Titanium alloy thin-wall part high-precision milling method and system based on dynamic compensation

Abstract

The invention provides a titanium alloy thin-wall part high-precision milling method and system based on dynamic compensation, and the method comprises the following steps: constructing a multi-physics field coupling model based on the constitutive relation and milling process parameters of a titanium alloy material; according to the method, a multi-physical field coupling model represented by dynamic parameters is constructed, structural feature recognition and coupling weight dynamic allocation are combined, and the model is iteratively optimized through a detection result after processing, so that the problem that a traditional prediction model is disjointed from an actual working condition is solved; by building a real-time monitoring and feedback system and combining a dynamic compensation closed-loop mechanism, non-interruption linkage of monitoring-adjustment in the machining process is achieved, and the problem that the hysteresis of a traditional compensation method is remarkable is solved; the initial compensation amount is optimized through a multi-objective genetic algorithm, a self-adaptive processing path strategy is adopted, dynamic parameter adjustment of structure adaptation is combined, a traditional single compensation strategy is replaced, and the problem of dynamic rigidity change in the complex curved surface and material removal process can be flexibly solved.

Description

Technical Field

[0001] This invention relates to the field of CNC machining technology, and in particular to a method and system for high-precision milling of thin-walled titanium alloy parts based on dynamic compensation. Background Technology

[0002] Titanium alloy thin-walled components, with their superior specific strength (strength-to-density ratio reaching 200-250 MPa·m³ / kg, far exceeding that of steel and aluminum alloys), excellent corrosion resistance, and good high-temperature mechanical properties, have become core structural components in aerospace, high-end equipment, and other fields. Examples include spacecraft cabin walls (typically 1.2-3 mm thick), aero-engine casings (diameters reaching 1-3 m, with some areas only 1 mm thick), and thin-walled frames for missile bodies. Lightweight design of these components can directly reduce equipment energy consumption and improve endurance or payload. However, their low stiffness and sensitivity to machining stress make deformation control difficult. Traditional methods mainly rely on finite element simulation to predict deformation and then perform static compensation, or on-line toolpath adjustment after machine measurement. However, certain problems still exist. 1. The prediction model is out of touch with the actual working conditions: Finite element simulation relies on simplified boundary conditions, which makes it difficult to accurately reflect the coupling effects of multiple physical fields such as cutting force, thermal stress, and vibration. 2. Significant compensation lag: Offline measurement and compensation cause processing interruptions and cannot adapt to real-time changes in deformation trends; Third, the compensation strategy is too simple: traditional mirror compensation or fixed parameter adjustment cannot cope with the dynamic stiffness changes of complex surfaces and material removal processes. To address this, a high-precision milling method and system for thin-walled titanium alloy parts based on dynamic compensation is proposed. Summary of the Invention

[0003] In view of this, the embodiments of the present invention aim to provide a high-precision milling method and system for thin-walled titanium alloy parts based on dynamic compensation, so as to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0004] To solve the above-mentioned technical problems, this application adopts a technical solution as follows: a high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation, comprising the following steps: Step 1: Based on the constitutive relation of titanium alloy material and milling process parameters, construct a multiphysics coupling model; Step 2: Based on the multiphysics coupling model, predict the deformation trend of titanium alloy thin-walled parts during the milling process and generate prediction results; Step 3: Based on the prediction results, calculate the initial compensation amount, and use a multi-objective genetic algorithm to optimize the initial compensation amount, balance processing efficiency and deformation control, and generate a pre-compensation instruction set; Step 4: Build a real-time monitoring and feedback system to collect dynamic machining data during the milling process and compare it with the prediction results to determine the deformation deviation; Step 5: Based on the deformation deviation, dynamically adjust the compensation strategy and processing parameters using the pre-compensation instruction set as a benchmark; Step 6: Based on the adjusted compensation strategy and machining parameters, generate an adaptive machining path and execute the milling operation; Step 7: Perform deformation detection on the processed titanium alloy thin-walled parts, and feed the detection results back to the multiphysics coupling model for iterative optimization.

[0005] As a further preferred embodiment of this technical solution, in step one, the method for constructing the multiphysics coupling model includes the following steps: Step 101: Obtain the mechanical and thermophysical parameters of the titanium alloy material and determine the characterization method of the material under different temperatures and stress states; Step 102: Select the key physical fields that affect machining deformation and define the coupling relationship between each physical field; the key physical fields include the cutting force field, temperature field and internal stress field; Step 103: Set the boundary conditions of the multiphysics coupling model; the boundary conditions include the contact conditions between the tool and the workpiece, the heat dissipation conditions, and the constraint conditions. Step 104: The multiphysics coupling model is calibrated using experimental data, and the material property parameters and field coupling coefficients are dynamically updated until the prediction error of the multiphysics coupling model meets the preset threshold. The multiphysics coupling model characterizes the machining process through dynamic parameters, including calculating the cutting heat generation rate in a way related to cutting power, making the thermal conductivity coefficient change dynamically with temperature, and using functions to characterize the softening effect of titanium alloy materials during machining.

[0006] As a further preferred embodiment of this technical solution, in step four, the real-time monitoring and feedback system includes a sensor for collecting dynamic processing data, wherein the sensor has a sampling frequency ≥ 2kHz, a measurement accuracy ≤ ±0.005mm, and a data processing delay ≤ 5ms.

[0007] As a further preferred embodiment of this technical solution, in step three, the multi-objective genetic algorithm optimizes the initial compensation amount in multiple dimensions with processing efficiency, deformation control accuracy, and tool wear as optimization objectives.

[0008] As a further preferred embodiment of this technical solution, in step five, the dynamic adjustment compensation strategy and machining parameters adopt an adaptive fuzzy PID algorithm. If the cutting force fluctuation or temperature change exceeds the set threshold, the feed rate, cutting depth or cooling medium flow rate will be automatically adjusted.

[0009] As a further preferred embodiment of this technical solution, in step six, the adaptive machining path adopts a layered cutting and spiral ramp feed strategy, and a collision detection mechanism is embedded in the adaptive machining path generation process to avoid the rigid and weak areas of the titanium alloy thin-walled parts.

[0010] As a further preferred embodiment of this technical solution, in step seven, the deformation detection of the processed titanium alloy thin-walled part is performed using a white light interferometer or a coordinate measuring machine, with a detection accuracy of not less than 0.001 mm.

[0011] To solve the above-mentioned technical problems, another technical solution adopted in this application is: a high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation, the system including: a model building module, a deformation prediction module, a pre-compensation generation module, a monitoring feedback module, a compensation adjustment module, a path generation and execution module, and a detection iteration module; The model building module is configured to construct a multiphysics coupling model based on the constitutive relation of titanium alloy material and milling process parameters. The deformation prediction module is configured to predict the deformation trend of titanium alloy thin-walled parts during milling based on a multiphysics coupling model and generate prediction results. The pre-compensation generation module is configured to calculate the initial compensation amount based on the prediction results, and optimize the initial compensation amount using a multi-objective genetic algorithm to balance processing efficiency and deformation control, thereby generating a pre-compensation instruction set. The monitoring and feedback module is configured to build a real-time monitoring and feedback system, collect dynamic machining data during the milling process, compare it with the prediction results, and determine the deformation deviation. The compensation adjustment module is configured to dynamically adjust the compensation strategy and processing parameters based on the deformation deviation and the pre-compensation instruction set. The path generation and execution module is configured to generate an adaptive machining path and execute milling operations based on the adjusted compensation strategy and machining parameters. The detection iteration module is configured to perform deformation detection on the processed titanium alloy thin-walled part, and feed the detection results back to the multiphysics coupling model for iterative optimization of the multiphysics coupling model.

[0012] As a further preferred embodiment of this technical solution, the model building module includes a structural feature recognition unit and a dynamic coupling weight allocation unit; The structural feature recognition unit is used to identify the structural type of the titanium alloy thin-walled part and extract structural feature parameters; The coupling weight dynamic allocation unit is connected to the structural feature identification unit and is used to dynamically adjust the coupling weights of the cutting force field, temperature field and internal stress field according to the distribution of rigid weak points of different structural types, so that the multi-physics coupling model can adapt to the deformation characteristics of different structures.

[0013] As a further preferred embodiment of this technical solution, the monitoring feedback module includes a multi-sensor unit and a data processing unit; The multi-sensor unit includes a strain sensor, an infrared temperature sensor, and a laser displacement sensor, which are used to collect cutting force fluctuations, temperature field distribution, and surface displacement data during the milling process. The data processing unit is used to receive the data collected by the multi-sensor unit, compare and analyze it with the prediction results generated by the deformation prediction module, and determine the deformation deviation.

