Processing defect real-time positioning and multi-physics field visualization system based on multi-source track

Abstract

The invention relates to the technical field of machining defect monitoring, and discloses a machining defect real-time positioning and multi-physics field visualization system based on a multi-source track. The system comprises a multi-source trajectory data acquisition module for acquiring a cutter motion trajectory sequence, a temperature distribution sequence and surface topography monitoring data to form a multi-source trajectory data set; the multi-physics field reconstruction module processes the data set to generate defect distribution characteristics containing thermal stress gradient, force distribution cumulant and defect fluctuation coefficient; the defect positioning analysis module calls the pre-training model for real-time analysis and outputs a defect position predicted value and a defect type identifier; the environment compensation correction module corrects a defect position prediction value by combining a temperature distribution sequence and a material thermal characteristic incidence relation; and a visualization strategy generation module generates a multi-physical field visualization strategy set containing thermal field rendering and force field superposition schemes according to the defect type identifier.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of processing defect monitoring, in particular to a processing defect real-time positioning and multi-physical field visualization system based on multi-source trajectories. BACKGROUND

[0002] In the modern manufacturing field, the workpiece processing quality directly determines the product performance and service life, and the timely identification and positioning of processing defects is the core link to ensure processing quality. With the development of processing technology towards high precision and high complexity, the workpiece processing process is affected by multiple factors coupling, and the traditional defect monitoring method has been difficult to meet the actual demand.

[0003] Traditional processing defect monitoring relies on a single data source, such as only collecting tool motion trajectory data or only monitoring workpiece surface temperature. Such single-source data collection method cannot fully reflect the interaction of various factors in the processing process. For example, tool motion trajectory deviation may cause abnormal workpiece surface topography, and temperature change may cause material thermal stress fluctuation, which in turn aggravates defect formation. If only a single data is relied on, the correlation between multiple factors is easily ignored, resulting in incomplete defect feature capture and lack of accurate data support for subsequent positioning analysis.

[0004] In terms of multi-physical field processing, traditional methods mostly use simple temperature distribution mapping, without integrating thermal stress changes, force distribution accumulation and defect fluctuations. In the processing process, the thermal stress gradient difference of different regions of the workpiece directly affects the probability of defect generation, and the force distribution accumulation reflects the defect development trend under long-term machining, while the traditional multi-physical field processing only focuses on temperature values, cannot generate multi-dimensional defect distribution features, and leads to lack of targeted features for subsequent defect positioning, making it difficult to accurately determine the defect type and location.

[0005] In the defect positioning link, traditional technology often uses offline analysis mode, i.e. detecting the workpiece after processing, which cannot realize real-time positioning. This lag leads to the continued processing of workpieces with defects, not only wasting raw materials and processing time, but also possibly causing chain quality problems in subsequent processes. Even some real-time monitoring technologies do not consider the influence of the processing environment on the positioning results, for example, temperature changes will change the thermal properties of the material, leading to deviation between the actual position and the predicted position of the defect, and the traditional positioning model does not include such compensation mechanism, further reducing the positioning accuracy.

[0006] In terms of visualization presentation, traditional methods mostly use single thermal field rendering, only showing temperature distribution, and cannot associate and present defect types with multi-physical field and force field. Operators need to make additional interpretation of the thermal field map based on experience to determine the defect type, which not only increases the operation difficulty, but also easily leads to defect misjudgment due to human judgment deviation, affecting the timeliness and accuracy of subsequent processing adjustment. SUMMARY

[0007] The purpose of this invention is to provide a real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides a real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories, the system comprising: The multi-source trajectory data acquisition module is configured to acquire a set of multi-source trajectory data of the target workpiece during the machining process. The set of multi-source trajectory data includes tool motion trajectory sequence, temperature distribution sequence and surface morphology monitoring data. The multiphysics reconstruction module is configured to perform multiphysics reconstruction processing on the multi-source trajectory data set to generate the defect distribution characteristics of the target workpiece, wherein the defect distribution characteristics include thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient. The defect location analysis module is configured to call a pre-trained defect location model to perform real-time defect analysis on the defect distribution characteristics and generate the defect location prediction value and defect type identifier of the target workpiece. The environmental compensation and correction module is configured to perform processing environment compensation and correction processing on the predicted defect location value to generate a corrected predicted defect location value. The processing environment compensation and correction processing is based on the correlation between the temperature distribution sequence and the thermal properties of the material. The visualization strategy generation module is configured to generate a multiphysics visualization strategy set based on the defect type identifier. The multiphysics visualization strategy set includes a thermal field rendering scheme and a force field superposition scheme.

[0009] Preferably, the multiphysics reconstruction module includes: The trajectory division unit is configured to divide the tool motion trajectory sequence into multiple trajectory sub-sequences according to a time window, and each trajectory sub-sequence corresponds to a machining cycle; The topology construction unit is configured to perform the following processing for each trajectory subsequence: construct a three-dimensional topology structure of the target workpiece based on the surface topology monitoring data, wherein the three-dimensional topology structure includes spatial distribution data of thermal displacement field, force displacement field and deformation displacement field; The coupling analysis unit is configured to perform coupling analysis processing on the three-dimensional topology and the trajectory subsequence to generate a multi-physics reconstruction result for the current time window. The multi-physics reconstruction result includes the spatial distribution matrix of thermal stress components, force distribution components and defect components. The cumulative calculation unit is configured to perform cumulative superposition processing on the multiphysics field reconstruction results of multiple consecutive time windows to calculate the thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient; wherein... The thermal stress gradient is the maximum rate of change of the thermal stress component along the workpiece surface. The cumulative force distribution is the integral of the force distribution components along the normal direction of the contact surface. The defect fluctuation coefficient is the ratio of the standard deviation to the average value of the defect component over a predetermined time interval.

[0010] Preferably, the coupling analysis unit includes: The thermal displacement mapping unit is configured to establish a thermal displacement-load mapping equation based on the correspondence between the thermal displacement field and the thermal load components in the trajectory subsequence, and obtain the first distribution function of the thermal stress components by solving the thermal displacement-load mapping equation. The contact stress calculation unit is configured to construct a contact stress calculation model based on the correlation characteristics between the force displacement field and the contact surface pressure. The contact stress calculation model includes dynamic correction parameters for the material's thermal expansion coefficient and elastic modulus. The defect iteration unit is configured to combine the spatial change rate of the deformation displacement field with the surface friction coefficient to establish a defect iteration calculation process, which includes a feedback correction mechanism for displacement increment and defect increment. The interpolation fusion unit is configured to perform spatial interpolation fusion processing on the output results of the first distribution function, the contact stress calculation model, and the defect iterative calculation process to generate three-dimensional multiphysics distribution data containing thermal stress, force distribution, and defect components.

[0011] Preferably, the defect location analysis module includes: The first feature analysis unit is configured to input the thermal stress gradient into the first feature analysis layer of the defect location model, and determine the distribution coordinates of the thermal stress concentration area and the thermal stress amplitude variation curve by calculating the thermal stress concentration factor. The second feature analysis unit is configured to input the cumulative force distribution into the second feature analysis layer of the defect location model, perform cumulative calculation of contact defect damage, and generate predicted values ​​of defect initiation probability and propagation rate of the contact surface. The third feature analysis unit is configured to input the defect fluctuation coefficient into the third feature analysis layer of the defect location model, and calculate the material loss thickness and surface roughness evolution data of the defect surface based on the surface defect degradation model. The integrated degradation unit is configured to integrate the thermal stress amplitude variation curve, the defect initiation probability, and the material loss thickness to generate a comprehensive defect index for the target workpiece, and determine the defect location prediction value based on the comparison result between the comprehensive defect index and the preset defect threshold. The region identification unit is configured with the spatial superposition result based on the distribution coordinates, the predicted expansion rate, and the surface roughness evolution data to identify the geometric location of thermal stress concentration areas, defect expansion paths, and high-risk loss areas.

[0012] Preferably, the environmental compensation and correction module includes: The temperature extraction unit is configured to extract extreme temperature values ​​and temperature change frequencies from the temperature distribution sequence, and calculate the dynamic adjustment of the material's thermal properties with temperature changes. A thermal stress compensation unit is configured to perform thermal stress compensation calculations on the thermal stress gradient based on the dynamic adjustment amount, and generate a corrected thermal stress gradient. The creep correction unit is configured to perform creep damage correction processing on the accumulated force distribution based on the correlation between the temperature change frequency and the material creep characteristics, and generate the corrected accumulated force distribution. The surface strength unit is configured to perform surface strength adaptation adjustment processing on the defect fluctuation coefficient based on the material hardness change data under extreme temperatures, and generate a corrected defect fluctuation coefficient. The recalculation unit is configured to input the corrected thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient into the defect location model for recalculation, generating a predicted defect location value after compensating for environmental factors.

[0013] Preferably, the thermal stress compensation unit includes: The thermal property acquisition unit is configured to acquire the initial thermal properties of the target workpiece at a reference temperature and the dynamic adjustment amount, and establish a thermal property-temperature correlation function. A thermal stress increment unit is configured to calculate the thermal stress increment based on the thermal characteristic-temperature correlation function, wherein the thermal stress increment is the product of the temperature change and the thermal characteristic change. The superposition unit is configured to superimpose the thermal stress increment onto the thermal stress gradient calculation process to generate a thermal stress gradient correction value that includes the influence of thermal stress. The relaxation compensation unit is configured to perform stress relaxation effect compensation processing on the thermal stress gradient correction value. The stress relaxation effect compensation processing is based on the product factor of the material stress relaxation curve and the temperature holding time.

