Digital twin full life cycle management method and system based on machining

Abstract

The invention discloses a digital twinning full life cycle management method and system based on machining, and belongs to the technical field of digital twinning in intelligent manufacturing, and the method comprises the steps: 1, obtaining a vibration spectrum of equipment, converting the vibration spectrum into a time sequence waveform, and constructing a multi-scale waveform database; 2, calculating a dynamic stiffness coefficient and a damping ratio based on the multi-scale waveform database, and generating a behavior feature vector; 3, constructing a local correction engine, detecting whether the behavior feature vector deviates from a preset reference threshold value, triggering local grid reconstruction of the digital twin model when the behavior feature vector deviates from the preset reference threshold value, and adjusting model parameters; 4, constructing a virtual load simulator, simulating equipment response based on the adjusted model parameters, and comparing the simulation response with the time sequence waveform; dynamic self-adaptive adjustment of the parameters of the digital twin model can be realized, the mapping precision of the digital twin model to the nonlinear and multivariable coupling state of the equipment is improved, and the timeliness and reliability of a prediction result are guaranteed.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology in intelligent manufacturing, and more specifically, to a method and system for full lifecycle management of digital twins based on machining. Background Technology

[0002] With the rapid development of intelligent manufacturing technology, digital twin technology has become a supporting technology for optimizing production efficiency and reducing failure risks in the field of mechanical processing because it can realize real-time mapping between physical entities and virtual models and full-process traceability. In the full life cycle management of mechanical processing equipment, the digital twin system can provide technical support for key links such as equipment operation status monitoring, process parameter adjustment, remaining life prediction and fault early warning by integrating functions such as real-time data acquisition, virtual simulation analysis and decision optimization, thereby improving the stability and controllability of the processing process.

[0003] However, existing digital twin systems still have certain limitations in practical applications in the field of machining. The main problem lies in the synchronization delay between model updates and changes in the physical entity's state. Current systems generally rely on periodic data acquisition or event-triggered data input to update the model. Although this can meet the basic requirements of conventional static machining scenarios, it is difficult to respond quickly to subtle changes in the physical entity's state in dynamic production environments, such as progressive wear during equipment operation and nonlinear drift of machining parameters. This leads to the continuous accumulation of deviations between the virtual model and the physical entity, resulting in problems such as decreased prediction accuracy and delayed fault warning response, which seriously affects the reliability of the entire lifecycle management.

[0004] The underlying technical root of the aforementioned synchronization delay problem lies in the limitations of the existing digital twin model's construction logic and update mechanism. On the one hand, the model construction phase often assumes that the characteristics of the physical entity satisfy a static or linear relationship. However, the actual machining process involves real-time fluctuations of multiple parameters such as cutting speed and temperature, and there is a complex nonlinear coupling effect between the equipment's operating state and the machining environment. This simplistic assumption leads to the model's inherent lack of adaptability to dynamic changes. On the other hand, the model update mechanism lacks the ability to continuously and adaptively correct the inherent dynamic behavior of the physical entity. It cannot adjust the model parameters in real time according to subtle changes in the entity's state, and relying solely on periodic or event-triggered updates is insufficient to compensate for the dynamic deviation between the model and the entity.

[0005] Especially in high-speed dynamic machining scenarios, when key parameters such as cutting speed and machining temperature fluctuate in real time, in order to ensure the high fidelity and real-time prediction accuracy of the digital twin model, the model needs to perform high-frequency parameter adjustments and simulation calculations. However, existing technologies are difficult to balance the contradiction between high fidelity and real-time responsiveness. High-frequency calculations can easily lead to overload of system computing resources, further exacerbating the delay in model updates, forming a vicious cycle of pursuing accuracy, resource overload, and worsening delays. Ultimately, this results in untimely fault prediction and delayed process optimization, failing to fully realize the important value of digital twin technology in the whole life cycle management. Summary of the Invention

[0006] To address the problems existing in the prior art, the purpose of this invention is to provide a digital twin full lifecycle management method and system based on machining, which can realize dynamic adaptive adjustment of digital twin model parameters, improve the mapping accuracy of digital twin model to the nonlinear and multivariate coupled state of equipment, and ensure the timeliness and reliability of prediction results.

[0007] To solve the above problems, the present invention adopts the following technical solution:

[0008] Firstly, a digital twin-based full lifecycle management method for machining includes:

[0009] Step 1: Obtain the vibration spectrum of the equipment and convert it into a time series waveform to build a multi-scale waveform database;

[0010] Step 2: Calculate the dynamic stiffness coefficient and damping ratio based on the multi-scale waveform database to generate behavioral feature vectors;

[0011] Step 3: Construct a local correction engine to detect whether the behavior feature vector deviates from a predetermined baseline threshold, and trigger local mesh reconstruction of the digital twin model when it deviates, and adjust the model parameters.

[0012] Step 4: Construct a virtual load simulator, simulate the device response based on the adjusted model parameters, compare the simulated response with the time series waveform, and calculate the deviation index.

[0013] Step 5: Construct a parameter converter to convert the deviation index into a compensation value and adjust the processing parameters;

[0014] Step 6: Build a lifecycle tracker to record correction and compensation events, and construct a behavior evolution map to predict the remaining lifespan of the equipment and potential failure points;

[0015] Step 7: Integrate the local correction engine, virtual load simulator, parameter converter, and lifecycle tracker, and combine them with behavioral feature vectors to build a collaborative management interface that displays the real-time model status and prediction suggestions, and allows users to customize thresholds and adjustment strategies.

[0016] Furthermore, step 1 also includes:

[0017] Step 11: Monitor the surface stress wave propagation characteristics and capture the original stress wave signal sequence;

[0018] Step 12: Apply the dispersion relation compensation mechanism to the original stress wave signal sequence to reconstruct the vibration spectrum;

[0019] Step 13: Convert the vibration spectrum into a time series waveform using the modal superposition principle;

[0020] Step 14: Construct a multi-scale waveform database based on time-series waveforms, where waveform segments are stored according to the device's operating status.

