A dynamic modeling method for machining deformation simulation of large and complex structural parts

Abstract

The present invention provides a dynamic modeling method for simulating deformation during machining of large and complex structural parts, which relates to the field of mechanical machining and manufacturing technology, and includes: constructing an intelligent model and performing dynamic optimization, including establishing a three-dimensional model, marking easily deformed areas, embedding tolerance information, and adaptive meshing; performing multi-source data-driven physical modeling, collecting data and correcting material properties, and constructing an intelligent constitutive model; carrying out multi-scale coupled dynamic simulation to achieve macro-micro model coupling and dynamic prediction of machining forces; performing digital twin verification and closed-loop control, quantifying deformation deviations, and optimizing machining parameters. The present invention can more accurately simulate the mechanical behavior of machining of large and complex structural parts and improve simulation accuracy by accurately marking easily deformed areas, embedding tolerance information, adaptive meshing, and driving multi-source data to correct material properties and construct an intelligent constitutive model.

Description

Technical Field

[0001] The present invention relates to the technical field of mechanical processing and manufacturing, and in particular to a dynamic modeling method for simulating deformation during processing of large and complex structural parts. Background Art

[0002] In the machining and manufacturing industry, large, complex structural components are widely used in key sectors such as aerospace, shipbuilding, and the automotive industry. These components typically feature complex shapes and high precision requirements. Their machining process is susceptible to numerous factors, including material properties, machining parameters, cutting forces, and thermal deformation, which can lead to machining deformation. Machining deformation not only reduces product dimensional accuracy and surface quality, impacting performance and reliability, but can also lead to product scrapping, increasing production costs and cycle time.

[0003] Traditional machining deformation simulation methods mainly rely on finite element analysis (FEA) for static or quasi-static simulation, which makes it difficult to accurately predict deformation behavior during dynamic machining. Traditional methods have obvious deficiencies in model construction, data processing, multi-scale coupling, real-time control, and data management. For example, it is difficult to accurately mark easily deformed areas during model construction, and the meshing lacks adaptive capabilities, resulting in low simulation accuracy. Material property corrections are inaccurate during data processing, multi-source data fusion is insufficient, and the constitutive model cannot reflect the actual mechanical behavior of the material. Multi-scale coupling is difficult, making it difficult to achieve accurate coordination of macro and micro models. Real-time monitoring and closed-loop control are lacking, making it impossible to dynamically adjust machining parameters based on deformation data. Data management is decentralized and insecure, lacking a unified data channel and a reliable evidence storage mechanism. Therefore, it is necessary to provide a dynamic modeling method for machining deformation simulation of large and complex structural parts to solve the above technical problems. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic modeling method for simulating the deformation of large and complex structural parts during machining. By constructing an intelligent model, multi-source data-driven physical modeling, multi-scale coupled dynamic simulation, digital twin verification and closed-loop control, and full-process data management, it is possible to accurately predict and effectively control the deformation of large and complex structural parts during machining, improve machining accuracy and production efficiency, and reduce production costs.

[0005] To solve the above technical problems, the present invention provides a dynamic modeling method for simulating deformation during machining of large and complex structural parts, comprising the following steps:

[0006] Step 1: Build an intelligent model and perform dynamic optimization, including establishing a 3D model, marking easily deformed areas, embedding tolerance information, and adaptive meshing;

[0007] Step 2: Conduct multi-source data-driven physical modeling to collect data, modify material properties, and build an intelligent constitutive model;

[0008] Step 3: Conduct multi-scale coupled dynamic simulation to achieve macro-micro model coupling and dynamic prediction of machining forces;

[0009] Step 4: Perform digital twin verification and closed-loop control to quantify deformation deviations and optimize processing parameters;

[0010] Step five: Implement full-process data management, establish real-time data channels and store data on the blockchain.

[0011] Preferably, in the construction of the intelligent model and the dynamic optimization, specifically:

[0012] Obtain design drawings of large and complex structural parts, use 3D modeling software to create a 3D geometric model based on the design drawings; calculate the wall thickness and curvature radius of each surface in the 3D geometric model;

[0013] A feature recognition algorithm is used to mark easily deformed areas on each surface of a 3D geometric model, and marking conditions for these areas are set. The marking conditions include surface areas where both the wall thickness and the curvature radius are less than a corresponding preset threshold. The Kriging interpolation method is used to embed the wall thickness fluctuation tolerance information.

