Intelligent scheduling system for tooling inventory throughout its entire lifecycle for cost optimization
The intelligent tool lifecycle management system solves the problems of low tool utilization and disordered inventory scheduling, and realizes refined management and cost optimization of the tool lifecycle, thereby improving resource utilization and processing quality.
Patent Information
- Application Number
- CN202611460545.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-20
- Publication Date
- 2026-11-17
AI Technical Summary
Existing technologies fail to effectively manage the entire lifecycle of cutting tools, resulting in low tool utilization, serious resource waste, inability to meet the needs of modern precision machining, disordered inventory scheduling, and high production costs.
The system acquires machining parameters through a data acquisition module, calculates the cutting edge radius through a working condition decoupling and tool status extraction module, performs status management and life assessment through a tool life cycle modeling module, conducts intelligent scheduling through an inventory optimization and scheduling decision module, and performs adaptive optimization through a closed-loop update module, thus building an intelligent management and control system.
It enables refined management of the entire tool lifecycle, improves the accuracy of tool condition detection, reduces waste, lowers production costs, and ensures machining quality and resource utilization.
Smart Images

Figure CN123113771A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of inventory tool management technology, and provides an intelligent scheduling system for the entire lifecycle of tool inventory oriented towards cost optimization, solving the technical problems mentioned in the background art. Background Technology
[0002] In the field of CNC machining, cutting tools are core consumable production resources. The level of control over tool wear, lifecycle management, and inventory scheduling directly affects the machining accuracy, production continuity, and overall production costs in the workshop. Traditional tool management and inventory scheduling technology systems have many inherent defects, especially the lack of refined control over the entire tool lifecycle. The overall technology is crude and lacks intelligence, making it difficult to adapt to the production needs of modern precision machining.
[0003] Current technologies lack a standardized lifecycle management model for cutting tools. They fail to differentiate dressing thresholds and maximum dressing times based on the performance differences of tools made of different materials such as cemented carbide and high-speed steel. Furthermore, they cannot distinguish the suitability conditions for first and second dressing operations and generally lack a post-dressing accuracy verification step. This management model fails to deeply explore the repairable value of cutting tools, easily leading to premature scrapping and excessive replacement of reusable tools, resulting in low tool reuse rates, significant material waste, and a lack of a comprehensive full-cycle data traceability system, relying heavily on manual, extensive management.
[0004] Current technologies mostly rely on a single machining time to determine tool life, failing to incorporate key data such as tool edge wear rate, cumulative usage time, and wear from multiple dressing cycles. This crude method of life prediction cannot adapt to the tool wear patterns under different machining loads and dressing conditions. It easily leads to problems such as machining with faulty tools, exceeding the tool lifespan and resulting in defective workpieces, or prematurely scrapping tools in good condition. This not only fails to guarantee machining quality but also further exacerbates production losses.
[0005] At the inventory scheduling level, the shortcomings of existing technologies are particularly prominent. They fail to incorporate intelligent scheduling based on the actual wear status, remaining lifespan, and remaining number of repair cycles of the cutting tools. They also lack multi-objective cost optimization mechanisms and standardized inventory classification and tiered management strategies. Furthermore, they cannot coordinate various production costs related to inventory storage, equipment downtime, tool replacement, and edge trimming. Tool resources are mismatched with processing tasks, and brand-new tools, trimmed tools, and tools nearing the end of their lifespan are mixed in scheduling, resulting in unreasonable resource allocation. Summary of the Invention
[0006] This invention provides an intelligent scheduling system for tool inventory throughout its entire lifecycle, oriented towards cost optimization, thus solving the technical problems mentioned in the background art.
[0007] This invention provides a cost-optimized intelligent scheduling system for tool inventory throughout its entire lifecycle, comprising:
[0008] The data acquisition module is used to acquire cutting force signals Fx, Fy, and Fz, as well as three types of machining parameters: feed rate f, depth of cut ap, and cutting speed v, during the tool machining process.
[0009] The working condition decoupling and tool state extraction module is used to eliminate the interference of machining conditions on cutting forces, extract tool wear characteristics, and calculate and output the tool cutting edge fillet radius r.
[0010] The tool lifecycle modeling module is used to statistically analyze the tool's usage time and state change patterns based on the cutting edge fillet radius r, generate wear state data and cutting edge dressing records, and realize traceable management of tool life. At the same time, it determines whether the tool meets the cutting edge dressing conditions based on the initial cutting edge fillet radius r, plans the number of cutting edge dressing times, tracks the state wear after dressing, and performs accuracy verification on the dressed cutting edge fillet radius.
[0011] The tool condition prediction and life assessment module is used to combine tool usage time, wear data and dressing records to predict the life of tools in stock and in use, output remaining life information, classify health status levels, and define the critical state of tool scrapping.
[0012] The inventory optimization and scheduling decision module is used to manage inventory access operations based on remaining life, critical scrap status, combined with the number of times the cutting edge can be repaired, production task requirements, and inventory constraints including the maximum inventory capacity of cutting tools and the minimum safety stock, and to generate scheduling, repair and scrapping schemes through comprehensive cost optimization.
[0013] The closed-loop update module is used to collect real-time cutting force and working condition data, and combine the cutting edge radius r, remaining life and scrap status data. Based on the deviation between the actual wear status and the prediction results, it updates the wear judgment criteria and life assessment basis, dynamically corrects the judgment and prediction results, and realizes adaptive updates.
[0014] Furthermore, the data acquisition module adopts a high-frequency synchronous acquisition mechanism, with the sampling frequency set according to the processing scenario: 1000Hz for fine processing, 500Hz for rough processing, and 200Hz for regular batch processing. It marks the corresponding timestamps of the two types of data to complete the time sequence alignment, uses the 3σ criterion to remove abnormal data, uses linear interpolation to complete missing data, and outputs standardized multi-source state data.
[0015] Furthermore, the working condition decoupling and tool state extraction module calculates the resultant force amplitude by vector synthesis of the synchronously acquired three-axis cutting force components Fx, Fy, and Fz, obtaining the measured scalar value of the cutting force; establishes a benchmark mapping model between the cutting force and working condition parameters, calculates the theoretical cutting force without wear interference under the current working condition, and obtains the pure tool wear characteristic quantity by subtracting the measured cutting force from the theoretical cutting force, thus filtering out the mechanical influence caused by working condition changes; and incorporates a fitting function between the tool wear characteristic and the cutting edge fillet radius, substituting the wear characteristic quantity into the fitting function to calculate the tool cutting edge fillet radius r.