[0021] Furthermore, step 2 also includes:

[0022] Step 21: Extract envelope parameters from the time series waveform to obtain the attenuation feature set;

[0023] Step 22: The attenuation feature set is processed using the stress-strain hysteresis relationship inversion method based on genetic algorithm, and the dynamic stiffness coefficient and damping ratio are calculated.

[0024] Step 23: Organize the dynamic stiffness coefficient and damping ratio into a state space vector to generate a behavior feature vector.

[0025] Furthermore, step 3 also includes:

[0026] Step 31: Calculate the instantaneous deviation of each dimension parameter in the behavior feature vector relative to the predetermined benchmark threshold, and generate a dynamic deviation index;

[0027] Step 32: Analyze the strain energy density distribution in the corresponding region of the digital twin model and identify the local mesh regions that need to be reconstructed;

[0028] Step 33 involves synchronously reconstructing the topology and node attributes of the local grid region and adjusting the parameters of the digital twin model.

[0029] Furthermore, step 4 also includes:

[0030] Step 41: Generate a virtual load spectrum and inject it into the dynamic equations of the digital twin model;

[0031] Step 42: Extract the displacement response of key nodes from the numerical solution of the dynamic equation and synthesize the simulated time series waveform;

[0032] Step 43: Spatiotemporally align the simulated time series waveform with the original time series waveform obtained in Step 1, and calculate the local waveform difference.

[0033] Step 44, and use the local waveform difference sequence to generate the overall deviation index through weighted fusion.

[0034] Furthermore, step 5 also includes:

[0035] Step 51: Classify the overall deviation index into different states and identify deviation pattern characteristics;

[0036] Step 52: Calculate the physical parameter compensation amount based on the preset processing dynamics reverse mapping function library;

[0037] Step 53: Generate a pre-compensated waveform by combining the dynamic response characteristics;

[0038] Step 54, and convert the pre-compensation waveform into machining parameter adjustment instructions to update the machining control parameters.

[0039] Furthermore, step 6 also includes:

[0040] Step 61: Associate and record correction events and compensation events, construct an event sequence chain and mark causal relationships;

[0041] Step 62: Extract state transition features and construct state nodes and transition paths in the behavior evolution graph;

[0042] Step 63: Calculate and predict the remaining lifespan of the equipment using a damage accumulation model on the critical path through a behavioral evolution graph;

[0043] Step 64: Locate potential fault points based on the damage accumulation calculation results and the stress distribution of the digital twin model.

[0044] Furthermore, step 7 also includes:

[0045] Step 71: Establish a dynamic data interface protocol between the local correction engine, virtual load simulator, parameter converter and lifecycle tracker, and define the data exchange format and triggering conditions;

[0046] Step 72: Based on the dynamic data interface protocol, map the output data of each component to the dynamic attributes of the visualization elements.

[0047] Furthermore, step 7 also includes:

[0048] Step 73: Receive user interaction instructions through visual elements to construct a feedback loop for the threshold adjustment strategy;

[0049] Step 74 involves using a threshold adjustment strategy to feed back closed-loop data to achieve adaptive synchronous updates of the operating parameters of each component.

[0050] Secondly, the present invention also provides a digital twin full lifecycle management system based on machining, comprising:

[0051] The waveform database module is used to acquire the vibration spectrum of the equipment and convert it into a time-series waveform to build a multi-scale waveform database.

[0052] The feature generation module calculates the dynamic stiffness coefficient and damping ratio based on a multi-scale waveform database, and generates behavioral feature vectors.

[0053] The local correction module is used to build a local correction engine, detect whether the behavior feature vector deviates from a predetermined baseline threshold, and trigger local mesh reconstruction of the digital twin model when it deviates, thereby adjusting the model parameters.

[0054] The response simulation module is used to build a virtual load simulator, simulate the device response based on the adjusted model parameters, and compare the simulated response with the time series waveform to calculate the deviation index.

[0055] The parameter compensation module is used to construct a parameter converter, convert the deviation index into a compensation value, and adjust the processing parameters;

[0056] The lifecycle tracking module is used to build a lifecycle tracker, record correction and compensation events, and construct a behavior evolution map to predict the remaining lifespan of the equipment and potential failure points.

[0057] The collaborative management module integrates a local correction engine, a virtual load simulator, a parameter converter, and a lifecycle tracker. It combines these with behavioral feature vectors to build a collaborative management interface that displays real-time model status and prediction suggestions, and allows users to customize thresholds and adjustment strategies.

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0059] (1) This scheme solves the problem of synchronization delay between traditional digital twin models and physical entities by monitoring the propagation characteristics of surface stress waves, constructing a multi-scale waveform database and generating behavioral feature vectors, and combining the local correction engine to detect parameter deviations in real time and trigger local mesh reconstruction. It realizes dynamic adaptive adjustment of model parameters, improves the mapping accuracy of digital twin models to nonlinear and multivariate coupled states of equipment, and ensures the timeliness and reliability of prediction results.

[0060] (2) This solution uses a virtual load simulator to accurately reproduce the equipment response. It calculates the overall deviation index by weighted fusion of spatiotemporal alignment and local waveform difference, avoiding the overload of computing resources caused by global model iteration. It reduces computing costs while ensuring high-fidelity simulation, and takes into account the real-time nature of model updates and system operating efficiency, providing efficient computing power support for scenarios such as dynamic optimization of process parameters and fault early warning.

[0061] (3) This scheme establishes a reverse mapping between the deviation index and the processing parameters through the parameter converter, generates a pre-compensation waveform by combining dynamic response characteristics and converts it into adjustment instructions, forming a closed-loop control of detection, deviation calculation and compensation adjustment. It can respond to unexpected factors such as material property fluctuations and cutting force changes in real time, dynamically optimize processing parameters, effectively reduce processing errors, improve product processing accuracy and quality stability, and reduce abnormal downtime in the production process.

