An intelligent monitoring and optimization system for CNC machine tools based on digital twins

Abstract

The present invention relates to the technical field of intelligent monitoring of machine tools, and discloses an intelligent monitoring and optimization system for CNC machine tools based on digital twins, wherein the present invention includes: a data acquisition module, which is used to collect discrete-time working condition data and distribute dynamic precision weights of CNC machine tools to obtain a dynamic precision distribution matrix; a calculation module, which is used to calculate the geometric simulation precision level and the physical simulation precision level; a construction module, which is used to construct an adaptive geometric grid model; an adaptive refinement module, which is used to adaptively refine the contact area for cutting force monitoring to obtain a tool-workpiece contact physical simulation model; a collaborative adjustment module, which is used to input the adaptive geometric grid model and the tool-workpiece contact physical simulation model into a working condition conversion state machine for collaborative adjustment to obtain an adaptive precision simulation result. The present invention ensures the asymptotic stability and simulation consistency convergence of the digital twin system in different processing stages, and improves the reliability and robustness of the system.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent monitoring of machine tools, and in particular to an intelligent monitoring and optimization system for CNC machine tools based on digital twins. Background Art

[0002] Digital twin technology, a key enabler of intelligent manufacturing, enables real-time monitoring, prediction, and optimization of manufacturing processes by building virtual models of physical entities. However, existing digital twin systems for CNC machine tools still face numerous technical challenges in practical application, particularly the difficult balance between real-time performance and simulation accuracy under complex machining conditions.

[0003] Traditional digital twin simulation systems generally employ fixed precision settings and static resource allocation strategies, failing to adapt to dynamic changes in actual machining conditions. This results in wasted computing resources and insufficient simulation accuracy during various stages, such as roughing, finishing, and complex surface machining. Furthermore, existing systems lack effective working condition perception mechanisms, unable to accurately identify real-time changes in key working condition parameters such as tool wear, cutting force variations, and spindle vibration. This results in significant deviations between the digital twin model and the physical entity, severely impacting the system's reliability and practicality. Summary of the Invention

[0004] The present invention provides an intelligent monitoring and optimization system for CNC machine tools based on digital twins. The present invention ensures the asymptotic stability and simulation consistency convergence of the digital twin system in different processing stages, thereby improving the reliability and robustness of the system.

[0005] In a first aspect, the present invention provides a digital twin-based intelligent monitoring and optimization system for CNC machine tools, comprising:

[0006] The data acquisition module is used to collect discrete time working condition data and allocate dynamic precision weights for CNC machine tools to obtain a dynamic precision allocation matrix;

[0007] A calculation module, configured to calculate a geometric simulation accuracy level and a physical simulation accuracy level based on the dynamic accuracy allocation matrix;

[0008] A construction module, configured to construct an adaptive geometric mesh model of the CNC machine tool according to the geometric simulation accuracy level;

[0009] An adaptive refinement module, configured to adaptively refine the contact area for cutting force monitoring based on the physical simulation accuracy level and the adaptive geometric grid model to obtain a tool-workpiece contact physical simulation model;

[0010] The collaborative adjustment module is used to input the adaptive geometric grid model and the tool-workpiece contact physical simulation model into the working condition conversion state machine for collaborative adjustment to obtain an adaptive precision simulation result.

[0011] In combination with the first aspect, in a first implementation of the first aspect of the present invention, the data acquisition module further includes:

[0012] A setting unit is used to set a discrete sampling period for the CNC machine tool to obtain a discrete sampling time sequence;

[0013] a synchronization unit for performing synchronous data acquisition and timestamp association on tool wear state parameters, cutting force change parameters, spindle speed fluctuation parameters, feed speed change parameters, and machined surface roughness parameters according to the discrete sampling time sequence to obtain an original working condition data set;

[0014] a vector construction unit, configured to perform normalization based on the original operating condition data set to obtain a normalized numerical sequence, and input the normalized numerical sequence into a multidimensional feature vector construction algorithm for vectorized combination and weight distribution calculation to obtain a standardized operating condition feature vector;

[0015] The weight allocation unit is used to perform dynamic precision weight allocation of the bounded uncertain diagonal matrix based on the standardized working condition eigenvector to obtain a dynamic precision allocation matrix.

[0016] In combination with the first aspect, in a second implementation of the first aspect of the present invention, the weight allocation unit further includes:

[0017] An initialization subunit, configured to initialize a bounded uncertain diagonal matrix structure and determine the number and dimension of simulation modules based on the standardized operating condition eigenvector to obtain an initial precision allocation matrix;

[0018] A constraint condition calculation subunit is used to perform bounded constraint condition calculation and dynamic weight allocation on the precision weights of each simulation module according to the initial precision allocation matrix to obtain a constraint precision weight sequence;

[0019] A threshold comparison subunit is used to input the standardized working condition feature vector into the working condition conversion detection algorithm to perform Euclidean distance calculation and threshold comparison at adjacent moments to obtain a working condition conversion trigger signal;

[0020] The matrix updating subunit is used to perform precision redistribution and matrix updating based on the working condition conversion trigger signal and the constraint precision weight sequence to obtain a dynamic precision allocation matrix.

[0021] In combination with the first aspect, in a third implementation of the first aspect of the present invention, the constraint condition calculation subunit is specifically configured to:

[0022] Setting bounded constraints on precision weights and limiting the value range of the diagonal elements of the initial precision allocation matrix to obtain precision weight constraint intervals;

[0023] Quantifying the basic computational overhead and evaluating the computational resource requirements of each simulation module based on the precision weight constraint interval to obtain a basic computational overhead coefficient;

[0024] Perform computing resource constraint mapping function calculation and total resource limit judgment based on the basic computing overhead coefficient and the precision weight constraint interval to obtain computing resource constraint conditions;

[0025] Based on the computing resource constraint conditions, precision weight optimization calculation and serialized output are performed to obtain a constrained precision weight sequence.

[0026] In combination with the first aspect, in a fourth implementation of the first aspect of the present invention, the matrix updating subunit is specifically configured to:

[0027] Based on the working condition conversion trigger signal, trigger state judgment and precision reallocation requirement determination are performed to obtain a precision adjustment instruction;

[0028] Recalculate the weights and update the values ​​of the constraint precision weight sequence according to the precision adjustment instruction to obtain an updated precision weight sequence;

[0029] Performing diagonal element replacement and matrix structure reconstruction based on the update precision weight sequence to obtain an update precision allocation matrix;

[0030] Numerical stability verification and bounded constraint condition verification are performed on the updated precision allocation matrix to obtain a dynamic precision allocation matrix.

[0031] In combination with the first aspect, in a fifth implementation of the first aspect of the present invention, the calculation module further includes:

[0032] An establishing unit is used to construct a working condition characteristic accuracy requirement mapping function and establish a geometric simulation and physical simulation coupling relationship based on the dynamic accuracy allocation matrix to obtain a geometric and physical simulation working condition accuracy mapping function;

[0033] a numerical value allocation unit, configured to perform normalized constraint calculation and numerical value allocation on the geometric simulation dynamic weight factor and the physical simulation dynamic weight factor according to the geometric and physical simulation working condition accuracy mapping function, to obtain a geometric and physical simulation dynamic weight factor combination;

[0034] A stage discrimination unit is used to input the geometric physical simulation dynamic weight factor combination into a processing stage identification algorithm to perform processing stage discrimination and weight adaptive adjustment to obtain a processing stage adjustment weight factor combination;

[0035] The grade calculation unit is used to calculate the geometric simulation accuracy grade and the physical simulation accuracy grade based on the weight factor combination adjusted in the processing stage to obtain the geometric simulation accuracy grade and the physical simulation accuracy grade.

[0036] In combination with the first aspect, in a sixth implementation of the first aspect of the present invention, the value allocating unit is specifically configured to:

[0037] Performing geometric simulation dynamic weight factor extraction and physical simulation dynamic weight factor extraction based on the geometric and physical simulation working condition accuracy mapping function to obtain geometric simulation weight factor values ​​and physical simulation weight factor values;

[0038] Performing sum constraint calculation and normalization verification on the geometric simulation weight factor value and the physical simulation weight factor value to obtain a normalized constraint condition;

[0039] Performing processing discrimination and weight value distribution calculation on the characteristic parameters of the processing stage according to the normalized constraint conditions to obtain a weight factor distribution value;

[0040] Based on the weight factor allocation values, geometric physics simulation weight pairing and numerical verification are performed to obtain a geometric physics simulation dynamic weight factor combination.

[0041] In combination with the first aspect, in a seventh implementation of the first aspect of the present invention, the building module is specifically configured to:

[0042] Based on the geometric simulation accuracy level, a quadtree voxel hybrid discretization strategy is constructed and tool wear perception parameters are set to obtain a strategy parameter combination;

[0043] According to the combination of the strategy parameters, a threshold value judgment and a quadtree recursive depth calculation and a voxel size determination are performed on the tool wear amount to obtain an adaptive discretization parameter;

[0044] Inputting the adaptive discretization parameters into the tool-workpiece contact area identification algorithm to perform envelope surface intersection calculation and area modeling to obtain a hierarchical grid discretization structure;

[0045] Geometric error accumulation control and grid re-division triggering judgment are performed on the hierarchical grid discretization structure to obtain an adaptive geometric grid model.

[0046] In combination with the first aspect, in an eighth implementation of the first aspect of the present invention, the adaptive refinement module is specifically configured to:

[0047] Based on the physical simulation accuracy level and the adaptive geometric grid model, a tool-workpiece contact state enhancement vector is constructed and parameters of normal contact force, tangential friction force, contact area, and contact pressure are extracted to obtain a tool-workpiece contact state enhancement vector;

[0048] Calculating the contact area accuracy grading standard and determining the mesh refinement multiple for the cutting force monitoring value according to the tool-workpiece contact state enhancement vector to obtain the area grading accuracy refinement parameter;

[0049] Based on the regional classification precision refinement parameters, real-time working condition identification, triggering and judgment and precision switching control are performed to obtain cutting force adjustment instructions;

[0050] Based on the cutting force adjustment instruction, radial basis function contact force field interpolation calculation and physical simulation accuracy correction are performed to obtain a tool-workpiece contact physical simulation model.

[0051] In conjunction with the first aspect, in a ninth implementation of the first aspect of the present invention, the collaborative adjustment module is specifically configured to:

[0052] Inputting the adaptive geometric grid model and the tool-workpiece contact physical simulation model into a working condition conversion state machine to construct a machining abnormality state, thereby obtaining a working condition conversion state machine control structure;

[0053] Based on the working condition conversion state machine control structure, a nested precision threshold adjustment mechanism is constructed and PID structure feedback controller parameters are set to obtain a nested precision adjustment feedback loop;

[0054] Performing a total resource limitation judgment and a processing quality impact weight distribution calculation on computing resource requirements according to the nested precision adjustment feedback loop to obtain a resource competition scheduling allocation plan;

[0055] The resource competition scheduling allocation scheme is combined with the precision control system convergence verification and simulation consistency verification to obtain adaptive precision simulation results.