[0014] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions: 1. This invention solves the problem of the disconnect between traditional prediction models and actual working conditions by constructing a multi-physics coupling model with dynamic parameter representation, combining structural feature identification and dynamic allocation of coupling weights, and iteratively optimizing the model through post-processing detection results. This enables the model to accurately adapt to the deformation characteristics of thin-walled parts with different structures. 2. This invention, by building a high-response real-time monitoring and feedback system and combining it with a dynamic compensation closed-loop mechanism, achieves uninterrupted linkage between "monitoring and adjustment" during the processing, thus solving the problem of significant lag in traditional compensation methods; 3. This invention optimizes the initial compensation amount through a multi-objective genetic algorithm, adopts an adaptive processing path strategy, and combines dynamic parameter adjustment for structural adaptation, replacing the traditional single compensation strategy. It can flexibly cope with the problem of dynamic stiffness changes during complex curved surfaces and material removal.

[0015] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

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

[0017] Figure 1This is a flowchart illustrating the high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to the present invention. Figure 2 This is a flowchart illustrating the method for constructing a multiphysics coupling model according to the present invention. Figure 3 This is a schematic diagram of the functional modules of the high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation according to the present invention. Detailed Implementation

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Example

[0019] Figure 1 This is a flowchart illustrating a high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation, according to an embodiment of the present invention. It should be noted that if substantially the same result is achieved, the method of this application is not necessarily identical. Figure 1 The process sequence shown is limited. Figure 1-Figure 2 The following is an example of a high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation, comprising the following steps: Step 1: Based on the constitutive relation of titanium alloy material and milling process parameters, construct a multiphysics coupling model; Specifically, firstly, the mechanical and thermophysical parameters of the titanium alloy are obtained, and its characteristics under different temperatures and stresses are determined experimentally. Then, the cutting force field, temperature field, and internal stress field are selected as key physical fields, and the coupling relationship between these fields is defined based on milling process parameters. Next, boundary conditions such as tool-workpiece contact, heat dissipation, and constraints are set. Then, the model accuracy is improved through dynamic parameters (such as the cutting heat generation rate related to cutting power, the thermal conductivity coefficient dynamically adjusted with temperature, and empirical functions characterizing material softening). Finally, the model is calibrated through multiple sets of milling experiments, and the material properties and field coupling coefficients are dynamically updated until the model's deformation trend prediction error is ≤5%. Step 2: Based on the multiphysics coupling model, predict the deformation trend of titanium alloy thin-walled parts during the milling process and generate prediction results; Specifically, firstly, the complete geometric information of the thin-walled part (including wall thickness distribution, cavity features, reinforcing rib dimensions, etc., which are completely consistent with the part to be machined) and the actual milling process parameters (cutting speed, feed rate, tool parameters, cooling method, etc., which are strictly consistent with the actual machining) are accurately imported. Next, the model is adaptively meshed. Fine meshes (0.15-0.2mm element size) are used in the cutting zone and areas with weak rigidity, while coarser meshes (1.5-2mm) are used in the non-machining zone and areas with strong rigidity. The cutting contact zone uses eight-node hexahedral thermo-mechanical coupling elements, and the overall mesh is a mixture of tetrahedral and hexahedral elements, with an element twist rate ≤12%. Then, set the solution parameters, ensuring the time step is such that the tool travel distance is ≤ 1 / 2 of the mesh element size in the cutting zone; solve in a coupled manner according to the "force-heat-stress" sequence; the convergence condition is that the relative value of the energy residual is ≤ 1 × 10⁻⁶. -4 Furthermore, the maximum temperature / stress change rate in the cutting zone for five consecutive cutting steps is ≤1.5%; When running the model, the cutting force field (outputting contact pressure and friction every 0.001s), temperature field (diffusion in high-temperature areas and cooling effects), and internal stress field (stress concentration at corners) are recorded in real time. The material removal is simulated and the deformation accumulation is tracked through the element life and death technology. When extracting deformation data, focus on recording rigid weak areas (such as the three-dimensional displacement of the curvature extreme point, accurate to 0.001mm) and functional reference surfaces (such as the positioning flatness, fitting 100 points to calculate the deviation), and generate deformation-time curves for key points (such as deformation fluctuation per cut and peak value of residual stress release). Finally, the results are integrated to output a visualized deformation cloud map, a force-temperature-deformation correlation curve, and quantitative data (maximum deformation, error range, fitting equation, etc.), and the confidence interval based on the ±5% error threshold is marked to provide a reference for the initial compensation calculation.

[0020] Step 3: Based on the prediction results, calculate the initial compensation amount, and use a multi-objective genetic algorithm to optimize the initial compensation amount, balance processing efficiency and deformation control, and generate a pre-compensation instruction set; Specifically, firstly, the initial compensation amount is calculated based on the predicted deformation data and set according to the mirror relationship (e.g., if the predicted Z-direction warpage at a certain point is 0.12mm, the compensation amount is -0.12mm); the curved area is fitted with a continuous compensation curve through cubic spline interpolation, and the transition is smooth with the curvature; a compensation coefficient K (1.1-1.2) is introduced in the rigid weak area (e.g., the edge of a cavity with a wall thickness of 1mm), and the compensation amount = K × predicted deformation amount to offset the springback; the compensation amount must be within the machine tool adjustment range (X / Y ± 0.001mm, Z ± 0.0005mm), and extreme deformation points (> 0.5mm) are marked as areas requiring secondary processing; Next, a multi-objective genetic algorithm was used for optimization, with machining efficiency (material removal rate Q), deformation control accuracy (RMSE≤0.01mm), and tool wear (cutting force integral) as objectives. Real-number encoding was adopted (the compensation amount of key points is a gene, chromosome length 3n), the initial population was 50, roulette wheel selection was used, the crossover probability was 0.7, and the mutation probability was 0.05. The fitness function was a weighted summation (F=ω1×(1 / Q)+ω2×RMSE+ω3×cutting force integral, the weights are dynamically adjusted according to the process), and iterated for 100 generations to converge to the Pareto optimal solution. In the optimization, contradictions were balanced: for example, large compensation amounts were decomposed into step-by-step compensation (0.2mm→0.15mm+0.05mm), or the feed rate was reduced when the cutting force exceeded the limit (0.12→0.1mm / r), taking into account accuracy, efficiency, and tool life. Finally, a pre-compensation instruction set is generated, presented in G-code format, containing compensation parameters for each axis, key point coordinate correction values, and trigger conditions. After simulated cutting verification, it is ensured that the deformation after compensation is ≤0.01mm and the single-piece machining time is ≤30min, laying the foundation for subsequent adjustments.

[0021] Step 4: Build a real-time monitoring and feedback system to collect dynamic machining data during the milling process and compare it with the prediction results to determine the deformation deviation; Specifically, firstly, a hardware architecture for a real-time monitoring and feedback system needs to be built to ensure accurate capture of dynamic data during the milling process. The core of the system includes a multi-type sensor array, a high-speed data acquisition unit, and a real-time processing module. For sensors, strain sensors (such as high-precision foil strain gauges with a sensitivity of 1.8 mV / V) are selected and attached to the tool holder and workpiece clamping area to collect dynamic changes in cutting force (resolution 0.1 N). An infrared temperature sensor (response time ≤ 2 ms, temperature range -20℃ to 1000℃) is fixed via a side-mounted bracket on the machine tool spindle, with a focused lens. In the cutting zone (50~80mm from the tool tip), the temperature at the cutting point is recorded in real time; a laser displacement sensor (measuring range 0~50mm, linearity ≤±0.002mm) is mounted on the machine tool column, using an oblique angle (45° to the workpiece surface) to avoid interference from cutting fluid reflection, and continuously monitors the displacement changes of key points on the surface of thin-walled parts (such as cavity edges and surface vertices); the sampling frequency of all sensors is uniformly set to 3kHz to ensure that high-frequency fluctuations in cutting force (such as the impact signal of the tool per revolution) and transient changes in temperature (such as the temperature jump during sudden feed) can be captured. The raw data collected by the sensors is transmitted to a high-speed data acquisition card (sampling rate 10MHz, 16-bit AD conversion) via shielded cable, and then sent to the real-time processing module of the industrial control computer (IPC) via EtherCAT real-time Ethernet protocol (transmission delay ≤1ms). The processing module first preprocesses the data: a 500Hz low-pass filter is used to remove high-frequency noise (filtering machine tool vibration interference) from the cutting force signal; the temperature signal is smoothed (the moving average window is set to 5 sampling points) to eliminate the jumps in the infrared signal; and a Kalman filter algorithm (the state equation is based on a rigid body motion model) is used to remove the occasional occlusion effect of cutting chips from the displacement signal. The preprocessed data needs to be converted into a standardized format. Among them, the cutting force data is converted into three-dimensional force components (Fx, Fy, Fz, in N), the temperature data output is the temperature of the center point of the cutting zone (T, in °C), and the displacement data output is the three-dimensional coordinate changes of key points (ΔX, ΔY, ΔZ, in mm), and a precise timestamp is added (synchronized to the pulse signal of the machine tool spindle encoder, with a time accuracy of ±1μs). Subsequently, the system performs a time-series alignment comparison between the dynamically processed data after real-time processing and the prediction results generated in step two; the comparison adopts a combination of "point-to-point" and "regional features" methods; Among them, the point-to-point comparison is for predefined key monitoring points (such as the deformation peak points marked in step two), and calculates the deviation value δ between the measured displacement ΔZactual and the predicted displacement ΔZpredicted at the same time, which is δ = ΔZactual - ΔZpredicted (for example, if the measured ΔZactual is 0.12 mm and the predicted ΔZpredicted is 0.10 mm, then δ = 0.02 mm). Regional feature comparison targets the entire surface or cavity. It uses the least squares method to fit the measured deformed surface to the predicted deformed surface, calculating the root mean square deviation (RMSD) between the two. For example, the RMSD for the cavity region is √[(1 / n)Σ(δi)]. 2 )], where n is the number of sampling points in the region, and δi is the deviation of the i-th point; At the same time, it is necessary to compare the consistency of the trends of cutting force and temperature; calculate the correlation coefficient between the measured cutting force curve and the predicted curve (≥0.85 required). If the correlation coefficient <0.85, it indicates that there is a systematic bias in the model's prediction of the force field; for temperature comparison, focus on the deviation of the highest temperature in the cutting zone (≤30℃ required). If it exceeds this threshold, it is determined that the temperature field prediction is distorted. Finally, the system determines the overall deformation deviation based on the comparison results and sets the deviation grading standard; When the maximum deviation of the key point is δ≤0.005mm and the RMSD of the area is ≤0.008mm, it is judged as "minor deviation" and no emergency adjustment is required; When 0.005mm < δ ≤ 0.01mm or 0.008mm < RMSD ≤ 0.015mm, it is judged as "medium deviation" and marked as an area that needs dynamic adjustment; When δ > 0.01 mm or RMSD > 0.015 mm, it is judged as "significant deviation", triggering an alarm and being handled with priority. All deviation data (including values, locations, and timestamps) are stored in the database in real time and displayed through a visual interface (such as real-time curve comparison charts and deviation heatmaps), providing a quantitative basis for the dynamic compensation adjustment in step five. For example, if the measured deformation of a certain curved surface area is 0.012 mm larger than the predicted value, the system will send the deviation value of 0.012 mm and the corresponding coordinates of that area to the compensation adjustment module as input parameters for real-time correction.