[0062] (4) This solution uses a lifecycle tracker to associate and record correction and compensation events, constructs a behavior evolution map, and combines a damage accumulation model to predict the remaining lifespan of the equipment and locate potential fault points. This enables visualized management and forward-looking operation and maintenance of the entire lifecycle of the equipment, provides early warning of fault risks and plans maintenance strategies, reduces production interruptions caused by sudden faults, lowers operation and maintenance costs, extends the service life of the equipment, and improves the continuity and economy of the overall production process. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0064] Figure 1 This is a flowchart of a digital twin full lifecycle management method based on machining, according to the present invention.

[0065] Figure 2 This is a flowchart illustrating the relationships between various modules in a digital twin lifecycle management system based on machining, as described in this invention. Detailed Implementation

[0066] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0067] Example 1

[0068] Please see Figure 1 A digital twin-based full lifecycle management method for machining includes:

[0069] Step 1: Obtain the vibration spectrum of the equipment and convert it into a time series waveform to build a multi-scale waveform database;

[0070] Step 1 further includes: monitoring the propagation characteristics of surface stress waves and capturing the original stress wave signal sequence; applying the dispersion relation compensation mechanism to the original stress wave signal sequence to reconstruct the vibration spectrum; converting the vibration spectrum into a time series waveform through the modal superposition principle; and constructing a multi-scale waveform database based on the time series waveform, in which waveform segments are stored according to the equipment operating status.

[0071] In this embodiment, during the signal acquisition stage, stress wave sensors are reasonably arranged on the surface of key components of the equipment to monitor the propagation characteristics of stress waves inside the components in real time. The sensors convert the physical propagation process of stress waves into electrical signals, thereby capturing a continuous sequence of original stress wave signals. The original stress wave signal sequence contains various state information such as material vibration and component interaction during equipment operation. However, during its propagation, dispersion phenomena will occur due to factors such as the heterogeneity of the equipment structure and changes in the propagation path, resulting in differences in the wave velocity of different frequency components and distortion of the signal waveform, which directly affects the accuracy of subsequent analysis.

[0072] To eliminate signal distortion caused by dispersion, a dispersion compensation mechanism needs to be applied to the original stress wave signal sequence. This compensation process is based on the propagation law of stress waves in a specific medium. By analyzing the dependence between the propagation speed and frequency of each frequency component of the signal, the phase and amplitude of different frequency components are adjusted in reverse so that each frequency component can be restored to the synchronous relationship in the state of non-dispersion propagation. This reconstructs a vibration spectrum that can truly reflect the vibration characteristics of the equipment. The reconstructed vibration spectrum presents the frequency distribution characteristics of the equipment vibration in the form of frequency and amplitude. However, in order to more intuitively reflect the change law of vibration over time, it needs to be converted into a time series waveform through the modal superposition principle. First, the inherent modal parameters of the equipment are identified through modal analysis, including natural frequency, mode shape and modal damping. Then, the reconstructed vibration spectrum is decomposed into frequency components corresponding to each inherent mode. Then, according to the vibration response law of each mode, the modal responses corresponding to different frequency components are superimposed in the time domain to finally form a time series waveform that can continuously reflect the evolution of equipment vibration over time.

[0073] Based on the time series waveforms obtained from the above processing, a multi-scale waveform database is further constructed. In the process of constructing the database, different scale levels are first divided according to the characteristics of waveform such as time length and frequency resolution. At the same time, combined with the operating status parameters of the equipment, such as load size, speed, and processing stage, the waveform segments at each scale are classified and labeled. The classification and storage are based on the normal operating status of the equipment, the operating status under different load conditions, and the possible initial abnormal states, to ensure that each waveform segment in the database corresponds to a specific equipment operating scenario.

[0074] In a preferred embodiment of the present invention, the method includes: step 2, calculating the dynamic stiffness coefficient and damping ratio based on a multi-scale waveform database, and generating a behavior feature vector;

[0075] Step 2 further includes: extracting envelope parameters from the time series waveform to obtain the attenuation feature set; processing the attenuation feature set using a stress-strain hysteresis relationship inversion method based on a genetic algorithm to calculate the dynamic stiffness coefficient and damping ratio; and organizing the dynamic stiffness coefficient and damping ratio into a state space vector to generate a behavior feature vector.

[0076] In this embodiment, the time-series waveform carries complete dynamic information about the equipment vibration, while the envelope can isolate the interference of high-frequency vibration components, highlighting the overall variation law of vibration amplitude over time, especially the key features of the equipment vibration decay process. By performing envelope extraction processing on the time-series waveform, a series of parameters that can quantify vibration decay characteristics can be obtained. These parameters include, but are not limited to, the decay rate of vibration amplitude, the peak decay amount in different time periods, the fitting curve coefficient of the envelope, and the fluctuation frequency during the decay process. These parameters together constitute the decay feature set, and their numerical changes directly map the changes in the internal dynamic characteristics of the equipment. The dynamic stiffness coefficient and damping ratio of the equipment reflect its dynamic characteristics. The parameters, together, determine the vibration response of the equipment under stress. The stress-strain hysteresis relationship is the direct external manifestation of these two parameters. Since direct measurement of stress and strain during equipment operation is technically challenging, and there is a clear intrinsic correlation between the attenuation feature set and the stress-strain hysteresis relationship, an inversion method is used to derive the dynamic stiffness coefficient and damping ratio from the attenuation feature set. To improve the accuracy and efficiency of the inversion, a genetic algorithm is introduced as an optimization tool to construct an inversion model for the stress-strain hysteresis relationship. The purpose of the inversion process is to establish an objective function. This function is based on the quantitative correlation between attenuation characteristics and dynamic parameters in vibration theory, transforming the inversion problem into an optimization problem. The objective function can be expressed as: ;

