Digital twinning-based tool magazine tool health state dynamic evaluation method and system

Abstract

The invention relates to the technical field of tool management, in particular to a tool magazine tool health state dynamic evaluation method and system based on digital twinning. When the tools are stored, tool service data of a tool magazine tool cabin are received from a tool delivery request end; cutter life cycle data are backtracked; taking the cutter service task list, the cutter service duration list, the cutter working parameters and the cutter execution task as constraints, retrieving a plurality of healthy cutter samples for statistical analysis, and obtaining a health factor threshold list; performing cutter damage evaluation based on the cutter health factor list and the health factor threshold list, and generating cutter predicted damage distribution information; and according to the cutter prediction damage distribution information, constructing a cutter digital twin model to execute cutter health state evaluation. And the accuracy and the reliability of cutter health state evaluation are improved.

Description

Technical Field

[0001] The present invention relates to the field of tool management technology, and in particular to a method and system for dynamically evaluating the health status of tool magazine tools based on digital twins. Background Art

[0002] In the field of tool health assessment, traditional approaches primarily monitor parameters that characterize tool health, such as temperature, vibration, force, and sound signals during tool operation. These parameters are then compared with preset parameter thresholds, and the tool's health status is determined based on the parameter differences. The parameter thresholds used in traditional approaches are static data, but the tool's state is constantly fluctuating during actual operation. Because static data cannot dynamically adapt to these fluctuations in tool status in real time, it is difficult to accurately assess tool health. This can lead to misjudgments of tool health, impacting the proper use and maintenance of the tool, increasing uncertainty and costs in the production process, and resulting in technical issues such as low accuracy and reliability in tool health assessment. Summary of the Invention

[0003] The present invention aims to solve the technical problem of low accuracy and reliability of tool health status assessment in the prior art by providing a method and system for dynamic assessment of tool health status of tool magazines based on digital twins.

[0004] The technical solution of the present invention to solve the above technical problems is as follows:

[0005] In the first aspect, the present invention provides a dynamic evaluation method for the health status of tool magazine tools based on digital twins, including: when the tool returns to the warehouse, receiving the tool service data of the tool compartment of the tool magazine from the tool out-of-warehouse request end, wherein the tool service data includes the tool health factor, the tool working parameters and the tool execution task; backtracking the tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list; with the tool service task list, the tool service time list, the tool working parameters and the tool execution task as constraints, retrieving several healthy tool samples for statistical analysis to obtain a health factor threshold list; performing tool damage evaluation based on the tool health factor list and the health factor threshold list to generate tool predicted damage distribution information; constructing a tool digital twin model according to the tool predicted damage distribution information to perform tool health status evaluation.

[0006] Optionally, the tool service data further includes tool working time, including: updating the tool life cycle data according to the tool execution task and the tool working time.

[0007] Optionally, with the tool service task list, the tool service time list, the tool working parameters and the tool execution task as constraints, several healthy tool samples are retrieved for statistical analysis to obtain a health factor threshold list, including: constructing a first-level constraint condition based on the tool service task list and the tool service time list, and constructing a second-level constraint condition based on the tool working parameters and the tool execution task; collecting several first-level healthy tool samples that meet the first-level constraint condition; collecting the several healthy tool samples that meet the second-level constraint condition from the several first-level healthy tool samples; performing a same-attribute central trend analysis on the health factors of the several healthy tool samples, obtaining a plurality of health factor attribute representative value distribution intervals, and constructing the health factor threshold list.

[0008] Among them, performing a same-attribute central trend analysis on the health factors of the several healthy tool samples to obtain a plurality of health factor attribute representative value distribution intervals and constructing the health factor threshold list includes: obtaining a plurality of health factor attribute rated areas; performing same-attribute intersection interval statistics on the plurality of health factor attribute representative value distribution intervals and the plurality of health factor attribute rated areas to obtain a plurality of health factor attribute fitting intervals and constructing the health factor threshold list.

[0009] Wherein, based on the tool service task list and the tool service time list, a first-level constraint condition is constructed, including: extracting the first tool service task type and the first task type service time from the tool service task list and the tool service time list; taking the first tool service task type as a constraint, collecting a one-to-one corresponding number of pre-service tool damage states, a number of post-service tool damage states and a number of tool service times; performing a consistency comparison on the several pre-service tool damage states and the several post-service tool damage states to obtain a number of consistency comparison results; based on the several consistency comparison results, extracting selected tools with inconsistent comparison results from the several tool service times. A service duration set; calculating a centralized value of the selected tool service duration set, and setting it as the sensitive duration of the first tool service task type; when the service duration of the first task type is less than the sensitive duration of the first tool service task type, deleting the first tool service task type and the first task type service duration from the tool service task list and the tool service duration list; when the service duration of the first task type is greater than or equal to the sensitive duration of the first tool service task type, making no changes; when the traversal of the tool service task list and the tool service duration list is completed, obtaining all updated tool service task lists and updated tool service duration lists, and constructing the first-level constraint conditions.