[0016] Furthermore, the tool lifecycle modeling module sets a tool dressing threshold range and compares the real-time cutting edge radius r with the initial standard radius. When the wear of r exceeds the standard allowable range but does not reach the scrap threshold, it is determined that the cutting edge dressing condition is met. The cutting edge dressing condition includes the first dressing condition and the second dressing condition. The first dressing condition is when a brand-new tool wears to the dressing threshold range, and the second dressing condition is when the tool wears to the dressing threshold range again after the first dressing. The maximum number of dressing times is set in combination with the tool material: 1-2 times for carbide tools and 2-3 times for high-speed steel tools. After each dressing, the tool reference radius and wear traceability data are updated synchronously.
[0017] Furthermore, the tool condition prediction and life assessment module sets four levels of tool health thresholds, corresponding to four health status levels: brand new and intact, slightly worn, requiring repair, and nearing scrap. Based on the cumulative tool usage time, wear from previous repairs, and the real-time attenuation rate of the cutting edge radius r, the remaining tool life is calculated through a multi-parameter fusion life calculation model. The state where the cutting edge radius r reaches the limit wear threshold is defined as the critical scrap state, and the limit wear threshold is set according to the tool material and machining accuracy requirements.
[0018] Furthermore, the inventory optimization and scheduling decision module constructs a multi-objective optimization model with four independent optimization objectives: inventory cost, equipment downtime cost, tool replacement cost, and cutting edge trimming cost. The inventory constraints also include procurement cycle and procurement cost. Based on the urgency of production tasks, tool inventory, remaining lifespan, and number of trimming cycles, the module determines the scheme selection preference and uses a non-dominated sorting genetic algorithm to obtain the Pareto optimal solution set. From the optimal solution set, the optimal scheduling combination that fits the current production scenario is selected to avoid production losses caused by premature tool scrapping or processing with defects.
[0019] Furthermore, the inventory optimization and scheduling decision module adopts a classification, partitioning, and hierarchical inventory storage and retrieval strategy; it classifies and stores brand-new unused tools, repaired and usable tools, and tools nearing the end of their service life into different partitions, and manages them hierarchically according to their remaining lifespan; it prioritizes scheduling repaired tools with sufficient remaining lifespan for routine machining tasks, and prioritizes scheduling brand-new tools for high-precision, high-load machining tasks.
[0020] Furthermore, the closed-loop update module is equipped with a periodic iterative update mechanism. The update node is determined according to the processing scenario: for regular continuous batch production scenarios, the node is the cumulative processing time of 8 hours, and for small-part high-frequency batch processing scenarios, the node is the cumulative processing time of 500 workpieces. Real-time processing data is collected within each update node cycle, and the deviation between the actual wear state and the predicted state is compared. Based on the deviation, the wear judgment algorithm parameters and life prediction model coefficients are adaptively corrected using the gradient descent method to gradually improve the detection and prediction accuracy.
[0021] Furthermore, after the tool edge trimming is completed, the tool lifecycle modeling module performs accuracy verification on the trimmed edge fillet radius r; if the verification meets the standard, the tool status is updated to trimmed and usable; if the verification fails, the tool is directly determined to be scrapped, and the tool's subsequent use cycle is terminated.
[0022] Furthermore, the inventory optimization and scheduling decision module is linked to the production task scheduling to dynamically match tool resources with processing procedures; for tools with no remaining repair cycles and a remaining lifespan less than the time required for a single batch of tasks, a scrap warning and replacement plan are generated in advance; the time required for a single batch of tasks is calculated and determined based on the processing time of a single piece in the corresponding process and the number of workpieces processed, and a tool procurement plan is generated at the same time to ensure that inventory meets production needs.
[0023] The beneficial effects of this invention are as follows:
[0024] This invention uses the tool cutting edge radius r as a unified detection and judgment benchmark, integrating it into the entire business process of condition identification, lifespan assessment, maintenance control, and inventory scheduling. This constructs an integrated intelligent tool lifecycle management system, effectively addressing industry pain points such as significant interference from operating conditions, inaccurate condition detection, coarse lifespan prediction, low tool utilization, disordered inventory scheduling, high production costs, and uncontrollable downtime risks in traditional tool management. It achieves data-driven, refined, and cost-optimized management of the entire tool process. The system employs a three-level calculation architecture—three-axis cutting force vector synthesis, a benchmark mapping model, and a cutting edge radius fitting function—along with dimensionless processing of wear characteristics, to accurately avoid issues related to feed rate, depth of cut, etc. The interference of cutting speed fluctuations on cutting force signals is eliminated, and the impact of differences in working conditions on wear judgment is removed. Pure wear characteristic quantities are extracted to calculate the r-value, significantly improving the accuracy of tool condition detection and the cross-working-condition versatility of the results. A full lifecycle modeling system is built based on the dynamic decay law of the r-value. Different dressing thresholds and maximum dressing times are configured according to the different tool material characteristics of cemented carbide and high-speed steel, clearly distinguishing between the judgment conditions for the first and second dressings. Simultaneously, after dressing, accuracy verification is carried out based on the r-value to complete a second quality check. The entire process of tool dressing, reuse, and scrapping is standardized, deeply exploring the repairable value of tools, improving tool reuse rate, and reducing new tool procurement consumption and ineffective scrapping losses. This system enables full traceability of tool wear, dressing, and loss data throughout the entire tool lifecycle, avoiding the traditional, experience-based, and crude management model. Based on the real-time decay rate of the r-value, the system integrates multi-dimensional data such as cumulative tool usage time and wear from each dressing cycle to construct a dimensionally consistent, multi-parameter fusion lifespan calculation model. Coupled with a four-level tool health status grading standard and differentiated extreme wear thresholds adapted to different materials and machining conditions, it accurately predicts the remaining tool lifespan and defines the critical state for tool obsolescence. Compared to the traditional single-duration determination method, it can accurately adapt to the tool lifespan decay patterns under different machining loads and dressing conditions, effectively avoiding premature tool obsolescence and machining with defects caused by lifespan prediction errors. To address the issue of stabilizing workpiece machining quality and reducing production defects, the system constructs four cost-independent multi-objective optimization systems based on the tool health status, remaining lifespan, and remaining number of trimming cycles corresponding to the r-value. It integrates various production costs related to inventory, equipment downtime, tool replacement, and edge trimming. Combining multiple constraints such as inventory capacity, safety stock, and procurement cycle, it employs a non-dominated sorting genetic algorithm to solve for the Pareto optimal solution set. It also selects an appropriate scheduling scheme based on dynamic preferences of the production scenario, and implements a smart inventory access strategy with classification, partitioning, and layering. This, along with production task scheduling, achieves differentiated matching between tools and machining scenarios, maximizing tool resource reuse while ensuring high-precision machining quality.Furthermore, the system possesses closed-loop adaptive update capabilities. By periodically collecting actual processing data and comparing it with predicted results, it adaptively corrects the wear judgment algorithm and life prediction model parameters based on deviation feedback, continuously improving the accuracy of condition detection and life prediction. This ensures the long-term stability and reliability of the system, ultimately achieving globally optimal control of overall production costs. Attached Figure Description
[0025] Figure 1 This is a flowchart illustrating the overall operation of the intelligent scheduling system for the entire lifecycle inventory of cutting tools, which is oriented towards cost optimization, according to the present invention.