[0077] This formula utilizes the stress-strain hysteresis relationship, where the attenuation characteristics of the strain response are jointly dominated by the dynamic stiffness coefficient and the damping ratio, and each measured parameter in the attenuation characteristic set... Each corresponds to a theoretically calculated value under specific dynamic parameters. The objective function achieves the optimal estimation of dynamic parameters by minimizing the weighted sum of squares of measured and theoretically calculated values. The construction of the objective function follows the physical laws of vibration decay and adapts to the global search characteristics of the genetic algorithm, ensuring the stability and accuracy of the inversion process. In the formula, E represents the objective function value, i.e., the sum of squared errors; n is the number of parameters in the decay feature set. Here, represents the weighting coefficient for the i-th decay characteristic parameter, used to adjust the importance of different parameters in the inversion process. is the measured value of the i-th attenuation characteristic parameter; is the theoretical value of the i-th attenuation characteristic parameter calculated based on the dynamic parameters, k is the dynamic stiffness coefficient, and c is the damping ratio.

[0078] During the inversion process, the dynamic stiffness coefficient and damping ratio are used as optimization variables in the genetic algorithm. Reasonable value ranges for these variables are set based on the equipment's structural characteristics and operating conditions. The constructed objective function serves as the fitness function. Through basic operations such as selection, crossover, and mutation in the genetic algorithm, the optimization variables are iteratively searched. In each iteration, the algorithm generates a set of candidate solutions for the dynamic stiffness coefficient and damping ratio, calculates the corresponding fitness function value, and selects candidate solutions with smaller errors to enter the next iteration. This process continues until the number of iterations reaches a preset threshold or the fitness function value is less than a set error threshold. The resulting candidate solution is then the optimal dynamic stiffness coefficient and damping ratio. Finally, the inverted dynamic stiffness coefficient and damping ratio are structured according to preset rules to form a state space vector. The dimension of this vector is determined based on the equipment's dynamic characteristics, covering key dynamic parameters in different directions and modes. Ultimately, this generates a behavioral feature vector that comprehensively and accurately characterizes the equipment's dynamic behavior.

[0079] In a preferred embodiment of the present invention, the method includes: step 3, constructing a local correction engine, detecting whether the behavior feature vector deviates from a predetermined benchmark threshold, and triggering local mesh reconstruction of the digital twin model when the deviation occurs, and adjusting the model parameters;

[0080] Step 3 further includes: calculating the instantaneous deviation of each dimension parameter in the behavior feature vector relative to a predetermined benchmark threshold, generating a dynamic deviation index; analyzing the strain energy density distribution of the corresponding region in the digital twin model, identifying the local mesh region that needs to be reconstructed; and synchronously reconstructing the topology and node attributes of the local mesh region to adjust the parameters of the digital twin model.

[0081] In this embodiment, the predetermined benchmark threshold is set based on the statistical analysis of the dynamic characteristics of the equipment under normal operating conditions. By analyzing the behavioral feature vectors under normal operating conditions in the multi-scale waveform database, the stable value range and fluctuation upper limit of each dimension parameter are extracted to form the benchmark threshold standard corresponding to each dimension. These benchmark thresholds are not fixed, but are dynamically calibrated according to the equipment's operating time, cumulative processing volume, and other life cycle information to ensure that they always match the normal state of the equipment at different stages. In the deviation detection process of the behavioral feature vector, the instantaneous deviation of the parameter of each dimension relative to the corresponding benchmark threshold needs to be calculated. The calculation of the instantaneous deviation needs to consider the dynamic change trend of the parameter, not only focusing on the static difference between the current value and the threshold, but also combining the rate of change of the parameter per unit time to avoid misjudgment caused by instantaneous fluctuations. On this basis, a dynamic deviation index is generated by comprehensively weighting the instantaneous deviation of each dimension. The weight allocation is determined according to the degree of influence of each dimension parameter on the overall operating state of the equipment. The dynamic parameters that affect the function of the equipment are assigned higher weights to ensure that the dynamic deviation index can truly reflect the overall difference between the equipment state and the benchmark state.

[0082] When the dynamic deviation index exceeds the set trigger threshold, the local correction engine initiates the local mesh reconstruction process. First, by analyzing the strain energy density distribution in the digital twin model, it locates the region in the model corresponding to the changes in the physical device's state. The strain energy density distribution can intuitively reflect the stress and deformation characteristics of each region of the model. Changes in the characteristics of a certain part of the physical device will directly correspond to abnormal fluctuations in the strain energy density of the corresponding region in the model. By identifying these abnormal fluctuation regions, the range of local meshes that need to be reconstructed can be accurately determined, avoiding unnecessary global mesh adjustments. After determining the local mesh region, the topology and node attributes of the region are reconstructed synchronously. The topology reconstruction adjusts the node connection relationship and distribution density of the mesh according to the actual structural changes of the physical device. In regions with significant characteristic changes, the mesh nodes are densified to improve model accuracy, while in regions with gradual changes, the mesh is appropriately simplified to control the amount of computation. The node attribute reconstruction updates the material mechanical parameters, boundary conditions, and other attribute information corresponding to the mesh nodes based on real-time data of behavioral feature vectors, so that the reconstructed local mesh can accurately map the current state of the physical device. Through the coordinated adjustment of the topology and node attributes, the parameters of the digital twin model are accurately updated, ensuring the dynamic consistency between the model and the physical device.

[0083] In a preferred embodiment of the present invention, the method includes: step 4, constructing a virtual load simulator, simulating the device response based on the adjusted model parameters, comparing the simulated response with the time series waveform, and calculating the deviation index;

[0084] Step 4 further includes: generating a virtual load spectrum and injecting it into the dynamic equation of the digital twin model; extracting the displacement response of key nodes from the numerical solution of the dynamic equation and synthesizing a simulated time series waveform; aligning the simulated time series waveform with the original time series waveform obtained in step 1 in time and space, calculating the local waveform difference degree; and using the local waveform difference degree sequence to generate an overall deviation index through weighted fusion.