[0014] The adaptive grid division adopts octree coarse division and sets voxel edges. The grid is encrypted by combining the Zienkiewicz-Zhu error estimation formula and the particle swarm optimization algorithm. The particle swarm velocity update formula in the particle swarm optimization algorithm is:

[0015] ;

[0016] in represents the velocity of the i-th particle at the k+1-th iteration, w is the inertia weight, and its value range is usually [0.4, 0.9]. Expressed as a learning factor, it adjusts the step size of the particle flying towards its own historical optimal position and the global optimal position, Represents a random number uniformly distributed in the interval [0, 1]; Represented as the historical optimal position of the i-th particle, represents the global optimal position, represents the position of the i-th particle at the k-th iteration.

[0017] Preferably, when performing multi-source data-driven physical modeling, a dynamometer and a laser displacement sensor are used to collect cutting data and displacement data; the dynamometer is Kistler9257B with a sampling rate of 10kHz, and the laser displacement sensor is Keyence with an accuracy of ±1μm; the extended Kalman filter algorithm is used to fuse data to correct material properties, and the state transfer matrix of the extended Kalman filter algorithm is for:

[0018] ;in Indicates the sampling time of collected data;

[0019] A nonlinear constitutive model based on LSTM is constructed, with cutting data and temperature as input and stress-strain curve as output; the cutting data includes cutting speed and feed rate;

[0020] The training process of the LSTM-based nonlinear constitutive model adopts the mean square error loss function, which is formulated as follows: ; Where L represents the loss function value, u represents the number of training samples, and b represents the number of training samples. represents the true stress-strain value of the bth sample, Predict the stress-strain value for the model of the bth sample and optimize the model parameters by minimizing the loss function.

[0021] Preferably, in the multi-scale coupled dynamic simulation, the macro-micro model includes a macro model and a micro crystal plasticity model, the macro model adopts Abaqus / Explicit, and the micro crystal plasticity model adopts VPSC, and the coupling formula is: , ;in, represents the macro stress, represents micro stress, V is the volume of the material, represents the macroscopic strain, represents the microscopic strain, represents local strain;

[0022] An alternating iterative method is used to transfer stress and strain, and the convergence accuracy is preset. An LSTM model based on the attention mechanism sets the time interval in advance to predict cutting force fluctuations.

[0023] The calculation formula for predicting cutting force fluctuations using the LSTM model based on the attention mechanism is: ;in represents the predicted cutting force, represents the attention weight of the s-th LSTM hidden layer output on the predicted cutting force, , hs represents the output of the LSTM hidden layer.

[0024] Preferably, in the digital twin verification and closed-loop control, industrial CT scanning is used to obtain point cloud data, and the ICP algorithm is used to align the point cloud data with the simulation model. The IPC algorithm formula is: , ; Where R represents the rotation matrix, t represents the translation vector, represents the jth point in the point cloud data, is the corresponding point in the simulation model, represents the centroid of the point cloud data, represents the centroid of the corresponding point set of the simulation model;

[0025] Then identify the out-of-tolerance area based on the consistency of the normal vector;

[0026] Design a multi-objective reward function: ; Among them, r represents the reward value, Indicates deformation deviation, Indicates the processing time deviation, Represents the energy consumption deviation; the processing parameters are adjusted online through the Actor-Critic network.

[0027] Preferably, in the Actor-Critic network, the Actor-Critic network includes an Actor network and a Critic network; the output of the Actor network is the adjustment amount of the processing parameters, the input of the Critic network is the current state and the output of the Actor network, and the output is a value evaluation of the current state and action. Through continuous iterative training, the Actor network can output the optimal processing parameter adjustment amount according to different states; the current state includes deformation deviation, processing time deviation, and energy consumption deviation.

[0028] Preferably, in the multi-scale coupled dynamic simulation, the crystal orientation distribution in the microscopic model is updated every time a set number of iterations are performed.