[0014] Preferably, the visualization strategy generation module includes: The rendering scheme unit configures the identifier for the thermal stress concentration area and calculates the thermal field rendering parameters, which include the color mapping scheme and the isotherm distribution density. The superposition scheme unit is configured to construct a force field superposition scheme based on the identifier of the defect expansion path. The force field superposition scheme includes force vector display ratio and transparency settings. An animation generation unit is configured with identifiers based on the high-risk loss areas to generate dynamic visual animations, the refresh frequency of which is adjusted according to defect evolution data. The integration unit is configured to integrate the thermal field rendering parameters, the force field superposition scheme, and the dynamic visualization animation to generate a set of visualization strategies that include display parameters.

[0015] Preferably, the rendering scheme unit includes: The feature extraction unit is configured to extract the geometric features of the thermal stress concentration region and calculate the region area and the maximum thermal stress value. The color mapping unit is configured to select a color gradient range based on the area of ​​the region, wherein the color gradient range is proportional to the area of ​​the region. The isotherm unit is configured to adjust the isotherm spacing based on the maximum thermal stress value, so that the isotherm density is inversely proportional to the thermal stress value. The parameter configuration unit configures the update rate of the color mapping to be dynamically adjusted according to the material's thermal conductivity, ensuring that the visualization is synchronized with real-time data. The generation unit is configured to generate a rendering parameter table that includes the color gradient range, isotherm spacing, and update rate.

[0016] Preferably, the system further includes: The verification data acquisition module is configured to acquire the actual defect size and defect evolution data of the target workpiece within a preset verification period. The deviation analysis module is configured to perform deviation analysis processing on the actual defect size and the predicted defect location to generate a first error correction coefficient. The time-domain comparison module is configured to perform time-domain comparison processing on the defect evolution data and the predicted expansion rate to generate a second error correction coefficient. The model optimization module is configured to adjust the weight parameters of the defect location model according to the first error correction coefficient and the second error correction coefficient, and generate an optimized defect location model. The application module is configured to apply the optimized defect location model to subsequent defect location tasks of the workpiece. The multi-source trajectory data acquisition module includes: The equipment has a built-in data acquisition unit, which is configured to read the actual motion data of the machining process from the CNC system in real time and generate a sequence of tool motion trajectories. A multi-sensor data acquisition unit is configured to include a vibration sensor and a power monitoring module. The vibration sensor is configured to acquire vibration data during the machining process, and the power monitoring module is configured to monitor the power or current data of the cutting tool and estimate the temperature distribution sequence based on the vibration data and power data. A topography scanner, configured to acquire surface topography monitoring data through optical scanning; The data synchronization unit is configured to perform time alignment processing on the tool motion trajectory sequence, temperature distribution sequence, and surface morphology monitoring data to form a synchronized multi-source trajectory data set.

[0017] Preferably, the multiphysics reconstruction module is further configured to calculate the curvature of the motion trajectory and the consistency of the feed rate based on the tool motion trajectory sequence, wherein the curvature of the motion trajectory is used to identify abrupt changes in the curvature of the trajectory, and the consistency of the feed rate is used to analyze the fluctuation of the actual feed rate relative to the commanded rate. The consistency of the motion trajectory curvature and feed rate is integrated into the defect distribution features to enhance defect location accuracy.

[0018] Compared with the prior art, the beneficial effects of the present invention are: At the data acquisition level, the system integrates tool motion trajectory sequences, temperature distribution sequences, and surface morphology monitoring data through a multi-source trajectory data acquisition module, overcoming the limitations of traditional single-source data acquisition. During machining, the tool motion trajectory is directly related to the workpiece machining accuracy, the temperature distribution reflects changes in the material's thermal state, and the surface morphology is the direct external manifestation of defects. The coordinated acquisition of these three data points can comprehensively capture the dynamic changes of each key factor during machining, fully presenting the multi-factor coupling process of defect formation, avoiding the omission of defect features due to the lack of single data, and providing richer and more comprehensive basic data support for subsequent defect analysis.

[0019] The defect distribution features generated by the multiphysics reconstruction module encompass thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient. Compared to traditional multiphysics processing methods that rely solely on temperature parameters, these multidimensional features can characterize defect attributes from different perspectives. The thermal stress gradient reflects the stress changes in different areas of the workpiece due to temperature differences; the cumulative force distribution reflects the superposition of force effects under long-term processing; and the defect fluctuation coefficient captures the dynamic changing trend of the defect state. The combination of these three features can more accurately and comprehensively reflect the defect distribution pattern, providing more targeted feature basis for subsequent defect localization.

[0020] The defect location analysis module calls a pre-trained defect location model to achieve real-time defect analysis, breaking the lag limitations of traditional offline analysis. During processing, the model can simultaneously process the defect distribution characteristics after reconstruction from multi-source trajectory data, quickly outputting predicted defect locations and defect type identifiers. This allows operators to monitor workpiece defects in real time, adjust processing parameters promptly, prevent defects from continuing to develop or causing quality problems in subsequent processes, and effectively improve the dynamic control capabilities of the processing.

[0021] The environmental compensation and correction module corrects the predicted defect location based on the correlation between temperature distribution sequences and material thermal properties, overcoming the shortcomings of traditional positioning methods that ignore environmental influences. In the processing environment, temperature changes alter thermal properties such as the coefficient of thermal expansion and elastic modulus of materials, leading to deviations between the actual defect location and the initial predicted location. This module establishes a correlation mechanism between temperature and material thermal properties to specifically correct the predicted value, reducing positioning errors caused by environmental factors, making defect location judgment more consistent with the actual processing scenario, and improving positioning accuracy.

[0022] The visualization strategy generation module generates a multiphysics visualization strategy set, including thermal field rendering schemes and force field overlay schemes, based on the defect type identifier, thus overcoming the limitations of traditional single thermal field presentation. Different types of defects exhibit different behaviors in thermal and force fields. By overlaying thermal and force fields and matching the corresponding visualization scheme with the defect type, operators can more intuitively identify the defect type, location, and associated thermal and force field characteristics without relying on complex interpretations based on experience. This reduces the difficulty of defect identification, improves information acquisition efficiency, and provides a clearer reference for processing adjustment decisions. Attached Figure Description

[0023] Figure 1 This is a timing diagram of the real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories described in this invention. Figure 2 A flowchart illustrating the operation of the multiphysics reconstruction module; Figure 3 A flowchart illustrating the operation of the thermal stress compensation unit; Figure 4 This is a comparison chart of environmental compensation errors. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1 This invention provides a real-time localization and multi-physics visualization system for machining defects based on multi-source trajectories. The system includes: an integrated multi-source trajectory data acquisition module, a multi-physics reconstruction module, a defect localization analysis module, an environmental compensation and correction module, and a visualization strategy generation module. Specific implementation methods are as follows: The multi-source trajectory data acquisition module is responsible for collecting tool motion trajectory sequences, temperature distribution sequences, and surface morphology monitoring data. These data are processed synchronously to form a multi-source trajectory dataset. The multiphysics reconstruction module performs multiphysics reconstruction processing on this dataset, generating defect distribution characteristics such as thermal stress gradient, force distribution accumulation, and defect fluctuation coefficient by analyzing the coupling relationship between trajectory subsequences and the three-dimensional topology. The defect location analysis module calls a pre-trained defect location model to perform real-time analysis of defect distribution characteristics, determining the predicted defect location and defect type identifier. The environmental compensation correction module compensates and corrects the predicted defect location based on the correlation between the temperature distribution sequence and the material's thermal properties, eliminating interference from environmental factors. The visualization strategy generation module generates thermal field rendering schemes and force field superposition schemes based on the defect type identifier, forming a multiphysics visualization strategy set for dynamically displaying the defect evolution process. All modules of the system are connected through a data bus to achieve real-time data stream processing, ensuring the synchronization and accuracy of defect location and visualization.

[0026] Example 1: See Figure 2 The implementation of the multiphysics reconstruction module begins with the structured processing of the original trajectory data. The trajectory partitioning unit receives the tool motion trajectory sequence from the data acquisition module. This sequence is a continuous data stream containing three-dimensional spatial coordinates and timestamps. The partitioning process adopts a sliding time window algorithm. The window length is dynamically set according to the interpolation cycle of the CNC system of the machining center. For example, in five-axis milling, the window covers the complete toolpath segment from the tool entering the workpiece to the tool exiting the workpiece. Each trajectory subsequence not only contains the position information of the tool tip, but also records process parameters such as feed rate and spindle speed. These parameters serve as boundary conditions for subsequent coupling analysis. The topology construction unit synchronously processes surface topology monitoring data, which originates from high-density point clouds acquired by an online laser scanner. After filtering and denoising, the point cloud data is used to construct a triangular mesh model of the workpiece surface using the Delaunay triangulation algorithm. This mesh model forms the geometric basis of the three-dimensional topology structure. On this mesh model, the thermal displacement field is calculated by spatially matching the temperature measurement points in the temperature distribution sequence with the mesh nodes, and converting the temperature change into a linear expansion using the material's thermal expansion coefficient, thus obtaining the displacement vector of each node under thermal load. The force displacement field is constructed based on the principles of contact mechanics, inversely calculating the contact pressure distribution based on the tool movement trajectory, and then combining the material's elastic modulus with Hooke's law to calculate the elastic deformation field. The deformation displacement field is directly derived from the surface topology differences in continuous scanning. Iterative nearest-point registration is performed on the mesh models of adjacent time frames to calculate the displacement components of plastic deformation. These three displacement fields are uniformly mapped onto the same mesh nodes, forming a comprehensive topology structure containing multi-physics information.