[0056] The technical solution provided by the present invention can accurately capture key working condition parameters such as tool wear, cutting force changes, and spindle speed fluctuations in real time through discrete-time working condition data acquisition and multi-dimensional feature vector construction. Compared with traditional continuous monitoring methods, it avoids data redundancy and improves the accuracy and efficiency of working condition identification. The dynamic accuracy weight allocation mechanism of the bounded uncertain diagonal matrix is ​​adopted to adaptively adjust the accuracy weight of each simulation module according to the real-time working condition, overcome the limitations of the traditional fixed accuracy setting, and realize dynamic optimization of accuracy allocation. The working condition feature-accuracy requirement mapping function is established to realize dynamic coupling adjustment of geometric simulation and physical simulation, solving the technical bottleneck of the two simulation modes being independent of each other and unable to coordinate optimization in traditional methods. The quadtree voxel hybrid discretization strategy for tool wear perception can adaptively adjust the grid density according to the wear state, significantly improving the accuracy and efficiency of geometric modeling compared with the traditional fixed grid division method. The contact area adaptive refinement technology for cutting force monitoring realizes high-precision modeling of key processing areas by constructing contact state enhancement vectors and hierarchical accuracy control, avoiding the waste of computing resources in non-critical areas. The collaborative adjustment mechanism of the working condition transition state machine and the computing resource competition scheduling strategy dynamically allocate computing resources according to the processing stage and quality requirements, achieving an optimal balance between simulation accuracy and real-time performance within limited edge computing capabilities. Lyapunov stability criterion verification and a nested precision threshold adjustment mechanism ensure the asymptotic stability and simulation consistency convergence of the digital twin system at different processing stages, improving the system's reliability and robustness. The processing stage identification algorithm and weight adaptive adjustment mechanism accurately distinguish different stages such as roughing, finishing, and complex surface processing, and adjust the precision strategy accordingly to meet the needs of diverse processing technologies.

[0057] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purposes and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0058] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of an embodiment of an intelligent monitoring and optimization system for CNC machine tools based on digital twins in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] The terms "including," "having," and any variations thereof, as used in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, system, product, or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to the process, system, product, or device.

[0062] To facilitate understanding of this embodiment, we first introduce in detail a CNC machine tool intelligent monitoring and optimization system based on digital twins disclosed in an embodiment of the present invention. Figure 1 As shown, this system includes:

[0063] Data acquisition module 001 is used to collect discrete time working condition data and allocate dynamic precision weights for CNC machine tools to obtain a dynamic precision allocation matrix;

[0064] Specifically, the setting unit sets the discrete sampling period for the CNC machine tool, pre-defining a high-frequency discrete time sampling period, such as sampling period T1 = 1 millisecond. By setting a fixed and stable discrete sampling sequence, a standardized time reference system is established. The synchronization unit performs real-time, synchronous data acquisition of tool wear parameters, cutting force variation parameters, spindle speed fluctuation parameters, feed rate variation parameters, and machined surface roughness parameters based on the set discrete sampling sequence. By adding precise timestamps to the collected data at each sampling moment, strict temporal correlation is achieved for multi-source, heterogeneous working condition parameters, ensuring that all working condition data points have a unified and alignable time reference, resulting in the original working condition data set. Based on the original working condition data set, the vector construction unit applies a normalization algorithm to address the dimensional and numerical range differences between different physical quantities, normalizing each type of working condition data to the [0, 1] interval. This eliminates computational bias caused by non-uniform physical quantity scales and ensures numerical stability in subsequent processing. The resulting numerical sequence is then fed into a multidimensional feature vector construction algorithm, which combines the different working condition parameters to generate a feature vector containing information about each working condition. The weight allocation mechanism is introduced in the vector construction process. Different weights are assigned according to the degree of influence of each working condition parameter on the overall processing quality and the dynamic behavior of the system. This gives the key working condition parameters a higher feature expression capability and obtains a standardized working condition feature vector with a unified scale and weight calibration. Based on the generated standardized working condition feature vector, the weight allocation unit constructs a bounded uncertain diagonal matrix as the basis for dynamic accuracy weight allocation. Construct an n×n diagonal matrix P(t), where the diagonal elements p ii( t) represents the accuracy weight of each simulation module under the current working condition, and sets the weight range constraint so that 0.1≤p ii (t)≤1.0 to prevent accuracy from dropping to unacceptable levels. During matrix construction, the real-time changes in the operating condition eigenvectors are combined with a dynamic detection mechanism based on characteristic Euclidean distance to monitor the degree of operating condition variation in real time. When the operating condition change reaches or exceeds the set threshold, weight redistribution is automatically triggered, ensuring that simulation resources dynamically adapt to changes in the operating condition, forming a dynamic precision allocation matrix that accurately matches the actual machine tool operating conditions.

[0065] The initialization subunit initializes the bounded, uncertain diagonal matrix structure based on the standardized operating condition eigenvector. The number of simulation modules is determined based on the dimension n of the eigenvector, generating an n×n diagonal matrix. The initial diagonal elements are set within a reasonable range, such as 0.5, to ensure a certain degree of balance and stability in the matrix's initial state. Furthermore, the initialization subunit pre-assigns preliminary weights based on the functional characteristics of the simulation modules and the engineering physics significance of the eigenvector components. For example, relatively high initial weights are assigned to modules representing key operating condition characteristics such as tool wear and cutting force variations to improve simulation accuracy in key areas, resulting in an initial precision allocation matrix. The constraint calculation subunit then calculates bounded constraints based on the diagonal elements of the initial precision allocation matrix, setting a lower weight limit of 0.1 and an upper weight limit of 1.0 to prevent distortion of simulation accuracy due to excessively low weights or excessive consumption of computing resources due to excessively high weights. The constraint calculation subunit dynamically adjusts the accuracy weights of each simulation module based on the different components of the standardized working condition feature vector. For example, when the tool wear characteristic increases significantly, the weights of modules related to tool wear are automatically increased, while the weights of modules with less impact on overall machining quality are reduced. A dynamic weight allocation mechanism generates a sequence of constraint accuracy weights. Simultaneously, the threshold comparison subunit integrates a working condition transition detection algorithm and performs real-time calculations of the Euclidean distance between two consecutive time steps of the standardized working condition feature vector to quantify the severity of the working condition change. The calculation is performed as ||W(t) - W(t-1)||, and a sensitivity threshold δ, for example, 0.15, is set. If the Euclidean distance exceeds this threshold, it indicates a significant change in the current working condition, triggering the threshold comparison subunit to generate a working condition transition trigger signal. By introducing a dynamic change detection mechanism at adjacent moments, the system responds promptly to sudden changes in the machine tool working condition or transitions between machining stages, avoiding the delayed response to working condition changes encountered in traditional static accuracy setting. The matrix update subunit combines the working condition transition trigger signal with the constraint accuracy weight sequence to perform accuracy redistribution and matrix update operations. When a sudden change in the operating conditions is detected, the matrix update subunit reconstructs the diagonal matrix elements according to the latest constraint precision weight sequence, adjusts the precision weights of each simulation module in a timely manner, and retains the overall diagonal structure of the matrix to ensure the numerical stability of the matrix calculation and the efficiency of the inverse operation, and obtains the updated dynamic precision allocation matrix.

[0066] After receiving the initial precision allocation matrix generated by the initialization subunit, the constraint calculation subunit sets the precision weight bounded constraint conditions and value range restrictions for the diagonal elements of the matrix, that is, the initial precision weights corresponding to each simulation module. Specifically, each weight value p ii (t) is limited to a reasonable range, for example, 0.1≤p ii(t)≤1.0, to avoid the simulation accuracy being out of control due to too low weight, and to prevent the unreasonable occupation of computing resources due to too high weight. Through this setting, the precision weight constraint interval of each simulation module is obtained. Based on the precision weight constraint interval, combined with the specific computational complexity characteristics of each simulation module, the basic computational overhead is quantified. For each simulation module, according to its different precision weights, the corresponding basic computational overhead coefficient C is quantified. i , the numerical value reflects the basic computing resource consumption required by the simulation module per unit time. The basic computing overhead coefficient is obtained through historical simulation data calibration, theoretical model estimation or online dynamic measurement to ensure that the evaluation results not only meet the actual computing needs of the module, but also adapt to the changing trend of computing resource consumption under different processing conditions. Through this process, the abstract simulation accuracy requirements are converted into specific calculable resource requirements. The constraint calculation subunit performs the calculation of the computing resource constraint mapping function based on the basic computing overhead coefficient and the accuracy weight constraint interval. The mapping function is expressed as the product relationship between resource consumption and accuracy weight. For each simulation module i, the resource consumption is formally expressed as p ii (t)×C i , by summing up the resource consumption of all modules, the total resource consumption R(t) is calculated. Introducing a total computing resource upper limit R max , represents the maximum total amount of computing resources that can be allocated by the current system, and then the resource limit judgment is performed: if R(t)≤R max , indicating that the current precision allocation scheme is feasible; on the contrary, if R(t)>R max , it is necessary to further optimize the precision weight of each module to reduce the total resource consumption and avoid system resource overload. The optimization calculation of the precision weight is performed based on the computing resource constraint conditions. The specific strategy adopts greedy algorithm, linear programming or heuristic search method to adjust the precision weight of each module within the constraint range, giving priority to ensuring the high precision requirements of modules related to key working conditions, while reducing the precision weight of secondary modules without affecting the overall simulation accuracy convergence, thereby maximizing the computational accuracy of the key simulation area while meeting the total resource constraints. After the optimization is completed, a new constraint precision weight sequence is generated and serialized and output for use by the matrix update subunit, driving the real-time update of the dynamic precision allocation matrix, and realizing efficient, accurate, and resource-controlled simulation precision scheduling for complex dynamic working conditions.

[0067] The matrix update subunit receives the operating condition transition trigger signal generated by the threshold comparison subunit in real time. Based on this trigger signal, it determines whether the current operating condition has undergone a drastic change exceeding the threshold. If the trigger signal is valid, the matrix update subunit dynamically determines whether precision reallocation is necessary based on the magnitude and type of the operating condition change, combined with a sensitivity analysis of the impact of each operating condition's characteristics on simulation accuracy. It then generates a precision adjustment instruction, which specifies the sequence of simulation modules to be adjusted, including the adjustment magnitude, priority ranking, and the target constraints for resource reallocation. Based on the precision adjustment instruction, the constraint precision weight sequence is recalculated. An adaptive update mechanism based on the operating condition change weight adjustment coefficient is employed. Through strategies such as weighted averaging, scaling, or incremental adjustment, the precision weights of each module are adjusted in conjunction with real-time operating condition data. This ensures that the weight changes reflect the operating condition trend while maintaining a certain degree of continuity and numerical smoothness, thus avoiding system instability caused by sudden changes. After the weight recalculation is complete, an updated precision weight sequence is generated. Based on the updated precision weight sequence, the diagonal elements of the original initial precision allocation matrix are replaced one by one, while the diagonal matrix structure with zero off-diagonal elements remains unchanged, and the matrix structure is reconstructed. The reconstructed matrix has the latest precision weight configuration, which can accurately map the precision requirements of each simulation module under the current working condition, and provide updated precision weight input for the joint optimization of geometric simulation and physical simulation. The matrix diagonalization principle is followed during the matrix reconstruction process to ensure that the complexity of matrix inversion and decomposition operations is reduced while maintaining the efficiency of numerical calculations. The newly generated updated matrix is ​​numerically stabilized and bounded constraint checked. Stability verification prevents simulation divergence or error accumulation problems caused by excessive changes in precision weights by detecting whether the spectral radius, condition number, and diagonal elements of the matrix meet the stability requirements; bounded constraint verification ensures that each diagonal element is still strictly limited to the preset weight range of 0.1 to 1.0, and the total resource consumption does not exceed the maximum resource limit R max , thus obtaining the final available dynamic precision allocation matrix.