[0010] Optionally, a tool damage assessment is performed based on the tool health factor list and the health factor threshold list to generate tool predicted damage distribution information, including: matching a tool damage assessment model from a tool damage assessment model library based on the tool execution task and tool model, wherein the tool damage assessment model is trained using machine learning through multiple sets of data, and any one of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label that identifies tool damage distribution information; the tool health factor list and the health factor threshold list are input into the tool damage assessment model, and the tool predicted damage distribution information is output.

[0011] Among them, the tool damage assessment model is trained using machine learning through multiple groups of data, and any group of the multiple groups of data includes tool health factor record data, health factor threshold record data and labels identifying tool damage distribution information of preset task types, including: randomly configuring tool health factor record data and health factor threshold record data based on the rated interval of tool health factor; collecting a number of tool damage distribution information for a preset tool model with the preset task type, the tool health factor record data and the health factor threshold record data as constraints; based on the several tool damage distribution information, performing damage concentration value calculation with the same attribute at the same position to obtain a label identifying tool damage distribution information; the tool damage assessment model, the preset tool model and the preset task type are associated and stored, and added to the tool damage assessment model library.

[0012] Optionally, based on the predicted damage distribution information of the tool, a tool digital twin model is constructed to perform tool health status assessment, including: obtaining a first execution task, wherein the first execution task has a predefined tool benchmark digital twin model; when the tool benchmark digital twin model is inconsistent with the tool digital twin model, the first execution task is marked as abnormal; when the tool benchmark digital twin model is consistent with the tool digital twin model, the first execution task is marked as normal; the abnormal mark or the normal mark, combined with the first execution task, is added to the tool health status assessment result and sent to the tool management end.

[0013] In a second aspect, the present invention provides a tool magazine tool health status dynamic assessment system based on digital twins, comprising:

[0014] A service data acquisition module is used to receive tool service data of the tool compartment of the tool magazine from the tool outbound request terminal when the tool is returned to the warehouse, wherein the tool service data includes tool health factors, tool working parameters and tool execution tasks;

[0015] A tool life backtracking module is used to backtrack tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list;

[0016] A healthy tool analysis module is used to retrieve a number of healthy tool samples for statistical analysis based on the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints to obtain a health factor threshold list;

[0017] a tool damage assessment module, configured to perform tool damage assessment based on the tool health factor list and the health factor threshold list, and generate tool damage prediction distribution information;

[0018] The health status assessment module is used to construct a tool digital twin model based on the tool damage distribution information and perform tool health status assessment.

[0019] By implementing the present invention, when a tool is returned to the tool storage, the tool service data of the tool compartment of the tool magazine can be received from the tool removal request terminal. The tool service data includes the tool health factor, tool working parameters, and tool execution task. This ensures that the acquired data is real-time information directly related to the actual service status of the tool, providing accurate and effective data support for subsequent steps.

[0020] By implementing the present invention, it is possible to trace back tool lifecycle data, wherein the tool lifecycle data includes a tool service task list and a tool service duration list. By tracing back the tool lifecycle data, it is possible to grasp the tool usage history from a macro perspective, providing a basis for the subsequent construction and analysis of constraint conditions based on the tool service history, making the evaluation more comprehensive and reliable.

[0021] By implementing the present invention, it is possible to retrieve a number of healthy tool samples for statistical analysis based on the tool service task list, the tool service time list, the tool working parameters, and the tool execution task, and obtain a health factor threshold list. By retrieving and analyzing healthy tool samples based on multi-dimensional constraint conditions, the obtained health factor threshold list can better fit the actual service conditions of the tool. Instead of a static fixed threshold, it dynamically adapts to the threshold range of different tool service conditions, thereby improving the accuracy and applicability of the threshold.

[0022] By implementing the present invention, it is possible to perform tool damage assessment based on the tool health factor list and the health factor threshold list, generate tool damage prediction distribution information, and compare and analyze the dynamic health factor threshold list with the actual tool health factor list to more accurately assess tool damage. The generated prediction damage distribution information provides a specific reference basis for subsequent in-depth assessment of tool health status.

[0023] By implementing the present invention, it is possible to construct a digital twin model of the tool based on the predicted damage distribution information of the tool to perform tool health status assessment, achieve accurate grasp and real-time monitoring of the tool health status, and provide scientific and effective decision-making support for tool management and maintenance.

[0024] In summary, the implementation of the present invention can improve the accuracy and reliability of tool health status assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A schematic diagram of the process of the digital twin-based dynamic assessment method for tool health status of tool magazines provided by the present invention;

[0026] Figure 2 This is a schematic diagram of the structure of the tool magazine tool health status dynamic assessment system based on digital twin provided by the present invention.