[0026] Figure 2 This is a flowchart of the data acquisition and standardized preprocessing process of the intelligent scheduling system for the entire lifecycle inventory of cutting tools for cost optimization, as described in this invention.
[0027] Figure 3 This is a flowchart of the working condition decoupling and cutting edge radius calculation of the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization according to the present invention;
[0028] Figure 4 This is a flowchart of the cutting edge dressing judgment and verification management process of the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization, as described in this invention.
[0029] Figure 5 This is a flowchart of the inventory optimization scheduling decision execution process of the intelligent inventory scheduling system for the entire life cycle of cutting tools, which is oriented towards cost optimization according to the present invention.
[0030] Figure 6 This is a flowchart of the closed-loop adaptive iterative update process of the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization according to the present invention. Detailed Implementation
[0031] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, features described in some examples may be combined in other examples.
[0032] like Figures 1-6 As shown, the implementation method of the intelligent scheduling system for tool inventory throughout its entire lifecycle, aimed at cost optimization, is as follows:
[0033] In one embodiment of the present invention, a cost-optimized intelligent scheduling system for the entire lifecycle inventory of cutting tools includes: a data acquisition module, used to acquire cutting force signals Fx, Fy, and Fz, as well as three types of machining parameters: feed rate f, depth of cut ap, and cutting speed v, during the tool machining process.
[0034] The data acquisition module adopts a high-frequency synchronous acquisition mechanism, with the sampling frequency set according to the processing scenario: 1000Hz for fine processing, 500Hz for rough processing, and 200Hz for regular batch processing. It marks the corresponding timestamps for the two types of data to complete the time sequence alignment, uses the 3σ criterion to remove abnormal data, uses linear interpolation to complete missing data, and outputs standardized multi-source status data.
[0035] The data acquisition module is deployed on CNC machining equipment and tool monitoring terminal to acquire three-axis cutting force signals Fx, Fy, and Fz during tool machining, as well as three types of machining condition parameters: feed rate f, depth of cut ap, and cutting speed v. All acquired data are aggregated and integrated in real time to form a multi-source dataset covering tool mechanical state and equipment operating conditions, providing raw data support for subsequent data processing and status analysis.
[0036] The data acquisition module employs a high-frequency synchronous acquisition mechanism, simultaneously activating the three-axis cutting force acquisition unit and the machine tool operating condition data acquisition unit to eliminate the time difference in acquisition between different types of data. The sampling frequency is matched and set according to the accuracy requirements of the machining scenario: 1000Hz for finishing scenarios to ensure the integrity of data acquisition for minor wear; 500Hz for roughing scenarios to balance data accuracy and equipment operating efficiency; and 200Hz for routine batch processing scenarios to adapt to normalized production rhythms.
[0037] During the data acquisition process, the system marks each set of synchronously acquired cutting force data and working condition data with a millisecond-level high-precision timestamp, so that the timestamps of the two types of data corresponding to the same processing time are completely matched, completing the time-series alignment preprocessing of multi-source heterogeneous data and avoiding subsequent data analysis errors caused by time-series misalignment.
[0038] After timing alignment is completed, the system executes a standardized data preprocessing procedure:
[0039] First, abnormal interference data is eliminated by using the 3σ criterion. The arithmetic mean and standard deviation of all cutting force data and working condition parameter data within a single acquisition cycle (1 minute) are calculated. The normal fluctuation range of the data is defined by combining the mean and three times the standard deviation. Abrupt data, electromagnetic interference noise data, and instantaneous abnormal fluctuation data that exceed the range are uniformly judged as abnormal data and directly eliminated.
[0040] To address the issue of localized data loss caused by momentary equipment lag, signal transmission fluctuations, and brief machine tool response delays, the system employs linear interpolation to complete the data. This involves accurately locating the time node corresponding to the missing data, retrieving two sets of valid data adjacent to the time node, calculating the data change gradient per unit time based on the numerical difference and time interval between the two sets of valid data, and combining the time span between the missing node and the nodes before and after it to calculate the corresponding value of the missing node, thus filling in the data gaps and ensuring the temporal continuity of data throughout the entire time period.
[0041] After removing outlier data and completing missing data, the system standardizes the units of measurement, numerical precision, and storage format for all data: cutting force values are uniformly retained to 3 decimal places, with the unit being N; feed rate is uniformly in mm / r, depth of cut is uniformly in mm, and cutting speed is uniformly in m / min. All time-series data are reordered and archived according to the time node order, completing the full data standardization process, and finally outputting standardized multi-source status data that is time-continuous, complete, and free from abnormal interference.
[0042] In one embodiment of the present invention, the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization further includes: a working condition decoupling and tool status extraction module, used to eliminate the interference of machining conditions on cutting forces, extract tool wear characteristics, and calculate and output the tool cutting edge radius r.
[0043] The working condition decoupling and tool state extraction module calculates the resultant force amplitude by vector synthesis of the synchronously acquired three-axis cutting force components Fx, Fy, and Fz, obtaining the measured scalar value of the cutting force; it establishes a benchmark mapping model between the cutting force and working condition parameters, calculates the theoretical cutting force without wear interference under the current working condition, and obtains the pure tool wear characteristic quantity by subtracting the measured cutting force from the theoretical cutting force, thus filtering out the mechanical influence caused by working condition changes; it has a built-in fitting function for tool wear characteristics and cutting edge fillet radius, and calculates the tool cutting edge fillet radius r by substituting the wear characteristic quantity into the fitting function.