[0068] A calculation module 002 is used to calculate the geometric simulation accuracy level and the physical simulation accuracy level based on the dynamic accuracy allocation matrix;

[0069] Specifically, the establishment unit, based on the dynamic precision allocation matrix, extracts the corresponding simulation precision requirements for different working condition characteristics and constructs a mapping relationship between working condition characteristics and simulation accuracy. The establishment unit utilizes key components of the working condition characteristic vector, such as tool wear, cutting force variation, spindle speed fluctuation, feed rate stability, and machined surface roughness, to extract the primary controlling factors affecting geometric and physical simulations, respectively. These physical quantities are quantitatively associated with the simulation precision requirements using a working condition characteristic precision requirement mapping function, resulting in a mapping formula for geometric and physical simulation precision requirements. Furthermore, the establishment unit further analyzes the coupling relationship between geometric and physical simulations, clarifying the interaction mechanism between the two at different machining stages and forming a working condition precision mapping function for geometric and physical simulations. The numerical allocation unit performs normalized constraint calculations on the dynamic weight factors for geometric and physical simulations. The initial weight allocation values ​​output by the mapping function are extracted, and then a normalization algorithm is used to compress the two types of simulation weights into a unified, comparable numerical range. For example, the weight factors are constrained to the interval [0, 1], with the sum of the two factors required to remain equal to 1. To ensure numerical stability and physical rationality of the system, the numerical allocation unit introduces constraint boundary conditions during the normalization process to prevent a simulation mode from losing accuracy due to excessively low weights. Through normalization constraints, a standardized geometric and physical simulation dynamic weight factor combination is obtained. The stage discrimination unit inputs this normalized geometric and physical simulation dynamic weight factor combination into the machining stage identification algorithm. Based on the changing trends of working condition characteristics and the actual machining process requirements, the stage identification algorithm performs machining stage identification and classification. The machining stage identification algorithm comprehensively determines the current machining stage of the machine tool, including roughing, semi-finishing, finishing, and complex surface machining, by combining key indicators such as cutting depth, feed rate, spindle load change rate, and surface roughness requirements. After identification, the stage discrimination unit adaptively adjusts the dynamic weight factor combination based on the characteristics of each machining stage. For example, in the roughing stage, the geometric simulation weight is increased to 0.7 and the physical simulation weight is reduced to 0.3; in the finishing stage, the physical simulation weight is appropriately increased to achieve higher local contact accuracy. Through the processing stage discrimination and weight adaptive adjustment mechanism, it is ensured that the simulation accuracy distribution can adapt to the changing needs of different processing conditions in real time, avoiding the degradation of simulation performance caused by stage conversion. The level calculation unit performs specific simulation accuracy level calculation based on the weight factor combination adjusted in the processing stage. The unit maps the weight factor to a specific accuracy level interval based on the adjusted weight and the preset accuracy level mapping rules. The accuracy level is set in the range of 1 to 10, and the larger the value, the higher the simulation accuracy. The level calculation unit determines the corresponding geometric simulation accuracy level and physical simulation accuracy level according to the numerical size of the current weight factor through table lookup or function mapping, thereby generating an accuracy level instruction.

[0070] The numerical allocation unit performs parameter analysis based on the geometric and physical simulation working condition accuracy mapping function. Specifically, it extracts the geometric simulation dynamic weight factor and the physical simulation dynamic weight factor that are highly correlated with the working condition characteristics from the mapping function, and calculates the geometric simulation weight factor value and the physical simulation weight factor value at the current moment respectively. The weight factor is extracted based on the contribution of each working condition characteristic in the simulation modeling, and comprehensively considers indicators such as tool wear status, cutting force change rate, spindle vibration amplitude, feed speed fluctuation and machining surface roughness to ensure that the extracted weight factor accurately reflects the different accuracy requirements for geometric morphology and physical contact modeling under the current machining state. After the extraction is completed, the sum constraint calculation and normalization verification are performed on the two preliminarily obtained weight factor values. The geometric simulation weight factor value α G and the physical simulation weight factor value α P Add and check whether the normalization condition of the sum value being 1 is met, that is, α G +α P =1, if there is a deviation, the numerical size of the two will be adjusted proportionally to normalize them, so as to ensure that different simulation modules have unified dimensions and comparability in resource allocation and precision control. Normalization verification not only ensures the consistency of numerical values, but also prevents extreme deviations in individual factor values ​​by setting reasonable boundary thresholds, ensuring that the simulation weight distribution is balanced between physical rationality and numerical stability, and forming normalization constraints. According to the normalization constraints, the processing stage characteristic parameters are processed and the weight value distribution is calculated. Based on the real-time collected working condition feature vector, the current processing stage of the machine tool is identified, including the rough processing stage, semi-finishing stage, finishing stage and complex surface processing stage. According to the different accuracy requirements for geometric simulation and physical simulation in different processing stages, combined with the normalization weight factor basis, the weight factor value is adjusted through the set stage-by-stage weight distribution rules. For example, in the rough processing stage, the geometric morphology changes dramatically, and the demand for geometric simulation accuracy is high, so α is appropriately increased. G to 0.7, while reducing α Pto 0.3; in the finishing stage, the reverse adjustment is made to improve the physical simulation accuracy requirements in order to obtain better contact force field and surface quality control. By allocating weight factors based on stage characteristics, the numerical allocation unit dynamically adapts to changes in working conditions and optimizes simulation resource configuration. The numerical allocation unit performs pairing and numerical verification of geometric simulation weights and physical simulation weights based on weight factor allocation values. The two types of simulation weights are combined according to the normalized values, and stability verification and boundary constraint verification are performed to ensure that the combined dynamic weight factors meet the system's preset accuracy control requirements and resource allocation upper limit. During the numerical verification process, it is examined whether the combined weights can maintain a dynamic balance between simulation accuracy and computing resource consumption, ensuring that there is no resource overflow or simulation performance degradation while ensuring machining quality. The numerical allocation unit outputs a geometric and physical simulation dynamic weight factor combination that meets the specifications.

[0071] A construction module 003 is used to construct an adaptive geometric mesh model of the CNC machine tool according to the geometric simulation accuracy level;

[0072] Specifically, the construction module initiates the construction of a quadtree-voxel hybrid discretization strategy based on the geometric simulation accuracy level and introduces a tool wear-aware parameter setting mechanism. The module sets the base voxel size and initial quadtree recursion depth based on the geometric simulation accuracy level. It then uses tool wear as a dynamic adjustment parameter to establish a strategy parameter combination. This combination includes a base meshing scale, an upper limit on the recursion depth, a voxel refinement factor, and a dynamic threshold for tool wear, enabling adaptive mesh density adjustment as tool wear progresses. Based on this strategy parameter combination, the construction module sequentially performs real-time threshold determinations on tool wear and dynamically calculates the specific values ​​for the quadtree recursion depth and voxel size based on the determinations. When tool wear is detected to be below 0.1 mm, a lower recursion depth and larger voxel size are set to reduce computational resource consumption. When wear exceeds the 0.1 mm threshold, the recursion depth is increased by two levels, and the voxel size is reduced to one-quarter of the base size. This ensures a higher mesh density and more detailed geometric description in critical contact areas as tool wear intensifies. In this way, the construction module obtains a set of adaptive discretization parameters that are updated in real time. Adaptive discretization parameters are input into the tool-workpiece contact area identification algorithm to calculate the intersection lines between the tool envelope and the workpiece surface. Intersection line extraction identifies the boundaries of the actual contact areas between the tool and workpiece, and a localized, high-density, hierarchical mesh discretization structure is constructed within these critical areas. During the modeling process, the surface is partitioned using a quadtree, with the hierarchical distribution adaptively adjusted based on local curvature and wear status. The volumetric region is then partitioned using a voxel grid, with the voxel refinement level changing synchronously with wear-sensing parameters to ensure modeling accuracy and analytical consistency within the three-dimensional contact area. This hierarchical meshing approach enables high-resolution modeling in critical machining areas while maintaining a lower mesh density in non-critical areas, effectively balancing simulation accuracy and computational resource consumption. Geometric error accumulation control is implemented within the hierarchical mesh discretization structure. By monitoring the geometric error trends between mesh nodes in real time, it is determined whether a preset geometric simulation error limit has been reached. When the accumulated error exceeds the threshold, a re-meshing process is triggered to automatically reconstruct the local mesh structure, thus avoiding error accumulation and accuracy degradation caused by long simulation times. Through dynamic monitoring of geometric errors and adaptive mesh reconstruction, the construction module can continuously maintain the high precision of the simulation model, and update the mesh density and distribution in real time according to the evolution of tool wear and changes in working conditions during the machining process, ultimately generating an adaptive geometric mesh model that conforms to the characteristics of the current working conditions.

[0073] An adaptive refinement module 004 is configured to adaptively refine the contact area for cutting force monitoring based on the physical simulation accuracy level and the adaptive geometric mesh model to obtain a tool-workpiece contact physical simulation model;

[0074] Specifically, the adaptive refinement module refines the tool-workpiece contact area based on the current physical simulation accuracy level and the adaptive geometric mesh model generated by the front-end construction module. Specifically, it extracts the contact state parameters between the tool and workpiece using a contact state enhancement vector construction algorithm. This enhancement vector contains four key components: normal contact force, tangential friction force, contact area, and contact pressure. Through real-time data acquisition and physical field measurement techniques, it captures the mechanical characteristics of the tool-workpiece contact process and constructs an enhancement vector that reflects the dynamic changes in the contact state. Based on the tool-workpiece contact state enhancement vector, the cutting force monitoring values ​​are calculated according to the contact area accuracy classification standard. The working condition complexity of each contact area is determined by analyzing the variation in normal force and tangential friction force, the expansion trend of the contact area, and the local peak of the contact pressure. Based on the complexity classification standard, the contact area is divided into multiple accuracy level intervals, such as low-load area, stable cutting area, and high-stress concentration area. Different intervals correspond to different mesh refinement factors. The mesh refinement factor is determined based on a dual approach of prioritizing accuracy and optimizing resources. Typically, low-load areas maintain the base mesh density, while high-stress concentration areas are refined to eight times the base mesh density. This results in a regional-level precision refinement parameter, which integrates the target refinement factor and the corresponding precision weight for each contact sub-region. Based on this regional-level precision refinement parameter, real-time working condition identification and triggering are performed. By analyzing the dynamic trends of the contact state enhancement vector and the rate of change of the cutting force monitoring value, the triggering conditions for precision switching are quickly identified. When the cutting force change rate exceeds a set threshold (e.g., 50 N / s) or a sudden change in contact pressure is detected, a precision switching control signal is automatically generated, which in turn generates a cutting force adjustment command. This command includes a mesh refinement start signal, a refinement factor update command, and a local region-prioritized encryption flag. This ensures that the precision adjustment process responds quickly to sudden working condition changes, avoiding loss of control or quality degradation due to simulation delays. Based on the cutting force adjustment command, a radial basis function interpolation algorithm is initiated to continuously process the contact force field. The sparsely sampled contact force data points are spatially interpolated to generate a continuous and smooth force field distribution. The interpolation error is kept within 5%, ensuring good spatial resolution and numerical stability of the physical simulation within the contact area. After the contact force field interpolation is completed, the module corrects the physical simulation accuracy by adjusting the local density of the physical simulation grid and the simulation step size based on the new force field distribution, ensuring high consistency between the simulation accuracy and the actual contact state. This results in a physical simulation model of tool-workpiece contact.