[0027] In the accompanying drawings, the components represented by the reference numerals are as follows:

[0028] Service data collection module 11, tool life backtracking module 12, healthy tool analysis module 13, tool damage assessment module 14, health status assessment module 15. DETAILED DESCRIPTION

[0029] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0030] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, features specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0031] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0032] Example 1, as Figure 1 As shown, an embodiment of the present invention provides a method for dynamically evaluating the health status of tool magazine tools based on digital twins, including:

[0033] S100: When a tool is returned to the tool storage, the tool service data of the tool compartment of the tool magazine is received from the tool out-of-warehouse request terminal, wherein the tool service data includes a tool health factor, a tool working parameter, and a tool execution task;

[0034] S200: Backtracking tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list;

[0035] S300: Using the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints, searching for a number of healthy tool samples for statistical analysis to obtain a health factor threshold list;

[0036] S400: Performing tool damage assessment based on the tool health factor list and the health factor threshold list to generate tool damage prediction distribution information;

[0037] S500: Based on the predicted damage distribution information of the tool, a tool digital twin model is constructed to perform tool health status assessment.

[0038] In step S100 of the embodiment of the present application, when the tool is returned to the warehouse, the tool service data of the tool compartment of the tool magazine is received from the tool out-of-warehouse request terminal, wherein the tool service data includes the tool health factor, tool working parameters and tool execution task.

[0039] When the tool completes its operation and returns to the warehouse, the system will obtain the tool service data in the tool compartment of the tool magazine from the tool out-of-warehouse request end (such as the machine tool control system, production management system, etc.). Specifically, the physical return of the tool to the warehouse can be used as the signal source, and the tool return status can be detected by sensors (such as position sensors, RFID readers) to trigger the data reception process. The data is then sent from the tool out-of-warehouse request end to the tool magazine management system via industrial Ethernet, fieldbus (such as Profinet, EtherCAT) or IoT communication protocol (such as MQTT). In the tool magazine, each tool corresponds to a tool compartment one-to-one, and each tool has a unique identifier, such as an RFID tag ID, barcode number, etc.

[0040] Tool health factors include the maximum temperature signal (such as 82°C) and the maximum vibration amplitude signal (12.5m / s 2 ), the X-axis maximum cutting force signal (such as 350N), and other parameter values ​​that characterize the health status of the tool can be collected in real time by the tool's built-in sensors or external monitoring equipment.

[0041] The tool working parameters are processing parameters such as cutting speed (m / min), feed rate (mm / r), and cutting depth (mm), which are derived from the operation records of the machine tool control system.

[0042] The tool execution task records the type of processing task undertaken by the tool, such as external cylindrical turning, end milling, drilling, etc., and is generated when the production management system assigns tasks.

[0043] For example, a tool with the model "M-500" performs a face milling task A. The type of its service data can be: tool health factor [maximum vibration amplitude 15m / s 2 , maximum temperature 75℃]; tool working parameters [cutting speed: 150m / min, feed rate: 0.1mm / r, cutting depth: 3mm], tool execution task [end milling].

[0044] In step S100 of the embodiment of the present application, the tool service data also includes the tool working time, including: updating the tool life cycle data according to the tool execution task and the tool working time.

[0045] Tool operating time is a key indicator of physical tool wear and is directly related to tool life. By recording actual cutting time, we can avoid the drawback of traditional static thresholds (such as fixed machining times) that fail to reflect actual load.

[0046] And the life cycle data (tool execution task list + time list) is dynamically updated with each time of returning to the warehouse, and a "digital record" of the tool can be formed. For example, the form of the life cycle data can be task B: turning, tool working time 2 hours; task C: turning, tool working time 1.5 hours; task D: turning, tool working time 0.5 hours, and so on. According to the tool execution task and the tool working time, the tool life cycle data is updated.

[0047] In step S200 of the embodiment of the present application, the tool life cycle data is traced back, wherein the tool life cycle data includes a tool service task list and a tool service time list.

[0048] The retrieval of a plurality of healthy tool samples for statistical analysis in step S300 of the embodiment is based on the tool life cycle data, and therefore, before step S300 is performed, the tool life cycle data needs to be traced back, that is, the recorded tool service task list and tool service time list of the target tool after entering the tool magazine tool compartment are extracted.

[0049] In step S300 of the embodiment of the present application, a plurality of healthy tool samples are retrieved for statistical analysis based on the tool service task list, the tool service time list, the tool working parameter and the tool execution task to obtain a healthy factor threshold list, including:

[0050] Based on the tool service task list and the tool service time list, a primary constraint condition is constructed, and based on the tool working parameter and the tool execution task, a secondary constraint condition is constructed;

[0051] A plurality of primary healthy tool samples satisfying the primary constraint condition are collected;

[0052] From the plurality of primary healthy tool samples, the plurality of healthy tool samples satisfying the secondary constraint condition are collected;

[0053] The healthy factors of the plurality of healthy tool samples are subjected to same attribute centralized trend analysis to obtain a plurality of healthy factor attribute representative value distribution intervals, and the healthy factor threshold list is constructed.

[0054] In the embodiment of the present application, this step is to filter matched healthy tool samples based on tool historical service data (i.e. the tool service task list and the tool service time list) and current task parameters (i.e. the tool working parameter and the tool execution task), generate a dynamic healthy factor threshold list through statistical analysis, and solve the problem that a traditional static threshold cannot adapt to working condition changes.