[0027] The coupled analysis unit is the core of multiphysics reconstruction, and its operation is performed window-by-window based on the trajectory subsequence. The thermal displacement mapping unit first establishes the thermal displacement-load mapping equation, which is essentially an equilibrium equation in thermoelasticity, considering the bidirectional coupling effect between the temperature and stress fields. The displacement-load mapping equation can be expressed as the following partial differential equation:

[0028] in: It is the gradient operator, representing the spatial derivative; It is the Laplace operator, representing the second spatial derivative; It is a displacement vector field that describes the displacement vectors at various points on the workpiece surface. It is a temperature field, representing the temperature distribution on the surface of the workpiece; and It is the Lamé constant, which is related to the elastic modulus and Poisson's ratio of the material, and characterizes the shear stiffness and volumetric deformation resistance, respectively. It is the thermal stress coefficient, defined as ,in It is the coefficient of thermal expansion of the material; It is a heat flux density vector, representing the heat load component extracted from the trajectory subsequence, and serves as the source term of the equation.

[0029] The solution process employs the finite element method, discretizing the continuous domain into mesh elements. A linear equation system is established on each element using the Galerkin weighted residual method. The coefficient matrix of the equation system includes thermal conductivity and elastic constants, and the right-hand side is the heat flux density boundary condition extracted from the trajectory subsequence. The solver uses the conjugate gradient method for iterative calculation, and the output is the distribution of thermal stress components on the mesh nodes, i.e., the first distribution function. The contact stress calculation unit runs in parallel. Its model is based on an extension of Hertzian contact theory, introducing a temperature influence factor to dynamically correct the material's elastic modulus and Poisson's ratio. The correction parameters are obtained in real-time from the material parameter table based on the average temperature value of the current time window. The contact area is determined by detecting the intersection of the tool trajectory and the workpiece mesh. The calculated contact stress distribution is applied as a Neumann boundary condition to the force-displacement field. The defect iteration unit handles nonlinear deformation problems, employing an explicit time integration algorithm. During initialization, both displacement and defect increments are set to zero. Within each time step, equivalent strain is calculated based on the spatial gradient of the current deformation displacement field. The stress field is then updated using the material constitutive relation, and the surface friction power is updated based on the relative velocity between the tool and the workpiece. The defect increment is calculated using a wear model and used as feedback to correct the deformation displacement field in the next time step. This iteration continues until the difference in defect increments between adjacent steps is less than a tolerance threshold. Finally, the interpolation and fusion unit integrates the outputs of the three units. Due to potential differences in the computational meshes of each unit, an inverse distance weighted interpolation algorithm is used before fusion to resample all physical fields onto a unified regular mesh. Thermal stress components, force distribution components, and defect components are organized into a three-dimensional array structure, with each mesh point storing the scalar or vector value of the corresponding physical quantity, forming a complete spatial distribution matrix.

[0030] The cumulative calculation unit processes data from continuous time windows. Its accumulation and superposition are not simple arithmetic sums but consider the superposition of physical processes. The calculation of the thermal stress gradient uses the central difference method, calculating the difference in thermal stress values ​​between adjacent points at each point in the 3D mesh, and taking the maximum value of the differences in all directions as the gradient value at that point. The thermal stress gradient field of the entire workpiece is thus generated. The calculation of the cumulative force distribution involves time integration. At each grid point, the normal force distribution components of the continuous time window are numerically integrated along the time axis using the trapezoidal rule. The result represents the cumulative load borne by that point throughout the entire processing. The calculation of the defect fluctuation coefficient requires a sequence of defect components within a predetermined time interval, which typically covers the entire processing cycle or a critical process stage. First, the average value and standard deviation of the defect components at each grid point within this time interval are calculated. Then, the standard deviation is divided by the average value to obtain the dimensionless fluctuation coefficient, which characterizes the stability of defect development. All cumulative results are ultimately organized into a defect distribution characteristic data set, which is passed as standardized interface data to the downstream defect location analysis module. The entire multiphysics reconstruction process employs a pipelined architecture, with processing of the current time window and data acquisition of the next window occurring in parallel to ensure system real-time performance. Internal caching mechanisms are incorporated within each module; when a computational unit takes an extended period, the time window overlap rate is automatically adjusted to maintain data flow continuity. All physics calculations utilize double-precision floating-point numbers, and the mesh size is adaptively adjusted based on the workpiece's geometric complexity, automatically refining the mesh in regions with significant curvature variations. The reconstruction results not only include scalar field data but also generate metadata recording computational parameters and boundary conditions, facilitating subsequent module traceability and analysis.

[0031] Example 2: The implementation process of the defect location analysis module begins with receiving and parsing the defect distribution features output by the multiphysics reconstruction module. This module calls the pre-trained defect location model to perform parallel analysis and processing of the input thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient. The first feature analysis unit specifically processes the thermal stress gradient data. This unit is the first convolutional neural network layer in the defect location model. Its structure contains multiple sets of separable convolutional kernels to extract spatial features at different scales. The thermal stress gradient data is input in the form of a two-dimensional matrix. Each element in the matrix corresponds to the gradient value of a grid point on the workpiece surface. The convolutional kernel performs sliding calculations on the matrix to generate a feature map. The calculation of the thermal stress concentration factor is completed through the local maximum region in the location feature map. The algorithm uses non-maximum suppression technology to eliminate noise interference. The final output distribution coordinates are represented by the grid index number and include the actual three-dimensional workpiece coordinates of that point. The thermal stress amplitude variation curve is generated by tracking the maximum gradient value of the same region in different time windows to generate a time series curve. The second feature analysis unit simultaneously processes the cumulative force distribution. This unit employs a physics-based damage accumulation calculation rule rather than a simple neural network, requiring the reading of fatigue characteristic parameters from a material library during the calculation process. Contact defect damage accumulation calculation decomposes the cumulative force distribution into multiple stress cycles. The amplitude of each cycle is identified using the rainflow counting method. The defect initiation probability is calculated based on a modified Miner's rule, summing the damage degree of each stress cycle and comparing it with the material durability limit to obtain the probability value. Defect propagation rate prediction applies fracture mechanics principles, treating the contact area as a microcrack initiation point, and calculating the rate value based on the stress intensity factor range and the material constant in Paris's law. The third feature analysis unit focuses on processing the defect fluctuation coefficient. This unit implements a surface defect degradation model that integrates Arcard's wear law and roughness transfer theory. The material loss thickness is calculated by multiplying and integrating the defect fluctuation coefficient with the sliding distance and contact pressure. Surface roughness evolution data is predicted by establishing a correlation model between the fluctuation coefficient and roughness parameters. The model parameters need to be calibrated for specific materials during the preprocessing stage.

[0032] The integrated degradation unit receives the output results from the first three units and performs data fusion. The thermal stress amplitude variation curve, defect initiation probability, and material loss thickness are first normalized to the same numerical range. The fusion algorithm adopts an adaptive weighted summation method, and the weight coefficients are dynamically adjusted according to the currently identified defect type. For example, when the thermal stress amplitude is significantly high, its weight is increased accordingly. The integrated defect index is calculated as a dimensionless scalar value. The comparison between this index and the preset defect threshold adopts a multi-level threshold strategy. Different thresholds correspond to different defect severity levels. The comparison result not only generates a Boolean defect presence identifier, but also outputs a confidence score. The defect location prediction value is output in the form of a point set, and each point contains three-dimensional coordinates and a defect probability value. The region identification unit finally performs a spatial information integration operation. This unit rasterizes the thermal stress concentration areas represented by distributed coordinates, the defect propagation paths described by the propagation rate prediction values, and the high-risk loss areas indicated by the surface roughness evolution data. The rasterization process discretizes the workpiece surface into a fine two-dimensional grid, and each grid cell is assigned a corresponding attribute value according to the spatial interpolation of the input data. The thermal stress concentration areas are identified as continuous high gradient value regions, the defect propagation paths are obtained by calculating the streamlines of the rate field, and the high-risk loss areas are delineated by setting a critical value for the material loss thickness. The final output geometric position information adopts the boundary representation method, and each region is defined by its contour polygon vertex sequence, along with time evolution attributes.

[0033] The entire defect localization analysis process employs a hierarchical decision-making mechanism. The bottom-level feature analysis unit generates initial hypotheses, while the higher-level fusion unit verifies and integrates these hypotheses. An internal cache queue manages data from different time steps, ensuring the continuity of time-series analysis. The defect localization model utilizes an online learning mechanism, enabling fine-tuning of network parameters based on feedback from the environmental compensation correction module. All analysis results are timestamped and labeled with confidence levels, and indexed to the original trajectory data for easy access by the visualization module and subsequent traceability analysis. This implementation method, through multi-feature fusion and spatial topology analysis, transforms abstract physical field data into specific defect location and type information, providing direct decision-making support for quality control during the manufacturing process.

[0034] The module's data flow processing adopts a pipelined architecture, with three feature analysis units capable of parallel computation to improve efficiency. Each unit contains multiple processing pipelines. The convolutional neural network of the first feature analysis unit employs a multi-scale feature pyramid structure, capable of simultaneously capturing localized subtle stress concentrations and macroscopic distribution patterns. Data augmentation techniques used during network training include elastic deformation, rotation, and scaling to enhance the model's robustness to different workpiece postures. The damage accumulation calculation of the second feature analysis unit incorporates real-time material condition monitoring data, updating material fatigue characteristic parameters in real time through embedded sensors, enabling the damage model to adapt to changes in material performance during processing. The calculation also considers load sequence effects and employs a bilinear accumulation rule to more accurately predict defect initiation life. The degradation model of the third feature analysis unit uses an incremental update mechanism, recalibrating the model coefficients based on new monitoring data after each processing cycle to maintain prediction accuracy over time.