[0029] Preferably, the Zienkiewicz-Zhu error estimation formula is: ;in represents the overall error estimate, represents the number of grid cells, Represents the error estimate of the e-th grid cell; when the overall error estimate is greater than its preset overall error allowable threshold, the particle swarm optimization algorithm is triggered to perform grid refinement.

[0030] Compared with related technologies, the dynamic modeling method for machining deformation simulation of large and complex structural parts provided by the present invention has the following beneficial effects:

[0031] 1. This invention can more accurately simulate the mechanical behavior of large and complex structural parts during machining, thereby improving simulation accuracy, by precisely marking easily deformed areas, embedding tolerance information, adaptively meshing, and driving the modification of material properties through multi-source data and constructing an intelligent constitutive model.

[0032] 2. The present invention realizes the effective combination of macro and micro models through multi-scale coupled dynamic simulation, considers the influence of microstructure on macro deformation, predicts cutting force fluctuations in advance, and provides comprehensive information for machining process control; and adopts digital twin verification and closed-loop control for real-time monitoring and feedback, quantifies deformation deviations and optimizes machining parameters, optimizes the machining process, and improves machining accuracy and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flowchart of a dynamic modeling method for simulating deformation during machining of large and complex structural parts provided by the present invention. DETAILED DESCRIPTION

[0034] 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.

[0035] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the appended claims, the singular forms "group," "class," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0036] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining."

[0037] Please refer to Figure 1 A dynamic modeling method for simulating deformation during machining of large and complex structural parts includes the following steps:

[0038] Step 1: Build an intelligent model and perform dynamic optimization, including establishing a 3D model, marking easily deformed areas, embedding tolerance information, and adaptive meshing;

[0039] Step 2: Conduct multi-source data-driven physical modeling to collect data, modify material properties, and build an intelligent constitutive model;

[0040] Step 3: Conduct multi-scale coupled dynamic simulation to achieve macro-micro model coupling and dynamic prediction of machining forces;

[0041] Step 4: Perform digital twin verification and closed-loop control to quantify deformation deviations and optimize processing parameters;

[0042] Step 5: Implement full-process data management, establish real-time data channels and store data on the blockchain. Specifically:

[0043] A real-time data channel is established based on the OPC UA protocol to enable data interaction between CAD, CAE, CNC and other systems; OPC UA servers and clients are installed in each system, and corresponding communication parameters are configured to ensure data interaction delay; Hyperledger Fabric blockchain technology is used to store key process parameters (such as cutting speed, feed rate, spindle speed, and processing time); after encrypting the key process parameters, they are packaged into transaction data and sent to the blockchain network; the nodes in the blockchain network verify and record the transaction data to form a tamper-proof distributed ledger.

[0044] It should be noted that the various steps complement each other to achieve accurate prediction and effective control of machining deformation of large and complex structural parts, as well as comprehensive management and optimization of the machining process, which plays an important role in improving machining quality, reducing costs, and enhancing data security and traceability.

[0045] Build intelligent models and dynamically optimize: A 3D model provides a foundational framework for subsequent analysis, accurately marking areas prone to deformation so that subsequent analysis focuses on these key areas. Embedding tolerance information allows the model to better reflect actual machining conditions. Adaptive meshing automatically adjusts mesh density based on the complexity and deformation characteristics of structural components, improving computational efficiency while ensuring accuracy, laying the foundation for accurate simulation of machining deformation.

[0046] Multi-source data-driven physical modeling: Collect multi-source data (such as cutting data, displacement data, and temperature data) to truly reflect various physical phenomena during the machining process. Modify material properties so that the model can more accurately describe the mechanical behavior of the material during actual machining. Construct an intelligent constitutive model to establish a precise relationship between input parameters (cutting parameters and temperature) and the output stress-strain curve, providing a reliable material model for subsequent simulation analysis.

[0047] Multi-scale coupled dynamic simulation: Macro-micro model coupling allows for comprehensive analysis of structural deformation at both macro and micro levels, considering the impact of microstructure on macro deformation and improving simulation accuracy. Dynamic prediction of machining forces can proactively capture force changes during machining, providing a basis for timely adjustment of machining parameters and avoiding deformation caused by excessive or fluctuating machining forces.