[0075] The collaborative adjustment module 005 is used to input the adaptive geometric grid model and the tool-workpiece contact physical simulation model into the working condition conversion state machine for collaborative adjustment to obtain an adaptive precision simulation result.

[0076] Specifically, the collaborative adjustment module inputs the adaptive geometric mesh model and the tool-workpiece contact physical simulation model into the working condition transition state machine. The state machine establishes a dynamic transition diagram for the machining state based on different working condition characteristics during the machining process, such as tool wear growth rate, cutting force fluctuation amplitude, feed rate stability, and spindle vibration characteristics. This constructs a complete working condition transition state machine control structure, including initialization, roughing, finishing, complex surface machining, and abnormal machining conditions. This control structure defines transition conditions and trigger thresholds between each state, enabling real-time identification and response to working condition changes. In particular, abnormal machining conditions, such as abnormal cutting force growth or a sharp deterioration in surface roughness, can be promptly captured and switched to the abnormal handling state, ensuring that the simulation model always reflects the real machining environment. Based on the working condition transition state machine control structure, a nested precision threshold adjustment mechanism is constructed, and the parameters of the PID structure feedback controller are set. The nested precision adjustment mechanism adopts a two-layer structure: the outer layer controls the overall precision distribution ratio between geometric and physical simulations, while the inner layer refines the precision details within each simulation module. By setting upper and lower limits for the outer and inner precision thresholds, the simulation accuracy is dynamically adjusted to adapt to changing working conditions. The feedback controller adopts a PID structure, with a proportional coefficient K set. p =0.8, integral coefficient K i =0.1, differential coefficient K d =0.05, and closed-loop control of simulation accuracy is achieved through real-time monitoring and adjustment of the precision deviation e(t). The nested adjustment feedback loop responds quickly to sudden changes in working conditions and adjusts the simulation accuracy configuration to avoid the accumulation of simulation deviations caused by system inertia or lag, thereby ensuring a high degree of consistency between the simulation results and the actual processing status. According to the control requirements of the nested precision adjustment feedback loop, real-time monitoring of the current simulation computing resource requirements is performed, and the total resource limit is judged to determine whether the current resource consumption exceeds the set resource upper limit R maxThe module then calculates resource competition scheduling and allocation based on the weights of machining quality impact. The module assigns weights to machining areas based on their importance, such as assigning a weight of 0.6 to the critical dimension machining area, 0.3 to the general surface machining area, and 0.1 to the non-contact area. Based on these weights, the simulation accuracy of critical areas is prioritized, limited computing resources are rationally allocated, and a resource competition scheduling and allocation plan is formed. Combined with the resource competition scheduling and allocation plan, the precision control system convergence verification and simulation consistency check are performed. Convergence verification, based on the Lyapunov stability criterion, ensures that the simulation system remains asymptotically stable under different operating conditions, with the simulation error approaching a stable value over time and meeting the engineering requirement of ||ε(∞)|| < 0.001. Simulation consistency check compares simulation results with actual machining monitoring data to verify that the deviation between simulation predictions and actual operating conditions is within an acceptable range, thereby ensuring the authenticity and reliability of the simulation system. Through this dynamic modeling and optimization control process, the collaborative adjustment module outputs high-precision, high-stability adaptive precision simulation results.

[0077] Optionally, the data acquisition module 001 further includes:

[0078] A setting unit is used to set a discrete sampling period for the CNC machine tool to obtain a discrete sampling time sequence;

[0079] a synchronization unit for performing synchronous data acquisition and timestamp association on tool wear state parameters, cutting force change parameters, spindle speed fluctuation parameters, feed speed change parameters, and machined surface roughness parameters according to the discrete sampling time sequence to obtain an original working condition data set;

[0080] a vector construction unit, configured to perform normalization based on the original operating condition data set to obtain a normalized numerical sequence, and input the normalized numerical sequence into a multidimensional feature vector construction algorithm for vectorized combination and weight distribution calculation to obtain a standardized operating condition feature vector;

[0081] The weight allocation unit is used to perform dynamic precision weight allocation of the bounded uncertain diagonal matrix based on the standardized working condition eigenvector to obtain a dynamic precision allocation matrix.

[0082] Specifically, the setting unit standardizes the sampling period for CNC machine tool operating condition data. By presetting a high-frequency discrete sampling period in the control system, such as T1 set to 1 millisecond, the sampling frequency ensures that the machine tool can meet the dynamic response requirements under variable operating conditions such as high-speed cutting and complex surface machining. Using a fixed discrete sampling period, the setting unit constructs a stable, continuous discrete sampling sequence and uses this discrete sequence as a unified time reference. Guided by the discrete sampling sequence output by the setting unit, the synchronization unit initiates synchronized data collection for tool wear parameters, cutting force variation parameters, spindle speed fluctuation parameters, feed rate variation parameters, and machined surface roughness parameters. To ensure temporal consistency between the collected data, the synchronization unit appends a unified timestamp to the collected multi-source data at each sampling moment, achieving precise alignment of multiple physical quantities on the time axis. Tool wear changes are measured in real time using a high-resolution displacement sensor, cutting force changes are recorded in real time using a three-force sensor array, spindle speed fluctuations are captured using a high-frequency vibration accelerometer to capture speed perturbations, feed rate variations rely on high-precision encoder signals, and surface roughness is measured using real-time scanning using an online optical measurement device. The synchronization unit encapsulates these original working condition data sets under the same time base to generate a set of original working condition data with a unified timestamp identifier. The vector construction unit normalizes each working condition data and standardizes all data to the [0,1] interval. Normalization uses linear scaling or maximum and minimum value normalization formulas to perform linear mapping based on the historical maximum and minimum values ​​of each working condition parameter to ensure that different physical quantities are numerically comparable and uniform. After the normalized numerical sequence is completed, the vector construction unit inputs these standardized data into the multidimensional feature vector construction algorithm for vectorized combination, and combines the tool wear normalization value, cutting force normalization value, speed fluctuation normalization value, feed speed normalization value and surface roughness normalization value in a fixed order into a working condition feature vector W=[w1, w2, w3, w4, w5]. During the vectorization process, the vector construction unit introduces a weight distribution calculation mechanism. According to the relative importance of each working condition parameter in the processing quality control and simulation accuracy control, the corresponding characteristic weight coefficient is set. For example, tool wear and cutting force are given higher weights, while feed rate and surface roughness are given secondary weights, and a standardized working condition characteristic vector with weight calibration and normalization characteristics is obtained. The weight distribution unit performs dynamic precision weight distribution of bounded uncertain diagonal matrices based on the standardized working condition characteristic vector W. The weight distribution unit initializes an n×n diagonal matrix P(t), where n is the dimension of the characteristic vector and the diagonal element p is the dimension of the characteristic vector. ii(t) represents the precision allocation weight of the simulation module corresponding to the i-th working condition feature. In order to ensure the numerical stability and computational efficiency of the matrix, the matrix is ​​set to a strict diagonal structure, and each diagonal element is limited to the constraint interval of [0.1, 1.0] to prevent the precision weight from being too low to cause simulation distortion or too high to cause resource waste. The weight allocation unit dynamically adjusts the corresponding p according to the changing trend of each eigenvalue in the standardized working condition feature vector. ii (t). For example, when tool wear is significant, the corresponding precision weight is automatically increased to enhance the simulation's resolution in wear-sensitive areas. When feed rate changes are small, the weight of the corresponding module is appropriately reduced to conserve computing resources. Furthermore, the weight allocation unit incorporates a working condition transition detection mechanism, continuously calculating the Euclidean distance ||W(t) - W(t-1)|| between feature vectors in adjacent time steps. When this distance exceeds a preset threshold δ, precision redistribution is triggered, further adjusting the weight matrix to account for significant changes in machining conditions, ensuring that the simulation precision configuration evolves synchronously with the actual working conditions. To avoid simulation instability caused by sudden weight changes, the weight allocation unit employs a progressive weight update strategy during dynamic adjustment. A buffer coefficient is set for each weight update to control the rate of weight change, ensuring that the system maintains numerical stability while tracking working condition changes. The final output is the dynamic precision allocation matrix P(t).

[0083] Optionally, the weight distribution unit further includes:

[0084] An initialization subunit, configured to initialize a bounded uncertain diagonal matrix structure and determine the number and dimension of simulation modules based on the standardized operating condition eigenvector to obtain an initial precision allocation matrix;

[0085] A constraint condition calculation subunit is used to perform bounded constraint condition calculation and dynamic weight allocation on the precision weights of each simulation module according to the initial precision allocation matrix to obtain a constraint precision weight sequence;

[0086] A threshold comparison subunit is used to input the standardized working condition feature vector into the working condition conversion detection algorithm to perform Euclidean distance calculation and threshold comparison at adjacent moments to obtain a working condition conversion trigger signal;

[0087] The matrix updating subunit is used to perform precision redistribution and matrix updating based on the working condition conversion trigger signal and the constraint precision weight sequence to obtain a dynamic precision allocation matrix.