[0085] In this embodiment, the generation of the virtual load spectrum needs to fully match the actual processing scenario of the equipment. It combines load characteristics recorded in a multi-scale waveform database under different operating states, and also references real-time processing parameters such as cutting speed and feed rate to construct a load sequence that truly reflects the stress state of the equipment. The virtual load spectrum not only includes static load components but also covers dynamic load components generated during processing due to material property fluctuations and dynamic changes in cutting force, ensuring the comprehensiveness and realism of the load simulation. The generated virtual load spectrum is then injected into the dynamic equations of the digital twin model adjusted in step 3. These dynamic equations incorporate the updated model parameters and can reflect... The current structural characteristics and dynamic state of the equipment are analyzed. The dynamic equations are solved using numerical calculation methods to obtain complete response data of the model under virtual load. The displacement response information of key nodes is extracted. The selection of key nodes is based on the structural characteristics and key processing areas of the equipment, including the connection parts of easily worn components and areas of concentrated stress. The displacement response of these nodes directly reflects the overall operating state and local characteristics of the equipment. The displacement responses of each key node are integrated in time series to synthesize a simulated time series waveform that can comprehensively reflect the dynamic behavior of the model. The format and dimensions of the simulated time series waveform are consistent with the original time series waveform obtained in step 1.

[0086] To ensure the accuracy of the comparison results, the simulated time series waveform and the original time series waveform need to be spatiotemporally aligned. In the time dimension, the time axis of the simulated waveform is adjusted based on the start time of the original signal acquisition to ensure that the two correspond one-to-one at the time nodes, eliminating the deviation caused by the difference between the signal acquisition and the start time of the simulation calculation. In the spatial dimension, the positional correspondence between the key nodes in the simulated waveform and the monitoring points of the original signal is calibrated to ensure that the two reflect the vibration response of the same part of the equipment. After the spatiotemporal alignment is completed, signal analysis methods are used to calculate the local waveform difference between the two in each corresponding time period. The local waveform difference is comprehensively reflected by quantifying key indicators such as amplitude deviation, frequency distribution consistency, and phase difference, forming a continuous local waveform difference sequence.

[0087] To comprehensively evaluate the overall fitting accuracy of the model, the local waveform difference sequences need to be weighted and fused to generate an overall deviation index. During the fusion process, weights are assigned according to the importance of the equipment operating status corresponding to each local time period. For critical processing stages of the equipment or periods with drastic load changes, higher weight coefficients are assigned to highlight their impact on the overall deviation. The overall deviation index can be calculated as follows: ;

[0088] This formula is based on the linear correlation between local differences and overall deviation. The contribution of each local waveform difference to the overall deviation is determined by the corresponding weighting coefficient. By using a weighted average, the influence of differences in different time periods can be balanced, resulting in a quantitative index that can objectively reflect the overall fitting effect of the model. In the formula, D represents the overall deviation index, and m is the length of the local waveform difference sequence, i.e., the number of local time periods. Let be the weight coefficient for the j-th local time period; The value represents the local waveform difference in the j-th local time period. The larger the overall deviation index, the greater the deviation between the simulated response and the actual equipment response, and the more the model needs further optimization and adjustment. The smaller the value, the higher the fit between the model and the physical equipment, and the more accurately it can reflect the actual operating status of the equipment.

[0089] In a preferred embodiment of the present invention, the method includes: step 5, constructing a parameter converter to convert the deviation index into a compensation value and adjusting the processing parameters;

[0090] Step 5 further includes: classifying the overall deviation index into states and identifying deviation pattern characteristics; calculating the physical parameter compensation amount based on the preset machining dynamics reverse mapping function library; generating a pre-compensation waveform by combining dynamic response characteristics; and converting the pre-compensation waveform into machining parameter adjustment instructions to update machining control parameters.

[0091] In this embodiment, the overall deviation index embodies the quantitative difference between the model simulation and the actual equipment response. The underlying causes of these differences, such as changes in equipment state and fluctuations in processing conditions, are significantly different. Therefore, the overall deviation index needs to be classified into different states first. This classification process requires considering multiple dimensions, including the numerical range of the deviation, its temporal evolution trend, and fluctuation frequency. For example, deviations can be categorized into different state types, such as low-amplitude persistent deviations, high-amplitude abrupt changes, and periodic fluctuations. Based on this state classification, deviation pattern characteristics are further identified. By associating historical data from a multi-scale waveform database with corresponding equipment state information, it becomes clear whether the deviation is caused by specific factors such as dynamic stiffness decay, changes in damping characteristics, load fluctuations, or process parameter drift. This results in clear pattern characteristics such as stiffness-dominated deviations, damping-affected deviations, and load-disturbance-induced deviations. The pre-defined processing dynamics inverse mapping function library is used to convert the deviation index into physical parameter compensation quantities. Supported by this library, which is built upon a large amount of machining experimental data and theoretical dynamic models, this library contains the correspondence between different deviation mode characteristics and physical parameter adjustment amounts. During machining, changes in physical parameters such as cutting force, machining temperature, and feed resistance will affect the equipment response through dynamic transmission, thus forming deviations. The inverse mapping function library is based on this positive interaction relationship. Through data fitting and theoretical derivation, it establishes an inverse mapping relationship, which can quickly retrieve the corresponding mapping function based on the identified deviation mode characteristics. When calculating the physical parameter compensation amount based on this function library, the quantitative index of the deviation mode characteristics needs to be used as input and substituted into the corresponding inverse mapping function. The output is the physical parameter adjustment range required to offset the current deviation. For example, for stiffness-dominated deviations, the required cutting force compensation value is calculated; for temperature-induced deviations, the required cooling parameter compensation amount is calculated, ensuring the pertinence and accuracy of the compensation amount.