[0044] The working condition decoupling and tool status extraction module receives standardized multi-source status data output from the data acquisition module. This data is used to eliminate interference from machining conditions on cutting forces, extract pure tool wear characteristics, calculate and output the tool edge fillet radius r, and provide core status indicators for subsequent full lifecycle management, as shown below:
[0045] 1. Triaxial cutting force vector synthesis
[0046] Since the collected Fx, Fy, and Fz are cutting force components in three orthogonal directions, they need to be converted into unified measured cutting force scalar values first. The system calculates the resultant force amplitude according to the vector synthesis formula: ;
[0047] In the formula, The measured resultant cutting force amplitude is expressed in N. , , These represent the cutting force components in the X, Y, and Z directions, all in Newtons (N). The resultant force amplitude is used as the representative value of the measured cutting force, comprehensively reflecting the mechanical load level during the cutting process.
[0048] 2. Decoupling of the benchmark mapping model from the operating condition
[0049] The system uses batch data statistical analysis to sort out the fluctuation law of cutting force when a single working condition parameter changes independently or when multiple working condition parameters change in conjunction. It summarizes the correlation characteristics between three types of parameters—feed rate, depth of cut, and cutting speed—and the value of cutting force, and establishes a benchmark mapping model between cutting force and working condition parameters. This model is used to calculate the theoretical cutting force under no wear interference and to filter out the mechanical effects caused by working condition changes.
[0050] The benchmark mapping model uses a multivariate linear correlation equation, in the following form: ;
[0051] In the formula: This is the standard theoretical cutting force under current operating conditions without tool wear interference, expressed in N; This refers to the feed rate, expressed in mm / r. This refers to the depth of cut, in mm. This refers to the cutting speed, expressed in m / min. This is the influence coefficient of the feed rate, with units of N·r / mm; This is the influence coefficient of the cutting depth, in N / mm; The coefficient representing the influence of cutting speed is expressed in N·min / m. This is a correction constant, measured in N, used to compensate for inherent operating errors of the equipment and minor environmental disturbances.
[0052] The weighting coefficients and correction constants of the benchmark mapping model were obtained through fitting using standard machining experiments: Under standard experimental conditions, using brand-new, wear-free cutting tools, at least 50 different combinations of working parameters were set, corresponding cutting force data were collected, and the equation coefficients were obtained through least squares fitting. The model fit goodness must meet the following requirements. Only then can it be put into use.
[0053] The aforementioned linear reference mapping model has sufficient fitting accuracy in the range of conventional cutting conditions with medium to low speeds and small to medium feed rates. For high-speed cutting (… m / min) or large feed rate ( For special operating conditions (mm / r), a piecewise linear model or an empirical model in exponential form can be used as a substitute to ensure the accuracy of the reference mapping.
[0054] By substituting the real-time collected operating parameters into the benchmark mapping model, the theoretical cutting force under the current operating condition is calculated. Then compare it with the actual cutting force measured on site. Perform the difference calculation to obtain the pure tool wear mechanical characteristic quantity. : ;
[0055] pure tool wear mechanical characteristics This refers to the change in cutting force caused solely by tool edge wear and minor deformation after completely eliminating interference from operating conditions. It is measured in N and directly reflects the degree of tool wear.
[0056] 3. Dimensionless transformation of wear characteristics and calculation of cutting edge fillet radius
[0057] To eliminate the influence of differences in cutting force levels under different working conditions on wear calculations and to ensure the cross-condition universality of the calculated cutting edge fillet radius, the system first performs dimensionless processing on the absolute force difference to obtain a dimensionless relative wear force characteristic quantity: ;
[0058] In the formula, For reference cutting force, the unit is N. The value is the actual measured cutting force of a brand new standard tool under the nominal reference working condition. The reference working condition is uniformly calibrated according to the tool model and the material of the workpiece being machined.
[0059] The system incorporates a fitting function for tool wear characteristics and cutting edge fillet radius. By substituting the dimensionless relative wear force characteristic into the fitting function, the real-time tool cutting edge fillet radius r is calculated. The fitting function takes the following form: ;
[0060] In the formula: The radius of the tool cutting edge fillet is in μm. It is a dimensionless characteristic quantity of relative wear force; The wear gradient coefficient is dimensionless and ranges from 0.8 to 1.5. The upper limit is used for cemented carbide tools and the lower limit is used for high-speed steel tools. The wear nonlinearity index is dimensionless and has a value of 1.2. This is the cutting edge deformation correction coefficient, dimensionless, with a value range of 0.1 to 0.3, used to compensate for numerical deviations caused by minute deformations of the cutting edge; This is a working condition adaptation constant, with the unit being μm and a value range of 5–15 μm. It is used to match the basic measurement errors of different processing scenarios.
[0061] The above-mentioned fitting function form was selected through comparative experiments with multiple candidate functions, including pure power functions, quadratic polynomials, exponential functions, and power functions combined with linear combinations. Experimental results show that the power function combined with linear combinations exhibits the highest goodness of fit throughout the tool's entire lifespan. It can simultaneously adapt to the entire stage of wear patterns, from initial slow wear to mid-term stable wear and later rapid wear.
[0062] During the calculation process, the system performs multiple rounds of iterative corrections based on the current stable working parameters to further reduce calculation errors and finally output accurate real-time tool cutting edge radius r value.
[0063] In one embodiment of the present invention, the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization further includes: a tool life cycle modeling module, which is used to statistically analyze the usage time and state change patterns of cutting tools based on the cutting edge radius r, generate wear state data and cutting edge dressing records, and realize traceable management of tool life; at the same time, it determines whether the tool meets the cutting edge dressing conditions based on the initial cutting edge radius r, plans the number of cutting edge dressing times, tracks the state loss after dressing, and performs accuracy verification on the dressing cutting edge radius.
[0064] The tool lifecycle modeling module sets the tool dressing threshold range and compares the real-time cutting edge fillet radius r with the initial standard fillet radius. When the wear of r exceeds the standard allowable range but does not reach the scrap threshold, it is determined that the cutting edge dressing condition is met. The cutting edge dressing condition includes the first dressing condition and the second dressing condition. The first dressing condition is when a brand new tool wears to the dressing threshold range, and the second dressing condition is when the tool wears to the dressing threshold range again after the first dressing. The maximum number of dressing times is set in combination with the tool material. For carbide tools, it is 1-2 times, and for high-speed steel tools, it is 2-3 times. After each dressing, the tool reference fillet radius and wear traceability data are updated synchronously.
[0065] After the tool edge dressing is completed, the tool lifecycle modeling module performs accuracy verification on the corner radius r of the dressed edge. If the verification meets the standard, the tool status is updated to the dressed usable status; if the verification fails, the tool is directly judged to be scrapped, and the tool's subsequent use cycle is terminated.