[0022] Step 5: Based on the deformation deviation, dynamically adjust the compensation strategy and processing parameters using the pre-compensation instruction set as a benchmark; Specifically, firstly, the system receives the deformation deviation data output from step four via a real-time data bus, including instantaneous deviation values ​​at key points (e.g., 0.008mm in the X direction and 0.012mm in the Z direction), root mean square deviation of the region (e.g., RMSD = 0.009mm in the cavity region), and deviation trend curve (e.g., the deviation increased linearly from 0.005mm to 0.012mm in the past 5 seconds). This data is written to the shared data area of ​​the system memory in real time for the compensation and adjustment module to call, and the data transmission delay is strictly controlled within 3ms to ensure synchronization with the milling process. Next, the compensation adjustment module initiates the corresponding adjustment logic based on the deviation level, as follows: For minor deviations (such as key point deviations ≤ 0.005mm), only the pre-compensation instruction set is fine-tuned: a correction coefficient (such as 0.95-1.05 times) is added to the original compensation amount. For example, if the original compensation amount for a certain curved surface point is -0.12mm and the measured deviation is 0.003mm (less than 0.005mm), then the corrected compensation amount is -0.12×1.02=-0.1224mm. Minor deviations are offset by fine-tuning the Z-axis coordinate of the toolpath. For moderate deviations (0.005mm < deviation ≤ 0.01mm), an adaptive fuzzy PID algorithm is activated: the algorithm takes the deviation value as input and calculates the real-time correction values ​​of the proportional coefficient (Kp), integral coefficient (Ki), and derivative coefficient (Kd) through a fuzzy rule base (e.g., "increase the adjustment amount if the deviation is large and the growth rate is fast"). For example, when the deviation in the Z direction reaches 0.007mm and increases by 0.002mm per second, Kp increases from 1.2 to 1.5, and Ki increases from 0.3 to 0.4. The PID calculation outputs the compensation correction value (e.g., additional compensation of -0.005mm), and the trigger threshold of the corresponding area in the pre-compensation instruction set is adjusted simultaneously (e.g., the trigger condition of Z=-5mm is advanced to Z=-4.8mm to ensure that the adjustment takes effect earlier). If a significant deviation (deviation > 0.01 mm) is detected or the deviation trend continues to increase (e.g., the deviation growth rate is > 0.003 mm / s for three consecutive sampling periods), the compensation strategy and processing parameters need to be adjusted simultaneously. In terms of compensation strategy, a "segmented progressive compensation" approach is adopted: the original one-time compensation is broken down into multiple step-by-step compensations. For example, the original plan was to compensate -0.15mm in one step, but it was changed to compensate -0.1mm in the first step, and then compensate the remaining -0.05mm after a 2-second interval, so as to avoid stress concentration caused by a large adjustment in one step. Regarding the adjustment of machining parameters, precise measures are taken to address the causes of deviations: if the deviation is caused by excessive fluctuations in cutting force (e.g., the force in the Fz direction suddenly increases from 2000N to 2500N, with a fluctuation exceeding 20%), the feed rate is reduced from 0.12mm / r to 0.10mm / r, and the depth of cut is reduced (from 0.5mm to 0.4mm) to stabilize the cutting force by reducing the material removal rate; if the deviation is caused by excessively high temperature (the temperature in the cutting zone reaches 600℃, exceeding the preset threshold of 550℃), the flow rate of the cooling medium is increased (from 3L / h to 5L / h), and the spray angle is adjusted (from 30° to 45° to enhance the cooling effect in the cutting zone) to suppress thermal deformation by reducing the temperature gradient; During the adjustment process, the system records the compensation strategy change log in real time (e.g., "10:23:15, segmented compensation enabled in zone 5, initial compensation amount -0.1mm") and the modified values ​​of machining parameters (e.g., "feed speed from 0.12→0.10mm / r, cooling flow rate 3→5L / h"). It also issues adjustment commands through a real-time communication interface with the machine tool's CNC system (e.g., OPCUA protocol), with a command execution delay of ≤5ms, ensuring that the adjustment takes effect immediately on the milling process. Simultaneously, the system evaluates the adjustment effect every 100ms, comparing the newly collected dynamic data with the predicted results. If the deviation decreases from 0.012mm to 0.007mm (entering the medium deviation range), the current adjustment strategy is maintained. If the deviation does not decrease or even increases, secondary optimization is triggered, calling up alternative parameter combinations (e.g., cutting speed reduced from 150m / min to 120m / min) for readjustment until the deviation stabilizes within 0.01mm. Finally, the compensation and adjustment module will feed back all adjustment data (including deviation values, corrected compensation amounts, and adjusted machining parameters) to the system database in real time. This will provide the latest parameters for subsequent path generation and serve as a basis for model iteration (e.g., recording that "when the cutting speed is 150m / min and the feed is 0.12mm / r, the Z-direction deviation is likely to exceed 0.01mm", for reference when optimizing the model in step seven). This will ensure that the dynamic adjustment forms a closed loop and continuously reduce the deviation between the actual machining and the predicted results.