[0055] In step S300 of the embodiment, based on the tool service task list and the tool service duration list, a first constraint condition is constructed, including:

[0056] From the tool service task list and the tool service duration list, a first tool service task type and a first task type service duration are extracted.

[0057] With the first tool service task type as a constraint, a one-to-one corresponding plurality of pre-service tool damage states, a plurality of post-service tool damage states and a plurality of tool service durations are collected.

[0058] The plurality of pre-service tool damage states and the plurality of post-service tool damage states are compared for consistency, and a plurality of consistency comparison results are obtained.

[0059] Based on the plurality of consistency comparison results, a selected tool service duration set with inconsistent comparison results is extracted from the plurality of tool service durations.

[0060] The central value of the selected tool service duration set is calculated and set as the first tool service task type sensitive duration.

[0061] When the first task type service duration is less than the first tool service task type sensitive duration, the first tool service task type and the first task type service duration are deleted from the tool service task list and the tool service duration list. When the first task type service duration is greater than or equal to the first tool service task type sensitive duration, no change is made.

[0062] When the tool service task list and the tool service duration list are traversed, an updated tool service task list and an updated tool service duration list are obtained, and the first constraint condition is constructed.

[0063] Suppose a tool T-600 needs to perform an "external turning" task with a service duration of 1.5 hours, and a first constraint needs to be constructed.

[0064] First, with "external turning" as a constraint, healthy tool samples of the same type (such as 10 tool samples of the same type) are retrieved from the historical database, and a plurality of pre-service tool damage states, a plurality of post-service tool damage states and a plurality of tool service durations of each group of samples are collected.

[0065] Among them, the plurality of service durations are obtained from the tool service duration list as described above, such as 1.0h, 1.5h, 2.0h, 2.5h, etc.

[0066] The tool damage state can be measured by the rake face wear VB (unit: mm) with an optical microscope or a 3D scanner before and after each use of the tool, with an accuracy of 0.01 mm, to obtain the tool damage state information.

[0067] The pre-service damage state can be VB = 0.05 mm, and the post-service damage state can be VB = 0.1 mm. If the change in the damage state before and after service is less than or equal to a threshold (e.g., VB change is less than or equal to 0.05 mm), it is determined to be "consistent", otherwise it is "inconsistent". The consistency comparison results are obtained.

[0068] Then, a selected tool service time set (e.g., 2.0 h, 2.5 h) is extracted from the tool service time set (e.g., 1.0 h, 1.5 h, 2.0 h, 2.5 h) for which the comparison result is inconsistent, and a central value of the selected tool service time set is calculated. The central value can be the mean value, such as (2.0+2.5) / 2 = 2.25 h. Then, the central value, i.e., 2.25 h, can be defined as the sensitive time of the "cylindrical turning" task (i.e., the first tool service task type).

[0069] Then, when the service time of the first task type (e.g., cylindrical turning) (e.g., 1.5 h) is less than the sensitive time of the first tool service task type (e.g., 2.25 h), the first tool service task type and the service time of the first task type are deleted from the tool service task list and the tool service time list, and when the service time of the first task type (e.g., 3.0 h) is greater than or equal to the sensitive time of the first tool service task type (e.g., 2.25 h), no change is made. If the tool reaches the scrap standard after performing a task, the service task corresponding to the task is removed.

[0070] When the tool service task list and the tool service time list are traversed, the updated tool service task list and the updated tool service time list are obtained, and the primary constraint condition is constructed. In the above example, all "cylindrical turning" tasks with a time < 2.25 h are removed from the tool service task list, and the remaining tasks are valid data with a time ≥ 2.25 h. Through the above method, a plurality of primary healthy tool samples satisfying the primary constraint condition can be obtained

[0071] Furthermore, it is necessary to construct secondary constraints based on the tool working parameters and the tool execution task, as well as the primary constraints, wherein the secondary constraints can be working parameters. Assume that the aforementioned tool T-600 needs to perform the "external turning" task, with a service time of 1.5 hours, and a secondary constraint needs to be constructed. The working parameters of the tool when performing the "external turning" task are a cutting speed of 150m / min and a feed rate of 0.2mm / r. Then, its secondary constraints can be a cutting speed of 150m / min±10% (135~165m / min) and a feed rate of 0.2mm / r±20% (0.16~0.24mm / r). Then, samples that simultaneously meet the secondary constraints are selected from several healthy tool samples at the first level, for example:

[0072] For example, if the working parameters corresponding to three groups of first-level healthy tool samples among five groups of first-level healthy tool samples are within the above range, then these groups become the healthy tool samples that meet the second-level constraint condition.

[0073] In step S300 of the embodiment of the present application, a central tendency analysis of the health factors of the healthy tool samples is performed to obtain a plurality of distribution intervals of representative values ​​of the health factor attributes, and to construct the health factor threshold list, including:

[0074] Obtain multiple health factor attribute rating areas;

[0075] Perform same-attribute intersection interval statistics on the multiple health factor attribute representative value distribution intervals and the multiple health factor attribute rated areas to obtain multiple health factor attribute fitting intervals and construct the health factor threshold list.