[0088] Specifically, the initialization subunit determines the number of simulation modules according to the dimension n of the feature vector. Each working condition characteristic quantity corresponds to an independent simulation module, so the simulation module dimension is directly equal to the length of the feature vector. Based on this, the initialization subunit constructs an n×n bounded uncertain diagonal matrix structure, which has non-zero elements only in the main diagonal position. Each diagonal element p ii (t) represents the accuracy allocation weight of the i-th simulation module in the initial state. In order to ensure the numerical stability of the system and the rationality of the simulation, the initial diagonal elements are set within a reasonable value range, for example, the initial value is set to 0.5, and the upper and lower limit constraints are set, taking 0.1≤p ii (t)≤1.0, which not only avoids the simulation distortion caused by too low initial precision, but also prevents the unnecessary waste of computing resources caused by too high initial setting. Through this diagonal initialization method, an initial precision allocation matrix with a stable numerical basis and the ability to adapt to the initial working conditions is obtained. The constraint calculation subunit performs more refined dynamic weight allocation and constraint setting based on the initial precision allocation matrix. The constraint calculation subunit performs more precise dynamic weight allocation and constraint setting for each diagonal element p ii(t) Calculate bounded constraints to ensure their values ​​are within a preset range. A dynamic adjustment mechanism based on standardized working condition feature vectors is also introduced. For example, if a characteristic shows a rapid growth trend within a specific sampling period, indicating that the working condition characteristic has an increasing impact on the stability of the machining process, the constraint calculation subunit will increase the accuracy weight of the corresponding simulation module accordingly. Conversely, if the change in a working condition characteristic is small, the corresponding module weight is appropriately reduced to free up computing resources. Through dynamic weight adjustment based on the real-time working condition status, the constraint calculation subunit generates a sequence of constraint accuracy weights. Each element is range-limited and working condition sensitivity-adjusted to ensure that the simulation accuracy distribution meets the current machining state requirements while reasonably balancing the system's computational load. The threshold comparison subunit detects working condition changes in real time and generates trigger signals. The standardized working condition feature vectors collected at two consecutive time steps are input into the working condition transition detection algorithm. The Euclidean distance ||W(t)−W(t−1)|| between the two is calculated. This distance quantifies the magnitude of the change between the current working condition and the previous state. The threshold comparison subunit sets a sensitivity threshold δ, approximately 0.15. If the calculated Euclidean distance exceeds this threshold, it indicates that the machine tool's machining state has changed significantly, posing potential risks such as a sudden change in cutting state, a sharp increase in tool wear, and abnormally increased spindle vibration. At this point, the threshold comparison subunit generates a working condition transition trigger signal, notifying the system to update and adjust the current simulation accuracy weight configuration. This threshold comparison mechanism, based on the magnitude of working condition change, enables rapid response to dynamic changes in the machining process, avoids simulation accuracy lags caused by sudden changes in working conditions, and improves the system's sensitivity and adaptability to abnormal working conditions. Upon receiving the working condition transition trigger signal, the matrix update subunit executes the precision reallocation and matrix update process. The matrix update subunit recalculates the weights based on the current constraint accuracy weight sequence and the working condition change trend. The weight recalculation adopts an adaptive update strategy, combining the weight value at the previous moment, the direction and magnitude of change of the current working condition characteristics, and smoothly updating the precision weights of each simulation module in accordance with the weighted average or weighted incremental adjustment method to avoid system instability caused by weight mutations. The updated weight sequence needs to meet the numerical range constraints and maintain a high degree of consistency with the current working condition change trend to ensure that the simulation model can follow the changes in the processing state in real time. The matrix update subunit replaces the diagonal elements of the original initial precision allocation matrix with the updated precision weight sequence, keeping the diagonal structure of the matrix unchanged, and completing the reconstruction process of the dynamic precision allocation matrix. After the matrix update is completed, the matrix update subunit verifies the numerical stability and boundary conditions of the new dynamic precision allocation matrix.Numerical stability verification checks the matrix's spectral radius, condition number, and other indicators to ensure that the matrix will not cause divergence or instability in the simulation system during numerical operations. Boundary condition verification checks that each diagonal element remains within the set range of 0.1 to 1.0, preventing weight out-of-bounds issues caused by abnormal data or numerical overflow during the update process. Through numerical verification and constraint verification, the dynamic precision allocation matrix is ​​guaranteed to have good numerical stability and physical rationality after the update.

[0089] Optionally, the constraint condition calculation subunit is specifically used to:

[0090] Setting bounded constraints on precision weights and limiting the value range of the diagonal elements of the initial precision allocation matrix to obtain precision weight constraint intervals;

[0091] Quantifying the basic computational overhead and evaluating the computational resource requirements of each simulation module based on the precision weight constraint interval to obtain a basic computational overhead coefficient;

[0092] Perform computing resource constraint mapping function calculation and total resource limit judgment based on the basic computing overhead coefficient and the precision weight constraint interval to obtain computing resource constraint conditions;

[0093] Based on the computing resource constraint conditions, precision weight optimization calculation and serialized output are performed to obtain a constrained precision weight sequence.

[0094] Specifically, the diagonal elements of the initial precision allocation matrix P(t) are subjected to bounded constraints on the precision weights. By setting each precision weight p ii The upper and lower limits of (t), such as 0.1 ≤ p ii (t) ≤ 1.0. The lower limit of 0.1 ensures that the simulation module still maintains basic accuracy even when resources are extremely tight, preventing the simulation quality from falling to an unacceptable level; the upper limit of 1.0 prevents the weight from being set too large, resulting in excessive concentration of resource allocation in a single module, which destroys the overall balance of the system. By imposing bounded constraints on the diagonal elements, an accuracy weight constraint interval is formed. Based on the accuracy weight constraint interval of each simulation module, combined with the computational complexity characteristics of the module itself, its basic computational overhead is quantified. The basic computational overhead refers to the computational resources required for the simulation module to execute a unit simulation step under standard accuracy conditions, which is related to factors such as the amount of data processed by the module, the complexity of numerical calculations, and the number of iterations. By performing statistical analysis on historical simulation data or estimating based on theoretical models, the system can assign a basic computational overhead coefficient C to each module. i, reflecting its resource consumption trend at different precision levels. High-precision modules have a larger computational overhead coefficient, while low-precision modules have a smaller computational overhead. Combined with the weight interval of each module, the minimum and maximum possible computational overhead prediction values ​​within the precision adjustment range are obtained. Based on the quantized basic computational overhead coefficient and the precision weight constraint interval, the computational resource constraint mapping function is calculated. This mapping function uses the current precision weight p of each module as the ii (t) and the corresponding basic calculation overhead coefficient C i The product of is used to represent the computing resource requirements of a single module. The resource requirements of all simulation modules are summed up to get the total computing resource consumption of the system at the current time step. At the same time, the system has a global total resource limit R max , represents the maximum amount of computing resources available to all simulation modules. By comparing R(t) with R max Compare and determine whether the current precision allocation scheme meets the resource constraints. If R(t) ≤ R max , indicating that the current resource configuration is feasible; on the contrary, if R(t)>R max If the current precision setting is too high, exceeding the system's resource load, further precision weight optimization is required to ensure resource allocation remains within constraints and prevent computational resource overflow, which could lead to system performance degradation or simulation failure. Precision weight optimization is performed based on computational resource constraints. Simulation modules are prioritized based on their machining quality sensitivity and resource consumption efficiency, prioritizing modules with the greatest impact on machining accuracy and surface quality to maintain high precision while appropriately reducing the precision weights of modules with less impact. The optimization algorithm utilizes linear programming, greedy strategies, or heuristic search methods to minimize total resource consumption while maintaining overall simulation quality. During the optimization process, each precision weight adjustment is performed within a pre-set constraint range to prevent adjustments from exceeding reasonable boundaries. Furthermore, the optimization process must consider the numerical stability of the simulation system to avoid difficulties in simulation convergence or inconsistent results caused by discontinuous changes in precision allocation. After the precision weight optimization calculation is completed, the optimized precision weights are serialized and output by module to form the final constrained precision weight sequence. The serialized output format is a one-dimensional vector, where each element corresponds to the final accuracy weight of a simulation module, making it convenient for subsequent modules to directly call and load new weight configurations, driving real-time updates of simulation accuracy.

[0095] Optionally, the matrix updating subunit is specifically configured to:

[0096] Based on the working condition conversion trigger signal, trigger state judgment and precision reallocation requirement determination are performed to obtain a precision adjustment instruction;

[0097] Recalculate the weights and update the values ​​of the constraint precision weight sequence according to the precision adjustment instruction to obtain an updated precision weight sequence;

[0098] Performing diagonal element replacement and matrix structure reconstruction based on the update precision weight sequence to obtain an update precision allocation matrix;

[0099] Numerical stability verification and bounded constraint condition verification are performed on the updated precision allocation matrix to obtain a dynamic precision allocation matrix.

[0100] Specifically, when the system receives a working condition change trigger signal generated by the threshold comparison subunit, the matrix update subunit initiates the trigger state judgment mechanism, performing a comprehensive analysis based on the magnitude and direction of changes in the working condition characteristic vector. The goal of this trigger state judgment is to determine whether the working condition has changed to a level that requires adjustment to simulation accuracy. By comparing the Euclidean distance corresponding to the trigger signal with a preset sensitivity threshold δ and combining it with working condition change characteristics such as the growth rate of tool wear, cutting force trends, and spindle speed fluctuations, the system determines whether a sudden event or nonlinear change trend has occurred during the machining process. If the judgment indicates a significant working condition change with a potential impact on simulation accuracy, the system determines that precision reallocation is necessary and generates a precision adjustment instruction. This instruction specifies the simulation module to be adjusted and includes the adjustment magnitude, priority order, and resource constraints. Based on the generated precision adjustment instruction, the current constraint precision weight sequence is recalculated and updated. During the weight recalculation process, a dynamic adjustment algorithm is introduced, using a weighted incremental adjustment mechanism or an exponential sliding average method. The precision weight of each simulation module is carefully adjusted, taking into account the magnitude of the working condition change and the module's impact on machining quality. For working condition characteristics that change dramatically and significantly impact machining quality, such as a sharp increase in tool wear or drastic fluctuations in cutting forces, the precision weights of the corresponding simulation modules should be appropriately increased to improve local simulation accuracy and ensure accurate modeling and simulation prediction of the machining process. For characteristics with smaller changes and limited impact on the overall machining process, the precision weights of the corresponding modules should be appropriately reduced to free up some computing resources, thereby optimizing the overall system resource allocation structure. During the weight update process, the system monitors in real time whether each weight value remains within the bounded interval [0.1, 1.0]. If the adjusted weight exceeds this bound, a clipping correction is performed to ensure that the updated weight sequence meets the numerical constraints. Through the above weight recalculation and update process, an updated precision weight sequence that dynamically adapts to the current working condition is obtained. Based on the updated precision weight sequence, diagonal element replacement and matrix structure reconstruction are performed. The diagonal elements of the initial precision allocation matrix P(t) are sequentially replaced with the updated precision weight values, maintaining the strictly diagonal matrix structure with off-diagonal elements zero, thereby generating a new updated precision allocation matrix. The updated precision allocation matrix is ​​verified for numerical stability and boundedness constraints. Numerical stability verification calculates the matrix's spectral radius and condition number to ensure that the matrix will not cause simulation convergence problems or numerical divergence during numerical operations. If the matrix condition number is detected to be too large or the spectral radius approaches or exceeds the stability boundary, further adjustments to the weight distribution are required to smooth the fluctuations in the precision weights, reduce the system's numerical sensitivity, and enhance the robustness of the simulation operations. At the same time, bounded constraint conditions are checked for each diagonal element of the matrix to ensure that all weight values ​​are strictly within the preset range. This prevents weight loss due to numerical drift or update anomalies, which could damage the simulation accuracy and the rationality of resource allocation.After completing numerical stability verification and bounded constraint checking, the system finally outputs a dynamic precision allocation matrix, which has the ability to dynamically update to adapt to changes in working conditions in real time, and maintains the optimal balance between simulation accuracy and computational efficiency under the conditions of numerical stability and resource constraints.