[0066] The tool lifecycle modeling module continuously collects and updates tool edge fillet radius (r) data in real time, statistically analyzes tool usage time and condition changes, generates wear status data and edge dressing records, and achieves traceable tool life management. Simultaneously, it determines tool dressing conditions, plans the number of dressing cycles, tracks post-dressing wear, and verifies the accuracy of the dressed edge, as shown below:
[0067] 1. Full-cycle data traceability and initial baseline acquisition
[0068] The system sequentially calculates the single-processing time and cumulative processing time of each tool, continuously recording the wear trend of the tool under different workpieces, different processing parameters, and different production batches. It completely preserves the full lifecycle status change record of each tool from its first use to its final scrapping. Based on the accumulated status data, the system automatically organizes and generates a tool wear status dataset and a cutting edge dressing ledger, sorting out the changing patterns of tool wear and lifespan decline, and achieving full-cycle data traceability and queryability.
[0069] The initial cutting edge radius r0 of each tool is used as the benchmark for wear comparison and is obtained as follows: For tools with factory calibration parameters, the factory-calibrated r0 value is directly used; for tools without factory calibration data, the cutting edge of a brand-new tool is measured at multiple points using an optical microscope or laser confocal microscope, and the arithmetic mean of more than 5 measurement points is taken as the initial r0, with a single-point measurement accuracy of not less than 1μm. If the deviation between different measurement points exceeds 3μm, the initial cutting edge consistency of the tool is deemed unqualified and it is not included in the system management scope.
[0070] 2. Determining Repair Conditions and Planning the Number of Repairs
[0071] The system sets the tool dressing threshold range and compares the real-time cutting edge fillet radius r with the initial standard fillet radius r0. When the wear of r exceeds the standard allowable range but does not reach the scrap threshold, it is determined that the cutting edge dressing condition is met.
[0072] The cutting edge dressing conditions are divided into initial dressing conditions and secondary dressing conditions: a brand new tool that has never undergone any dressing operation will trigger the initial dressing judgment when the wear value falls into the dressing threshold range during continuous machining; a tool that has already completed one dressing operation will trigger the secondary dressing judgment when it wears to the dressing threshold range again during subsequent continuous machining.
[0073] The dressing threshold range is set differently depending on the tool model, material, and machining accuracy requirements. Taking the machining of 45# steel with a 10mm diameter carbide end mill as an example, the initial cutting edge radius r0 = 15μm, and the dressing threshold range is set to 30μm~50μm, that is, the dressing judgment is triggered when the cutting edge radius exceeds the initial value by 100%~233%.
[0074] The system takes into account the differences in physical properties of different tool materials and sets the maximum number of dressing times differently: Carbide tools have high hardness but weak toughness, and are prone to chipping and breakage during grinding and dressing, with limited repair space. The maximum number of dressing times is set to 1-2 times. Tools with light wear can be dressed twice, while tools with heavy wear only support a single dressing. High-speed steel tools have excellent toughness and stable wear resistance, and can be repeatedly ground and repaired without being easily damaged. The maximum number of dressing times is set to 2-3 times, which is suitable for high-frequency dressing and repair needs.
[0075] 3. Accuracy verification and status update after trimming
[0076] After each tool edge dressing operation is completed, the system immediately collects the measured value of the finished edge fillet radius and conducts accuracy verification: the industry-standard machining accuracy parameters corresponding to each type of tool are pre-entered as the basis for verification and comparison, the measured edge fillet radius r value after dressing is accurately compared with the standard parameters, the range of numerical deviation is calculated, and it is determined whether the edge accuracy of the dressed tool meets the production requirements of the corresponding machining scenario.
[0077] If the verification results show that the deviation of the cutting edge fillet radius r is within the standard allowable range, and the tool's cutting edge accuracy and wear condition meet the subsequent machining standards, the system will immediately update the tool's life cycle status, changing the tool status from pending repair to repairable. Simultaneously, the measured repaired cutting edge fillet radius will be updated to the new reference fillet radius, replacing the original reference parameter, as the basis for comparison in subsequent wear detection. At the same time, complete traceability data for this repair will be recorded, including repair operation time, material loss from cutting edge grinding, cutting edge accuracy parameters after repair, machining conditions after repair, and cumulative machining time, and updated to the tool life cycle log.
[0078] If the verification result shows that the deviation of the cutting edge radius r exceeds the standard allowable range, and the accuracy of the tool cutting edge after repair cannot meet the basic machining requirements, and there is no value or feasibility in repairing it again, the system will directly determine that the tool has reached the scrap condition, immediately lock the tool's usage rights, terminate all subsequent use cycles of the tool, generate a standardized tool scrap record, update the tool's full life cycle ledger synchronously, and complete the scrapping and archiving of unqualified repaired tools.
[0079] In one embodiment of the present invention, the intelligent scheduling system for the entire life cycle of tool inventory for cost optimization further includes: a tool status prediction and life assessment module, which is used to combine tool usage time, wear data and maintenance records to predict the life of tools in stock and in use, output remaining life information, classify health status levels, and define the critical state of tool scrapping.
[0080] The tool condition prediction and life assessment module sets four levels of tool health thresholds, corresponding to four health status levels: brand new and intact, slightly worn, requiring dressing, and nearing scrap. Based on the cumulative tool usage time, wear from each dressing operation, and the real-time attenuation rate of the cutting edge radius r, the remaining tool life is calculated through a multi-parameter fusion life calculation model. The state where the cutting edge radius r reaches the limit wear threshold is defined as the critical scrap state, and the limit wear threshold is set according to the tool material and machining accuracy requirements.
[0081] The tool condition prediction and life assessment module retrieves the cumulative usage time, wear status data at each stage, and full data of all cutting edge dressing records accumulated throughout the tool's entire life cycle. It performs life estimation calculations for both tools stored in warehouses and tools continuously in use on-site, outputs remaining life information, classifies health status levels, and defines the critical state for scrapping, as shown below:
[0082] 1. Tool health status classification
[0083] The system sets four levels of tool health thresholds, corresponding to four health status levels. The range values are set differently based on the tool type, machining material, and process requirements, enabling refined classification and judgment of tool operating status. Taking the machining of 45# steel with a 10mm diameter carbide end mill as an example, the classification and judgment logic is as follows:
[0084] Brand new and in good condition: the deviation between the fillet radius of the cutting edge and the initial standard radius is within the range of 0 to 10%, and there are no abnormal vibrations, abnormal fluctuations in cutting force, or other adverse working conditions during the machining process;
[0085] Slight wear condition: The deviation of the cutting edge radius is within the range of 10% to 100%, which does not have a negative impact on the dimensional accuracy and surface roughness of the workpiece, and the workpiece processing quality fully meets the standards;
[0086] Dressing required: The deviation of the cutting edge radius falls within the dressing threshold range (100%~233%, corresponding to 30μm~50μm), and the machining accuracy fluctuates slightly but still meets the basic production quality requirements.