[0048] Digital twin verification and closed-loop control: Quantify deformation deviations, compare actual machining results with simulation models, and accurately assess machining quality. By optimizing machining parameters and adjusting the machining process in real time based on deformation deviations, closed-loop control of the machining process is achieved, continuously improving machining accuracy and reducing scrap rates.

[0049] Full-process data management: By establishing a real-time data channel based on the OPC UA protocol, efficient data interaction between CAD, CAE, CNC and other systems is achieved, ensuring real-time data sharing and collaborative work in each link, thereby improving production efficiency; Hyperledger Fabric blockchain technology is used to store key process parameters to ensure data security, non-tamperability and traceability, providing strong support for quality traceability, process improvement and production management.

[0050] In this application, in building an intelligent model and performing dynamic optimization, specifically:

[0051] Obtaining design drawings of large, complex structural components, using 3D modeling software to create a 3D geometric model based on the design drawings provides the foundation for the entire simulation process. Calculating the wall thickness and curvature radius of each surface in the 3D geometric model quantifies the geometric characteristics of the structural component, providing specific data for subsequent marking of areas prone to deformation.

[0052] A feature recognition algorithm is used to mark easily deformed areas on each surface of a 3D geometric model, and marking conditions for these areas are set. The marking conditions include surface areas where both the wall thickness and the curvature radius are less than a corresponding preset threshold. The Kriging interpolation method is used to embed the wall thickness fluctuation tolerance information.

[0053] It should be further explained that for complex structures in three-dimensional geometric models, a pre-trained convolutional neural network is used for feature recognition. This convolutional neural network is trained using a large number of structural component models containing easily deformed areas. It can automatically identify and mark the easily deformed areas in the model, providing a basis for subsequent analysis and processing. The use of feature recognition algorithms and convolutional neural networks to mark easily deformed areas can accurately locate the parts of the structural components that are prone to deformation, allowing subsequent analysis and processing to focus on these key areas, improving the pertinence and accuracy of the simulation.

[0054] The Kriging interpolation method is used to embed wall thickness fluctuation tolerance information. Specifically, interpolation points are selected at key locations in the structural component model (such as boundaries of easily deformed areas and stress concentration areas) and evenly distributed across the model surface. A Gaussian semivariogram is chosen as the basis for Kriging interpolation, and its parameters, such as range and nugget value, are determined through cross-validation. Based on the selected interpolation points and the determined semivariogram parameters, the wall thickness fluctuation values ​​at each model point are calculated and embedded into the 3D geometric model, making the model more realistically reflect dimensional variations during actual processing. By embedding wall thickness fluctuation tolerance information using the Kriging interpolation method, dimensional fluctuations during actual processing are taken into account. By selecting interpolation points at key locations and on the model surface, combining the Gaussian semivariogram and cross-validation to determine parameters, the wall thickness fluctuation values ​​are calculated and embedded, making the model more realistically reflect dimensional variations during actual processing, improving the model's accuracy and reliability. Considering that wall thickness fluctuation is an inevitable factor in actual processing, incorporating it into the model better simulates the uncertainty in the processing and provides more realistic boundary conditions for subsequent simulation analysis.

[0055] The adaptive grid division adopts octree coarse division and sets the voxel edge. The grid is encrypted by combining the Zienkiewicz-Zhu error estimation formula and the particle swarm optimization algorithm. The particle swarm velocity update formula in the particle swarm optimization algorithm is:

[0056] ;

[0057] in represents the velocity of the i-th particle at the k+1-th iteration, w is the inertia weight, and its value range is usually [0.4, 0.9], which is used to balance the global search and local search capabilities of the particle. It is expressed as the learning factor, which is generally taken as , respectively adjust the step size of the particle flying towards its own historical optimal position and the global optimal position, Represents a random number uniformly distributed in the interval [0, 1], used to increase the randomness of the search; Represented as the historical optimal position of the i-th particle, represents the global optimal position, represents the position of the i-th particle at the k-th iteration; by using octree coarse partitioning and combining the Zienkiewicz-Zhu error estimation formula with the particle swarm optimization algorithm for grid encryption, the grid density can be adaptively adjusted according to the complexity of the model and the error distribution. While ensuring the calculation accuracy, unnecessary waste of computing resources is avoided and the calculation efficiency is improved.