[0023] Step 6: Based on the adjusted compensation strategy and machining parameters, generate an adaptive machining path and execute the milling operation; Specifically, firstly, based on the compensation strategy adjusted in step five (such as segmented progressive compensation and regional differential compensation) and machining parameters (such as feed rate 0.10 mm / r, depth of cut 0.4 mm, and cooling flow rate 5 L / h), the adaptive machining path generation algorithm is activated. The algorithm first reads the 3D model of the thin-walled part and the geometric features of the remaining area to be machined (such as the uncut cavity depth of 1.2 mm and the surface curvature radius of 50 mm), and determines the layered cutting parameters in combination with the adjusted depth of cut. The remaining machining depth of 1.2 mm is divided into 3 layers, and each layer is cut... 0.4mm (consistent with the adjusted cutting depth), with an overlap of 0.1mm for interlayer transition to avoid delamination marks; the tool entry strategy adopts a helical ramp entry, setting the helical radius to 1 / 2 (5mm) of the tool diameter and the ramp angle to 30°, ensuring that the tool gradually contacts the workpiece from a safe height (Z=5mm), reducing the impact on thin-walled parts—especially in areas with weak rigidity (such as the edge with a wall thickness of 1mm), the tool entry speed is reduced by 30% (from the original 0.12mm / r to 0.084mm / r), and workpiece vibration is avoided by slow entry; During path generation, the algorithm incorporates a collision detection mechanism in real time. Based on the tool radius (10mm) and the workpiece's real-time geometric model (dynamically updated during machining), the AABB (axis-aligned bounding box) collision detection algorithm is used. Every 1mm of path generated, the distance between the tool and the non-machined area of ​​the workpiece (such as machined surfaces or fixtures) is checked. If the distance is ≤0.5mm (safety threshold), the path is automatically offset (X / Y direction fine adjustment ±0.1mm) or the Z-axis is raised (increased by 0.05mm) to ensure that the tool only moves in the area to be machined. At the same time, the path is corrected according to the segmented compensation instructions in the compensation strategy. For example, during the third layer of cutting, the Z-axis coordinate of the corresponding area needs to be superimposed with the first compensation amount of -0.1mm. After the cutting of this layer is completed, after an interval of 2 seconds (consistent with the time interval of segmented compensation), the remaining compensation amount of -0.05mm is superimposed in the subsequent tool compensation path. The step-by-step execution of compensation is achieved through the time segmentation of the path. After the path is generated, it needs to be converted into a G-code instruction set that can be recognized by the machine tool's CNC system. The instructions include the coordinate information of each path segment (e.g., G01X100.2Y80.5Z-3.4F600, where F600 corresponds to the feed rate converted from the adjusted feed speed of 0.10mm / r), the tool feed mode marker (e.g., G03I5.0J0.0K-0.4 indicates helical tool feed), and the compensation trigger point (e.g., G65P1002Z-3.4 indicates that the second compensation segment is activated at Z=-3.4mm). The generated instruction set is first imported into the virtual machine tool. The simulation software is used to perform a full-process simulation of cutting. The simulation software will simulate the movement of the tool along the path and check for overcutting (e.g., the tool cuts into the machined surface by 0.02mm), undercutting (e.g., 0.03mm of uncut material remains at the bottom of the cavity), or collisions (e.g., the distance between the tool and the fixture is 0.3mm, which is less than the safety threshold of 0.5mm). If a problem is found, the path generation algorithm is returned to readjust (e.g., the Z-axis of the path in the collision area is raised by 0.2mm) until the simulation result shows that the path completely fits the area to be machined, and the compensation command takes effect in segments according to the preset time. After confirming the path is correct, the G-code instruction set is sent to the CNC system via the machine tool's real-time communication interface (such as EtherCAT). The system parses the instructions and drives each axis to perform the milling operation. During execution, the machine tool spindle operates according to the adjusted parameters (rotation speed 3000 r / min, corresponding to a cutting speed of 150 m / min), and the cooling system operates synchronously at a flow rate of 5 L / h and a spray angle of 45°. The tool cuts layer by layer along the generated adaptive path. The first layer starts from Z=-0.4mm, cutting into the workpiece in a helical infeed manner, along the curved surface contour. The tool moves at a feed rate of 0.10 mm / r, while the cutting force is monitored in real time by a strain sensor on the tool holder (ensuring Fz ≤ 2000 N, matching the adjusted parameters). During the second layer of cutting, the Z-axis coordinate of the path is automatically superimposed with the initial compensation amount of -0.1 mm, that is, the actual cutting depth extends from Z = -0.4 mm to Z = -0.8 mm, and pauses for 2 seconds between layers to wait for the compensation effect to stabilize. After the third layer of cutting is completed, the tool compensation path is executed, and the remaining compensation amount of -0.05 mm is superimposed to perform fine trimming on local deformation-sensitive areas (such as cavity corners). During execution, the real-time monitoring and feedback system continuously collects tool position (via laser displacement sensor, accuracy ±0.001mm), cutting zone temperature (infrared sensor, real-time display 550℃, below the threshold 600℃) and surface displacement data, comparing it with the theoretical position in the path command every 10ms. If the measured tool position deviates from the theoretical path by ≤0.003mm (within the allowable range), execution continues. If the deviation reaches 0.005mm (approaching the warning threshold), the system automatically sends a deceleration command (feed speed reduced to 0.08mm / r) and simultaneously fine-tunes the subsequent path coordinates (X / Y direction correction ±0.002mm). If the deviation suddenly increases to 0.01mm (exceeding the warning threshold), a pause command is immediately triggered. After checking and confirming that there are no abnormalities (such as no tool wear or workpiece loosening), the path segment is regenerated and execution continues from the deviation point, ensuring that the entire milling operation is strictly performed according to the adaptive path and that the compensation strategy and adjusted parameters are accurately implemented.

[0024] Step 7: Perform deformation detection on the processed titanium alloy thin-walled parts, and feed the detection results back to the multiphysics coupling model for iterative optimization.

[0025] Specifically, firstly, the processed titanium alloy thin-walled parts need to undergo pretreatment before inspection to ensure the accuracy of the inspection data. After the workpiece is removed from the machine tool, compressed air is used to blow away residual cutting chips and cooling media from the surface. Then, anhydrous ethanol is used to wipe the processed surface (especially curved surfaces, cavities, and other characteristic areas) to remove oil or oxide layers and avoid impurities affecting the inspection accuracy. Subsequently, the workpiece is placed stably on the inspection platform and fixed using the same positioning reference as during processing (such as using the three positioning holes on the edge of the workpiece as the reference) with a magnetic chuck or vacuum clamp (the clamping force is controlled at 50-100N to avoid secondary deformation due to excessive clamping force), ensuring that the workpiece does not shift or vibrate during the inspection process. Next, the high-precision inspection equipment is started, and the inspection plan is selected according to the characteristics of the workpiece, as follows: For areas with high requirements for flatness and roughness (such as assembly reference surfaces), a white light interferometer (resolution 0.0001mm, measurement range 50mm×50mm) is used for scanning, with a scanning step size of 0.01mm, to acquire three-dimensional surface topography data, focusing on recording micron-level deformation undulations; For overall dimensions and form and position errors (such as surface profile and cavity position), a coordinate measuring machine (measurement accuracy 0.001mm, detection speed 5mm / s) is used to collect coordinate data according to the preset detection path (covering all key features of the workpiece, such as setting a sampling point every 10mm). This includes the warping of the thin-walled part edge, the actual value of the cavity depth, the curvature deviation of the surface, etc. Each feature is measured three times, and the average value is taken as the final detection result. After the inspection is completed, the data is standardized to form a format comparable to the prediction results in step two. The coordinate data collected by the coordinate measuring machine is imported into the CAD software and compared with the design model to generate a "measured-design" deviation cloud map, marking the actual deformation of each sampling point (e.g., the measured Z-coordinate of a cavity corner is -2.012mm, the design value is -2.000mm, and the deformation is 0.012mm). The three-dimensional topography data from the white light interferometer is filtered using an algorithm (to remove high-frequency noise caused by surface roughness) to extract the macroscopic deformation trend and transform it into a deformation curve along the length of the workpiece (e.g., linear deformation increasing from 0.005mm on the left to 0.015mm on the right). At the same time, the deformation errors of key features are summarized, and the maximum deformation of the rigid weak area (e.g., the edge with a wall thickness of 1mm), the flatness error of the functional reference surface (e.g., 0.008mm), and the surface contour error (e.g., 0.01mm) are calculated to form a complete inspection report. Subsequently, the detection results are fed back to the multiphysics coupling model, initiating the iterative optimization process of the model. The specific process is as follows: First, the actual deformation measured was compared with the model's predicted deformation, and the deviation distribution between the two was calculated. If the actual deformation in a certain area (0.012 mm) was 33% greater than the predicted value (0.009 mm), it was determined that there was a deviation in the model prediction for that area, and the source of the deviation needed to be traced. By analyzing the measured cutting force and temperature data of that area (collected in step four), it was found that the actual cutting force was 15% higher than the model's predicted value. It was speculated that the strain hardening coefficient (B value in the Johnson-Cook model) in the material constitutive model was set too small, resulting in an overestimation of the material's resistance to deformation. To address this issue, the model parameters were dynamically updated, and the B value was adjusted from the original 1200 MPa to 1300 MPa (the optimal adjustment was determined by fitting the error between the measured data and the model output using the least squares method). At the same time, the force-heat conversion coefficient was corrected (increased from 0.7 to 0.75) to match the correlation between the actual cutting force and temperature. After updating the parameters, the milling process of the workpiece is re-simulated using the optimized model, and the new prediction results are compared with the detection data. If the relative error between the newly predicted deformation (0.011 mm) and the actual deformation (0.012 mm) is reduced to 8% (original error 33%), the optimization effect of other feature areas is verified. If there is still an area error exceeding 10% (e.g., 12% error in curved areas), the corresponding field coupling coefficient (e.g., the influence weight of temperature field on internal stress field) is further adjusted, and the iterative process of "parameter adjustment - simulation prediction - error comparison" is repeated. After each iteration, the change log of model parameters is recorded (e.g., "2025 / 7 / 30, B value 1200 → 1300 MPa, force-thermal coefficient 0.7 → 0.75") until the prediction error of the model for all key features is ≤5% (meeting the preset threshold). At this time, the model iteration optimization is determined to be completed, and the optimized parameters are solidified as the new model benchmark for the next batch of milling deformation prediction of titanium alloy thin-walled parts, forming a closed-loop improvement mechanism of "machining-detection-optimization".