[0087] Nearing the end of its service life: The deviation of the cutting edge radius is close to the limit wear threshold, the machining accuracy is significantly reduced, and the stability of the workpiece machining quality continues to decrease.
[0088] 2. Calculation of remaining useful life
[0089] The system calculates the remaining tool life based on the cumulative tool usage time, wear from each dressing cycle, and the real-time corner radius (r) decay rate, using a multi-parameter fusion life calculation model. The model adopts a modified Taylor tool life equation. To eliminate the influence of the magnitude of operating parameters and ensure dimensional consistency, the operating parameters are dimensionless, as shown in the following formula: ;
[0090] In the formula: The remaining tool life, in minutes, refers to the effective duration for which the tool can continue to operate stably under the current wear condition and machining conditions. This is the tool life constant, in minutes, and is related to the tool material. It is 250 for carbide tools and 160 for high-speed steel tools. This is a dimensionless material correction factor, adapted to the material of the workpiece: 1.0 for machining 45# steel, 1.4 for machining aluminum alloy, and 0.7 for machining stainless steel. It can be further calibrated and corrected through cutting tests. The dressing attenuation factor is dimensionless and is used to reflect the attenuation characteristics of tool life after dressing: 0.85 after the first dressing, 0.72 after the second dressing, and 0.6 after the third dressing. The values are based on the dressing life test statistics of the same type of tool. , , These are the current cutting speed, feed rate, and depth of cut, respectively. , , These are the reference cutting speed, reference feed rate, and reference depth of cut, respectively, which are consistent with the nominal reference working conditions of the reference mapping model; , , These are the influence indices of cutting speed, feed rate, and depth of cut, respectively. They are all dimensionless quantities and are taken as 1.2, 0.8, and 0.5, respectively. This represents the cumulative time the tool has been used, in minutes.
[0091] During the calculation process, the system dynamically corrects the calculation results by combining the attenuation rate of the real-time blade fillet radius r, further improving the accuracy of life prediction.
[0092] 3. Delineation of the critical state for scrapping
[0093] The system defines the state where the cutting edge fillet radius r reaches the limit wear threshold as the critical scrap state. The limit wear threshold is set differently according to the tool material and machining accuracy requirements.
[0094] In the precision machining of cemented carbide tools, the limit wear threshold is set to 65μm to ensure the machining quality of high-precision workpieces.
[0095] For rough machining scenarios using carbide tools, the limit wear threshold is appropriately relaxed to 80μm to meet the requirements of rough machining processes.
[0096] For conventional machining scenarios using high-speed steel tools, the limit wear threshold is set to 90μm, while for roughing scenarios it can be relaxed to 110μm.
[0097] When the system detects that the real-time corner radius r of the cutting tool reaches the limit wear threshold of the corresponding machining scenario, it determines that the tool is in a critical scrap state and locks the tool's usage permission, restricting the tool from continuing to be used in machining and production, so as to avoid unqualified tools affecting production quality.
[0098] In one embodiment of the present invention, the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization further includes: an inventory optimization and scheduling decision module, which is used to manage inventory access operations based on remaining life, critical scrap status, combined with the number of times the cutting edge can be repaired, production task requirements, and inventory constraints including the maximum inventory capacity of cutting tools and the minimum safety stock, and to generate scheduling, repair and scrapping schemes through comprehensive cost optimization.
[0099] The inventory optimization and scheduling decision module constructs a multi-objective optimization model with four independent optimization objectives: inventory cost, equipment downtime cost, tool replacement cost, and cutting edge trimming cost. Inventory constraints also include procurement cycle and procurement cost. Based on the urgency of production tasks, tool inventory, remaining lifespan, and number of trimming cycles, the module determines the scheme selection preference. A non-dominated sorting genetic algorithm is used to obtain the Pareto optimal solution set. From the optimal solution set, the optimal scheduling combination that fits the current production scenario is selected to avoid production losses caused by premature tool scrapping or processing with defects.
[0100] The inventory optimization and scheduling decision module adopts a classification, partitioning and hierarchical inventory storage and retrieval strategy; it classifies and stores brand new unused tools, repaired and usable tools and tools nearing the end of their lifespan into different partitions, and manages them hierarchically according to their remaining lifespan; it prioritizes scheduling repaired tools with sufficient remaining lifespan for regular machining tasks, and prioritizes scheduling brand new tools for high-precision and high-load machining tasks.
[0101] The inventory optimization and scheduling decision module is linked to the production task scheduling to dynamically match tool resources with processing procedures. For tools with no remaining dressing times and a remaining lifespan less than the time required for a single batch of tasks, a scrap warning and replacement plan are generated in advance. The time required for a single batch of tasks is calculated and determined based on the processing time of a single piece in the corresponding process and the number of workpieces processed. At the same time, a tool procurement plan is generated to ensure that inventory meets production needs.
[0102] The inventory optimization and scheduling decision module integrates the remaining life data, scrapping threshold judgment criteria, and remaining number of times the cutting edge can be repaired for all tools in the system. Combined with the current production task requirements and inventory constraints of the workshop, it uniformly manages the entire process of tool storage and retrieval, including tool entry, exit, retrieval, and return. Through multi-objective optimization, it generates scheduling, repair, and scrapping plans to reduce overall production costs while ensuring production quality and schedule.
[0103] 1. Construction of a multi-objective optimization model
[0104] The system constructs a multi-objective optimization model with four independent optimization objectives: inventory cost, equipment downtime cost, tool replacement cost, and cutting edge dressing cost. The goal is to achieve the optimal overall cost. The four objectives are defined as follows:
[0105] Inventory costs This includes costs for tool storage and safekeeping, rust prevention and maintenance, and ledger management, which are positively correlated with the quantity of tools in stock and the duration of storage.
[0106] Equipment downtime costs These include equipment standby losses, production delay losses, and order delay compensation costs, which are positively correlated with downtime caused by untimely tool supply.
[0107] Tool replacement cost This includes the cost of tool disassembly and assembly, labor costs, and the cost of purchasing new tools.
[0108] Edge trimming cost This includes the cost of polishing consumables, labor costs for finishing, and post-finishing testing costs.