[0101] Optionally, the calculation module 002 further includes:

[0102] An establishing unit is used to construct a working condition characteristic accuracy requirement mapping function and establish a geometric simulation and physical simulation coupling relationship based on the dynamic accuracy allocation matrix to obtain a geometric and physical simulation working condition accuracy mapping function;

[0103] a numerical value allocation unit, configured to perform normalized constraint calculation and numerical value allocation on the geometric simulation dynamic weight factor and the physical simulation dynamic weight factor according to the geometric and physical simulation working condition accuracy mapping function, to obtain a geometric and physical simulation dynamic weight factor combination;

[0104] A stage discrimination unit is used to input the geometric physical simulation dynamic weight factor combination into a processing stage identification algorithm to perform processing stage discrimination and weight adaptive adjustment to obtain a processing stage adjustment weight factor combination;

[0105] The grade calculation unit is used to calculate the geometric simulation accuracy grade and the physical simulation accuracy grade based on the weight factor combination adjusted in the processing stage to obtain the geometric simulation accuracy grade and the physical simulation accuracy grade.

[0106] Specifically, the establishment unit constructs a working condition feature precision requirement mapping function based on the precision weight allocation information corresponding to each simulation module in the dynamic precision allocation matrix P(t). By analyzing the correlation between the diagonal elements in the dynamic precision allocation matrix and the working condition feature vectors, the establishment unit extracts the sensitivity parameters of key working condition features, such as tool wear, cutting force fluctuation, spindle speed variation, feed rate variation, and surface roughness, to the simulation precision requirement. Taking tool wear as an example, the establishment unit determines its dynamic impact coefficient on simulation accuracy based on its weight fluctuation trend in the dynamic precision allocation matrix. This mathematically maps the physical working condition features with the simulation precision requirement, forming a working condition feature precision requirement mapping function. To enhance the integrity and coordination of the simulation model, the establishment unit simultaneously establishes a coupling relationship between geometric and physical simulations, clarifying the degree of mutual dependence and precision weight allocation ratio under the influence of different working condition features, and forming a geometric-physical simulation working condition precision mapping function. After obtaining the geometric-physical simulation working condition precision mapping function, the numerical allocation unit initiates the extraction and normalization constraint calculation of the dynamic weight factors of the geometric and physical simulations. According to the accuracy requirement value output by the mapping function, the dynamic weight factor α required for geometric simulation is extracted G(t) The dynamic weight factor α required for physical simulation P(t), and then normalize both to ensure that α G(t) +α P(t) =1, maintaining the consistency and controllability of simulation resources and precision allocation. During the normalization process, the numerical allocation unit introduces constraint boundaries to prevent the weight of a single simulation mode from being too high or too low due to excessive changes in a certain working condition feature, thereby destroying the overall performance and numerical stability of the simulation. Through normalized constraint calculation, a combination of geometric and physical simulation dynamic weight factors is obtained. The stage discrimination unit inputs the geometric and physical simulation dynamic weight factor combination output by the numerical allocation unit into the processing stage identification algorithm to perform processing stage discrimination and weight adaptive adjustment. The processing stage identification algorithm is based on the changing trend of working condition features and processing process characteristics. By analyzing parameters such as tool wear rate, cutting force fluctuation amplitude, spindle vibration change rate, feed speed fluctuation stability and surface roughness requirements, it comprehensively judges the specific stage of the current processing process, such as rough machining stage, semi-finishing stage, finishing stage or complex surface machining stage. According to the process requirements and precision requirements of different stages, the stage discrimination unit adaptively adjusts the geometric and physical simulation dynamic weight factor combination after preliminary normalization. For example, in the rough machining stage, the geometric simulation weight α is increased. G(t) To 0.7, reduce the physical simulation weight α P(t) To 0.3, in order to enhance the shape capture and surface forming quality; while in the finishing stage, the physical simulation weight is appropriately increased to 0.6 to enhance the simulation accuracy control of the contact force field and surface details. Through the weight adaptive adjustment mechanism based on the judgment of the processing stage, it is ensured that the simulation system can reasonably allocate simulation accuracy and computing resources in different processing stages, realize intelligent response and precise adaptation to variable processing conditions, and form a combination of weight factors for adjustment in the processing stage. The level calculation unit performs dynamic calculation of the geometric simulation accuracy level and the physical simulation accuracy level based on the combination of weight factors for adjustment in the processing stage. According to the preset accuracy level division standard, the weight factor value is mapped to a specific level interval. The accuracy level is set in an integer range of 1 to 10. The higher the level value, the higher the simulation accuracy and the greater the computing resource demand. The level calculation unit uses a linear mapping or interval segmented mapping algorithm to convert the current geometric simulation weight α G(t) and physical simulation weight α P(t) Mapped to the corresponding geometric simulation accuracy level P G(t) Compared with the physical simulation accuracy level P P(t)For example, when αG(t) = 0.7, the corresponding geometric simulation accuracy level is 7, and when αP(t) = 0.3, the corresponding physical simulation accuracy level is 3. To ensure the rationality of accuracy level allocation and the numerical stability of the system, the level calculation unit introduces a level smoothing mechanism to avoid frequent accuracy level jumps due to minor fluctuations in operating conditions. The geometric simulation accuracy level and physical simulation accuracy level output by the level calculation unit serve as direct control parameters for subsequent simulation processes such as adaptive geometric meshing, contact area mesh refinement, and force field interpolation modeling, driving the system to achieve accurate, stable, and efficient simulation modeling and dynamic control under different operating conditions.

[0107] Optionally, the value allocating unit is specifically configured to:

[0108] Performing geometric simulation dynamic weight factor extraction and physical simulation dynamic weight factor extraction based on the geometric and physical simulation working condition accuracy mapping function to obtain geometric simulation weight factor values ​​and physical simulation weight factor values;

[0109] Performing sum constraint calculation and normalization verification on the geometric simulation weight factor value and the physical simulation weight factor value to obtain a normalized constraint condition;

[0110] Performing processing discrimination and weight value distribution calculation on the characteristic parameters of the processing stage according to the normalized constraint conditions to obtain a weight factor distribution value;

[0111] Based on the weight factor allocation values, geometric physics simulation weight pairing and numerical verification are performed to obtain a geometric physics simulation dynamic weight factor combination.

[0112] Specifically, after dynamically monitoring the current working condition characteristics of the machine tool and constructing a geometric-physical simulation working condition accuracy mapping function based on the working condition characteristic vector, the dynamic weight factors of the geometric simulation and the physical simulation are extracted. This mapping function takes working condition characteristic quantities such as tool wear status, cutting force fluctuation amplitude, spindle speed variation trend, feed rate stability, and machined surface roughness variation as input. By analyzing the contribution of these characteristics to the accuracy requirements of geometric and physical simulations, it outputs preliminary geometric simulation weight factors and physical simulation weight factors, respectively. The geometric simulation weight factor values ​​and the physical simulation weight factor values ​​are subjected to sum value constraint calculation and normalization verification. The two weight factor values ​​are added to obtain the total weight sum, and then the respective weight factors are normalized to ensure that the sum of the weight factors of the two simulation modes is always 1. This also prevents the absolute value of the weight factor from being directly affected by the amplitude of the working condition characteristic variation, thereby improving the stability of the numerical calculation and the robustness of the simulation system. During the normalization process, the system sets upper and lower bounds, for example, ensuring that both factor values ​​are between [0.1, 0.9]. This prevents extreme operating conditions from causing a simulation mode's weight to be too low or too high, thereby impacting overall simulation performance. Through sum constraints and normalization verification, normalized constraints that meet numerical stability requirements and physical plausibility are obtained. Based on these constraints, the system combines the characteristic parameters of each processing stage to identify the processing stage and calculate weight distribution. The system analyzes the relationship between the changing trends and thresholds of various indicators in the current working condition characteristic vector to identify the specific stage of the machine tool processing process. If the tool wear rate is low, the cutting force fluctuation is small, the spindle speed changes steadily, and the surface roughness index does not deteriorate significantly, the process is identified as the roughing stage. If tool wear increases, cutting force fluctuations increase, feed rate fluctuations become more severe, and surface roughness requirements increase, the process is identified as the finishing stage. When surface complexity increases and the local curvature of the machined surface changes dramatically, the process enters the complex surface processing stage. Based on the identification results, the system dynamically adjusts the weight factor distribution ratio based on the different simulation accuracy requirements of each processing stage. For example, during the rough machining phase, the system tends to increase the geometric simulation weight factor to 0.7 and reduce the physical simulation weight factor to 0.3 to enhance the simulation accuracy of large-scale shape changes; while during the fine machining phase, the physical simulation weight factor is increased to 0.6 and the geometric simulation weight factor is reduced to 0.4 to enhance the simulation control of the contact force field and surface microscopic characteristics. By combining the working condition characteristics and the processing stage identification results, the system can adaptively complete the weight factor numerical allocation calculation and obtain the weight factor allocation value that is highly adapted to the current processing environment, ensuring the optimal configuration of simulation accuracy and computing resources under different working conditions. Based on the weight factor allocation value, the geometric simulation and physical simulation weights are paired and numerically verified. The geometric simulation weight factor and the physical simulation weight factor after numerical allocation are combined into a complete dynamic weight factor pair, and the pairing results are verified for consistency and stability.The system checks whether the combined weight factors still meet the normalization condition of a sum of 1 and are within the set range. It also compares the current pairing result with the previous pairing result to evaluate the rate of change of the weight factors, ensuring smooth changes in the weight factors and avoiding frequent and drastic jumps that could cause simulation system oscillation or decreased convergence. If the weight change exceeds the set threshold, the system automatically introduces an adjustment factor to smooth the weights, such as using an exponentially smoothed weighted moving average to reduce the risk of sudden changes in weights. After normalization verification and numerical smoothing, the system outputs a stable and reliable combination of dynamic weight factors for geometric and physical simulations.

[0113] Optionally, the construction module 003 is specifically used to:

[0114] Based on the geometric simulation accuracy level, a quadtree voxel hybrid discretization strategy is constructed and tool wear perception parameters are set to obtain a strategy parameter combination;

[0115] According to the combination of the strategy parameters, a threshold value judgment and a quadtree recursive depth calculation and a voxel size determination are performed on the tool wear amount to obtain an adaptive discretization parameter;

[0116] Inputting the adaptive discretization parameters into the tool-workpiece contact area identification algorithm to perform envelope surface intersection calculation and area modeling to obtain a hierarchical grid discretization structure;

[0117] Geometric error accumulation control and grid re-division triggering judgment are performed on the hierarchical grid discretization structure to obtain an adaptive geometric grid model.