[0058] In this application, when performing multi-source data-driven physical modeling, a dynamometer and a laser displacement sensor are used to collect cutting data and displacement data; the extended Kalman filter algorithm is used to fuse the data to correct material properties, and the state transfer matrix of the extended Kalman filter algorithm is used. for:

[0059] ;in Indicates the sampling time of collected data;

[0060] A nonlinear constitutive model based on LSTM is constructed, with cutting data and temperature as input and stress-strain curve as output; the cutting data includes cutting speed and feed rate;

[0061] By using the extended Kalman filter algorithm to fuse data and correct material properties, in actual processing, material properties will change due to the influence of multiple factors. Traditional static material properties cannot accurately describe this dynamic change. The extended Kalman filter algorithm can comprehensively consider information from different data sources and dynamically update and correct material properties through the state transition matrix to make the material properties more consistent with actual processing conditions. The state transition matrix includes sampling time, which means that the algorithm can take into account the influence of time factors on material properties, thereby better simulating the dynamic mechanical behavior of materials during processing.

[0062] The training process of the LSTM-based nonlinear constitutive model adopts the mean square error loss function, which is formulated as follows: ; Where L represents the loss function value, u represents the number of training samples, and b represents the number of training samples. represents the true stress-strain value of the bth sample, Predict the stress-strain value of the model for the bth sample and optimize the model parameters by minimizing the loss function;

[0063] By constructing a nonlinear constitutive model based on LSTM (long short-term memory) networks, it is possible to process the complex nonlinear relationship between cutting data, temperature, and stress-strain curves. During the processing of large and complex structural parts, the mechanical behavior of the material is affected by a combination of factors and exhibits highly nonlinear characteristics. The LSTM network has the ability to memorize and process sequential data, and can capture the long-term dependencies between these factors, thereby more accurately predicting the stress-strain response of the material.

[0064] In this application, in the multi-scale coupled dynamic simulation, the macro-micro model includes a macro model and a micro crystal plasticity model. The macro model adopts Abaqus / Explicit, and the micro crystal plasticity model adopts VPSC. Through homogenization theory coupling, the coupling formula is: , ;in, represents the macro stress, represents micro stress, V is the volume of the material, represents the macroscopic strain, represents the microscopic strain, represents local strain;

[0065] By using Abaqus / Explicit for the macroscopic model and VPSC for the microscopic crystal plasticity model, the mechanical behavior of the structural component is analyzed at the macroscopic overall structure and microscopic crystal levels, respectively. The macroscopic model can describe the overall deformation trend and mechanical response of the structural component, while the microscopic model reveals the plastic deformation mechanism of the material at the crystal scale. The two are coupled through homogenization theory, and the relationship between macroscopic stress and strain is established using coupling formulas. This allows the deformation of the microscopic crystal to affect the mechanical properties of the macroscopic structure, and the macroscopic mechanical conditions can also be fed back into the deformation analysis of the microscopic crystal, truly reflecting the influence of the microstructure on the macroscopic deformation.

[0066] An alternating iterative method is used to transfer stress and strain, enabling continuous information exchange between the macro and micro models, gradually approaching the true mechanical state. During the iteration process, the stress and strain results of the macro model are transmitted to the micro model. The micro model calculates the micro stress and strain based on this information, and then feeds the results back to the macro model. This cycle enables the model to more accurately simulate the mechanical changes during the machining process. A preset convergence accuracy of 1e-5 ensures that the iteration process stops after reaching a certain accuracy requirement. The LSTM model based on the attention mechanism sets a pre-set time interval (50ms) to predict cutting force fluctuations.

[0067] The calculation formula for predicting cutting force fluctuations based on the LSTM model of the attention mechanism is: ;in represents the predicted cutting force, represents the attention weight of the s-th LSTM hidden layer output on the predicted cutting force, , hs represents the output of the LSTM hidden layer; through the attention mechanism, the LSTM model can automatically learn the importance of the hidden layer output at different times for predicting the cutting force, and assign corresponding attention weights to each hidden layer output, so that the model can focus on key information when processing complex cutting force data, thereby more accurately predicting cutting force fluctuations.