[0109] The objective function vector for multi-objective optimization is: ;
[0110] The model constraints include: ;
[0111] In the formula: This represents the current tool inventory. For maximum inventory capacity, Minimum safety stock; For the procurement and delivery cycle of cutting tools, To ensure that procurement and replenishment can match the production schedule, production tasks are scheduled to run on time.
[0112] 2. Model Solving and Scheme Selection
[0113] The system employs the Non-Dominated Sorting Genetic Algorithm (NSGA-II) to solve the multi-objective optimization model. The algorithm parameters and their selection criteria are as follows:
[0114] With a population size of 100, a balance can be achieved between solution accuracy and computational efficiency in scheduling scenarios with a scale of up to 50 tools. Tests have shown that the number of iterations to converge to the Pareto front is stable within 300 generations, taking into account both computational speed and solution diversity.
[0115] The crossover probability is set to 0.8: This is a commonly recommended value for genetic algorithms, which can maintain population diversity while ensuring the genetic characteristics of excellent solutions;
[0116] Setting the mutation probability to 0.2 effectively avoids the algorithm getting stuck in local optima and ensures global search capability;
[0117] The maximum number of iterations is set to 500: This ensures that the algorithm converges fully. For small-scale scheduling scenarios, the number of iterations can be appropriately reduced to improve efficiency.
[0118] The decision variables are encoded as the usage status of each tool (in use / in stock / repair / scrapped), repair time, and scheduling priority. The algorithm outputs a Pareto optimal solution set through non-dominated sorting and crowding distance calculation. Each solution in the solution set is a non-dominated solution, meaning it is impossible to optimize a particular objective without degrading other objectives.
[0119] The system selects the final scheduling scheme from the Pareto optimal solution set based on the preferences of the current production scenario. The preference level is dynamically adjusted according to the urgency of the production task, inventory balance, and tool life status. The quantitative adjustment rules are as follows:
[0120] 1. Urgency Level Classification: Production tasks are classified into 5 levels based on the time remaining until order delivery, with higher levels indicating greater delivery pressure.
[0121] Level 1: Delivery cycle ≥ 7 days, low delivery pressure;
[0122] Level 2: Delivery cycle of 3-7 days, with less delivery pressure;
[0123] Level 3: Delivery cycle 1-3 days, moderate delivery pressure;
[0124] Level 4: Delivery cycle of 12-24 hours, with significant delivery pressure;
[0125] Level 5: Delivery cycle < 12 hours, extremely high delivery pressure.
[0126] 2. Preference Weight Adjustment: Base preference weights are set for the four objectives. Initially, the base weights for inventory cost, equipment downtime cost, tool replacement cost, and cutting edge trimming cost are all 0.25. These weights are dynamically adjusted based on scenario factors.
[0127] For each level of urgency that increases, the preference weight for equipment downtime costs is increased by 20%, while the weights of other targets are proportionally reduced to ensure that the total weight is 1.
[0128] When the tool inventory is below 120% of the safety stock, the preference weight for inventory cost is increased by 15%, while the weights of other targets are decreased simultaneously.
[0129] When the average remaining life of in-service tools is less than 30% of the total life, the combined preference weight for tool replacement and overhaul costs is increased by 15%, prioritizing the availability of tools.
[0130] By dynamically adjusting preference weights, the system can select the optimal scheduling scheme that best suits the current scenario from the Pareto solution set, reasonably allocate the usage scenarios and usage time of new tools and refurbished tools, standardize the timing of tool refurbishment and scrapping standards, avoid premature scrapping of tools in good condition to avoid material waste, and at the same time avoid production losses caused by the continuous processing of worn tools with defects, which may lead to workpiece scrapping and equipment malfunctions.
[0131] 3. Categorized, zoned, and tiered inventory management strategy
[0132] The system adopts a classified, partitioned, and hierarchical inventory storage and retrieval strategy. Based on the tool usage status, the warehouse is divided into three independent physical storage areas, which store brand new unused tools, repaired and usable tools, and tools that are close to being scrapped. This achieves partitioned and isolated storage of tools in three different states, eliminating problems such as scheduling chaos, status statistics errors, and retrieval errors caused by mixed storage of tools. Each storage area is set with a separate status identifier and inventory ledger.
[0133] Each zone implements refined hierarchical management rules, dividing storage into different levels based on the remaining lifespan of individual tools. Tools within the same storage area are arranged in descending order of remaining lifespan, with tools having longer remaining lifespans placed in the priority access level and tools with shorter remaining lifespans placed in the spare level. The system simultaneously completes hierarchical registration in the electronic inventory ledger, clearly recording the model, specifications, inventory quantity, remaining lifespan, and availability status of tools at each level, achieving refined and visual management of inventory tools.
[0134] The system executes differentiated intelligent scheduling rules to accurately match tool status with machining task type: For routine machining tasks with conventional machining accuracy requirements, stable cutting load, and standardized batch production, the system prioritizes the use of repaired tools that have been dressed, passed accuracy testing, and have sufficient remaining life, making full use of repairable tool resources, reducing the consumption of brand-new tools, and controlling production material costs; For high-end machining tasks with stringent machining accuracy requirements, high cutting load intensity, and high workpiece quality standards, the system prioritizes the use of brand-new tools that have never been used, have intact cutting edges, and stable machining performance, avoiding the impact of minor cutting edge deviations of dressed tools on the machining quality of high-end workpieces, and ensuring the workpiece pass rate in high-precision machining scenarios.
[0135] 4. Production task linkage and early warning procurement
[0136] The system is linked to the production task scheduling, and synchronizes information on all processing procedures, production batches, processing progress and delivery nodes in real time. It establishes a dynamic matching mechanism between tool resources and processing procedures. Based on the processing accuracy, load intensity and production time of each procedure, it automatically adapts the tool resources in the corresponding state, so that tools with different performance and different remaining life are matched with suitable processing procedures.
[0137] The system continuously scans the remaining number of dressing cycles and remaining lifespan of inventory and in-use tools, and performs risk prediction calculations in conjunction with production scheduling data: the processing time required for a single batch of processing tasks is calculated by multiplying the processing time of a single piece in the corresponding process by the total number of workpieces planned to be processed in this batch, which serves as the basis for determining tool compatibility.