[0026] In one embodiment, specifically: the method for constructing a multiphysics coupling model in step one includes the following steps: Step 101: Obtain the mechanical and thermophysical parameters of the titanium alloy material and determine the characterization method of the material under different temperatures and stress states; Specifically, firstly, basic parameters are obtained by combining standard tests with database queries, as follows: Regarding mechanical properties, in accordance with GB / T 228.1-2021 standard for tensile testing of metallic materials, room temperature (20℃) tensile tests were conducted on titanium alloy specimens on a universal testing machine to determine the elastic modulus (approximately 110 GPa for TC4 titanium alloy), Poisson's ratio (approximately 0.34), yield strength (approximately 860 MPa for TC4), ultimate strength (approximately 950 MPa for TC4), and strain hardening index. Simultaneously, high-temperature tensile tests (using a testing machine with a heating furnace) were conducted to determine the mechanical parameters at typical processing temperatures such as 200℃, 400℃, and 600℃ to supplement the performance data at high temperatures. For thermophysical parameters, referring to the GB / T 22588-2008 standard for the determination of thermal conductivity, the thermal conductivity of titanium alloys at different temperatures was determined using the laser flash method (e.g., approximately 7.4 W / (m·K) for TC4 at room temperature and approximately 11 W / (m·K) at 600℃); the specific heat capacity was determined by differential scanning calorimetry (DSC) (e.g., approximately 520 J / (kg·K) for TC4); and according to the GB / T 4339-2008 standard for the determination of thermal expansion coefficient, the linear expansion coefficient from 20℃ to 600℃ was obtained using a thermal dilatometer (e.g., approximately 9.0 × 10⁻⁶ for TC4). -6 / ℃), and integrates the basic thermophysical parameters in the material handbook as a supplement; Then, design variable experiments are conducted to determine how to characterize the material's properties under different temperatures and stress states. The specific process is as follows: Determine the temperature range and stress state boundaries: The temperature range covers the typical range of titanium alloy milling (20℃-800℃, including room temperature and high temperature in the cutting zone); the stress state covers the main types in the milling process, including uniaxial tensile stress (simulating surface tensile deformation), shear stress (simulating chip formation zone), and composite stress (simulating stress superposition at cavity corners). Conduct multi-condition tests: At different temperatures (e.g., 20℃, 300℃, 600℃), apply different stress states to titanium alloy samples (using customized fixtures to achieve tensile, shear, and combined loading), and collect data such as stress-strain curves, yield limit, and plastic strain. Establish a property characterization model: Based on experimental data, constitutive equations are used to fit the relationship between material properties and temperature and stress; for example, the Johnson-Cook constitutive model modified for titanium alloy milling scenarios (enhancing adaptability under high temperature and high strain rate) is selected, in the form of: σ=(A+Bε n )(1+C lnε * (1-T) *m ) Where σ is the flow stress (MPa), which characterizes the material’s ability to resist plastic deformation; ε is the plastic strain (dimensionless), which reflects the degree of plastic deformation of the material; ε is a dimensionless strain rate, defined as the actual strain rate ε'(s) -1 ) and reference strain rate ε0 (taken as 1s) -1 The ratio of ε to ε o =ε' / ε0), used to describe the strain rate hardening effect; T * For dimensionless temperature, it is defined as (T-T0) / (T m -T0), where T is the actual temperature (°C), T0 is the reference room temperature (taken as 20°C), and T m The value is the melting point of titanium alloy (e.g., TC4 is about 1668℃), ranging from 0 (room temperature) to 1 (melting point), used to quantify the temperature softening effect; A, B, n, C, and m are material constants: A is the static yield strength (MPa), B is the strain hardening coefficient (MPa), n is the strain hardening exponent (dimensionless), C is the strain rate sensitivity coefficient (dimensionless), and m is the temperature sensitivity coefficient (dimensionless). All constants are obtained by testing titanium alloys at different temperatures (20℃-800℃) and strain rates (10...). 1 -10 4 s -1 The data were obtained by fitting the tensile / shear test data under the given conditions. This model couples plastic strain (ε) and strain rate (ε) * ) and temperature (T) * The influence of ) is accurately characterized to represent the mechanical response characteristics of titanium alloys during milling (high strain rate, wide temperature range), providing a reliable material constitutive basis for multiphysics coupling models; Finally, the above parameters and constitutive model are substituted into a simplified finite element simulation (such as single-tooth cutting simulation). The material deformation trend obtained from the simulation is compared with the plastic deformation characteristics (such as chip morphology and strain of the machined surface) observed in the actual cutting test. If the deviation is within the allowable range (such as ≤10%), the effectiveness of the parameters and characterization method is confirmed; otherwise, the model constants are corrected through supplementary experiments until the accuracy requirements are met.

[0027] Step 102: Select the key physical fields that affect machining deformation and define the coupling relationship between each physical field; the key physical fields include the cutting force field, temperature field, and internal stress field; Specifically, firstly, the cutting force field, temperature field, and internal stress field are clearly selected as the key physical fields affecting the deformation of thin-walled titanium alloy parts during machining. The reason is that during milling, the cutting force directly causes elastic / plastic deformation of the workpiece, the temperature field generates thermal stress through thermal expansion and contraction, and the internal stress field is the final stress state caused by the combined action of force and heat. The coupling effect of the three is the core cause of the deformation of thin-walled parts. Among them, the cutting force field refers to the force distribution state in the contact area between the tool and the workpiece, including the main cutting force, feed resistance and back force. Its magnitude is related to the cutting parameters (cutting speed, feed rate), tool geometric parameters (rake angle, clearance angle) and material properties (such as flow stress). Its spatial distribution can be characterized by the cutting force model (such as the Kienzle formula). The temperature field refers to the temperature distribution in the cutting zone and the workpiece as a whole. The main heat source is the work done by the cutting force (plastic deformation work and friction work). The temperature gradient will directly affect the mechanical properties of the material (such as a decrease in yield strength). Internal stress field refers to the stress distribution inside a workpiece caused by cutting force, temperature change and material removal, including mechanical stress caused by cutting force and thermal stress caused by temperature gradient. When its accumulation exceeds the material yield limit, it will lead to permanent deformation. Then, the coupling relationship between the three fields is defined as follows: Coupling of cutting force field and temperature field: During the cutting process, the work done by the cutting force (main cutting force × cutting speed + feed resistance × feed speed) is converted into cutting heat through energy conversion. By setting the force-heat conversion coefficient (based on the plastic work conversion efficiency of the material, usually taken as 60%-80% in titanium alloy milling), the energy value of the cutting force field is mapped to the heat source intensity of the temperature field (heat input per unit time), that is, "the greater the cutting force and the higher the cutting speed, the more significant the temperature rise in the cutting zone". Coupling of temperature field and internal stress field: Non-uniform distribution of the temperature field (such as the temperature difference between the local high temperature in the cutting zone and the low temperature of the workpiece matrix) will lead to differences in the thermal expansion of the material, through the coefficient of thermal expansion (approximately 9.0 × 10⁻⁶ for titanium alloys). -6 The thermal strain is calculated by taking the temperature gradient (°C) and then combined with the material's elastic modulus (which decreases as the temperature rises, such as the elastic modulus of TC4 dropping to 80% of that at room temperature at 600°C) to convert it into thermal stress. This is then incorporated into the stress superposition calculation of the internal stress field, meaning that "the greater the temperature gradient, the more significant the contribution of thermal stress to the internal control stress field." Coupling of internal stress field and cutting force field: The accumulation of internal stress field will change the local stiffness of the workpiece (such as the decrease of the elastic modulus of the material in the stress concentration area), which will lead to changes in the contact state between the tool and the workpiece (such as the relative position shift between the workpiece surface and the tool after deformation). Then, the calculated value of the cutting force field will be adjusted by the stiffness-force correction coefficient (based on the fitting relationship between the internal stress amplitude and stiffness attenuation). That is, "the greater the internal stress, the lower the local stiffness of the workpiece, and the cutting force will produce adaptive fluctuations under the same cutting parameters". Direct coupling of cutting force field and internal stress field: The cutting force directly generates mechanical stress on the surface of the workpiece. This stress is compared with the material flow stress (calculated by the modified Johnson-Cook constitutive model). If it exceeds the yield limit, plastic deformation occurs, forming residual stress and incorporating it into the internal stress field. Its distribution is positively correlated with the spatial gradient of the cutting force (such as the force concentration area near the tool tip).

