization benchmark. This is then combined with the optimal parameter community dispersion weight to form an equivalent accounting value of resource consumption, which can compare the resource consumption of different tasks relative to historical norms. For example, when the median value of polishing for the same type of task is low, the slightly higher polishing consumption of the current task will be more clearly identified.

[0173] In detail, because internal enterprise management ultimately needs cost results that can be entered into budgets, ledgers, and cost centers, rather than just process proxy indicators, multiplying and summing the unit consumption rate of the enterprise system with the total consumption of each optimal schedule item by item can convert engineering status quantities into manageable cost values. For example, the same retesting time will correspond to higher management and accounting costs in a workshop with a higher outsourced testing rate.

[0174] In detail, the method for obtaining the measured values ​​of local residual strain energy is as follows: First, multi-source inversion is performed on the residual thermal image temperature difference, ultrasonic anomaly indication, geometric warping residual, and local re-measurement results in the actual detection data. Then, it is converted into measured values ​​of local residual strain energy according to a scale consistent with the prediction model. If necessary, historical calibration samples can be used to establish a mapping relationship from measured features to local residual strain energy. For example, if the residual thermal image temperature difference and ultrasonic anomaly at a certain location are both high, the measured value of local residual strain energy after inversion should also be increased accordingly.

[0175] In detail, the estimation methods for measured uncertainty variance and prediction uncertainty variance are as follows: measured uncertainty variance is estimated by discrete fluctuations from more than three repeated tests, and prediction uncertainty variance is estimated by perturbation analysis of key input parameters or statistical estimation of historical prediction errors. When the sample size is insufficient, the empirical variance of similar tasks can be used as the initial value. For example, if the prediction error of a certain type of patch repair has been stable in the past 10 tasks, then this statistical error can be used as the prior for prediction uncertainty variance.

[0176] In detail, the construction method for similar historical resource sample sets is as follows: Multiple criteria are used for screening, including spare part type, base material system, patch type, damage area range, thickness transition degree, testing methods, repair equipment, and work team conditions. At least the main structural and technological characteristics must be consistent before a candidate historical sample set is formed. For example, the repair task of thick-walled, high-curvature patches for composite material shells should not be mixed with small-area repair tasks for flat plates in the same historical sample set.

[0177] In detail, the equivalent resource consumption value should be explained as follows: it should be clearly defined as a comprehensive deviation index relative to the historical median resource consumption level of similar tasks. A value greater than 1 indicates that the overall resource consumption is higher than the historical median level, a value close to 1 indicates that it is close to the historical normal level, and a value less than 1 indicates that it is lower than the historical normal level. For example, if the polishing and retesting costs of a certain task are both higher than the historical median level, the equivalent resource consumption value will be significantly higher.

[0178] In detail, the way the unit consumption rate of the enterprise system is mapped to the total consumption of each optimal schedule is as follows: the grinding time rate, retesting time rate, secondary heating time rate, and fixture adjustment time rate are respectively assigned to their respective hourly rates, the volume of consumables is assigned to the corresponding unit volume material rate, and finally calculated according to the latest rate table for the same settlement period. For example, although grinding and retesting are both time-based tasks, their equipment depreciation and labor structures are different, so different unit consumption rates should be used.

[0179] In detail, the posterior local residual strain energy is incorporated into the final output as follows: After obtaining the posterior local residual strain energy, the frequency of repeated retests and the risk of local warping related to the local residual strain energy should be recalibrated, the total cost of optimal scheduling for each item should be updated, and finally, the equivalent accounting value of resource consumption and management accounting costs should be recalculated. For example, if the posterior local residual strain energy is significantly lower than the predicted value, the retest time and the secondary heating time should be adjusted accordingly.

[0180] It should be noted that the implementation of a unified coordinate system, unified unit system, and unified time reference is as follows: the entire process uses the same three-dimensional tooling coordinate system; lengths are uniformly expressed in millimeters; time is uniformly expressed in seconds or minutes and fixed in the output aperture; temperature is uniformly expressed in degrees Celsius; and material thermophysical properties are uniformly retrieved according to the same batch database version. Pulse thermal imaging data, phased array ultrasonic data, and repair process logs must be written in the same timestamp format to ensure that there are no unit mismatches or time drifts during registration of multi-source data. For example, if the thermal imaging record uses milliseconds while the process log uses seconds, a unified conversion should be performed before importing into the system.