[0068] It should be noted that its attention weight By calculating the similarity between the hidden layer output hs and a learnable vector u, it is obtained through the Softmax function, that is, The input of the model is the previous cutting force data. After being processed by the LSTM layer, the attention weight of each moment is calculated through the attention mechanism, and the weighted sum of the hidden layer output is used to obtain the predicted cutting force. During model training, the learning rate is set to 0.001 and the number of training rounds is set to 500. By continuously optimizing the model parameters, the accuracy of cutting force prediction is improved, providing a basis for adjusting the machining process.

[0069] In this application, during digital twin verification and closed-loop control, industrial CT scanning (accuracy ±5μm) is used to acquire point cloud data. The ICP algorithm is used to align the point cloud data with the simulation model. The IPC algorithm formula is: , ; Where R represents the rotation matrix, t represents the translation vector, represents the jth point in the point cloud data, is the corresponding point in the simulation model, represents the centroid of the point cloud data, Represents the centroid of the corresponding point set of the simulation model; by calculating the rotation matrix R and the translation vector t, the two are accurately aligned in space, providing a basis for subsequent accurate comparison;

[0070] Then, based on the consistency of normal vectors, the out-of-tolerance area is identified. Specifically, the normal vector of each point in the point cloud data is calculated with the normal vector of the corresponding point in the simulation model. By comparing the angle θ between the normal vectors, when cosθ < 0.95, it indicates that there is an out-of-tolerance in the area, and the area is marked as an out-of-tolerance area.

[0071] Design a multi-objective reward function: ; Among them, r represents the reward value, Indicates deformation deviation, Indicates the processing time deviation, Represents energy consumption deviation. Machining parameters are adjusted online via an actor-critic network, with feed rates ranging from 500-2000 mm / min and spindle speeds from 8000-15000 rpm. By utilizing this network, the actor network outputs parameter adjustments based on the current state (deformation deviation, machining time deviation, and energy consumption deviation), while the critic network evaluates the current state and actions. Through continuous iterative training, the actor network learns to output the optimal parameter adjustments under different conditions, enabling real-time optimization of the machining process. For example, when deformation deviation is large, the feed rate and spindle speed are adjusted to minimize deformation; when machining time is excessive, the feed rate is appropriately increased to shorten it.

[0072] In the present application, in the Actor-Critic network, the Actor-Critic network includes an Actor network and a Critic network; the output of the Actor network is the adjustment amount of the processing parameters, the input of the Critic network is the current state and the output of the Actor network, and the output is a value evaluation of the current state and action. Through continuous iterative training, the Actor network can output the optimal processing parameter adjustment amount according to different states; the current state includes deformation deviation, processing time deviation, and energy consumption deviation; it should be noted that the Actor-Critic network is mainly used in this application to intelligently optimize processing parameters and realize multi-objective dynamic adjustment of the processing process; according to the real-time deformation deviation, processing time deviation and energy consumption deviation, the Actor network outputs the appropriate processing parameter adjustment amount, and the Critic network evaluates the effects of these adjustments, and continuously iterates training so that the Actor network can find the optimal adjustment plan under various states, thereby improving processing quality and efficiency and reducing energy consumption.

[0073] In this application, in the multi-scale coupled dynamic simulation, the crystal orientation distribution in the microscopic model is updated every 10 iterations.

[0074] In this application, the Zienkiewicz-Zhu error estimation formula is: ;in represents the overall error estimate, represents the number of grid cells, It represents the error estimate of the e-th grid cell, which is used to evaluate the accuracy of the current grid division. When the overall error estimate is greater than the preset overall error threshold, the particle swarm optimization algorithm is used to refine the grid.

[0075] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.