[0028] Step 103: Set the boundary conditions for the multiphysics coupling model; the boundary conditions include the contact conditions between the tool and the workpiece, the heat dissipation conditions, and the constraint conditions. Specifically, firstly, the contact conditions between the tool and the workpiece are set, and the mechanical and thermal properties of the contact interface during the cutting process are defined, including: The contact arc length between the tool rake face and the chip is limited (calculated based on the cutting thickness and the tool rake angle; for example, when the feed rate is 0.1 mm / r, the contact arc length is approximately 0.3-0.5 mm), and the contact width between the flank face and the machined surface is limited (the product of the tangent of the tool clearance angle and the height of the residual cutting area). The rake face and the chip adopt a mixed friction model. The friction coefficient of the sliding zone is calibrated by cutting test (e.g., the friction coefficient of TC4 titanium alloy is 0.6-0.8 when the cutting speed is 100-300m / min). The bonding zone adopts a shear yield model (the critical shear stress of the bonding zone is set to be equal to 0.7 times the material flow stress). Based on Hertzian contact theory, the contact pressure on the flank face is parabolic, and the maximum pressure value is related to the perpendicular component of the cutting force and the contact area (through iterative optimization via finite element simulation). Define the thermal conductivity coefficient of the contact interface (considering the effect of contact pressure, such as the thermal conductivity coefficient increasing from 200 W / (m²) when the pressure increases to 500 MPa). 2 ·K) increased to 400W / (m 2 ·K) represents the heat transfer efficiency between the tool and the workpiece; Next, heat dissipation conditions are set, and the heat transfer paths and efficiency at different interfaces are quantified, as follows: The heat transfer between the workpiece and the fixture was modeled using contact thermal resistance. The heat exchange coefficient between the fixture (e.g., aluminum alloy fixture) and the titanium alloy workpiece was determined by thermal imaging experiments (approximately 150-300 W / (m²)). 2 ·K), and dynamically adjusted according to clamping pressure (for every 100MPa increase in pressure, the heat exchange coefficient increases by 10%-15%). The convective heat dissipation between the workpiece and the air is based on Newton's law of cooling. A convective heat dissipation coefficient is set, taking into account the ambient temperature (20-25℃) and the airflow velocity in the cutting zone (e.g., at a spindle speed of 3000 r / min, the airflow velocity is approximately 5-8 m / s). The convective heat dissipation coefficient is then taken as 20-50 W / (m²). 2 ·K); The heat dissipation of the cutting tool ignores the heat accumulation inside the tool (the thermal conductivity of high-speed steel tools is about 40W / (m·K), and that of carbide tools is about 80-100W / (m·K)). The surface outside the contact area between the tool and the workpiece is set as the adiabatic boundary (because the heat dissipation of the tool has little effect on the temperature field of the workpiece). Finally, constraints are set to simulate the actual fixed state of the workpiece in the fixture, specifically including: Geometric constraints: Limit the displacement freedom of the clamping area (e.g., 5-10mm width at the workpiece edge), with translational constraints in the X and Y directions (allowing micro-displacements of ±0.01mm to simulate clamp elasticity), and the Z direction completely fixed (limiting warping). Rotational freedom is set according to the clamping stiffness (e.g., a stiffness of 10 for rotational constraints around the X-axis). 5 N·m / rad); Elastic constraint treatment: For flexible supports in non-clamping areas (such as auxiliary ejector pins), a spring-damping model is used to simulate the constraint stiffness. The spring coefficient is determined through modal testing (e.g., the stiffness of the ejector pin contact point is approximately 5 × 10⁻⁶). 4 (N / m), to avoid simulation distortion caused by stress concentration due to rigid constraints; Boundary update mechanism: The geometric range of the constraint region is updated in real time as the workpiece material is removed during the milling process; By refining the boundary conditions as described above, the multiphysics coupling model can accurately reflect the contact state, energy transfer, and workpiece constraint characteristics during the milling process, thereby improving the model's prediction accuracy.

[0029] Step 104: The multiphysics coupling model is calibrated using experimental data, and the material property parameters and field coupling coefficients are dynamically updated until the prediction error of the multiphysics coupling model meets the preset threshold. Specifically, firstly, TC4 titanium alloy thin-walled specimens (including curved surfaces / cavities) were selected, and with cutting speed (80-300m / min), feed rate (0.05-0.2mm / r), and depth of cut (0.3-1mm) as variables, a full factorial design was adopted, and each set of parameters was repeated 3 times to obtain calibration data; Then, a multi-physics acquisition system was built, which used a triaxial piezoelectric force meter (10kHz sampling) to collect cutting force, an infrared thermal imager (≥640×512 pixels) to record temperature field, a laser displacement sensor (2kHz) and a coordinate measuring machine to collect deformation data, and simultaneously recorded the process parameters of the CNC system. Next, iterative optimization is used to calibrate the parameters, which include material properties (such as Johnson-Cook model parameters) and field coupling coefficients. The root mean square error (RMSE) between the predicted and measured values ​​is used as the target, and the particle swarm optimization (PSO) algorithm is used for iterative adjustment with an initial step size of 5% (gradually reduced to 1%), prioritizing the updating of high-sensitivity parameters. During the verification, the deformation prediction error must be ≤5%, the cutting zone temperature ≤8%, and the cutting force ≤10%. If the standards are not met, extreme working condition tests are added to finally form a high-precision model.

[0030] The multiphysics coupling model characterizes the machining process through dynamic parameters, including calculating the cutting heat generation rate in a way related to cutting power, making the thermal conductivity change dynamically with temperature, and using functions to characterize the softening effect of titanium alloy materials during machining. Specifically, the multiphysics coupling model accurately characterizes the time-varying properties of titanium alloy milling through the following dynamic parameter settings: The specific process for calculating the cutting heat generation rate using a method related to cutting power is as follows: Cutting heat generation rate (the amount of cutting heat generated per unit time, expressed in W / m²) 3 It is directly related to the energy input during the milling process, and the calculation formula is: q=η·P / V_c Where P is the cutting power (W), determined by the product of the main cutting force F_c (N) and the cutting speed v_c (m / s) (P=F_c·v_c); V_c is the volume of the cutting zone (m³). 3 The calculation is based on the cutting width, cutting depth, and tool contact arc length; η is the energy conversion coefficient (0.6-0.8 in titanium alloy milling, because 60%-80% of the plastic deformation work and friction work are converted into heat energy); This method enables the cutting heat generation rate to change dynamically with real-time cutting force and cutting speed (e.g., when v_c increases during high-speed cutting, P increases and q increases simultaneously), avoiding the temperature field prediction deviation caused by the traditional constant heat source assumption. The specific process by which the thermal conductivity coefficient dynamically changes with temperature is as follows: The thermal conductivity (k) of titanium alloys is not a constant value; it increases nonlinearly with increasing temperature. Its dynamic characteristics can be characterized by a polynomial function. k(T) = k0 + a·T + b·T 2 Where T is temperature (°C); k0 is the thermal conductivity coefficient at room temperature (20°C) (e.g., k0 = 7.4 W / (m·K) for TC4 titanium alloy); a and b are fitting coefficients (determined through high-temperature thermal conductivity tests; in TC4, a = 0.012 W / (m·K·°C) and b = 5 × 10⁻⁶). -6 W / (m·K·℃ 2 )); For example, when the temperature in the cutting zone rises to 600℃, k(600)≈11W / (m·K), which is about 49% higher than that at room temperature. The model can dynamically adjust the energy distribution of the heat conduction path accordingly, which is more in line with the actual heat transfer law. The softening effect of titanium alloy materials during processing was characterized using a function, and the specific process is as follows: By combining the modified Johnson-Cook constitutive model, the material softening effect is quantified through temperature-related terms: Softening factor S(T) = 1 - T m Where T is the dimensionless temperature ((T-T0) / (T) m -T0), m is the temperature sensitivity coefficient (0.6-0.8 for TC4 titanium alloy). As temperature increases, T increases and S(T) decreases, leading to a material flow stress σ=(A+Bε) n The decrease of )(1+C lnε)·S(T) directly reflects the softening characteristic of "decreased material strength and increased plasticity at high temperatures"; for example, when the temperature in the cutting zone rises from 20℃ to 600℃, T * ≈0.36, S(T)≈1-0.36 07 The flow stress is approximately 0.78, which is about 22% lower than that at room temperature, accurately reflecting the degradation of the mechanical properties of titanium alloys during high-temperature milling.

[0031] In one embodiment, specifically: in step four, the real-time monitoring and feedback system includes a sensor for acquiring dynamic processing data, wherein the sensor has a sampling frequency ≥ 2 kHz, a measurement accuracy ≤ ± 0.005 mm, and a data processing delay ≤ 5 ms; Specifically, the sensor array of the real-time monitoring and feedback system must meet the following requirements: sampling frequency ≥ 2kHz, measurement accuracy ≤ ±0.005mm, and data processing delay ≤ 5ms. This includes: Laser displacement sensor: mounted on the machine tool column, with a 45° angled beam to avoid interference, measuring range 0-50mm, sampling frequency ≥2kHz, overall accuracy ±0.003mm, monitoring surface displacement; Strain sensor: Full-bridge foil strain gauge attached to the tool holder and fixture, three-dimensional measurement, sensitivity 2.0mV / V, sampling frequency 3kHz, temperature compensation, force error ≤±2N (corresponding to deformation error ±0.002mm), monitoring cutting force; Infrared temperature sensor: mounted on the spindle side, focusing on the cutting area (focal length 50-80mm), response time ≤1ms, sampling frequency 2.5kHz, filtering to resist interference, temperature measurement accuracy ±3℃ (corresponding to thermal deformation error ≤±0.001mm), monitoring temperature.