[0181] It should be noted that the handling method for low-confidence detection points is as follows: when the pulsed thermal imaging signal-to-noise ratio is lower than the preset lower limit, the phased array ultrasonic signal-to-noise ratio is lower than the preset lower limit, or the coordinates of two adjacent adhesive surfaces abnormally overlap, the point should be marked as a low-confidence detection point. Low-confidence detection points should not be directly involved in the calculation of local porosity index and local tangential vector, but should be filled by interpolation of adjacent effective points or local window mean. For example, when thermal image distortion is caused by reflection at the patch edge, the ultrasonic results can be retained and the thermal image results can be filled by neighborhood mean.

[0182] It should be noted that the linkage between the strain energy offset on the thick-walled side and the rework direction is as follows: when the strain energy offset on the thick-walled side increases towards the thick-walled side, the frequency of repeated scanning and cooling dwell time on that side should be increased first; when the strain energy offset on the thick-walled side increases towards the thin-walled side, it should be checked whether there are abnormal heat dissipation boundaries or abnormal fixture constraints before deciding whether to adjust the work step sequence. For example, if a task should theoretically be concentrated on the thick-walled side, but the actual post-test results are biased towards the thin-walled side, it often indicates that the field boundary conditions are inconsistent with the design boundary conditions.

[0183] It should be noted that the calibration method between the local residual strain energy proxy and the actual rework resources is as follows: Select no fewer than 20 historical repair tasks, statistically analyze the correspondence between local residual strain energy, grinding time, number of retests, secondary heating time, and filler material usage, and establish a monotonic calibration curve or piecewise mapping table. For subsequent new tasks, first obtain the relative local residual strain energy using the formula, and then convert it into a rework intensity level that more closely reflects the actual site conditions through the calibration curve. For example, the relative local residual strain energy can be divided into three levels: low, medium, and high, corresponding to different retesting strategies and material allowances.

[0184] It should be noted that the implementation methods for intelligent scheduling initialization and iterative control are as follows: the total population size is preferentially set to 50, the maximum number of iteration rounds is preferentially set to 100, the selection strategy adopts a combination of retaining the best individual and random sampling, the exchange and mutation ratio is set according to the historical task complexity, and the dwell time weight is updated in small steps after each iteration. For example, for complex surface tasks, the mutation ratio can be increased to avoid the repair execution order encoding from prematurely converging to a local optimum.

[0185] It should be noted that the version management method for similar historical resource sample sets and enterprise system unit consumption rates is as follows: historical samples are updated on a quarterly rolling basis, while enterprise system unit consumption rates are locked by financial version number. Once the current task enters the cost output stage, both the historical sample set version number and the rate version number must be written into the report to ensure traceability for subsequent review. For example, if the rate version changes when the same task is recalculated in different quarters, it should be clearly stated that the change in management accounting costs comes from rate updates rather than changes in process parameters.

[0186] It should be noted that the secondary closed-loop implementation method after posterior correction is as follows: After the initial calculation of the total cost of the optimal scheduling for each item, the actual detection data is immediately read to form the posterior local residual strain energy. Then, the repetition frequency of retesting, the amount of compensation for local contour trimming, and the allocation ratio of cooling dwell time are corrected according to the posterior local residual strain energy, generating the final version of the cost management schedule and the final version of the cost output. For example, if the initial prediction of the number of retests is 3, and the risk decreases after posterior correction, the final version can be adjusted to 2.

[0187] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.

Claims

1. An intelligent management system for spare parts repair engineering costs, characterized in that, include: The data alignment module acquires the three-dimensional contour scan data, pulsed thermal imaging data, phased array ultrasonic data and repair process log of the spare part to be repaired, and registers them to the center line of the same adhesive surface to establish the contour elongation coordinates of each adhesive surface position. The parameter solving module obtains physical property parameters and curing parameters, and solves the local bending stiffness, curing mismatch deformation, heat dissipation delay and porosity influence coefficient at each adhesive surface position based on the contour elongation coordinates. The strain energy calculation module calculates the local residual strain energy based on the local bending stiffness, the curing mismatch deformation, the heat dissipation delay, and the porosity influence coefficient, and calculates the strain energy concentration point and the geometric center point. The thick-walled side strain energy offset is obtained from the offset distance of the strain energy concentration point relative to the geometric center point. The consumption mapping module obtains the operating parameters of the processing equipment and converts the thick-walled side strain energy offset into five comprehensive consumption evaluation indicators: grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume. The intelligent scheduling module acquires initial coded variables, encodes the repair execution order and cooldown dwell time allocation, generates a multi-parameter dispersion evaluation function and optimizes it to generate a cost management schedule and the total cost of each optimal schedule. The cost output module acquires actual testing data and enterprise accounting parameters, integrates the actual testing data for post-test correction, and outputs the equivalent accounting value of resource consumption and management accounting cost.