[0076] In the embodiment of the present application, in order to perform a central trend analysis of the health factors of the healthy tool samples, obtain a plurality of distribution intervals of representative values ​​of the health factor attributes, and construct the health factor threshold list, it is first necessary to obtain the health factor data of several (e.g., 3 groups) healthy tool samples (e.g., the maximum vibration amplitude m / s during cutting). 2 , maximum temperature °C, maximum cutting force on X axis (N)

[0077] For example, in the above example, it is necessary to collect the health factor data of 3 healthy tool samples, for example, sample A (12.5m / s 2 , 78°C, 280N), sample B (11.8m / s 2 , 82℃, 270N), sample C (13.2m / s2, 75℃, 290N).

[0078] Exemplarily, the process of constructing the health factor threshold interval is introduced by taking the maximum vibration amplitude as an example.

[0079] First, the tool material and design standards are required to set the rated range of vibration amplitude: [10~15m / s

[0080] ](i.e. normal working range).

[0081] Then, the maximum vibration amplitude of the three groups of healthy tool samples was statistically analyzed. The mean value was (12.5+11.8+13.2) / 3≈12.5 m / s2; the standard deviation was σ≈0.68 m / s2; the distribution range was the mean ±2σ, that is, [12.5-1.36, 12.5+1.36]=[11.14, 13.86] m / s 2 .

[0082] Next, we need to calculate the intersection of the rated area and the distribution interval. In this example, the distribution interval [11.14, 13.86] ∩ the rated area [10, 15] = [11.14, 13.86] m / s2. However, if the distribution interval exceeds the rated area (e.g. [8, 16]), the intersection is [10, 15].

[0083] The intersection of the rated area and the distribution interval, that is, the fitting interval, is used as the maximum vibration amplitude threshold.

[0084] Using the same method, we can calculate thresholds for health factors such as maximum temperature and maximum cutting force, generating a list of these health factor thresholds. For example, when the rated range for maximum temperature is [60, 90]°C, the fitting interval is [75, 82]°C. When the rated range for maximum X-axis cutting force is [200, 350]N, the fitting interval is [270, 290]N.

[0085] Through the above method, the health factor threshold list can be constructed.

[0086] In step S400 of the embodiment of the present application, tool damage assessment is performed based on the tool health factor list and the health factor threshold list to generate tool damage prediction distribution information, including:

[0087] matching a tool damage assessment model from a tool damage assessment model library based on the tool execution task and tool model, wherein the tool damage assessment model is trained using machine learning through multiple sets of data, and any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information;

[0088] The tool health factor list and the health factor threshold list are input into the tool damage assessment model, and the tool predicted damage distribution information is output.

[0089] In step S400 of the embodiment of the present application, a machine learning model is used to analyze the tool health factor and threshold data, predict the damage distribution of various parts of the tool (such as the amount of tool surface wear), and provide damage status input for digital twin modeling.

[0090] In step S400 of the embodiment of the present application, the tool damage assessment model is trained using machine learning using multiple sets of data, wherein any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information, including:

[0091] Based on the rated interval of tool health factor, randomly configure tool health factor recording data and health factor threshold recording data;

[0092] Based on the preset task type, the tool health factor record data and the health factor threshold record data as constraints, a plurality of tool damage distribution information is collected for a preset tool model;

[0093] Based on the plurality of tool damage distribution information, performing damage concentration value calculation for the same position and attribute, and obtaining a label identifying the tool damage distribution information;

[0094] The tool damage assessment model, the preset tool model and the preset task type are stored in association and added into the tool damage assessment model library.

[0095] The input data of the tool damage assessment model includes a health factor list and a health factor threshold list. The output data is the tool damage distribution information. Among them, the health factor list is the health factor data of the specific tool currently executing the task, such as 12.5m / s 2 , 78°C, 280N; the health factor threshold list is obtained through the above steps. The predicted tool damage distribution information can include the wear location and degree of the tool. Through machine learning model analysis, a set of predicted damage distribution information is generated: the predicted wear amount of the rake face wear (VB) is 0.25mm; the predicted wear amount of the flank face wear (VC) is 0.18mm. The sample data used to train the tool damage assessment model must be no less than 5,000 sets (covering different tool models, processing materials, and cutting parameters).

[0096] The prediction is based on the fact that if the maximum vibration amplitude is close to the threshold, it indicates that the tool stability is reduced; and if the maximum tool temperature is too high, it will accelerate the softening of the tool material and increase the wear rate.

[0097] Among them, some tool damage distribution information is collected. Under the constraints of the preset task type and tool health factor record data and the health factor threshold, before and after each use of the tool, the tool front face wear VB (unit: mm) and the back face wear VC (unit: mm) are measured using an optical microscope or a 3D scanner, respectively, with an accuracy of 0.01 mm, to obtain some tool damage distribution information.