[0092] The physical parameter compensation amount only specifies the adjustment range. It is also necessary to generate a pre-compensation waveform based on the dynamic response characteristics of the equipment. The dynamic response characteristics include the vibration attenuation law of the equipment under different parameter adjustments, the response delay time, and the coupling relationship between parameter adjustment and response change. When generating the pre-compensation waveform, the physical parameter compensation amount needs to be converted into a continuous waveform signal that changes with time based on these characteristics. This waveform needs to complement the current dynamic response of the equipment. For example, if the equipment response has lag, the pre-compensation waveform needs to start the adjustment a certain time in advance; if the response fluctuates, the pre-compensation waveform needs to adopt a smooth transition trend to avoid new equipment response anomalies caused by parameter abrupt changes and ensure the stability of the compensation process.

[0093] The pre-compensation waveform is essentially a time-evolution signal of physical parameters, which needs to be further converted into machining parameter adjustment commands. This conversion process must follow the control protocol and parameter coding rules of the machining equipment, mapping the amplitude, frequency, phase, and other characteristics of the pre-compensation waveform to specific machining parameter control dimensions. For example, the amplitude change of the pre-compensation waveform is converted into a percentage adjustment of the cutting speed, and the time series of the waveform is mapped to different stages of the machining process. The converted machining parameter adjustment commands must have clear execution parameters, including adjustment start time, adjustment magnitude, and duration. Finally, these commands are sent to the equipment's control system to update the machining control parameters and achieve real-time compensation of the machining process. Through this series of processes, the parameter converter completes a closed loop from deviation quantification to parameter adjustment, ensuring that the machining parameters can be dynamically optimized according to the actual state of the equipment, continuously reducing the deviation between the model and the physical equipment, and guaranteeing machining quality and efficiency.

[0094] In a preferred embodiment of the present invention, step 6 is included: constructing a lifecycle tracker, recording correction and compensation events, and constructing a behavior evolution map to predict the remaining lifespan of the device and potential failure points;

[0095] Step 6 further includes: associating and recording correction events and compensation events, constructing event sequence chains and marking causal relationships; extracting state transition features, constructing state nodes and transition paths in the behavior evolution graph; using the behavior evolution graph to calculate and predict the remaining life of the equipment through a damage accumulation model on the critical path; and locating potential fault points based on the damage accumulation calculation results combined with the stress distribution of the digital twin model.

[0096] In this embodiment, the event recording process needs to fully capture the key information of each correction and compensation event, including the event trigger time, the corresponding dynamic deviation index threshold, the area range of local mesh reconstruction of the model, the specific value of the physical parameter compensation, the content of the machining parameter adjustment instruction, and the effect feedback data after the event execution. Based on this, by analyzing the chronological order of events, the transmission relationship of parameter changes, and the correlation of response results, the causal logic between different events is marked. For example, a correction event caused by a certain stiffness-dominant deviation forms a direct causal relationship with the subsequent targeted cutting force compensation event. Then, discrete events are linked together with the causal chain along the time axis to construct a complete event sequence chain, ensuring that the cause and effect of equipment state changes are traceable.

[0097] The extraction of state transition features is based on the equipment state change data recorded in the event sequence chain. By analyzing the dimensional changes of behavioral feature vectors before and after the execution of different events, the fluctuation patterns of dynamic deviation index, and the adjustment range of processing parameters, the features of equipment transitioning from one state to another are extracted, including state duration, parameter change threshold, response delay time, etc. The construction of the behavioral evolution map is based on these state transition features. Each stable operating state of the equipment is defined as a state node in the map. The node attributes include the baseline value of the behavioral feature vector in that state, the corresponding processing condition information, and the cumulative running time. The transition path between state nodes is established according to the actual state transition process. The path attributes mark the event type that triggers the transition, the conditions required for the transition, and the damage increment during the transition process. Finally, a behavioral evolution map that can intuitively reflect the state evolution law of the entire life cycle of the equipment is formed, clearly presenting the possible paths of the equipment evolving from the initial state through different intermediate states to the fault state.

[0098] The prediction of remaining lifetime relies on the critical path and damage accumulation model in the behavioral evolution graph. The critical path refers to the state transition path in the graph that has the greatest impact on equipment life and is most likely to lead to failure; it is usually the path with the fastest damage accumulation rate or the largest cumulative damage. The damage accumulation model is constructed based on material fatigue theory and the actual damage laws of equipment operation. During each state transition, the equipment will generate a certain damage increment due to factors such as load and parameter adjustment. These damage increments gradually accumulate over time. When the cumulative damage reaches the equipment's maximum allowable damage threshold, the equipment enters a failure state. Based on this logic, the calculation of remaining lifetime can be expressed as: ;

[0099] In the formula, the maximum permissible damage The cumulative damage is determined by the material properties of the equipment, structural design parameters, and safety factor. It is the sum of damage increments at each state transition stage on the critical path. The current damage accumulation rate r is calculated from the damage increments and corresponding time spans in the recent state transition process. By the ratio of the difference between the two to the rate, the time required for the device to reach the maximum allowable damage threshold from the current state can be obtained.

[0100] The location of potential failure points needs to be combined with the damage accumulation calculation results and the stress distribution characteristics of the digital twin model. The damage accumulation calculation results clearly identify the damage concentration areas in each state transition stage of the equipment, while the digital twin model can accurately simulate the stress distribution of each part of the equipment through real-time updated parameters. The stress concentration areas are often the key parts where damage is prone to occur and accumulate. By superimposing the areas where the cumulative damage value exceeds the set warning threshold with the high value areas of stress distribution in the model, those parts that have both high cumulative damage and are in stress concentration are the potential failure points of the equipment. By locating these parts, the key focus of equipment operation and maintenance can be identified in advance, providing accurate location guidance for preventive maintenance and avoiding production interruptions caused by sudden failures.