2. The intelligent management system for spare parts repair engineering costs according to claim 1, wherein, The data alignment module establishes the contour elongation coordinates of each adhesive surface position, including: Extract the coordinates of the adhesive surface location, local wall thickness, local bevel angle, actual heat dissipation boundary, normalized thermal image defect response value, pulsed thermal image signal-to-noise ratio, normalized ultrasonic defect response value, and phased array ultrasonic signal-to-noise ratio. The contour elongation coordinates of each adhesive surface are calculated by summing the spatial straight-line distances between the coordinates of adjacent adhesive surfaces. The local tangential vector is calculated by using the ratio of the difference between the position coordinates of adjacent adhesive surfaces to the straight-line distance in space; The local surface curvature is calculated by using the ratio of the difference between adjacent local tangential vectors to the difference between the corresponding contour elongation coordinates. The width of the local bevel joint is calculated by dividing the local wall thickness by the tangent function value of the local bevel angle. Extract the shortest spatial distance from the coordinates of the adhesive surface to the actual heat dissipation boundary, and use it as the heat dissipation path length; The local porosity index is calculated by multiplying the normalized thermal image defect response value by the pulsed thermal image signal-to-noise ratio, adding the product of the normalized ultrasonic defect response value by the phased array ultrasonic signal-to-noise ratio, and then dividing by the sum of the pulsed thermal image signal-to-noise ratio and the phased array ultrasonic signal-to-noise ratio.

3. The intelligent management system for spare parts repair engineering costs according to claim 2, wherein, The parameter solving module calculates the local bending stiffness, curing mismatch deformation, heat dissipation delay, and porosity influence coefficient at each adhesive surface location based on the profile elongation coordinates, including: The following parameters are obtained: single-layer thickness, first principal axis elastic modulus, second principal axis elastic modulus, in-plane shear modulus, fiber layup angle, local equivalent Poisson's ratio, difference in thermal expansion coefficient between patch and base material, local curing temperature difference, total curing chemical shrinkage coefficient, local degree of curing and local thermal diffusivity. The ratio of the single-layer thickness to the local wall thickness is calculated as the single-layer thickness weight; Using the trigonometric function values ​​of the first principal axis elastic modulus, the second principal axis elastic modulus, the in-plane shear modulus, and the fiber layup angle, combined with the single-layer thickness weight, the local equivalent elastic modulus is calculated. The local bending stiffness is calculated by dividing the product of the local equivalent elastic modulus and the cube of the local wall thickness by the product of the difference between the local equivalent Poisson's ratio squared and the value of twelve. The amount of curing mismatch deformation is calculated by multiplying the difference in thermal expansion coefficients between the patch and the base material by the local curing temperature difference, and then adding the product of the total curing chemical shrinkage coefficient and the local curing degree. The local heat dissipation delay time is calculated by dividing the square of the heat dissipation path length by the local thermal diffusivity. The heat dissipation delay is calculated by dividing the local heat dissipation delay time by the average local heat dissipation delay time at each adhesive surface location. The porosity influence coefficient is calculated by adding the value one to the local porosity index.

4. The intelligent management system for spare parts repair engineering costs according to claim 3, wherein, The strain energy calculation module calculates the local residual strain energy, and obtains the thick-wall side strain energy offset from the offset distance of the strain energy concentration point relative to the geometric center point, including: The local residual strain energy is calculated by multiplying the local bending stiffness, the local oblique bonding width, the square of the curing mismatch deformation, the heat dissipation delay, and the porosity influence coefficient. The geometric center point is calculated by summing the product of the contour elongation coordinates and the local bevel bonding width, and then dividing by the sum of the local bevel bonding widths. The product of the contour elongation coordinates and the local residual strain energy is summed and then divided by the sum of the local residual strain energy to calculate the strain energy concentration point. The absolute value of the difference between the strain energy concentration point and the geometric center point is obtained as the offset distance. The offset distance is divided by the difference between the maximum and minimum values ​​of the profile elongation coordinates to calculate the thick-walled side strain energy offset.