[0098] Among them, based on the damage distribution information of the aforementioned tools, the concentration value calculation of damage with the same attributes at the same position is performed to obtain labels that identify the tool damage distribution information, and the labels are grouped according to the wear position (front blade, back blade) and attributes (VB, VC wear), such as grouping all front blade VB data into one group. Then the mean is calculated for each group of data, such as the mean VB wear of the front blade is 0.2mm, and the mean VC wear of the back blade is 0.15mm. The scattered damage data is converted into labels that can be used for model training to ensure that the labels can reflect the concentration trend and distribution range of damage with the same attributes at the same position, and then the labeled data is used to train the tool damage assessment model.

[0099] Alternatively, the tool damage assessment model can be built using a multilayer perceptron (MLP) neural network. The number of neurons in the input layer is the same as the number of input features; there are two hidden layers, each with 16 neurons, using ReLU as the activation function; the output layer has two neurons (VB wear prediction value, VC wear prediction value); the loss function uses mean squared error (MSE); and the optimizer uses Adam (with a learning rate of 0.001, β1 = 0.9, and β2 = 0.999).

[0100] In the training of the tool damage assessment model, the batch size is set to 32; the number of training rounds is 200; the validation set ratio is 20%; if the validation loss does not decrease for 10 consecutive rounds, the training is stopped to prevent overfitting. When the MSE of the validation set decreases by less than 0.001 for 10 consecutive rounds, the model is judged to have converged and the tool damage assessment model is obtained.

[0101] In the embodiment of the present application, due to the significant differences in wear mechanisms and health factor thresholds for different tool models, materials (such as carbide / ceramic), and task types (such as turning / milling), it is necessary to train multiple independent tool damage assessment models according to the above training method based on the different health factor records and health factor threshold records corresponding to different tool models and task types, and labels identifying tool damage distribution information, to correspond to different tool models and task types.

[0102] Due to the complexity of tool working conditions in actual applications, associating the tool damage assessment model with the tool model and task type allows the tool damage assessment model to adapt to the uniqueness of different tools and tasks. Therefore, it is necessary to associate the tool damage assessment model, the preset tool model and the preset task type and store them in the tool damage assessment model library. That is, first extract the three types of information: tool damage assessment model, preset tool model and preset task type, and then establish an association relationship between the three. For example, use the tool model and task type as index keys and store the corresponding model as a value. Finally, according to this association structure, add the data to the tool damage assessment model library so that the matching assessment model can be quickly retrieved based on the tool model and task type.

[0103] Finally, the tool health factor list and the health factor threshold list are input into the tool damage assessment model to output the tool predicted damage distribution information (such as the rake face wear VB is 0.2 mm, and the flank face wear VC is 0.15 mm).

[0104] In step S500 of the embodiment of the present application, a tool digital twin model is constructed based on the tool damage distribution information to perform tool health status assessment, including:

[0105] Obtaining a first execution task, wherein the first execution task has a predefined tool reference digital twin model;

[0106] When the tool reference digital twin model is inconsistent with the tool digital twin model, marking the first execution task as abnormal;

[0107] When the tool reference digital twin model is consistent with the tool digital twin model, marking the first execution task as normal;

[0108] The abnormal flag or the normal flag is combined with the first execution task, added to the tool health status assessment result, and sent to the tool management end.

[0109] In an embodiment of the present application, a benchmark digital twin model is obtained. For example, based on the current first execution task (such as "external cylindrical turning"), a predefined tool benchmark digital twin model (such as a standard three-dimensional model of a T-700 turning tool, including material properties, geometric parameters and ideal wear threshold) is retrieved from a model library.

[0110] The predicted damage distribution information generated by S400 (such as rake face wear VB of 0.2mm and flank face wear VC of 0.15mm) is input into the benchmark model to drive the model parameter update and generate a digital twin model that reflects the current tool status (such as color marking of the tool face wear area).

[0111] If the wear amount and damage location of the actual model exceed the ideal threshold of the benchmark model (e.g., VB>0.3mm), it is judged as "inconsistent" and the task is marked as "abnormal"; if all parameters are within the range of the benchmark model, it is judged as "consistent" and the task is marked as "normal"

[0112] Finally, the identification result (abnormal / normal) and the first execution task information (such as tool model, task type, etc.) are packaged and sent to the tool management end for operator monitoring or decision-making (such as triggering a tool change warning).

[0113] Among them, the method of obtaining the tool reference digital twin model is an existing technology and is well known to those skilled in the art, so it will not be described here in detail. Figure 2 As shown, based on the same inventive concept as the method for dynamic evaluation of tool health status of a tool magazine based on digital twins provided in the first embodiment, the embodiment of the present invention further provides a dynamic evaluation system for tool health status of a tool magazine based on digital twins, comprising:

[0114] The service data collection module 11 is used to receive the tool service data of the tool compartment of the tool magazine from the tool outbound request terminal when the tool is returned to the warehouse, wherein the tool service data includes the tool health factor, tool working parameters and tool execution task;

[0115] A tool life backtracking module 12 is used to backtrack tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list;

[0116] A healthy tool analysis module 13 is configured to retrieve a number of healthy tool samples for statistical analysis based on the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints, and obtain a health factor threshold list;

[0117] a tool damage assessment module 14, configured to perform tool damage assessment based on the tool health factor list and the health factor threshold list, and generate tool damage prediction distribution information;

[0118] The health status assessment module 15 is used to construct a tool digital twin model to perform tool health status assessment based on the tool damage distribution information.