[0118] Specifically, after obtaining the geometric simulation accuracy level, the system starts to build a quadtree-voxel hybrid discretization strategy. The quadtree-voxel hybrid discretization strategy is based on hierarchical refinement and spatial partitioning. It recursively divides the two-dimensional geometric features of the workpiece surface through the quadtree data structure, and uniformly discretizes the three-dimensional volume features inside the workpiece through the voxel grid. The system sets the maximum depth D of the quadtree recursion according to the current geometric simulation accuracy level. max And the basic voxel unit size L base , where D max It is positively correlated with the precision level PG. The higher the PG, the higher the D max The deeper the setting, the baseThe smaller the setting, the more it can meet the requirements of high-precision modeling. At the same time, the system introduces a tool wear perception parameter setting module, which conducts perception modeling for the tool wear amount w1 monitored in real time, and uses the wear amount as an important basis for dynamically adjusting the grid density. By comprehensively considering the geometric simulation accuracy level and tool wear perception parameters, the system forms a complete set of discrete strategy parameter combinations, covering the maximum recursive depth of the quadtree, the basic voxel size, the wear threshold setting, and the local grid encryption conditions. According to the above strategy parameter combination, a threshold judgment is made on the tool wear amount collected in real time, and the quadtree recursive depth and voxel unit size are dynamically determined based on this. The system presets a wear amount threshold. For example, when w1 < 0.1mm, the quadtree recursive depth is set to D1, and the voxel size is set to L1, ensuring a medium grid density under normal wear conditions and balancing the simulation accuracy and computational cost; when w1 ≥ 0.1mm, the tool enters the severe wear stage, and the system automatically increases the quadtree recursive depth to D2 (D2 > D1), and reduces the voxel size to L2 (L2 < L1) to improve the grid accuracy in the contact area, carefully capture the geometric profile changes and microscopic surface defects caused by tool wear, and ensure the continuity of high-precision simulation modeling and the accuracy of local detail description. By dynamically adjusting the recursive depth and voxel size based on the wear amount in this way, the system realizes the generation of adaptive discretization parameters, ensuring that the grid division strategy can be automatically optimized as the tool wear evolves during the machining process. The adaptive discretization parameters are input into the tool-workpiece contact area recognition algorithm to calculate the intersection line of the workpiece surface envelope and the tool motion trajectory. Based on the current tool trajectory and the workpiece surface geometric model, an efficient intersection line extraction algorithm, such as the adaptive space division method accelerated by the octree, is used to extract the actual contact boundary between the tool and the workpiece at each moment. Through the intersection line extraction, the system accurately identifies the true contact area between the tool and the workpiece, and conducts local area modeling based on the recognition result. For the identified contact area, the system performs hierarchical grid discretization according to the wear state and simulation accuracy level: within the contact area, a higher quadtree recursive depth and a smaller voxel size are used for local encryption to ensure that the key contact area has a higher grid resolution and detail capture ability; in the non-contact area, a lower grid density is maintained to save computational resources and shorten the simulation time. The system obtains a hierarchical grid discretization structure. To ensure the simulation accuracy and stability, geometric error accumulation control and grid redivision trigger judgment are performed on the generated hierarchical grid discretization structure. The change trend of geometric errors between grid nodes is monitored in real time, especially paying attention to the error accumulation situation of grid nodes in the contact area due to factors such as tool wear and workpiece morphology changes during continuous machining. The system sets an error threshold, such as 0.001mm. When it is detected that the local or overall geometric error accumulation exceeds this threshold, the grid redivision mechanism is automatically triggered.During the re-meshing process, the system dynamically adjusts the quadtree recursion depth and voxel size based on the current tool wear status and machining stage characteristics, regenerating the mesh locally or globally to ensure that the mesh model remains within an acceptable accuracy range and prevent simulation accuracy degradation or convergence issues caused by mesh degradation. By introducing dynamic error monitoring and triggered mesh reconstruction mechanisms, the system can continuously maintain high-precision simulation modeling during the machining process, significantly improving the reliability and real-time performance of simulation results.

[0119] Optionally, the adaptive refinement module 004 is specifically configured to:

[0120] Based on the physical simulation accuracy level and the adaptive geometric grid model, a tool-workpiece contact state enhancement vector is constructed and parameters of normal contact force, tangential friction force, contact area, and contact pressure are extracted to obtain a tool-workpiece contact state enhancement vector;

[0121] Calculating the contact area accuracy grading standard and determining the mesh refinement multiple for the cutting force monitoring value according to the tool-workpiece contact state enhancement vector to obtain the area grading accuracy refinement parameter;

[0122] Based on the regional classification precision refinement parameters, real-time working condition identification, triggering and judgment and precision switching control are performed to obtain cutting force adjustment instructions;

[0123] Based on the cutting force adjustment instruction, radial basis function contact force field interpolation calculation and physical simulation accuracy correction are performed to obtain a tool-workpiece contact physical simulation model.

[0124] Specifically, based on the physical simulation accuracy level and the current adaptive geometric mesh model, the system initiates the construction process of the tool-workpiece contact state enhancement vector. This contact state enhancement vector is designed as a four-dimensional vector containing four fundamental physical quantities: normal contact force, tangential friction force, contact area, and contact pressure. To accurately extract these quantities, high-frequency dynamic force sensors, surface contact probes, and stress measurement modules embedded in the CNC machine tool collect micromechanical data of the contact area in real time. The normal contact force is calculated from the average force applied by the tool perpendicular to the workpiece surface. The tangential friction force is extracted by measuring the change in tool resistance in the cutting direction. The contact area is calculated based on the number of contact nodes and the cumulative element area in the adaptive geometric mesh. The contact pressure is calculated by dividing the normal contact force by the contact area, reflecting the stress concentration in the local contact area. By extracting these parameters, the constructed contact state enhancement vector can comprehensively describe the mechanical properties of the tool-workpiece contact area and dynamically reflect the contact evolution trend during machining. Based on the tool-workpiece contact state enhancement vector, the contact area accuracy grading standard and mesh refinement factor are calculated based on the real-time monitored cutting force data. The system establishes a multi-level grading standard based on the physical quantities in the enhancement vector. For example, areas with normal contact force below 100 N are defined as level 1 precision, while areas with tangential friction greater than 300 N or contact pressure greater than 50 MPa are defined as level 5 precision. Different precision levels correspond to different mesh refinement factors, typically ranging from 1x to 8x. The mesh refinement factor is determined by prioritizing high pressure, high friction, and high contact area. This ensures a higher mesh density in areas with the most complex physical contact characteristics, thereby improving the local resolution and force field description accuracy of the simulation model in critical areas. The system continuously outputs regional grading precision refinement parameters through real-time updates of the contact state enhancement vector. Based on these grading precision refinement parameters, the system performs real-time working condition identification trigger judgment and precision switching control. Based on the contact state enhancement vector and the cutting force change rate, the real-time working condition identification module uses a combination of threshold judgment and trend analysis to detect typical working condition anomalies during machining, such as sudden contact force changes, drastic changes in contact area, or excessive contact pressure. If an abnormal operating condition is detected, the system automatically generates a precision switching control signal based on preset trigger logic, activating the cutting force adjustment command generation module. The cutting force adjustment command includes control parameters such as the target mesh refinement level, the precision switching direction (refining or coarsening), and the local refinement area range. This ensures that the system can adjust the simulation model in the shortest possible time, rapidly responding to changes in operating conditions and avoiding modeling distortion or simulation convergence issues caused by insufficient simulation accuracy. After the cutting force adjustment command is issued, the system performs contact force field interpolation calculations based on radial basis functions and performs physical simulation accuracy corrections.The system constructs an RBF interpolation network based on the sampled discrete contact force data points, selects high-order continuous basis functions such as Gaussian functions and thin plate spline functions, and automatically adjusts the interpolation parameters according to the actual contact area node distribution and mechanical characteristics to generate a continuous and smooth force field distribution in the contact area. RBF interpolation can compensate for the local interpolation errors caused by the uneven distribution of sampling points, and can also effectively capture the local extreme values ​​and gradient changes implicit in the force field, thereby improving the physical realism and numerical stability of the simulation force field. After completing the contact force field interpolation, the system corrects the physical simulation accuracy based on the latest force field distribution, dynamically adjusts the density and shape function order of the unit grid in the contact area, so that the simulation model can better match the actual contact force distribution characteristics, thereby improving the reliability of the simulation results and the accuracy of engineering predictions. Through dynamic accuracy adjustment and local grid refinement mechanisms driven by working condition characteristics, the system outputs an updated tool-workpiece contact physical simulation model.

[0125] Optionally, the collaborative adjustment module 005 is specifically configured to:

[0126] Inputting the adaptive geometric grid model and the tool-workpiece contact physical simulation model into a working condition conversion state machine to construct a machining abnormality state, thereby obtaining a working condition conversion state machine control structure;

[0127] Based on the working condition conversion state machine control structure, a nested precision threshold adjustment mechanism is constructed and PID structure feedback controller parameters are set to obtain a nested precision adjustment feedback loop;

[0128] Performing a total resource limitation judgment and a processing quality impact weight distribution calculation on computing resource requirements according to the nested precision adjustment feedback loop to obtain a resource competition scheduling allocation plan;

[0129] The resource competition scheduling allocation scheme is combined with the precision control system convergence verification and simulation consistency verification to obtain adaptive precision simulation results.

[0130] Specifically, an adaptive geometric mesh model and a physical simulation model of tool-workpiece contact are fed into a working condition transition state machine to construct a module for identifying and managing abnormal machining states. The working condition transition state machine uses key physical variables during machining as driving signals, such as tool wear rate, cutting force fluctuation amplitude, abnormal contact pressure increase, and abnormal spindle vibration. By monitoring and analyzing these indicators in real time, a set of machining states is established, including normal machining state, roughing stage, finishing stage, abnormal tool wear state, abnormal cutting force fluctuation state, and abnormal deterioration of machined surface quality. The transition conditions and logic between each state are defined. The transition conditions are determined based on multi-feature combined thresholds. For example, if tool wear exceeds a preset threshold and frequent cutting force fluctuations occur simultaneously, an abnormal wear state is identified; if the average cutting force value suddenly changes and spindle vibration increases simultaneously, an abnormal cutting state is identified. Through this state identification and transition mechanism driven by adaptive mesh and physical simulation data, a working condition transition state machine control structure is constructed that reflects the dynamic evolution of machining processes. Based on this working condition transition state machine control structure, a nested precision threshold adjustment mechanism is implemented and the parameters of a PID feedback controller are set, forming a nested precision adjustment feedback loop. The nested precision adjustment mechanism is divided into two levels: the outer layer and the inner layer. The outer layer adjusts the overall simulation precision strategy according to the processing state switch. For example, it automatically improves the physical simulation precision level when switching from rough processing to fine processing. The inner layer makes fine-grained adjustments to parameters such as local grid density, force field interpolation precision, and time step length to ensure smooth transition of simulation local detail precision during working condition changes. In order to achieve dynamic control of simulation precision changes, the system equips each level of the adjustment mechanism with an independent PID feedback controller. By setting the proportional coefficient K p , integral coefficient K i , differential coefficient K d , monitor the deviation between the simulation accuracy error e(t) and the set target accuracy in real time, and dynamically adjust the accuracy parameters based on the feedback control law. The PID controller parameters need to be adaptively set according to different processing states. For example, in the normal processing stage, set K p Small to ensure smooth regulation, and increase K when abnormal conditions are detected p Improve the response speed and adjust K appropriately i With K dThis suppresses system oscillation and over-adjustment. Through this multi-level, dynamically adjustable nested precision feedback control mechanism, the system can finely control simulation accuracy based on real-time changes in machining conditions, ensuring that simulation results evolve synchronously with the actual machining state. Based on the real-time precision adjustment requirements output by the nested precision adjustment feedback loop, the system performs a total resource limit judgment of computing resource requirements and a weighted allocation calculation for machining quality impact, and constructs a resource competition scheduling and allocation scheme. Specifically, the system calculates the resource consumption of the current simulation task and compares it with the total resource upper limit. If the resource consumption of the current simulation task exceeds the total resource upper limit, the resource scheduling mechanism is triggered to reallocate resources. To ensure the rationality of simulation accuracy configuration under limited resources, a machining quality impact weight allocation model is introduced. Key quality indicators such as machining dimensional accuracy, surface roughness, and contact stress distribution uniformity are assigned different weights according to their importance. For example, a weight of 0.5 for dimensional accuracy, 0.3 for surface roughness, and 0.2 for contact stress distribution are assigned to simulation modules that have a greater impact on machining quality. The resource competition scheduling process utilizes a multi-objective optimization algorithm to optimize overall simulation quality metrics while ensuring that total resource consumption remains within limits. This generates a resource competition scheduling plan and specifies parameter configurations for each simulation module, such as the accuracy level, grid density, and time step. In conjunction with this resource competition scheduling plan, the system performs convergence verification of the precision control system and simulation consistency checks. This convergence verification, based on the Lyapunov stability criterion, analyzes the temporal trend of the simulation accuracy error to ensure that the error sequence e(t) gradually converges to a stable range—that is, ||e(t)|| tends to zero or converges to an acceptable range as t increases—to prevent simulation divergence due to error accumulation. Simulation consistency checks evaluate the simulation model's fit to the actual machining process by comparing the simulation predictions with actual machining monitoring data, such as cutting force, tool wear, and surface roughness. The fitting error is required to be below a preset threshold, for example, the error between the simulated and measured cutting forces must not exceed 5%. Through this dual guarantee of convergence verification and consistency checks, the system ensures that the updated adaptive precision simulation model maintains numerical stability while maintaining high reliability and prediction accuracy.