[0101] In a preferred embodiment of the present invention, step 7 is included: integrating a local correction engine, a virtual load simulator, a parameter converter, and a lifecycle tracker, and combining them with behavioral feature vectors to build a collaborative management interface that displays real-time model status and prediction suggestions, and allows users to customize thresholds and adjustment strategies.

[0102] Step 7 further includes: establishing a dynamic data interface protocol between the local correction engine, virtual load simulator, parameter converter, and lifecycle tracker, defining the data exchange format and triggering conditions; mapping the output data of each component to the dynamic attributes of the visualization elements based on the dynamic data interface protocol; constructing a threshold adjustment strategy feedback loop by receiving user interaction instructions through the visualization elements; and using the data from the threshold adjustment strategy feedback loop to achieve adaptive synchronous updates of the operating parameters of each component.

[0103] In this embodiment, the foundation of component integration is the establishment of a dynamic data interface protocol. This protocol needs to clearly define the data interaction content, format standards, and transmission rules between each component, taking into account the functional characteristics of the local calibration engine, virtual load simulator, parameter converter, and lifecycle tracker. The data exchange format must balance machine readability and data integrity, uniformly standardizing the encoding methods of data such as behavioral feature vectors, deviation indices, compensation amounts, and lifecycle prediction values ​​to ensure unambiguous data parsing between different components. The definition of triggering conditions must be based on the business logic associations of each component. For example, after the local calibration engine outputs the model parameter adjustment results, it automatically triggers the virtual load simulator to start a new round of response simulation; after the parameter converter generates processing parameter adjustment instructions, it synchronously triggers the lifecycle tracker to record compensation events, forming an automatic linkage mechanism between components to avoid data transmission delays or confusing triggering logic.

[0104] Based on the dynamic data interface protocol, the output data of each component will be mapped to the dynamic attributes of the visual elements in the collaborative management interface, realizing the intuitive presentation of abstract data. The parameters of each dimension of the behavioral feature vector can be mapped to the pointer position or numerical display of the dynamic dashboard in the interface, reflecting the changes in the dynamic characteristics of the equipment in real time. The local mesh reconstruction area of ​​the digital twin model can be marked by color highlighting or outline thickening in the 3D view of the model in the interface, clearly showing the adjustment parts of the model. The changing trend of the deviation index can be mapped to the dynamic update of the line graph, allowing users to intuitively perceive the degree of fit between the simulated response and the actual response. The remaining life prediction value and potential failure point information can be presented in the form of numerical labels and model annotations, clarifying the key points of equipment operation and maintenance. The dynamic attributes of these visual elements are synchronized with the component output data in real time, ensuring that the content displayed on the interface can accurately reflect the system operating status and the actual situation of the equipment, providing intuitive basis for user decision-making.

[0105] The collaborative management interface not only displays the status but also supports user interaction through visual elements, constructing a feedback loop for threshold adjustment strategies. Users can customize key parameters such as the trigger threshold of the dynamic deviation index, the status classification standard of the deviation index, and the adjustment range of processing parameters through interactive controls such as sliders, input boxes, or drop-down menus on the interface. They can also select different adjustment strategies such as model correction intensity and compensation priority according to actual production needs. After receiving user interaction commands, the interface converts them into standardized control signals and transmits them to the corresponding functional components through a dynamic data interface protocol, while recording the command content and trigger time. The feedback loop lies in the real-time feedback of the command execution effect. After each component adjusts its operating logic according to the user-defined parameters, the changes in its output results will be presented in real time through visual elements. Users can further optimize the adjustment strategy based on the feedback results, forming a complete closed loop of user interaction, component response, status feedback, and strategy optimization, improving the system's adaptability to different production scenarios.

[0106] The interactive data and component operation status data generated by the feedback loop will serve as the basis for the system's adaptive synchronous updates, ensuring the consistency and synergy of the operating parameters of each component. When a user adjusts a key threshold or strategy, the system will analyze the impact of the adjustment on each component and automatically update the related parameters of the relevant components. For example, when a user increases the trigger threshold of the dynamic deviation index, the detection standard of the local correction engine will be adjusted accordingly. At the same time, the load spectrum generation threshold of the virtual load simulator and the deviation status classification threshold of the parameter converter will also be updated adaptively to avoid system operation logic conflicts caused by the adjustment of parameters of a single component. This adaptive synchronous update mechanism does not require users to manually intervene in the parameters of each component. It realizes the automatic transmission and collaborative optimization of parameters through data interface protocols, ensuring that the local correction engine, virtual load simulator, parameter converter and lifecycle tracker always work collaboratively based on consistent operating standards, maintaining the overall stability and operating efficiency of the system, and giving full play to the collaborative value of full lifecycle management.

[0107] Example 2:

[0108] Please see Figure 2 Based on Example 1, this example provides a digital twin full lifecycle management system for machining, including:

[0109] The waveform library module is used to acquire the vibration spectrum of the equipment and convert it into a time-series waveform to build a multi-scale waveform database;

[0110] The feature generation module calculates the dynamic stiffness coefficient and damping ratio based on a multi-scale waveform database, and generates behavioral feature vectors.

[0111] The local correction module is used to build a local correction engine, detect whether the behavior feature vector deviates from a predetermined baseline threshold, and trigger local mesh reconstruction of the digital twin model when it deviates, thereby adjusting the model parameters.

[0112] The response simulation module is used to build a virtual load simulator, simulate the device response based on the adjusted model parameters, and compare the simulated response with the time series waveform to calculate the deviation index.

[0113] The parameter compensation module is used to construct a parameter converter, convert the deviation index into a compensation value, and adjust the processing parameters;

[0114] The lifecycle tracking module is used to build a lifecycle tracker, record correction and compensation events, and construct a behavior evolution map to predict the remaining lifespan of the equipment and potential failure points.

[0115] The collaborative management module integrates a local correction engine, a virtual load simulator, a parameter converter, and a lifecycle tracker. It combines these with behavioral feature vectors to build a collaborative management interface that displays real-time model status and prediction suggestions, and allows users to customize thresholds and adjustment strategies.