[0032] In one embodiment, specifically: in step three, the multi-objective genetic algorithm optimizes the initial compensation amount in multiple dimensions with processing efficiency, deformation control accuracy and tool wear as optimization objectives; Specifically, the optimization of the initial compensation amount by the multi-objective genetic algorithm is a multi-dimensional balancing process that takes into account machining efficiency, deformation control accuracy, and tool wear. Its core lies in selecting the optimal solution from a large number of initial compensation schemes by simulating the "selection-crossover-mutation" mechanism of biological evolution. The specific process is as follows: First, clarify the quantification methods for optimization variables and objectives. Optimization variables are key parameters of the initial compensation amount, including the three-dimensional compensation coordinates (ΔX, ΔY, ΔZ) of each characteristic region of the thin-walled part (such as surface vertices, cavity edges, and weak rigid areas), the time node for compensation execution (such as the compensation trigger layer depth in layered cutting), and the step-by-step ratio of the compensation amount (such as the ratio of the first and second compensations in segmented compensation). These variables are encoded into chromosomes using real numbers (e.g., a chromosome can be represented as [ΔX1, ΔY1, ΔZ1, t1, k1, ΔX2, ΔY2, ΔZ2, t2, k2…], where t is the trigger time and k is the step-by-step ratio). The population size is set to 80 to ensure that a sufficient number of compensation schemes are covered. The quantification of the three optimization objectives must be directly related to actual processing requirements, as detailed below: Processing efficiency: The material removal rate Q (mm) per unit time. 3 / s) Quantization, Q = cutting width × cutting depth × feed rate / 60, the goal is to maximize Q (e.g., if the single-piece machining time is required to be ≤30min); if the initial compensation amount is too large, the tool path will be repeated (e.g., multiple tool replenishment), which will reduce Q. Therefore, the algorithm needs to find a balance between the compensation amount and the path length. Deformation control accuracy is measured by the root mean square error (RMSE) (mm) between the actual deformation after compensation and the design value. RMSE = √[(1 / n)Σ(actual deformation - design value)] 2 The goal is to minimize RMSE (e.g., ≤0.01mm); insufficient initial compensation will increase RMSE, while excessive compensation will introduce new deformations, so precise control is required. Tool wear: characterized by the cumulative work W (J) of the cutting force, W=∫(Fx 2 +Fy 2 +Fz 2 )^(1 / 2)·vdt (v is the feed rate, t is the machining time), the goal is to minimize W (e.g., ≤5000J is required); unreasonable compensation leads to increased cutting force fluctuations (e.g., the impact caused by frequent tool path adjustments), which in turn increases W, and the cutting force needs to be stabilized by optimizing the compensation. Next, a fitness function is constructed to achieve multi-objective fusion; the three objectives are normalized and integrated into a single-objective fitness function using a weighted coefficient method. F=ω1×(1 / Q)+ω2×RMSE+ω3×(W / W0); Where ω1+ω2+ω3=1 (the weights are dynamically adjusted according to the process priority, such as ω2=0.5 when high precision is required, and ω1=0.4 when high efficiency is required), W0 is the allowable cumulative power of the tool (such as 5000J); the smaller the fitness value, the better the solution, and the algorithm selects high-quality individuals through the fitness value; During the algorithm iteration process, population evolution is achieved through selection, crossover, and mutation operations, as detailed below: Selection: A tournament selection method is used to randomly select 5 individuals from the population, and the 2 individuals with the lowest fitness are selected to enter the next generation to ensure the preservation of superior genes; Crossover: For the selected parent chromosomes, some genes are exchanged at a random crossover point (such as the 3rd compensation parameter). The crossover probability is set to 0.8. For example, after the crossover of parent A's [ΔX1=0.12, ΔY1=0.08] and parent B's [ΔX1=0.10, ΔY1=0.10], the offspring will have [ΔX1=0.12, ΔY1=0.10], thus increasing population diversity. Mutation: Randomly perturb 10% of the genes in the chromosome (e.g., ΔX1 mutates within ±0.005 mm), with the mutation probability set to 0.05, to avoid the population getting trapped in local optima; During the iteration process, it is necessary to dynamically balance the conflict between the three objectives. For example, in an initial compensation scheme, a large compensation amount (ΔZ=-0.15mm) can reduce RMSE to 0.008mm (good accuracy), but it leads to a 20% increase in toolpath, a 15% decrease in Q (poor efficiency), and an increase in cutting force fluctuations that causes W to increase by 10% (high tool wear). The algorithm will generate a new scheme through crossover mutation, splitting the compensation amount into ΔZ1=-0.1mm (first compensation) and ΔZ2=-0.05mm (secondary compensation). The path length is only 5%, Q is reduced by 5%, and the cutting force fluctuations are reduced, causing W to decrease by 8%. Finally, RMSE=0.009mm (still meeting accuracy), and the fitness value is better. When the algorithm stops iterating to the 50th generation (the preset maximum number of generations) or when the fitness value change rate is ≤1% for 10 consecutive generations, it extracts the Pareto optimal solution (i.e., the solution that cannot improve one objective without harming the other) from the final population. Based on the actual process requirements (such as prioritizing accuracy), one of the solutions is selected as the optimized compensation amount. This compensation amount must meet the following requirements: machining efficiency ≥ 90% of the original solution (Q ≥ original Q × 0.9), deformation control accuracy RMSE ≤ 0.01mm, and tool wear W ≤ W0 × 1.1. Finally, it is converted into specific parameters in the pre-compensation instruction set to achieve multi-dimensional optimization of the initial compensation amount.

[0033] In one embodiment, specifically: in step five, the dynamic adjustment of compensation strategy and machining parameters adopts an adaptive fuzzy PID algorithm. If the cutting force fluctuation or temperature change exceeds the set threshold, the feed rate, cutting depth or cooling medium flow rate will be automatically adjusted. Specifically, the dynamic adjustment of compensation strategies and processing parameters adopts an adaptive fuzzy PID algorithm, the core of which is "fuzzy inference decision-making + PID precise control": The inputs are deformation deviation and rate of change, cutting force fluctuation, and temperature change; the outputs are PID parameter correction values ​​and machining parameter adjustment commands such as feed rate, depth of cut, and cooling flow rate. The fuzzy controller divides the input into 7 fuzzy subsets, infers based on the rule base, and generates precise instruction correction parameters after defuzzification; When the cutting force fluctuates by more than 20%, the feed rate should be reduced first, and the depth of cut should be reduced if necessary; when the temperature change exceeds 50°C, the cooling flow rate should be increased first, and the cutting speed should be reduced if necessary. A closed-loop cycle is completed every 10ms, which quickly responds to dynamic deviations and achieves deformation control and processing stability optimization.

[0034] In one embodiment, specifically: in step six, the adaptive machining path adopts a layered cutting and spiral ramp feed strategy, and a collision detection mechanism is embedded in the adaptive machining path generation process to avoid the rigid and weak areas of the titanium alloy thin-walled part; Specifically, adaptive processing path generation focuses on rigid features and processing stability, and achieves protective processing of weak areas through a collaborative strategy; Among them, the layered cutting is divided into a weak rigid area (≤1.5mm) and a strong rigid area (≥3mm) according to the wall thickness. The former has a layer thickness of 0.2-0.3mm / layer (not exceeding 50% of the wall thickness), and the latter has a layer thickness of 0.5mm / layer. A 0.1mm overlap is set between layers to eliminate traces. Helical ramp infeed replaces vertical infeed, with a helical radius of 1 / 3 to 1 / 2 of the tool diameter and an angle of 15° to 30°. The infeed speed is reduced by 30%, which reduces impact and controls the peak cutting force. The collision detection adopts "pre-detection + real-time correction". The AABB algorithm is used to detect once every 5mm. When the distance is ≤0.3mm, the path is adjusted. For weak areas, the safety distance is increased to 0.5mm. When approaching, the speed is reduced and the cutting depth is reduced. When there is a risk, the process is paused and a backup path is used to ensure that the cutting force is ≤1000N and the deformation is ≤0.005mm, thus balancing efficiency and accuracy.