[0035] The multi-source information fusion algorithm of the integrated degradation unit is based on an improvement of DS evidence theory, capable of handling uncertainties and conflicts in the output results of different feature units. This unit also incorporates a contradiction detection mechanism, automatically triggering a recalculation process when there is a significant deviation in the defect location indicated by different features. The spatial overlay algorithm of the region identification unit employs hierarchical grid management, using different resolution grids for regions of varying importance, optimizing memory usage while ensuring computational accuracy. The identification results also undergo morphological post-processing, including region closure and small region removal, resulting in smoother and more reasonable output geometric boundaries. The module's interfaces with other parts of the system use standardized data formats, with both input and output data carrying complete metadata descriptions, including data source, processing time, and coordinate system information. The module has an internal anomaly handling mechanism; when the input data quality is poor, it automatically degrades to a conservative prediction mode and records warning information through a log system. The computational complexity of the entire defect location analysis module has been optimized, meeting real-time requirements on a standard industrial computer, providing reliable technical support for online quality monitoring of the manufacturing process.

[0036] Example 3: See Figure 3The implementation of the environmental compensation and correction module involves performing processing environment compensation and correction processing on the predicted defect location value to generate a corrected predicted defect location value. The processing environment compensation and correction processing is based on the correlation between temperature distribution sequence and material thermal properties. The environmental compensation and correction module includes a temperature extraction unit, a thermal stress compensation unit, a creep correction unit, a surface strength unit, and a recalculation unit. These units work together to eliminate the influence of processing environment fluctuations on defect location. The temperature extraction unit is configured to extract extreme temperature values ​​and temperature change frequencies from the temperature distribution sequence, and calculate the dynamic adjustment of the material's thermal properties with temperature changes; the thermal stress compensation unit is configured to perform thermal stress compensation calculations on the thermal stress gradient based on the dynamic adjustment, and generate a corrected thermal stress gradient; the creep correction unit is configured to perform creep damage correction processing on the cumulative force distribution based on the correlation between the temperature change frequency and the material's creep characteristics, and generate a corrected cumulative force distribution; the surface strength unit is configured to perform surface strength adaptation adjustment processing on the defect fluctuation coefficient based on the material hardness change data at extreme temperatures, and generate a corrected defect fluctuation coefficient; the recalculation unit is configured to input the corrected thermal stress gradient, the corrected cumulative force distribution, and the corrected defect fluctuation coefficient into the defect location model for recalculation, and generate a predicted defect location value after compensating for environmental factors.

[0037] In specific implementation, the temperature extraction unit extracts extreme temperature values ​​and temperature change frequencies from the temperature distribution sequence of the multi-source trajectory data set. Extreme temperature values ​​are obtained by scanning the temperature time series to identify global maximum and minimum points, and the temperature change frequency is calculated by analyzing the periodic components of the temperature sequence using Fast Fourier Transform. Based on the extracted extreme temperature values ​​and temperature change frequencies, the temperature extraction unit calculates the dynamic adjustment amount of the material's thermal properties with temperature changes. This dynamic adjustment amount is achieved by querying a material thermal property database, which stores calibrated values ​​of the coefficient of thermal expansion and elastic modulus at different temperatures. The dynamic adjustment amount is calculated as the difference between the thermal property value at the current temperature and the thermal property value at the reference temperature. In some embodiments, the temperature extraction unit uses a sliding window algorithm to update the extreme temperature values ​​in real time, with the window size synchronized with the processing cycle to ensure that the dynamic adjustment amount reflects instantaneous environmental changes. A low-pass filter is introduced in the calculation of the temperature change frequency to eliminate high-frequency noise and improve the accuracy of frequency estimation. The thermal stress compensation unit receives the dynamic adjustment amount output by the temperature extraction unit and performs thermal stress compensation calculations on the thermal stress gradient. Specifically, the thermal stress compensation calculation first establishes a thermal characteristic-temperature correlation function using a thermal characteristic acquisition unit. This function fits the relationship between parameters such as the material's thermal expansion coefficient and elastic modulus and temperature in polynomial form. Subsequently, the thermal stress increment unit calculates the thermal stress increment based on this function. This increment is the product of the temperature change and the thermal characteristic change. The calculation process considers the spatial distribution of the temperature gradient and is performed point-by-point on the grid. The core of the compensation calculation is to integrate the thermal stress increment into the analysis of the thermal stress gradient. By quantifying the additional effect of temperature change on thermal stress, the original thermal stress gradient value is corrected to more accurately reflect the stress state under actual thermal load. The thermal stress compensation unit includes a thermal characteristic acquisition unit, a thermal stress increment unit, a superposition unit, and a relaxation compensation unit. The thermal characteristic acquisition unit is configured to acquire the initial thermal characteristics and dynamic adjustment amount of the target workpiece at the reference temperature, and establish a thermal characteristic-temperature correlation function. The thermal characteristic-temperature correlation function is expressed in polynomial form and obtained by fitting experimental data using the least squares method. The thermal stress increment element is configured to calculate the thermal stress increment based on the thermal characteristic-temperature correlation function. The thermal stress increment is the product of the temperature change and the thermal characteristic change. The calculation process takes into account the spatial distribution of the temperature gradient. The superposition element is configured to superimpose the thermal stress increment into the calculation process of the thermal stress gradient, generating a thermal stress gradient correction value that includes the influence of thermal stress. The superposition operation adopts vector addition and is performed point by point on the grid.Specifically, the superposition operation employs vector addition, performed point-by-point on the workpiece surface grid to ensure spatial consistency. The thermal stress increment is an additional stress value calculated based on the thermal characteristic-temperature correlation function, representing the additional thermal stress contribution caused by temperature fluctuations. During superposition, this value is weighted and merged with the original thermal stress gradient value, with the weighting coefficient dynamically adjusted according to the current temperature distribution and material properties. The generated thermal stress gradient correction value not only includes the initial gradient information but also incorporates the compensation effect of the thermal stress increment, thus more comprehensively capturing the actual distribution of thermal stress under changes in the processing environment and providing more reliable feature input for subsequent defect localization. The relaxation compensation unit is configured to perform stress relaxation effect compensation processing on the thermal stress gradient correction value. This stress relaxation effect compensation processing is based on the product factor of the material stress relaxation curve and the temperature holding time, which is calibrated through accelerated aging experiments. It can be understood that the core of thermal stress compensation calculation lies in quantifying the additional impact of temperature fluctuations on thermal stress and eliminating time accumulation errors through incremental superposition and relaxation compensation. Specifically, the product factor in the stress relaxation effect compensation processing is a key parameter based on the material stress relaxation curve and the temperature holding time, used to quantify the influence of stress relaxation on the thermal stress gradient correction value. The material stress relaxation curve, obtained through experimental measurements, describes the stress decay over time under constant strain. The product factor, a time-temperature superposition factor calibrated through accelerated aging experiments, correlates the temperature holding time with the stress relaxation rate, specifically manifested as a scaling factor used to adjust the intensity of the relaxation effect. During compensation, the relaxation compensation unit applies the product factor to the thermal stress gradient correction value, calculates the attenuation caused by stress relaxation using an integral model, and adjusts the correction value accordingly to offset the underestimation of stress caused by material creep and relaxation, ensuring that the compensated thermal stress gradient more closely matches the actual time-temperature history behavior during processing.

[0038] The following formula is used to quantify the change in thermal stress in the calculation of the thermal stress increment element: Δσ thermal =∫(α(T)·ΔT)dA Where, Δσ thermal Let α(T) represent the thermal stress increment, α(T) represent the temperature-dependent coefficient of thermal expansion, ΔT represent the temperature change, and A represent the workpiece surface region. The meanings of all characters in the formula are as follows: Δσ thermalThe formula defines the thermal stress increment in Pascals; α(T) is the coefficient of thermal expansion, varying with temperature T, measured in Kelvin; ΔT is the temperature change, calculated using the difference between the current and reference temperatures, also measured in Kelvin; and dA represents the area element, with the integral covering the entire workpiece surface. This formula ensures that the thermal stress increment calculation considers the nonlinear variation of the coefficient of thermal expansion and spatial inhomogeneity. In practice, the thermal stress increment element discretizes the temperature distribution sequence into grid data, calculates the local ΔT value at each grid point, and obtains the α(T) value through interpolation. The integration process employs numerical integration algorithms such as Gaussian quadrature. In the calculation of the thermal stress gradient correction value for the superimposed element, the original thermal stress gradient and the thermal stress increment are combined using a weighted coefficient adjusted based on temperature sensitivity, which is derived from the material's thermal conductivity. Specifically, the derivation process is based on the correlation between material thermal conductivity and thermal diffusivity: higher thermal conductivity leads to faster heat transfer, resulting in a more uniform temperature distribution and reducing the impact of local temperature changes on thermal stress, thus reducing temperature sensitivity; conversely, lower thermal conductivity makes temperature changes more concentrated in local areas, leading to higher temperature sensitivity. During the derivation, the system retrieves thermal conductivity data from a material database and combines this with extreme temperature values ​​and frequency of change in the temperature distribution sequence to calculate the dynamic adjustment of thermal properties with temperature changes. Weighting coefficients are dynamically allocated based on temperature sensitivity values. For example, in thermal stress compensation calculations, higher weights are assigned to highly sensitive areas to more accurately correct the thermal stress gradient and ensure that the predicted defect location is not affected by environmental fluctuations. The entire process is integrated into the establishment of the thermal property-temperature correlation function, indirectly quantifying temperature sensitivity through material thermal conductivity to optimize the environmental compensation effect. The stress relaxation effect compensation of the relaxation compensation unit uses an integral model, with the product factor representing the time-temperature superposition factor, and the temperature holding time and relaxation rate are correlated through the Arrhenius equation. In some embodiments, the thermal stress compensation unit employs an iterative optimization process to dynamically adjust the coefficients of the thermal characteristic-temperature correlation function to adapt to different material types. The relaxation compensation process introduces an error feedback mechanism; when the deviation between the thermal stress gradient correction value and the measured data exceeds a threshold, the product factor is automatically recalibrated. The creep correction unit performs creep damage correction processing on the cumulative force distribution based on the correlation between temperature change frequency and material creep characteristics. Creep characteristics are modeled using creep constitutive equations, such as the Norton power-law model. The creep damage correction processing includes calculating the cumulative creep strain and adding it as a correction term to the original cumulative force distribution. The surface strength unit performs surface strength adaptation adjustment processing on the defect fluctuation coefficient based on material hardness change data at extreme temperatures. Hardness change data is obtained through real-time hardness sensors or pre-stored hardness-temperature curves. The adaptation adjustment process uses a linear scaling model, multiplying the defect fluctuation coefficient by the normalized hardness value.The recalculation unit takes the corrected thermal stress gradient, the corrected cumulative force distribution, and the corrected defect fluctuation coefficient as inputs, and re-executes the forward calculation process of the defect location model to generate a predicted defect location value after compensating for environmental factors. The recalculation process maintains the model structure unchanged, but the input features now include environmental compensation information, thereby improving the robustness of defect prediction. Optionally, the environmental compensation correction module can be configured to process compensation tasks for multiple workpieces in parallel, accelerating real-time response through distributed computing. It can be understood that the environmental compensation correction module effectively isolates environmental noise through a multi-level compensation mechanism, ensuring the stability of the defect location system under varying temperature conditions. In specific implementation, the output defect location prediction value of the recalculation unit is checked for consistency with the uncompensated result. The verification method uses spatial overlap analysis; when the overlap is lower than a preset threshold, a recompensation cycle is triggered. The dynamic adjustment calculation of the temperature extraction unit also considers the material phase transition effect, treating the phase transition temperature point as a special extreme value to avoid errors caused by sudden changes in thermal characteristics. In the relaxation compensation process of the thermal stress compensation unit, the material stress relaxation curve is measured using a dynamic mechanical analyzer, and the product factor is represented as a piecewise function to adapt to different temperature ranges. The creep damage correction process of the creep correction unit introduces a damage accumulation model, such as Robinson's damage rule, to calculate the creep life consumption ratio. The adaptation adjustment process of the surface strength unit also considers the influence of residual surface stress and corrects the hardness value using X-ray diffraction data. In some embodiments, the overall process of the environmental compensation correction module is embedded in a real-time operating system, with the compensation cycle synchronized with data acquisition to ensure low-latency output. All compensation parameters are stored in non-volatile memory, supporting offline calibration and online updates. In specific implementations, the thermal characteristic acquisition unit of the thermal stress compensation unit uses a multivariate regression method to introduce additional variables, such as heating rate, to improve function accuracy when establishing the thermal characteristic-temperature correlation function. The vector addition operation of the superposition unit is parallelized on the GPU to process high-resolution mesh data. The product factor of the relaxation compensation unit is dynamically predicted through a machine learning model, with model training data derived from historical processing records. The implementation of the environmental compensation correction module significantly improves the reliability of defect localization in thermal fluctuation environments through a refined compensation chain.