[0076] It should be understood that the present invention is not limited to the exact construction described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A dynamic modeling method for simulating deformation during machining of large and complex structural parts, characterized in that: The following steps are involved: Step 1: Build an intelligent model and perform dynamic optimization, including establishing a 3D model, marking easily deformed areas, embedding tolerance information, and adaptive meshing; In the construction of the intelligent model and dynamic optimization, specifically: Obtain design drawings of large and complex structural parts, use 3D modeling software to create a 3D geometric model based on the design drawings; calculate the wall thickness and curvature radius of each surface in the 3D geometric model; A feature recognition algorithm is used to mark easily deformed areas on each surface of a 3D geometric model, and marking conditions for these areas are set. The marking conditions include surface areas where both the wall thickness and the curvature radius are less than a corresponding preset threshold. The Kriging interpolation method is used to embed the wall thickness fluctuation tolerance information. The adaptive grid division adopts octree coarse division and sets voxel edges, and combines the Zienkiewicz-Zhu error estimation formula and particle swarm optimization algorithm to perform grid encryption; Step 2: Conduct multi-source data-driven physical modeling to collect data, modify material properties, and build an intelligent constitutive model; In the multi-source data driven physical modeling, a dynamometer and a laser displacement sensor are used to collect cutting data and displacement data; an extended Kalman filter algorithm is used to fuse the data and correct material properties; A nonlinear constitutive model based on LSTM is constructed, with cutting data and temperature as input and stress-strain curve as output; the cutting data includes cutting speed and feed rate; Step 3: Conduct multi-scale coupled dynamic simulation to achieve macro-micro model coupling and dynamic prediction of machining forces; Step 4: Perform digital twin verification and closed-loop control to quantify deformation deviations and optimize processing parameters; Step five: Implement full-process data management, establish real-time data channels and store data on the blockchain.

2. A dynamic modeling method for simulating deformation during machining of large and complex structural parts according to claim 1, characterized in that: In the multi-scale coupled dynamic simulation, the macro model uses Abaqus / Explicit, and the micro crystal plasticity model uses VPSC. Through homogenization theory coupling, the coupling formula is: , ;in, represents the macro stress, represents micro stress, V is the volume of the material, represents the macroscopic strain, represents the microscopic strain, represents local strain; An alternating iterative method is used to transfer stress and strain, and the convergence accuracy is preset. An LSTM model based on the attention mechanism sets the time interval in advance to predict cutting force fluctuations. The calculation formula for predicting cutting force fluctuations using the LSTM model based on the attention mechanism is: ;in represents the predicted cutting force, represents the attention weight of the s-th LSTM hidden layer output on the predicted cutting force, , hs represents the output of the LSTM hidden layer.

3. The method for dynamic modeling of deformation simulation of large complex structural parts according to claim 1, characterized in that: In the digital twin verification and closed-loop control, industrial CT scanning is used to obtain point cloud data, and the ICP algorithm is used to align the point cloud data with the simulation model. The IPC algorithm formula is: , ; Where R represents the rotation matrix, t represents the translation vector, represents the jth point in the point cloud data, is the corresponding point in the simulation model, represents the centroid of the point cloud data, represents the centroid of the corresponding point set of the simulation model; Then identify the out-of-tolerance area based on the consistency of the normal vector; Design a multi-objective reward function: ; Among them, r represents the reward value, Indicates deformation deviation, Indicates the processing time deviation, Represents the energy consumption deviation; the processing parameters are adjusted online through the Actor-Critic network.

4. The method for dynamic modeling of deformation simulation of large complex structural parts according to claim 1, characterized in that: In the Actor-Critic network, the output of the Actor network is the adjustment of the processing parameters, the input of the Critic network is the current state and the output of the Actor network, and the output is the value evaluation of the current state and action. Through continuous iterative training, the Actor network can output the optimal processing parameter adjustment according to different states; The current status includes deformation deviation, processing time deviation, and energy consumption deviation.

5. The method for dynamic modeling of deformation simulation of large complex structural parts according to claim 1, characterized in that: In the multi-scale coupled dynamic simulation, the crystal orientation distribution in the microscopic model is updated every time a set number of iterations are performed.

6. The method for dynamic modeling of deformation simulation of large complex structural parts according to claim 1, characterized in that: The Zienkiewicz-Zhu error estimation formula is: ;in represents the overall error estimate, represents the number of grid cells, Represents the error estimate of the e-th grid cell; when the overall error estimate is greater than its preset overall error allowable threshold, the particle swarm optimization algorithm is triggered to perform grid refinement.

Images