[0035] In one embodiment, specifically: in step seven, the deformation detection of the processed titanium alloy thin-walled part is performed using a white light interferometer or a coordinate measuring machine, with a detection accuracy of not less than 0.001 mm; Among them, the white light interferometer is mainly used to detect the micro-deformation and local morphological deviation of the surface of thin-walled parts, and is especially suitable for areas with high flatness requirements (such as assembly reference surfaces with Ra≤0.8μm) or high surface profile accuracy (such as curvature radius tolerance ±0.005mm). Coordinate measuring machines (CMMs) are primarily used to detect the overall macroscopic deformation of thin-walled parts, such as edge warping, cavity position, and overall dimensional deviation. They are particularly suitable for detecting thin-walled parts with complex three-dimensional features (such as mesh cavities and variable curvature surfaces).

[0036] Figure 3 This is a functional module diagram of a high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation, according to an embodiment of this application. Figure 3 As shown, a high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation is presented. The system includes: a model building module, a deformation prediction module, a pre-compensation generation module, a monitoring and feedback module, a compensation adjustment module, a path generation and execution module, and a detection and iteration module. The model building module is configured to construct a multiphysics coupling model based on the constitutive relation of titanium alloy material and milling process parameters. The deformation prediction module is configured to predict the deformation trend of thin-walled titanium alloy parts during milling based on a multiphysics coupling model and generate prediction results. The pre-compensation generation module is configured to calculate the initial compensation amount based on the prediction results, and use a multi-objective genetic algorithm to optimize the initial compensation amount, balance processing efficiency and deformation control, and generate a pre-compensation instruction set. The monitoring and feedback module is configured to build a real-time monitoring and feedback system, collect dynamic machining data during the milling process, compare it with the prediction results, and determine the deformation deviation. The compensation adjustment module is configured to dynamically adjust the compensation strategy and processing parameters based on the deformation deviation and the pre-compensation instruction set. The path generation and execution module is configured to generate an adaptive machining path and execute milling operations based on the adjusted compensation strategy and machining parameters. The detection iteration module is configured to perform deformation detection on the processed titanium alloy thin-walled parts, and feed the detection results back to the multiphysics coupling model for iterative optimization of the multiphysics coupling model.

[0037] In one embodiment, the model building module specifically includes a structural feature recognition unit and a dynamic weight allocation unit. The structural feature recognition unit is used to identify the structural type of thin-walled titanium alloy parts and extract structural feature parameters; The coupling weight dynamic allocation unit is connected to the structural feature identification unit. It is used to dynamically adjust the coupling weights of the cutting force field, temperature field and internal stress field according to the distribution of rigid weak points of different structural types, so that the multi-physics coupling model can adapt to the deformation characteristics of different structures.

[0038] In one embodiment, the monitoring feedback module specifically includes a multi-sensor unit and a data processing unit. The multi-sensor unit includes a strain sensor, an infrared temperature sensor, and a laser displacement sensor, which are used to collect data on cutting force fluctuations, temperature field distribution, and surface displacement during the milling process. The data processing unit receives the data collected by the multi-sensor unit, compares and analyzes it with the prediction results generated by the deformation prediction module, and determines the deformation deviation.

[0039] In summary, the high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation provided in this embodiment of the invention forms a closed-loop system of "modeling-prediction-compensation-monitoring-adjustment-execution-iteration" through the collaboration of various modules, thereby achieving high-precision milling and precise deformation control of thin-walled titanium alloy parts.

[0040] For further details regarding the implementation techniques of each module in the high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation in the above embodiments, please refer to the description in the high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation in the above embodiments, which will not be repeated here.

[0041] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For system-type embodiments, since they are basically similar to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation, characterized in that, Includes the following steps: A multiphysics coupling model is constructed based on the constitutive relation of titanium alloy materials and milling process parameters. Based on a multiphysics coupling model, the deformation trend of thin-walled titanium alloy parts during milling is predicted, and the prediction results are generated. Based on the prediction results, the initial compensation amount is calculated, and a multi-objective genetic algorithm is used to optimize the initial compensation amount, balance processing efficiency and deformation control, and generate a pre-compensation instruction set. A real-time monitoring and feedback system was established to collect dynamic machining data during the milling process and compare it with the prediction results to determine the deformation deviation. Based on the deformation deviation, the compensation strategy and processing parameters are dynamically adjusted using the pre-compensation instruction set as a reference. Based on the adjusted compensation strategy and machining parameters, an adaptive machining path is generated, and the milling operation is executed. Deformation detection is performed on the processed titanium alloy thin-walled parts, and the detection results are fed back to the multiphysics coupling model for iterative optimization.

2. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The method for constructing a multiphysics coupling model includes the following steps: To obtain the mechanical and thermophysical parameters of titanium alloy materials and determine the characterization methods of the materials under different temperatures and stress states; Key physical fields affecting machining deformation are selected, and the coupling relationships between these fields are defined. These key physical fields include the cutting force field, temperature field, and internal stress field. Set the boundary conditions for the multiphysics coupling model; the boundary conditions include the contact conditions between the tool and the workpiece, the heat dissipation conditions, and the constraint conditions. The multiphysics coupling model is calibrated using experimental data, and the material property parameters and field coupling coefficients are dynamically updated until the prediction error of the multiphysics coupling model meets the preset threshold. The multiphysics coupling model characterizes the machining process through dynamic parameters, including calculating the cutting heat generation rate in a way related to cutting power, making the thermal conductivity coefficient change dynamically with temperature, and using functions to characterize the softening effect of titanium alloy materials during machining.

3. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The real-time monitoring and feedback system includes sensors for collecting dynamic processing data. The sampling frequency of the sensors is ≥2kHz, the measurement accuracy is ≤±0.005mm, and the data processing delay is ≤5ms.

4. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The multi-objective genetic algorithm optimizes the initial compensation amount in multiple dimensions, with machining efficiency, deformation control accuracy, and tool wear as optimization objectives.

5. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The dynamic adjustment compensation strategy and machining parameters adopt an adaptive fuzzy PID algorithm. If the cutting force fluctuation or temperature change exceeds the set threshold, the feed rate, cutting depth or cooling medium flow rate will be automatically adjusted.

6. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The adaptive machining path employs a layered cutting and helical ramp feed strategy, and a collision detection mechanism is embedded in the adaptive machining path generation process to avoid the rigid and weak areas of the titanium alloy thin-walled parts.

7. The high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to claim 1, characterized in that: The deformation detection of the processed titanium alloy thin-walled parts is performed using a white light interferometer or a coordinate measuring machine, with a detection accuracy of not less than 0.001 mm.

8. A high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation, applied to the high-precision milling method for thin-walled titanium alloy parts based on dynamic compensation according to any one of claims 1-7, characterized in that, The system includes: a model building module, a deformation prediction module, a pre-compensation generation module, a monitoring and feedback module, a compensation adjustment module, a path generation and execution module, and a detection and iteration module; The model building module is configured to construct a multiphysics coupling model based on the constitutive relation of titanium alloy material and milling process parameters. The deformation prediction module is configured to predict the deformation trend of titanium alloy thin-walled parts during milling based on a multiphysics coupling model and generate prediction results. The pre-compensation generation module is configured to calculate the initial compensation amount based on the prediction results, and optimize the initial compensation amount using a multi-objective genetic algorithm to balance processing efficiency and deformation control, thereby generating a pre-compensation instruction set. The monitoring and feedback module is configured to build a real-time monitoring and feedback system, collect dynamic machining data during the milling process, compare it with the prediction results, and determine the deformation deviation. The compensation adjustment module is configured to dynamically adjust the compensation strategy and processing parameters based on the deformation deviation and the pre-compensation instruction set. The path generation and execution module is configured to generate an adaptive machining path and execute milling operations based on the adjusted compensation strategy and machining parameters. The detection iteration module is configured to perform deformation detection on the processed titanium alloy thin-walled part, and feed the detection results back to the multiphysics coupling model for iterative optimization of the multiphysics coupling model.

9. The high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation according to claim 8, characterized in that: The model building module includes a structural feature recognition unit and a dynamic weight allocation unit. The structural feature recognition unit is used to identify the structural type of the titanium alloy thin-walled part and extract structural feature parameters; The coupling weight dynamic allocation unit is connected to the structural feature identification unit and is used to dynamically adjust the coupling weights of the cutting force field, temperature field and internal stress field according to the distribution of rigid weak points of different structural types, so that the multi-physics coupling model can adapt to the deformation characteristics of different structures.

10. The high-precision milling system for thin-walled titanium alloy parts based on dynamic compensation according to claim 8, characterized in that: The monitoring feedback module includes a multi-sensor unit and a data processing unit; The multi-sensor unit includes a strain sensor, an infrared temperature sensor, and a laser displacement sensor, which are used to collect cutting force fluctuations, temperature field distribution, and surface displacement data during the milling process. The data processing unit is used to receive the data collected by the multi-sensor unit, compare and analyze it with the prediction results generated by the deformation prediction module, and determine the deformation deviation.

Images