[0039] See Figure 4This chart presents key data visualization results from the environmental compensation and correction module within a multi-physics visualization system for real-time machining defect localization based on multi-source trajectories. The chart, presented as a grouped bar chart, clearly compares the defect location errors in six monitored areas of the workpiece before and after environmental compensation. The vertical axis quantifies the defect location error, while the horizontal axis divides the workpiece monitoring areas. The intuitive contrast between light and dark gray bars visualizes the effectiveness of the environmental compensation mechanism. This chart validates the core logic of the environmental compensation and correction module: by extracting extreme temperatures and frequency of change from the temperature distribution sequence, it dynamically adjusts parameters such as thermal stress gradient and cumulative force distribution, correcting the positioning deviation caused by environmental factors, making the defect location prediction more consistent with the actual machining scenario. This chart not only intuitively quantifies the effect of environmental compensation but also highlights the module's crucial value in improving defect location accuracy and providing precise decision-making basis for machining quality control in complex machining environments.

[0040] Example 4: The implementation of the visualization strategy generation module involves generating a multiphysics visualization strategy set based on defect type identifiers. The multiphysics visualization strategy set includes a thermal field rendering scheme and a force field overlay scheme. The visualization strategy generation module includes a rendering scheme unit, an overlay scheme unit, an animation generation unit, and an integration unit. The rendering scheme unit is configured to calculate thermal field rendering parameters for the identifiers of thermal stress concentration areas. The thermal field rendering parameters include a color mapping scheme and isotherm distribution density. The overlay scheme unit is configured to construct a force field overlay scheme based on the identifiers of the defect expansion path. The force field overlay scheme includes force vector display ratio and transparency settings. The animation generation unit is configured to generate dynamic visualization animations based on the identifiers of high-risk loss areas. The dynamic visualization animations adjust their refresh rate according to defect evolution data. The integration unit is configured to integrate the thermal field rendering parameters, the force field overlay scheme, and the dynamic visualization animations to generate a visualization strategy set containing display parameters. In specific implementation, the rendering scheme unit further includes a feature extraction unit, a color mapping unit, an isotherm unit, a parameter configuration unit, and a generation unit. The feature extraction unit is configured to extract the geometric features of the thermal stress concentration area and calculate the area and the maximum thermal stress value. The color mapping unit is configured to select the color gradient range based on the area and the color gradient range is proportional to the area. The isotherm unit is configured to adjust the isotherm spacing based on the maximum thermal stress value and make the isotherm density inversely proportional to the thermal stress value. The parameter configuration unit is configured to dynamically adjust the update rate of the color mapping based on the material's thermal conductivity to ensure that visualization is synchronized with real-time data. The generation unit is configured to generate a rendering parameter table containing the color gradient range, isotherm spacing, and update rate.

[0041] The feature extraction unit receives the identification data of thermal stress concentration areas from the defect location analysis module. This identification data exists in the form of a sequence of polygon vertex coordinates. The feature extraction unit calculates the area of ​​the polygon by calculating the pixel coverage of the polygon region, and simultaneously scans the thermal stress distribution matrix to obtain the maximum thermal stress within the area. The color mapping unit maps the calculated area to a predefined color gradient range; the larger the area, the wider the assigned color band. The color gradient range uses a continuous gradient from blue to red in the HSV color space. The isotherm unit uses the maximum thermal stress as a reference and calculates the isotherm spacing using logarithmic scaling. Higher maximum thermal stress values ​​result in an automatically increased time interval to avoid overly dense isotherms. The parameter configuration unit reads the material's thermal conductivity coefficient in real time, with the update rate set to be linearly correlated with the thermal conductivity coefficient; a higher thermal conductivity coefficient corresponds to a higher color refresh rate. The generation unit organizes the above parameters into a structured table for storage, as shown in the table below: Table 1: Thermal Field Rendering Parameter Configuration Table

[0042] In specific implementations, the color gradient range selection algorithm of the color mapping unit includes a saturation adjustment step, automatically enhancing color saturation to improve visual contrast when the area exceeds a threshold. The spacing adjustment of the isotherm unit introduces a minimum spacing constraint to prevent isotherms from becoming too sparse when the thermal stress value is too low. The update rate calculation of the parameter configuration unit also considers the refresh capability of the display device; when the calculation rate exceeds the device's limit, a frame dropping strategy is used to maintain smoothness. In some embodiments, the area calculation of the feature extraction unit uses a grid discretization method, projecting the polygonal region onto a regular grid and counting the number of non-empty grids. The maximum thermal stress value is obtained by sampling within the region and applying a maximum value filtering algorithm. The HSV color space conversion of the color mapping unit uses the standard RGB to HSV conversion formula and is calibrated for the color gamut of industrial displays. The isotherm generation of the isotherm unit uses a moving cube algorithm to extract isosurfaces from the thermal stress scalar field, and the spacing adjustment is achieved by modifying the threshold step size of the isosurface extraction. Specifically, the thermal stress scalar field is three-dimensional grid data generated by the multiphysics reconstruction module, with each grid point storing the thermal stress value. The moving cube algorithm generates continuous isotherms by traversing each cube cell in the mesh and interpolating isosurfaces (i.e., surfaces with specific thermal stress values) based on the thermal stress values ​​at the cell vertices. Spacing adjustment is achieved by modifying the threshold step size for isosurface extraction: the threshold step size defines the interval of thermal stress difference between adjacent isosurfaces during extraction. When it is necessary to increase the isotherm spacing, the system increases the threshold step size, allowing the isosurface extraction to cover a wider range of thermal stresses and reducing the isotherm density; conversely, decreasing the threshold step size reduces the spacing and increases the isotherm density. This adjustment is based on the maximum thermal stress value, ensuring that the isotherm density is inversely proportional to the thermal stress value. For example, a larger step size is used in high thermal stress areas to avoid visual overlap, while a smaller step size is used in low stress areas to capture details. The algorithm is integrated into the parameter configuration of the isotherm cells, achieving adaptive optimization of the isotherm distribution by dynamically adjusting the threshold step size, thus improving the clarity and real-time performance of the thermal field rendering. The material thermal conductivity coefficient of the parameter configuration cells is dynamically loaded from the material database, and the linear relationship coefficient of the update rate is calibrated through display performance testing. The rendering parameter table of the generated unit is output as a JSON file, containing parameter names, values, and unit information for the visualization engine to parse.