[0119] Furthermore, the service data collection module 11 includes the following execution steps:

[0120] The tool life cycle data is updated according to the tool execution task and the tool working time.

[0121] Furthermore, the healthy tool analysis module 13 includes the following execution steps:

[0122] Based on the tool service task list and the tool service time list, a first-level constraint condition is constructed; based on the tool working parameters and the tool execution task, a second-level constraint condition is constructed;

[0123] Collecting a number of first-level healthy tool samples that meet the first-level constraint conditions;

[0124] Collecting the plurality of healthy tool samples that meet the second-level constraint conditions from the first-level healthy tool samples;

[0125] A central tendency analysis of the same attributes is performed on the health factors of the healthy tool samples to obtain a plurality of distribution intervals of attribute representative values ​​of the health factors, and to construct the health factor threshold list.

[0126] The health factors of the healthy tool samples are subjected to a central tendency analysis of the same attributes to obtain a plurality of representative value distribution intervals of the health factor attributes, and the health factor threshold list is constructed, including:

[0127] Obtain multiple health factor attribute rating areas;

[0128] Perform same-attribute intersection interval statistics on the multiple health factor attribute representative value distribution intervals and the multiple health factor attribute rated areas to obtain multiple health factor attribute fitting intervals and construct the health factor threshold list.

[0129] Wherein, based on the tool service task list and the tool service time list, a first-level constraint condition is constructed, including:

[0130] Extracting a first tool service task type and a first task type service duration from the tool service task list and the tool service duration list;

[0131] Taking the first tool service task type as a constraint, collecting a number of tool damage states before service, a number of tool damage states after service, and a number of tool service times that correspond to each other;

[0132] Performing consistency comparison on the damage states of the plurality of cutting tools before service and the damage states of the plurality of cutting tools after service to obtain a plurality of consistency comparison results;

[0133] Based on the plurality of consistency comparison results, extracting a set of selected tool service times having inconsistent comparison results from the plurality of tool service times;

[0134] Calculating a central value of the selected tool service time set and setting it as the first tool service task type sensitive time;

[0135] When the service time of the first task type is less than the sensitive time of the first tool service task type, the first tool service task type and the service time of the first task type are deleted from the tool service task list and the tool service time list; when the service time of the first task type is greater than or equal to the sensitive time of the first tool service task type, no change is made;

[0136] When the traversal of the tool service task list and the tool service duration list is completed, all updated tool service task lists and updated tool service duration lists are obtained, and the first-level constraint condition is constructed.

[0137] Furthermore, the tool damage assessment module 14 includes the following execution steps:

[0138] Performing a tool damage assessment based on the tool health factor list and the health factor threshold list to generate tool damage prediction distribution information includes:

[0139] matching a tool damage assessment model from a tool damage assessment model library based on the tool execution task and tool model, wherein the tool damage assessment model is trained using machine learning through multiple sets of data, and any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information;

[0140] The tool health factor list and the health factor threshold list are input into the tool damage assessment model, and the tool predicted damage distribution information is output.

[0141] The tool damage assessment model is trained using machine learning through multiple sets of data, wherein any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information, including:

[0142] Based on the rated interval of tool health factor, randomly configure tool health factor recording data and health factor threshold recording data;

[0143] Based on the preset task type, the tool health factor record data and the health factor threshold record data as constraints, a plurality of tool damage distribution information is collected for a preset tool model;

[0144] Based on the plurality of tool damage distribution information, performing damage concentration value calculation for the same position and attribute, and obtaining a label identifying the tool damage distribution information;

[0145] The tool damage assessment model, the preset tool model and the preset task type are stored in association and added into the tool damage assessment model library.

[0146] Furthermore, the health status assessment module 15 includes the following execution steps:

[0147] Obtaining a first execution task, wherein the first execution task has a predefined tool reference digital twin model;

[0148] When the tool reference digital twin model is inconsistent with the tool digital twin model, marking the first execution task as abnormal;

[0149] When the tool reference digital twin model is consistent with the tool digital twin model, marking the first execution task as normal;

[0150] The abnormal flag or the normal flag is combined with the first execution task, added to the tool health status assessment result, and sent to the tool management end.

[0151] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0152] Those skilled in the art will appreciate that embodiments of the present invention may provide methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0153] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0154] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxesFigure 1 The function specified in one or more boxes.

[0155] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0156] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0157] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A dynamic evaluation method for tool health status of tool magazine based on digital twin, characterized by: include: When the tool is returned to the warehouse, the tool service data of the tool compartment of the tool magazine is received from the tool out-of-warehouse request end, wherein the tool service data includes the tool health factor, the tool working parameters and the tool execution task; Reviewing tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list; Using the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints, a number of healthy tool samples are retrieved for statistical analysis to obtain a health factor threshold list; Performing tool damage assessment based on the tool health factor list and the health factor threshold list to generate tool damage prediction distribution information; Based on the tool predicted damage distribution information, a tool digital twin model is constructed to perform tool health status assessment.