5. The intelligent management system for spare parts repair engineering costs according to claim 4, wherein, The cost mapping module converts the thick-walled side strain energy offset into five comprehensive cost evaluation indicators: grinding time rate, retesting time rate, secondary heating time rate, fixture adjustment time rate, and filling material volume. The following parameters are obtained: contour elongation distance, equipment volume removal rate, probe moving scanning speed, base material density, base material specific heat capacity, local heating power, local surface normal vector, adjacent local surface normal vector, and fixture movement angular velocity. Multiply the curing mismatch deformation amount, the heat dissipation delay and the porosity influence coefficient, and then divide by the local beveled adhesive width to calculate the local warping deformation curvature; Multiply the square of the local bevel bonding width by the local warping curvature, and then divide by the value of eight to calculate the local contour trimming compensation amount. Multiply the local bevel bonding width, the contour elongation distance, and the local contour trimming compensation amount, and then divide by the equipment volume removal rate to calculate the grinding time rate. Divide the local residual strain energy by the average local residual strain energy at each adhesive surface location, and add a value of 1 to calculate the retest scan repetition frequency; multiply the retest scan repetition frequency by the profile elongation distance step distance, and then divide by the probe moving scan speed to calculate the retest time consumption rate. The secondary heating time rate is calculated by multiplying the base material density, the base material specific heat capacity, the local bevel bonding width, the local wall thickness, the profile elongation distance, the local curing temperature difference, and the heat dissipation delay, and then dividing by the local heating power. Obtain the dot product of the local surface normal vector and the adjacent local surface normal vector, and calculate the inverse cosine value of the dot product as the clamp attitude deflection angle; divide the clamp attitude deflection angle by the clamp movement angular velocity to calculate the clamp adjustment time rate; The volume of the filling material is calculated by multiplying the local bevel bonding width, the contour extension distance, and the local contour trimming compensation amount.

6. The intelligent management system for spare parts repair engineering costs according to claim 5, wherein, The intelligent scheduling module generates a multi-parameter dispersion evaluation function and optimizes it to generate a cost management schedule and the total cost of each optimal schedule, including: Obtain the dwell time weight; The dwell time weight is converted into a cooling dwell time allocation ratio using the natural exponential function; the cooling dwell time is calculated by multiplying the cooling dwell time allocation ratio by the sum of the secondary heating time consumption rate. Using the base of the natural logarithm, combined with the cumulative value of the cooling dwell time and the negative exponent formed by the local heat dissipation delay time at the previous position, the local residual strain energy at the previous position is converted into cross-segment strain energy conduction. The corrected local residual strain energy is calculated by adding the local residual strain energy to the cross-segment strain energy conduction. Based on the corrected local residual strain energy, the grinding time rate, the retesting time rate, the secondary heating time rate, the fixture adjustment time rate, and the filling material volume are updated and calculated as comprehensive cost evaluation indicators. The corresponding population probability is calculated based on the distribution ratio of the various comprehensive cost assessment indicators in the population, and the parameter community dispersion weight is calculated based on the population probability. The multi-parameter dispersion evaluation function is calculated by summing the product of the range normalization results of the comprehensive cost evaluation indicators and the corresponding parameter community dispersion weights. The cost management schedule and the total cost of each optimal schedule at the optimal optimization node are generated by optimizing the population median difference based on the multi-parameter dispersion evaluation function.

7. The intelligent management system for spare parts repair engineering costs according to claim 6, wherein, The cost output module integrates actual test data for post-hoc correction and outputs the equivalent accounting value of resource consumption and management accounting cost, including: Analyze the actual test data to extract the measured values ​​of local residual strain energy and the variance of the measured uncertainty; Extract the predicted value of the local residual strain energy and the variance of the prediction uncertainty corresponding to the local residual strain energy; Obtain the central value of similar historical related resource samples and the unit consumption rate of the enterprise system; The numerator is obtained by summing the quotient obtained by dividing the predicted value of the local residual strain energy by the variance of the prediction uncertainty and the quotient obtained by dividing the measured value of the local residual strain energy by the variance of the measured uncertainty. The quotient obtained by dividing the numerical value by the predicted uncertainty variance is summed with the quotient obtained by dividing the numerical value by the measured uncertainty variance to obtain the denominator. Divide the numerator by the denominator to calculate the a posteriori local residual strain energy. Multiply the parameter community dispersion weight in the optimal iteration state by the corresponding total cost of the optimal scheduling of each item, divide by the corresponding position value of the historical related resource sample of the same type, and sum the division results of all item categories to calculate the equivalent accounting value of resource consumption. The management accounting cost is calculated by multiplying the unit cost rate of the enterprise system by the total cost of the corresponding optimal scheduling for each item, and summing the results of the multiplication for all item categories.

Images