[0043] The overlay unit receives identification data of the defect propagation path, which is represented as a series of continuous spatial points, each associated with force vector data and propagation rate value. The force field overlay scheme constructed by the overlay unit calculates the force vector display ratio, defined as the ratio of the force vector magnitude to the reference force value, which is set based on the material's yield strength. Transparency is set based on the depth value or overlay level of the path points; deeper path points use higher transparency to reflect a three-dimensional sense of hierarchy. In specific implementation, the calculation of the force vector display ratio incorporates a dynamic scaling factor. When the force vector magnitude exceeds a threshold, the display ratio is automatically compressed to avoid visual overlap. Specifically, the dynamic scaling factor is an adaptively adjusted parameter used to automatically scale the vector display ratio based on the size range of real-time force vector data and the display context, avoiding visual overlap or distortion. This factor is obtained through the statistical distribution of the force vector magnitude: the system first analyzes the magnitude sequence of all force vectors within the current time window, calculates their maximum, minimum, and average values, and then dynamically generates a scaling coefficient based on the spatial size and resolution of the display area. For example, when the magnitudes of force vectors differ significantly, the scaling factor compresses the display ratio of excessively large vectors while magnifying excessively small vectors, ensuring that all vectors are uniformly visible in the visualization. The main purpose of introducing a dynamic scaling factor is to optimize the visualization effect of the force field overlay scheme. It automatically adjusts the vector display based on force changes during processing, preventing vectors from becoming too long or too short due to force value fluctuations, thus affecting recognition. Transparency settings employ alpha blending technology, mapping the height values ​​of path points to the transparency channel. The overlay scheme unit also includes a vector smoothing step, performing spline interpolation on the force vector sequence to generate smooth vector lines, enhancing visualization coherence. Optionally, the overlay scheme unit can be configured to use different force vector color codes for different defect types, such as yellow vectors for fatigue defects and red vectors for overload defects. The animation generation unit generates dynamic visualization animations based on the identification of high-risk loss areas. These high-risk loss areas are defined by the probability map output by the defect location analysis module. The animation generation unit calculates the refresh frequency based on the time series of defect evolution data, which includes the material loss thickness change rate and surface roughness change gradient. The dynamic visualization animation employs keyframe animation technology. The keyframe interval is adaptively adjusted according to the defect change rate, with shorter intervals for larger change rates. The animation generation unit also includes an interpolation algorithm to generate smooth transition frames between keyframes. The interpolation algorithm uses spherical linear interpolation to maintain the geometric consistency of the spatial path. It can be understood that the dynamic adjustment of the animation refresh rate is crucial for ensuring real-time performance; the frequency calculation is synchronized with the data acquisition cycle to avoid screen tearing.

[0044] The integration unit integrates the rendering parameter table output by the rendering scheme unit, the force field superposition scheme constructed by the superposition scheme unit, and the dynamic visualization animation generated by the animation generation unit. The integration process includes three steps: parameter format unification, coordinate system unification, and timing synchronization. Parameter format unification converts the outputs of different units into a unified XML description file. Coordinate system unification transforms the local coordinates of each element to the global workpiece coordinate system. Timing synchronization uses a timestamp alignment mechanism to ensure precise matching between physical field data and animation frames. The final visualization strategy set generated by the integration unit is a structured configuration file containing all parameters of the thermal field rendering scheme, vector and transparency settings for the force field superposition scheme, and frame sequence information and refresh rules for the dynamic visualization animation. In specific implementation, the format unification step of the integration unit uses XSLT transformation to convert the JSON-formatted rendering parameter table into XML nodes, embeds the vector data of the force field superposition scheme in SVG path format, and stores the animation frame sequence as a time-encoded list. Coordinate system unification uses a homogeneous transformation matrix to transform the local coordinates output by each unit to a global coordinate system with the workpiece base point as the origin. The transformation matrix is ​​determined by the workpiece clamping parameters. The timing synchronization module uses a network time protocol to align data streams, ensuring that each frame of the dynamic visualization animation strictly corresponds to the output timestamp of the multiphysics reconstruction module. Optionally, the integration unit can add a visualization preview function to verify the rendering effect before generating the formal strategy; only after successful verification will the final strategy set be output. In practice, the integration process also includes resource optimization steps, compressing large-size textures or high-frequency animation data in the visualization strategy set. The compression algorithm uses lossy compression based on visual importance, reducing storage and transmission overhead while maintaining visual quality. The integration unit's output interface supports multiple industry standard protocols, such as OPCUA, facilitating integration with third-party visualization platforms. The overall process of the visualization strategy generation module adopts a pipeline architecture, with each unit processing visualization tasks for different artifacts in parallel, and data flow and load balancing achieved through message queues.

[0045] Example 5: The verification data acquisition module collects the actual defect size and defect evolution data of the target workpiece within a preset verification cycle. The preset verification cycle is set to perform a comprehensive verification every five processing cycles, based on the total processing time. The actual defect size is obtained by a coordinate measuring machine (CMM) through contact scanning of the workpiece surface during processing breaks, with a measurement point density of one point per square millimeter. The defect evolution data is continuously recorded by a high-speed industrial camera fixed in the processing chamber at a rate of 1,000 frames per second, showing the surface morphology changes of specific areas of the workpiece. In specific implementation, the trigger logic of the verification data acquisition module is interconnected with the main control program of the CNC system. When the processing cycle counter reaches the preset modulus value, the verification process is automatically started. The probe path of the CMM is planned according to the defect location prediction value predicted by the defect location analysis module, prioritizing the scanning of the predicted defect area and its surrounding transition zone. The shooting focal length and illumination intensity of the high-speed industrial camera are dynamically adjusted according to the reflective characteristics of the workpiece material. The acquired image sequence is transmitted in real time to the image processing unit for defect contour extraction and size quantization. The deviation analysis module performs deviation analysis between the actual defect size and the predicted defect location. The actual defect size is represented as 3D point cloud data, and the predicted defect location is the coordinate set output by the defect localization model. The deviation analysis process first matches each actual defect point to the nearest predicted point using a nearest neighbor algorithm, then calculates the Euclidean distance of all matched point pairs and takes the average as the position error. The first error correction coefficient is finally calculated as the reciprocal of the position error value and normalized. The temporal comparison module performs temporal comparison between the defect evolution data and the predicted propagation rate. The defect evolution data is a time series of defect area extracted from high-speed camera image sequences, and the predicted propagation rate is the predicted defect propagation rate output by the defect localization analysis module. The temporal comparison process uses a dynamic time warping algorithm to align the two time series and calculates their Pearson correlation coefficient as a consistency measure. The second error correction coefficient is obtained by mapping the correlation coefficient to the range of zero to one using an S-shaped function. Specifically, the Pearson correlation coefficient is applied to the temporal comparison module to quantify the degree of linear correlation between the defect evolution data and the predicted propagation rate. In the specific calculation process, the defect evolution data is a time series of defect areas on the workpiece surface acquired from a high-speed industrial camera, reflecting the dynamic changes of the actual defects. The predicted propagation rate is a theoretical value output by the defect localization analysis module based on multiphysics reconstruction, representing the expected trend of defect propagation. The Pearson correlation coefficient is calculated by aligning the points of the two time series using a dynamic time warping algorithm to eliminate nonlinear distortions on the time axis, and then analyzing the coordinated changes in the sequence data: first, the covariance of the two sequences is calculated to measure whether their directions of change are consistent; then, it is divided by the product of their respective standard deviations to normalize and obtain the correlation coefficient, with a value range of -1 to 1. A value close to 1 indicates a strong positive correlation, meaning that the actual defect propagation is highly synchronized with the prediction; a value close to -1 indicates a strong negative correlation, indicating prediction deviation; and a value close to 0 indicates no linear relationship.Its function is to provide objective correlation indicators for model optimization. A second error correction coefficient is generated through an sigmoid function mapping, used to adjust the weight parameters of the defect localization model, enhancing the model's adaptability to temporal dynamics, thereby improving the accuracy and robustness of defect prediction and reducing misjudgments caused by environmental fluctuations or measurement noise. The model optimization module adjusts the weight parameters of the defect localization model based on the first and second error correction coefficients. The adjustment process uses a stochastic gradient descent algorithm with momentum. The first error correction coefficient is used to scale the gradient of the position correlation loss, and the second error correction coefficient is used to adjust the weights of the time series prediction loss. The optimized defect localization model is deployed to the production environment only after it has passed full validation set testing. The application module applies the optimized defect localization model to subsequent workpiece defect localization tasks. Model loading is achieved through a dedicated model service interface. The multi-source trajectory data set of each new workpiece is input into the latest version of the defect localization model to generate real-time prediction results. The multi-source trajectory data acquisition module includes a built-in data acquisition unit, a multi-sensor data acquisition unit, a topography scanner, and a data synchronization unit. The built-in data acquisition unit reads encoder data from the servo motor and interpolation data from the G-code interpreter in real time from the CNC system's PLC via the OPCUA protocol, generating a tool motion trajectory sequence after coordinate transformation. The multi-sensor data acquisition unit includes a vibration sensor mounted on the spindle head and a power monitoring module integrated into the driver. The vibration sensor uses an IEPE-type accelerometer to collect vibration acceleration data in the XYZ directions during machining. The power monitoring module monitors the three-phase current and voltage of the spindle motor using a Hall effect sensor, based on the vibration... Dynamic and power data are used to estimate the temperature distribution sequence through a thermal generation model. The thermal generation model converts mechanical vibration energy and electrical power loss into heat flux density, which is then input into the finite element thermal model of the workpiece to solve for the temperature field. The topography scanner uses blue light structured light projection technology to perform non-contact three-dimensional scanning of the workpiece surface during the tool retraction gap to acquire surface topography monitoring data. The data synchronization unit uses hardware trigger signals to coordinate all data sources. The sampling clocks of the vibration sensor and power monitoring module are phase-locked with the interpolation clock of the CNC system. The exposure time of the topography scanner is strictly synchronized with the tool retraction signal. Finally, a timestamp interpolation algorithm is used to align all data streams to a unified time base axis to form a synchronized multi-source trajectory data set.