2. The method according to claim 1, wherein The tool service data also includes tool working time, including: The tool life cycle data is updated according to the tool execution task and the tool working time.

3. The method according to claim 1, wherein Using the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints, a number of healthy tool samples are retrieved for statistical analysis to obtain a health factor threshold list, including: Based on the tool service task list and the tool service time list, a first-level constraint condition is constructed; based on the tool working parameters and the tool execution task, a second-level constraint condition is constructed; Collecting a number of first-level healthy tool samples that meet the first-level constraint conditions; Collecting the plurality of healthy tool samples that meet the second-level constraint conditions from the first-level healthy tool samples; A central tendency analysis of the same attributes is performed on the health factors of the healthy tool samples to obtain a plurality of distribution intervals of attribute representative values ​​of the health factors, and to construct the health factor threshold list.

4. The method according to claim 3, wherein Performing a central tendency analysis of the attributes of the health factors of the healthy tool samples to obtain a plurality of distribution intervals of representative values ​​of the health factor attributes, and constructing the health factor threshold list, including: Obtain multiple health factor attribute rating areas; Perform same-attribute intersection interval statistics on the multiple health factor attribute representative value distribution intervals and the multiple health factor attribute rated areas to obtain multiple health factor attribute fitting intervals and construct the health factor threshold list.

5. The method according to claim 3, wherein Based on the tool service task list and the tool service time list, a first-level constraint condition is constructed, including: Extracting a first tool service task type and a first task type service duration from the tool service task list and the tool service duration list; Taking the first tool service task type as a constraint, collecting a number of tool damage states before service, a number of tool damage states after service, and a number of tool service times that correspond to each other; Performing consistency comparison on the damage states of the plurality of cutting tools before service and the damage states of the plurality of cutting tools after service to obtain a plurality of consistency comparison results; Based on the plurality of consistency comparison results, extracting a set of selected tool service times having inconsistent comparison results from the plurality of tool service times; Calculating a central value of the selected tool service time set and setting it as the first tool service task type sensitive time; When the service time of the first task type is less than the sensitive time of the first tool service task type, the first tool service task type and the service time of the first task type are deleted from the tool service task list and the tool service time list; when the service time of the first task type is greater than or equal to the sensitive time of the first tool service task type, no change is made; When the traversal of the tool service task list and the tool service duration list is completed, all updated tool service task lists and updated tool service duration lists are obtained, and the first-level constraint condition is constructed.

6. The method according to claim 1, wherein Performing a tool damage assessment based on the tool health factor list and the health factor threshold list to generate tool damage prediction distribution information includes: matching a tool damage assessment model from a tool damage assessment model library based on the tool execution task and tool model, wherein the tool damage assessment model is trained using machine learning through multiple sets of data, and any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information; The tool health factor list and the health factor threshold list are input into the tool damage assessment model, and the tool predicted damage distribution information is output.

7. The method according to claim 6, wherein The tool damage assessment model is trained using machine learning through multiple sets of data, wherein any set of the multiple sets of data includes tool health factor record data of a preset task type, health factor threshold record data, and a label identifying tool damage distribution information, including: Based on the rated interval of tool health factor, randomly configure tool health factor recording data and health factor threshold recording data; Based on the preset task type, the tool health factor record data and the health factor threshold record data as constraints, a plurality of tool damage distribution information is collected for a preset tool model; Based on the plurality of tool damage distribution information, performing damage concentration value calculation for the same position and attribute, and obtaining a label identifying the tool damage distribution information; The tool damage assessment model, the preset tool model and the preset task type are stored in association and added into the tool damage assessment model library.

8. The method according to claim 1, wherein Based on the tool damage prediction distribution information, a tool digital twin model is constructed to perform tool health status assessment, including: Obtaining a first execution task, wherein the first execution task has a predefined tool reference digital twin model; When the tool reference digital twin model is inconsistent with the tool digital twin model, marking the first execution task as abnormal; When the tool reference digital twin model is consistent with the tool digital twin model, marking the first execution task as normal; The abnormal flag or the normal flag is combined with the first execution task, added to the tool health status assessment result, and sent to the tool management end.

9. The tool magazine tool health status dynamic assessment system based on digital twin is characterized by: include: A service data acquisition module is used to receive tool service data of the tool compartment of the tool magazine from the tool outbound request terminal when the tool is returned to the warehouse, wherein the tool service data includes tool health factors, tool working parameters and tool execution tasks; A tool life backtracking module is used to backtrack tool life cycle data, wherein the tool life cycle data includes a tool service task list and a tool service time list; A healthy tool analysis module is used to retrieve a number of healthy tool samples for statistical analysis based on the tool service task list, the tool service time list, the tool working parameters, and the tool execution task as constraints to obtain a health factor threshold list; a tool damage assessment module, configured to perform tool damage assessment based on the tool health factor list and the health factor threshold list, and generate tool damage prediction distribution information; The health status assessment module is used to construct a tool digital twin model based on the tool damage distribution information and perform tool health status assessment.

Images