[0116] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A method for machining-based digital twin full life cycle management, characterized by, Comprise: Step 1, acquire the vibration spectrum of the device and convert it into a time series waveform, build a multi-scale waveform database; Step 2, calculate the dynamic stiffness coefficient and damping ratio based on the multi-scale waveform database, generate the behavior feature vector; Step 3, build a local correction engine, detect whether the behavior feature vector deviates from the predetermined baseline threshold, and trigger the local grid reconstruction of the digital twin model when it deviates, adjust the model parameters; Step 4, build a virtual load simulator, simulate the device response based on the adjusted model parameters, and compare the simulated response with the time series waveform to calculate the deviation index; Step 5, build a parameter converter, convert the deviation index into a compensation value, and adjust the processing parameters; Step 6, build a life cycle tracker, record the correction and compensation events, and build an behavior evolution atlas to predict the remaining life and potential failure points of the device; Step 7, integrate the local correction engine, virtual load simulator, parameter converter and life cycle tracker, and combine the behavior feature vector to build a collaborative management interface to display real-time model status and prediction suggestions, and allow users to customize thresholds and adjustment strategies.

2. The digital twin full life cycle management method based on machining according to claim 1, characterized in that, Step 1 also includes: Step 11, monitor the surface stress wave propagation characteristics, capture the original stress wave signal sequence; Step 12, apply the dispersion relation compensation mechanism to the original stress wave signal sequence to reconstruct the vibration spectrum; Step 13, convert the vibration spectrum into a time series waveform through the modal superposition principle; Step 14, build a multi-scale waveform database according to the time series waveform, which stores waveform segments according to the running state of the device.

3. The method of claim 2, wherein, Step 2 also includes: Step 21, extract envelope parameters from the time series waveform to obtain the attenuation feature set; Step 22, use the stress-strain hysteresis relationship inversion method based on genetic algorithm to process the attenuation feature set to calculate the dynamic stiffness coefficient and damping ratio; Step 23, and organize the dynamic stiffness coefficient and damping ratio into a state space vector to generate the behavior feature vector.

4. The method of claim 3, wherein, Step 3 also includes: Step 31, calculate the instantaneous deviation of each dimension parameter in the behavior feature vector relative to the predetermined baseline threshold to generate the dynamic deviation index; Step 32, analyze the strain energy density distribution of the corresponding area in the digital twin model to identify the local grid area that needs to be reconstructed; Step 33, and synchronize the reconstruction of the topological structure and node attributes of the local grid area to adjust the parameters of the digital twin model.

5. The method of claim 4, wherein, Step 4 also includes: Step 41, generate a virtual load spectrum and inject it into the dynamics equation of the digital twin model; Step 42, extract the displacement response of the key nodes from the numerical solution of the dynamics equation to synthesize the simulated time series waveform; Step 43, perform spatio-temporal alignment between the simulated time series waveform and the original time series waveform obtained in step 1 to calculate the local waveform difference degree; Step 44, and use the local waveform difference degree sequence to generate the overall deviation index through weighted fusion.

6. The method of claim 5, wherein, Step 5 also includes: Step 51, classify the overall deviation index to identify the deviation mode characteristics; Step 52, calculate the physical parameter compensation amount based on the preset processing dynamics inverse mapping function library; Step 53, generate a pre-compensation waveform combined with the dynamic response characteristics; Step 54, convert the pre-compensation waveform into processing parameter adjustment instructions, and update the processing control parameters.

7. The method of claim 6, wherein, Step 6 also includes: Step 61, associate and record the correction event and the compensation event, build an event sequence chain and mark the causal relationship; Step 62, extract state transition features, and build state nodes and transition paths in the behavior evolution graph; Step 63, use the behavior evolution graph to calculate the predicted remaining life of the equipment through the damage accumulation model on the key path; Step 64, locate the potential fault point according to the damage accumulation calculation result combined with the stress distribution of the digital twin model.

8. The method of claim 7, wherein, Step 7 also includes: Step 71, establish a dynamic data interface protocol between the local correction engine, the virtual load simulator, the parameter converter, and the life cycle tracker, define the data exchange format and the trigger condition; Step 72, map the output data of each component to the dynamic attributes of the visualization elements based on the dynamic data interface protocol.

9. The method of claim 8, wherein, Step 7 also includes: Step 73, receive user interaction instructions through the visualization elements to build a threshold adjustment strategy feedback loop; Step 74, use the data of the threshold adjustment strategy feedback loop to realize adaptive synchronous update of the running parameters of each component.

10. A machining-based digital twin full life cycle management system applied to a machining-based digital twin full life cycle management method according to any one of claims 1-9, characterized in that, Comprise: Waveform library building module, for obtaining the vibration spectrum of the equipment and converting it into time series waveform, and building a multi-scale waveform database; Feature generation module, based on the multi-scale waveform database, calculate the dynamic stiffness coefficient and damping ratio, and generate the behavior feature vector; Local correction module, for building a local correction engine, detecting whether the behavior feature vector deviates from the predetermined baseline threshold, and triggering local grid reconstruction of the digital twin model when it deviates, adjusting the model parameters; Response simulation module, for building a virtual load simulator, simulating the device response based on the adjusted model parameters, and comparing the simulation response with the time series waveform to calculate the deviation index; Parameter compensation module, for building a parameter converter, converting the deviation index into a compensation value, and adjusting the processing parameters; Life tracking module, for building a life cycle tracker, recording correction and compensation events, and building a behavior evolution graph to predict the remaining life of the equipment and potential fault points; Collaborative management module, for integrating the local correction engine, the virtual load simulator, the parameter converter, and the life cycle tracker, and combining the behavior feature vector to build a collaborative management interface, display real-time model state and prediction suggestions, and allow users to customize thresholds and adjustment strategies.

Images