[0046] In some embodiments, the coordinate measuring machine of the verification data acquisition module is equipped with a temperature drift compensation function, which corrects measurement errors in real time through an ambient temperature sensor; the image processing unit of the high-speed industrial camera uses a deep learning segmentation network to extract defect regions from the background, and the network model has been trained on a set of labeled images containing various defect types before use. The nearest neighbor matching algorithm of the deviation analysis module uses a KD tree structure to accelerate the search process, and predicted points without matching actual points are considered false alarm errors and included in the overall error calculation. The dynamic time warping algorithm of the time domain comparison module allows for nonlinear scaling between two time series, and the constraint window width of the warping path is set to one-tenth of the sequence length to balance flexibility and computational efficiency. The momentum stochastic gradient descent algorithm of the model optimization module sets the initial learning rate to 0.01 and the momentum coefficient to 0.9, and the optimization process continues until the comprehensive error index on the validation set no longer decreases for ten consecutive iterations. The model service interface of the application module supports gRPC remote calls, allowing multiple processing centers in a distributed deployment to share the same optimized defect localization model instance. The multi-source trajectory data acquisition module's built-in data acquisition unit records process parameters such as feed rate and spindle rate for subsequent analysis when reading G-code interpolation data. The vibration sensor's installation position is determined through modal analysis to avoid machine tool structural nodes. The power monitoring module's current sensor has an accuracy of 0.5 class, capable of distinguishing minute power fluctuations caused by tool wear. The structured light projection pattern of the topography scanner is coded to overcome workpiece surface reflection interference. The hardware trigger signal of the data synchronization unit is generated by a dedicated synchronization controller, with jitter controlled at the microsecond level. Optionally, the verification data acquisition module can be configured to use an online probe to collect partial verification data during machining. The online probe is installed in the tool magazine and is switched to the spindle by a robot arm when needed to perform on-machine measurement. Optionally, the deviation analysis module, in addition to calculating positional error, can also calculate the ratio of the predicted error to the actual error of defect area as a third error correction coefficient for model optimization. Optionally, the data synchronization unit can integrate a GPS timing module for data synchronization and comparative analysis between multiple machine tools. In some embodiments, the model optimization module retains multiple versions of the historical model when adjusting weight parameters, allowing for rapid rollback to a previous stable version when the new model degrades on the validation set. Vibration sensor data from the multi-source trajectory data acquisition module undergoes wavelet denoising before being used for temperature estimation, and readings from the power monitoring module are converted based on the motor efficiency curve to obtain the actual power output. It can be understood that the entire implementation of Embodiment 5 constitutes a complete closed-loop optimization system. By continuously collecting actual processing data and comparing it with prediction results, the performance of the defect location model is iteratively optimized, enabling the system to adapt to the effects of long-term factors such as tool wear and material batch variations.In practical implementation, the preset verification cycle of the verification data acquisition module can be dynamically adjusted according to the production cycle, extending the verification interval during high-load production and shortening it during trial production. The error calculation of the deviation analysis module distinguishes between systematic and random errors, using only the systematic error portion to generate the first error correction coefficient. The Pearson correlation coefficient calculation of the time-domain comparison module removes obvious outliers in the data to avoid distortion. The optimization process of the model optimization module is executed on a dedicated GPU computing server to avoid occupying computing resources on the processing site. The model update of the application module adopts a hot-swap method; after the new model is loaded, the old model is not immediately deleted, but it is run in parallel for a period of time for result comparison. The switch is only completed after the performance of the new model is confirmed to be stable. The data synchronization unit of the multi-source trajectory data acquisition module also records the timestamp and quality flag of each data packet, and can automatically enable interpolation compensation or mark the data as unavailable when an anomaly occurs in a data source. In practice, the data exchange between the verification data acquisition module, deviation analysis module, time domain comparison module, model optimization module and application module is achieved through shared memory and message queues to ensure high throughput and low latency. Each sub-unit of the multi-source trajectory data acquisition module has a self-diagnostic function, regularly reports its own health status, and immediately notifies the host computer system once a fault is detected.

[0047] Example 6: The implementation of the multiphysics reconstruction module is further configured to calculate the curvature of the motion trajectory and the consistency of the feed rate based on the tool motion trajectory sequence. The curvature of the motion trajectory is used to identify abrupt changes in the curvature, and the consistency of the feed rate is used to analyze the fluctuation of the actual feed rate relative to the commanded rate. The curvature of the motion trajectory and the consistency of the feed rate are integrated into the defect distribution features to enhance the defect positioning accuracy. In specific implementation, the curvature of the motion trajectory is calculated by processing continuous path points in the tool motion trajectory sequence. The path points contain three-dimensional coordinates and temporal information. The curvature calculation uses differential geometry to solve for the reciprocal of the radius of curvature at each point on the trajectory line. The curvature calculation process first performs B-spline curve fitting on the path point sequence to eliminate measurement noise, and then calculates the first and second derivatives of the curve. The final curvature value is obtained through the formula of the ratio of the derivative modulus. The identification of abrupt changes in the curvature of the motion trajectory is achieved by analyzing the first difference sequence of curvature values. When the absolute value of the difference exceeds the adaptive threshold, it is marked as an abrupt change point. The adaptive threshold is dynamically adjusted according to the statistical variance of historical curvature data. Feed rate consistency analysis is achieved by comparing the degree of agreement between the actual feed rate and the commanded feed rate. The actual feed rate is calculated by dividing the displacement difference between adjacent path points by the time difference, while the commanded feed rate is obtained in real time from the G-code parser of the CNC system. Feed rate consistency is quantified as the fluctuation of the actual feed rate sequence relative to the commanded feed rate sequence. The fluctuation is comprehensively evaluated by calculating the standard deviation and correlation coefficient of the two sequences within the sliding window.

[0048] The integration of motion trajectory curvature and feed rate consistency into the defect distribution features is manifested as an extension operation of the feature vector. The original defect distribution features contain scalar field data such as thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient. The extended defect distribution features add two dimensions: the motion trajectory curvature field and the feed rate consistency field. In practice, the motion trajectory curvature field maps the curvature values ​​at discrete path points to a three-dimensional mesh identical to the thermal stress gradient using spatial interpolation. The feed rate consistency field is mapped to the mesh of the contact area between the tool and the workpiece based on the machining time. The integration method employs feature stitching technology, stitching the data matrices of the curvature field and consistency field with the original defect distribution feature matrix in the channel dimension to form an enhanced multi-channel defect distribution feature tensor. It can be understood that abrupt changes in motion trajectory curvature correspond to regions of rapid changes in the tool's movement direction, often accompanied by instantaneous fluctuations in cutting force. Feed rate consistency deviation reflects the degree of deviation between the actual machining process and the theoretical process, which may lead to uneven material removal. In some embodiments, the curvature calculation of the motion trajectory is corrected in conjunction with tool geometry parameters; for example, the curvature calculation of a ball end mill needs to consider tip radius compensation. Feed rate consistency analysis distinguishes the fluctuation characteristics of different motion stages, using different weighting coefficients for fluctuations in the rapid positioning stage and the finishing stage. After the enhanced defect distribution feature tensor is input into the defect localization model, the model automatically learns the correlation between curvature features, speed consistency features, and defect types through a convolutional neural network. High-frequency features of curvature abrupt change points are transmitted to the deep network through residual connections, and the temporal features of feed rate consistency are extracted through long short-term memory network units.

[0049] The algorithm for identifying abrupt changes in motion trajectory curvature includes a multi-scale analysis process. It detects abrupt changes on trajectory curves with varying degrees of smoothness, and finally merges the detection results from all scales to reduce the probability of missed detections. The adaptive sliding window size for feed rate consistency analysis is related to the process duration of the machining operation. A large window is used to analyze long-term trends in the roughing stage, while a small window is used to capture instantaneous fluctuations in the finishing stage. Optionally, motion trajectory curvature data can be used to correct the calculation weight of thermal stress gradients, increasing the calculation sensitivity of thermal stress gradients in high-curvature regions to capture local stress concentration phenomena. Optionally, the feed rate consistency index can be used as a confidence weight for the cumulative force distribution; when consistency is poor, the confidence of the cumulative force distribution in the corresponding region is reduced. It can be understood that the introduction of motion trajectory curvature and feed rate consistency provides kinematic supplementary information for defect localization, allowing defect prediction to move beyond static physical field analysis. In some embodiments, the motion trajectory curvature field is normalized to eliminate magnitude differences between different machining paths, and the feed rate consistency field is Z-score normalized to conform to Gaussian distribution characteristics. Before generating the final defect distribution features, the multiphysics reconstruction module performs outlier detection on the motion trajectory curvature field and feed rate consistency field, using the isolated forest algorithm to identify and remove noisy data points that significantly deviate from the main distribution. In practice, the spatial coordinates of curvature abrupt change points in the motion trajectory are spatially correlated with the defect location predictions output by the defect localization model to establish a statistical relationship model between curvature abrupt changes and defect initiation. The specific process begins by calculating the motion trajectory curvature based on the tool motion trajectory sequence and identifying curvature abrupt change points—these points correspond to areas of abrupt changes in the tool's motion direction, often accompanied by instantaneous fluctuations in cutting force and localized stress concentrations. The model collects the spatial distribution characteristics of curvature abrupt change points (such as abrupt change intensity, frequency, and location) and actual defect initiation records (such as the location of microcracks or material loss) from historical machining data. Statistical learning methods, such as regression analysis or machine learning classifiers, are used to train the mapping relationship between abrupt change point features and defect probabilities. For example, linear regression or support vector machines are used to analyze the positive correlation between abrupt change point density and defect initiation probability, or clustering algorithms are used to identify curvature patterns of frequently occurring defects. Its function is to integrate kinematic dynamic characteristics into defect distribution features, supplementing static physical field information such as thermal stress gradient and cumulative force distribution. This allows the defect location model to provide earlier and more accurate warnings of potential defect areas, making it particularly suitable for scenarios with complex toolpaths in high-precision machining, thereby improving the overall reliability and adaptability of the system. The fluctuation pattern of feed rate consistency is analyzed in the time domain against the predicted defect propagation rate to identify the impact of abnormal machining rhythm on defect propagation. The multiphysics reconstruction module effectively enhances the ability of defect distribution features to represent machining dynamics by introducing two feature dimensions: motion trajectory curvature and feed rate consistency, providing richer input information for the defect location analysis module.