[0131] In an embodiment of the present invention, discrete-time working condition data acquisition and multidimensional feature vector construction enable real-time and accurate capture of key working condition parameters such as tool wear, cutting force variations, and spindle speed fluctuations. Compared to traditional continuous monitoring methods, this approach avoids data redundancy and improves the accuracy and efficiency of working condition identification. A dynamic precision weight allocation mechanism using a bounded, uncertain diagonal matrix adaptively adjusts the precision weights of each simulation module based on the real-time working condition, overcoming the limitations of traditional fixed precision settings and enabling dynamic optimization of precision allocation. A working condition feature-precision requirement mapping function is established to dynamically couple geometric and physical simulations, resolving the technical bottleneck of traditional methods where the two simulation modes are independent and cannot be optimized in a coordinated manner. A quadtree voxel hybrid discretization strategy for tool wear sensing adaptively adjusts mesh density based on wear status, significantly improving the accuracy and efficiency of geometric modeling compared to traditional fixed meshing methods. The contact area adaptive refinement technology for cutting force monitoring achieves high-precision modeling of critical machining areas by constructing contact state enhancement vectors and hierarchical precision control, avoiding the waste of computing resources in non-critical areas. The collaborative adjustment mechanism of the working condition transition state machine and the computing resource competition scheduling strategy dynamically allocate computing resources according to the processing stage and quality requirements, achieving an optimal balance between simulation accuracy and real-time performance within limited edge computing capabilities. Lyapunov stability criterion verification and a nested precision threshold adjustment mechanism ensure the asymptotic stability and simulation consistency convergence of the digital twin system at different processing stages, improving the system's reliability and robustness. The processing stage identification algorithm and weight adaptive adjustment mechanism accurately distinguish different stages such as roughing, finishing, and complex surface processing, and adjust the precision strategy accordingly to meet the needs of diverse processing technologies.

[0132] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned system embodiments and will not be repeated here.

[0133] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the system described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0134] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent monitoring and optimization system for CNC machine tools based on digital twins, characterized in that: include: The data acquisition module is used to collect discrete time working condition data and allocate dynamic precision weights for CNC machine tools to obtain a dynamic precision allocation matrix; A calculation module, configured to calculate a geometric simulation accuracy level and a physical simulation accuracy level based on the dynamic accuracy allocation matrix; The calculation module also includes: an establishment unit, which is used to construct a working condition characteristic accuracy requirement mapping function and establish a geometric simulation and physical simulation coupling relationship based on the dynamic accuracy allocation matrix to obtain a geometric and physical simulation working condition accuracy mapping function; a numerical allocation unit, which is used to perform normalized constraint calculation and numerical allocation on the geometric simulation dynamic weight factor and the physical simulation dynamic weight factor according to the geometric and physical simulation working condition accuracy mapping function to obtain a geometric and physical simulation dynamic weight factor combination; a stage discrimination unit, which is used to input the geometric and physical simulation dynamic weight factor combination into the processing stage identification algorithm to perform processing stage discrimination and weight adaptive adjustment to obtain a processing stage adjustment weight factor combination; a level calculation unit, which is used to perform geometric simulation accuracy level calculation and physical simulation accuracy level calculation based on the processing stage adjustment weight factor combination to obtain a geometric simulation accuracy level and a physical simulation accuracy level; A construction module, configured to construct an adaptive geometric mesh model of the CNC machine tool according to the geometric simulation accuracy level; An adaptive refinement module, configured to adaptively refine the contact area for cutting force monitoring based on the physical simulation accuracy level and the adaptive geometric grid model to obtain a tool-workpiece contact physical simulation model; The collaborative adjustment module is used to input the adaptive geometric grid model and the tool-workpiece contact physical simulation model into the working condition conversion state machine for collaborative adjustment to obtain an adaptive precision simulation result.

2. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 1 is characterized in that: The data acquisition module also includes: A setting unit is used to set a discrete sampling period for the CNC machine tool to obtain a discrete sampling time sequence; a synchronization unit for performing synchronous data acquisition and timestamp association on tool wear state parameters, cutting force change parameters, spindle speed fluctuation parameters, feed speed change parameters, and machined surface roughness parameters according to the discrete sampling time sequence to obtain an original working condition data set; a vector construction unit, configured to perform normalization based on the original operating condition data set to obtain a normalized numerical sequence, and input the normalized numerical sequence into a multidimensional feature vector construction algorithm for vectorized combination and weight distribution calculation to obtain a standardized operating condition feature vector; The weight allocation unit is used to perform dynamic precision weight allocation of the bounded uncertain diagonal matrix based on the standardized working condition eigenvector to obtain a dynamic precision allocation matrix.

3. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 2 is characterized in that: The weight distribution unit further includes: An initialization subunit, configured to initialize a bounded uncertain diagonal matrix structure and determine the number and dimension of simulation modules based on the standardized operating condition eigenvector to obtain an initial precision allocation matrix; A constraint condition calculation subunit is used to perform bounded constraint condition calculation and dynamic weight allocation on the precision weights of each simulation module according to the initial precision allocation matrix to obtain a constraint precision weight sequence; A threshold comparison subunit is used to input the standardized working condition feature vector into the working condition conversion detection algorithm to perform Euclidean distance calculation and threshold comparison at adjacent moments to obtain a working condition conversion trigger signal; The matrix updating subunit is used to perform precision redistribution and matrix updating based on the working condition conversion trigger signal and the constraint precision weight sequence to obtain a dynamic precision allocation matrix.

4. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 3 is characterized in that: The constraint condition calculation subunit is specifically used for: Setting bounded constraints on precision weights and limiting the value range of the diagonal elements of the initial precision allocation matrix to obtain precision weight constraint intervals; Quantifying the basic computational overhead and evaluating the computational resource requirements of each simulation module based on the precision weight constraint interval to obtain a basic computational overhead coefficient; Perform computing resource constraint mapping function calculation and total resource limit judgment based on the basic computing overhead coefficient and the precision weight constraint interval to obtain computing resource constraint conditions; Based on the computing resource constraint conditions, precision weight optimization calculation and serialized output are performed to obtain a constrained precision weight sequence.

5. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 3 is characterized in that: The matrix updating subunit is specifically used for: Based on the working condition conversion trigger signal, trigger state judgment and precision reallocation requirement determination are performed to obtain a precision adjustment instruction; Recalculate the weights and update the values ​​of the constraint precision weight sequence according to the precision adjustment instruction to obtain an updated precision weight sequence; Performing diagonal element replacement and matrix structure reconstruction based on the update precision weight sequence to obtain an update precision allocation matrix; Numerical stability verification and bounded constraint condition verification are performed on the updated precision allocation matrix to obtain a dynamic precision allocation matrix.

6. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 1 is characterized in that: The value allocation unit is specifically used for: Performing geometric simulation dynamic weight factor extraction and physical simulation dynamic weight factor extraction based on the geometric and physical simulation working condition accuracy mapping function to obtain geometric simulation weight factor values ​​and physical simulation weight factor values; Performing sum constraint calculation and normalization verification on the geometric simulation weight factor value and the physical simulation weight factor value to obtain a normalized constraint condition; Performing processing discrimination and weight value distribution calculation on the characteristic parameters of the processing stage according to the normalized constraint conditions to obtain a weight factor distribution value; Based on the weight factor allocation values, geometric physics simulation weight pairing and numerical verification are performed to obtain a geometric physics simulation dynamic weight factor combination.

7. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 1 is characterized in that: The building blocks are specifically used for: Based on the geometric simulation accuracy level, a quadtree voxel hybrid discretization strategy is constructed and tool wear perception parameters are set to obtain a strategy parameter combination; According to the combination of the strategy parameters, a threshold value judgment and a quadtree recursive depth calculation and a voxel size determination are performed on the tool wear amount to obtain an adaptive discretization parameter; Inputting the adaptive discretization parameters into the tool-workpiece contact area identification algorithm to perform envelope surface intersection calculation and area modeling to obtain a hierarchical grid discretization structure; Geometric error accumulation control and grid re-division triggering judgment are performed on the hierarchical grid discretization structure to obtain an adaptive geometric grid model.

8. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 1 is characterized in that: The adaptive refinement module is specifically used for: Based on the physical simulation accuracy level and the adaptive geometric grid model, a tool-workpiece contact state enhancement vector is constructed and parameters of normal contact force, tangential friction force, contact area, and contact pressure are extracted to obtain a tool-workpiece contact state enhancement vector; Calculating the contact area accuracy grading standard and determining the mesh refinement multiple for the cutting force monitoring value according to the tool-workpiece contact state enhancement vector to obtain the area grading accuracy refinement parameter; Based on the regional classification precision refinement parameters, real-time working condition identification, triggering and judgment and precision switching control are performed to obtain cutting force adjustment instructions; Based on the cutting force adjustment instruction, radial basis function contact force field interpolation calculation and physical simulation accuracy correction are performed to obtain a tool-workpiece contact physical simulation model.

9. The CNC machine tool intelligent monitoring and optimization system based on digital twin according to claim 1 is characterized in that: The collaborative adjustment module is specifically used to: Inputting the adaptive geometric grid model and the tool-workpiece contact physical simulation model into a working condition conversion state machine to construct a machining abnormality state, thereby obtaining a working condition conversion state machine control structure; Based on the working condition conversion state machine control structure, a nested precision threshold adjustment mechanism is constructed and PID structure feedback controller parameters are set to obtain a nested precision adjustment feedback loop; Performing a total resource limitation judgment and a processing quality impact weight distribution calculation on computing resource requirements according to the nested precision adjustment feedback loop to obtain a resource competition scheduling allocation plan; The resource competition scheduling allocation scheme is combined with the precision control system convergence verification and simulation consistency verification to obtain adaptive precision simulation results.

Images