[0050] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories, characterized in that, The system includes: The multi-source trajectory data acquisition module is configured to acquire a set of multi-source trajectory data of the target workpiece during the machining process. The set of multi-source trajectory data includes tool motion trajectory sequence, temperature distribution sequence and surface morphology monitoring data. The multiphysics reconstruction module is configured to perform multiphysics reconstruction processing on the multi-source trajectory data set to generate the defect distribution characteristics of the target workpiece, wherein the defect distribution characteristics include thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient. The defect location analysis module is configured to call a pre-trained defect location model to perform real-time defect analysis on the defect distribution characteristics and generate the defect location prediction value and defect type identifier of the target workpiece. The environmental compensation and correction module is configured to perform processing environment compensation and correction processing on the predicted defect location value to generate a corrected predicted defect location value. The processing environment compensation and correction processing is based on the correlation between the temperature distribution sequence and the thermal properties of the material. The visualization strategy generation module is configured to generate a multiphysics visualization strategy set based on the defect type identifier. The multiphysics visualization strategy set includes a thermal field rendering scheme and a force field superposition scheme.

2. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 1, characterized in that, The multiphysics reconstruction module includes: The trajectory division unit is configured to divide the tool motion trajectory sequence into multiple trajectory sub-sequences according to a time window, and each trajectory sub-sequence corresponds to a machining cycle; The topology construction unit is configured to perform the following processing for each trajectory subsequence: construct a three-dimensional topology structure of the target workpiece based on the surface topology monitoring data, wherein the three-dimensional topology structure includes spatial distribution data of thermal displacement field, force displacement field and deformation displacement field; The coupling analysis unit is configured to perform coupling analysis processing on the three-dimensional topology and the trajectory subsequence to generate a multi-physics reconstruction result for the current time window. The multi-physics reconstruction result includes the spatial distribution matrix of thermal stress components, force distribution components and defect components. The cumulative calculation unit is configured to perform cumulative superposition processing on the multiphysics field reconstruction results of multiple consecutive time windows to calculate the thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient; wherein... The thermal stress gradient is the maximum rate of change of the thermal stress component along the workpiece surface. The cumulative force distribution is the integral of the force distribution components along the normal direction of the contact surface. The defect fluctuation coefficient is the ratio of the standard deviation to the average value of the defect component over a predetermined time interval.

3. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 2, characterized in that, The coupling analysis unit includes: The thermal displacement mapping unit is configured to establish a thermal displacement-load mapping equation based on the correspondence between the thermal displacement field and the thermal load components in the trajectory subsequence, and obtain the first distribution function of the thermal stress components by solving the thermal displacement-load mapping equation. The contact stress calculation unit is configured to construct a contact stress calculation model based on the correlation characteristics between the force displacement field and the contact surface pressure. The contact stress calculation model includes dynamic correction parameters for the material's thermal expansion coefficient and elastic modulus. The defect iteration unit is configured to combine the spatial change rate of the deformation displacement field with the surface friction coefficient to establish a defect iteration calculation process, which includes a feedback correction mechanism for displacement increment and defect increment. The interpolation fusion unit is configured to perform spatial interpolation fusion processing on the output results of the first distribution function, the contact stress calculation model, and the defect iterative calculation process to generate three-dimensional multiphysics distribution data containing thermal stress, force distribution, and defect components.

4. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 1, characterized in that, The defect location and analysis module includes: The first feature analysis unit is configured to input the thermal stress gradient into the first feature analysis layer of the defect location model, and determine the distribution coordinates of the thermal stress concentration area and the thermal stress amplitude variation curve by calculating the thermal stress concentration factor. The second feature analysis unit is configured to input the cumulative force distribution into the second feature analysis layer of the defect location model, perform cumulative calculation of contact defect damage, and generate predicted values ​​of defect initiation probability and propagation rate of the contact surface. The third feature analysis unit is configured to input the defect fluctuation coefficient into the third feature analysis layer of the defect location model, and calculate the material loss thickness and surface roughness evolution data of the defect surface based on the surface defect degradation model. The integrated degradation unit is configured to integrate the thermal stress amplitude variation curve, the defect initiation probability, and the material loss thickness to generate a comprehensive defect index for the target workpiece, and determine the defect location prediction value based on the comparison result between the comprehensive defect index and the preset defect threshold. The region identification unit is configured with the spatial superposition result based on the distribution coordinates, the predicted expansion rate, and the surface roughness evolution data to identify the geometric location of thermal stress concentration areas, defect expansion paths, and high-risk loss areas.

5. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 1, characterized in that, The environmental compensation and correction module includes: The temperature extraction unit is configured to extract extreme temperature values ​​and temperature change frequencies from the temperature distribution sequence, and calculate the dynamic adjustment of the material's thermal properties with temperature changes. A thermal stress compensation unit is configured to perform thermal stress compensation calculations on the thermal stress gradient based on the dynamic adjustment amount, and generate a corrected thermal stress gradient. The creep correction unit is configured to perform creep damage correction processing on the accumulated force distribution based on the correlation between the temperature change frequency and the material creep characteristics, and generate the corrected accumulated force distribution. The surface strength unit is configured to perform surface strength adaptation adjustment processing on the defect fluctuation coefficient based on the material hardness change data under extreme temperatures, and generate a corrected defect fluctuation coefficient. The recalculation unit is configured to input the corrected thermal stress gradient, cumulative force distribution, and defect fluctuation coefficient into the defect location model for recalculation, generating a predicted defect location value after compensating for environmental factors.

6. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 5, characterized in that, The thermal stress compensation unit includes: The thermal property acquisition unit is configured to acquire the initial thermal properties of the target workpiece at a reference temperature and the dynamic adjustment amount, and establish a thermal property-temperature correlation function. A thermal stress increment unit is configured to calculate the thermal stress increment based on the thermal characteristic-temperature correlation function, wherein the thermal stress increment is the product of the temperature change and the thermal characteristic change. The superposition unit is configured to superimpose the thermal stress increment onto the thermal stress gradient calculation process to generate a thermal stress gradient correction value that includes the influence of thermal stress. The relaxation compensation unit is configured to perform stress relaxation effect compensation processing on the thermal stress gradient correction value. The stress relaxation effect compensation processing is based on the product factor of the material stress relaxation curve and the temperature holding time.

7. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 4, characterized in that, The visualization strategy generation module includes: The rendering scheme unit configures the identifier for the thermal stress concentration area and calculates the thermal field rendering parameters, which include the color mapping scheme and the isotherm distribution density. The superposition scheme unit is configured to construct a force field superposition scheme based on the identifier of the defect expansion path. The force field superposition scheme includes force vector display ratio and transparency settings. An animation generation unit is configured with identifiers based on the high-risk loss areas to generate dynamic visual animations, the refresh frequency of which is adjusted according to defect evolution data. The integration unit is configured to integrate the thermal field rendering parameters, the force field superposition scheme, and the dynamic visualization animation to generate a set of visualization strategies that include display parameters.

8. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 7, characterized in that, The rendering scheme unit includes: The feature extraction unit is configured to extract the geometric features of the thermal stress concentration region and calculate the region area and the maximum thermal stress value. The color mapping unit is configured to select a color gradient range based on the area of ​​the region, wherein the color gradient range is proportional to the area of ​​the region. The isotherm unit is configured to adjust the isotherm spacing based on the maximum thermal stress value, so that the isotherm density is inversely proportional to the thermal stress value. The parameter configuration unit configures the update rate of the color mapping to be dynamically adjusted according to the material's thermal conductivity, ensuring that the visualization is synchronized with real-time data. The generation unit is configured to generate a rendering parameter table that includes the color gradient range, isotherm spacing, and update rate.

9. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 1, characterized in that, The system also includes: The verification data acquisition module is configured to acquire the actual defect size and defect evolution data of the target workpiece within a preset verification period. The deviation analysis module is configured to perform deviation analysis processing on the actual defect size and the predicted defect location to generate a first error correction coefficient. The time-domain comparison module is configured to perform time-domain comparison processing on the defect evolution data and the predicted expansion rate to generate a second error correction coefficient. The model optimization module is configured to adjust the weight parameters of the defect location model according to the first error correction coefficient and the second error correction coefficient, and generate an optimized defect location model. The application module is configured to apply the optimized defect location model to subsequent defect location tasks of the workpiece. The multi-source trajectory data acquisition module includes: The equipment has a built-in data acquisition unit, which is configured to read the actual motion data of the machining process from the CNC system in real time and generate a sequence of tool motion trajectories. A multi-sensor data acquisition unit is configured to include a vibration sensor and a power monitoring module. The vibration sensor is configured to acquire vibration data during the machining process, and the power monitoring module is configured to monitor the power or current data of the cutting tool and estimate the temperature distribution sequence based on the vibration data and power data. A topography scanner, configured to acquire surface topography monitoring data through optical scanning; The data synchronization unit is configured to perform time alignment processing on the tool motion trajectory sequence, temperature distribution sequence, and surface morphology monitoring data to form a synchronized multi-source trajectory data set.

10. The real-time localization and multiphysics visualization system for machining defects based on multi-source trajectories according to claim 1, characterized in that: The multiphysics reconstruction module is also configured to calculate the curvature of the motion trajectory and the consistency of the feed rate based on the tool motion trajectory sequence. The curvature of the motion trajectory is used to identify abrupt changes in the curvature of the trajectory, and the consistency of the feed rate is used to analyze the fluctuation of the actual feed rate relative to the commanded rate. The consistency of the motion trajectory curvature and feed rate is integrated into the defect distribution features to enhance defect location accuracy.

Images