[0138] For tools that have exhausted all their service life cycles and whose remaining usable lifespan is less than the total processing time required for a single batch of work, it is determined that they cannot fully support the production operation and pose a potential risk of failure during processing. The system generates multi-level scrap warnings for such tools in advance, differentiating warning levels based on the size of the lifespan deficit; the larger the deficit, the higher the warning priority. Simultaneously, a tool replacement plan adapted to the production schedule is developed, specifying the tool replacement time, replacement tool model, and replacement operation procedure, allowing sufficient time for replacement and debugging to avoid equipment downtime and workpiece scrap caused by sudden tool failure during production.
[0139] The system simultaneously combines current tool inventory levels, tool demand in subsequent production schedules, tool procurement cycles, and procurement costs to automatically generate a rational tool procurement plan: it calculates existing inventory gaps, combines estimated tool consumption for multiple future production batches, matches procurement cycle lengths to determine procurement timing, and determines procurement quantities and models based on procurement cost control requirements. This accurately replenishes inventory gaps, ensuring that tool inventory resources can continuously meet the needs of subsequent production tasks and maintain the continuity and stability of the production process.
[0140] In one embodiment of the present invention, the intelligent scheduling system for the entire life cycle inventory of cutting tools for cost optimization further includes: a closed-loop update module, which is used to collect real-time cutting force and working condition data, combine the cutting edge radius r, remaining life and scrap status data, update the wear judgment criteria and life assessment basis based on the deviation between the actual wear status and the prediction results, dynamically correct the judgment and prediction results, and realize adaptive update.
[0141] The closed-loop update module is set up with a periodic iterative update mechanism. The update node is determined according to the processing scenario: for regular continuous batch production scenarios, the node is the cumulative processing time of 8 hours, and for small part high-frequency batch processing scenarios, the node is the cumulative processing time of 500 workpieces. Real-time processing data is collected within each update node cycle, and the deviation value between the actual wear state and the predicted state is compared. Based on the deviation value, the wear judgment algorithm parameters and life prediction model coefficients are adaptively corrected using the gradient descent method to gradually improve the detection and prediction accuracy.
[0142] The closed-loop update module continuously collects dynamically updated triaxial cutting force data and operating parameter data during the machining process. Combined with the tool edge radius r, remaining lifespan information, and critical scrap condition data obtained from real-time system calculations, it iteratively optimizes the system's built-in judgment rules and evaluation standards. By comparing the deviation between actual field-measured data and system-predicted data, it dynamically corrects the tool condition judgment logic and lifespan prediction calculation results, continuously optimizing the system's detection accuracy and evaluation precision. This achieves adaptive iterative updates to the tool's full lifecycle condition evaluation system, ensuring the system's long-term stability and accuracy, as shown below:
[0143] 1. Periodic update trigger mechanism
[0144] The system establishes a two-dimensional periodic iterative update mechanism, setting two types of freely switchable update trigger nodes: cumulative processing time node and cumulative number of processed workpieces node, adapting to production scenarios with different production rhythms and processing types:
[0145] In typical continuous batch production scenarios, the system uses cumulative processing time as the update node. An automatic system iteration update is triggered every 8 hours of continuous processing. Based on the typical wear pattern of carbide cutting tools, the increase in the corner radius after 8 hours of continuous processing is approximately 5-10 μm, exceeding the measurement error range. This provides statistically significant deviation data, allowing users to adjust the time interval according to the actual scenario.
[0146] In scenarios involving high-frequency batch processing of small parts, the update node is the number of workpieces processed. An iterative update process is triggered every 500 standardized workpieces processed. For this scenario, the tool wear corresponding to 500 workpieces is sufficient to support model parameter correction, ensuring the effectiveness of the iteration. In high-precision machining scenarios, the update cycle can be appropriately shortened according to production needs, increasing the frequency of system iteration and optimization.
[0147] 2. Deviation Calculation and Adaptive Parameter Correction
[0148] Once the system reaches the set update node, it automatically collects complete real-time machining data for the current cycle, including all-time three-axis cutting force data, dynamic working condition parameter data, actual tool cutting edge radius data, and task machining time data, and summarizes them into a complete cycle machining dataset.
[0149] The system retrieves the wear state prediction results and remaining life estimation data of each tool at the initial stage of this cycle. It then compares the theoretical data predicted by the system with the actual tool state data measured on site item by item, and calculates two types of deviation data: tool wear deviation value and remaining life estimation deviation value. The deviation value is the algebraic difference between the theoretical value and the measured value.
[0150] Based on the calculated deviation values, the system employs an adaptive parameter correction method based on gradient descent to iteratively update parameters such as the threshold and weight coefficients of the wear determination algorithm, as well as the wear attenuation coefficient and duration conversion coefficient of the life prediction model. The parameter correction formula is as follows: ;
[0151] In the formula: These are the algorithm parameters before correction. These are the corrected parameters; The learning rate ranges from 0.01 to 0.1. The larger the deviation, the higher the learning rate should be, and the smaller the deviation, the lower the learning rate should be, to ensure smooth convergence of the correction process. This is the system's predicted value. These are the actual measured values.
[0152] For the wear and tear coefficient in the life prediction model, an exponentially weighted moving average method is used for updating, taking into account both the stability of historical data and the timeliness of new data: ;
[0153] In the formula: The wear attenuation coefficient before correction. These are the corrected coefficients; This is the wear attenuation coefficient obtained from the actual measurement in this cycle, which is the wear increment of the cutting edge fillet radius corresponding to a unit processing time; The smoothing coefficient is set to 0.2 to ensure that new data effectively updates the model while avoiding drastic parameter changes caused by single data fluctuations.
[0154] The criteria for determining tool wear status are iteratively optimized in sync with the updated wear determination algorithm parameters, and the reference criteria for tool life assessment are upgraded in sync with the updated life prediction model coefficients. The system automatically retains records of every parameter correction and iterative operation log, recording the parameter values, deviation data, and update time for each correction. Through continuous periodic iterative correction, the error between the system's predicted data and actual machining data is constantly reduced, steadily improving the accuracy of tool condition detection and remaining life prediction, thereby achieving continuous optimization of system performance.
[0155] Example 1:
[0156] Taking a routine production task of machining 1000 pieces of 45# steel with a 10mm diameter carbide end mill as an example, the complete process of the system from data acquisition to scheduling plan output is shown below:
[0157] 1. Data Acquisition and Preprocessing: A standard batch machining scenario was used, with a sampling frequency of 200Hz, to simultaneously acquire three-axis cutting forces and operating parameters. The baseline operating condition was set as follows: m / min, mm / r, mm, corresponding reference cutting force N. Data collected at a specific processing moment: N, N, N, operating parameters m / min, mm / r, mm. The data was verified to be normal by the 3σ criterion, and standardized data was output after time alignment.
[0158] 2. Decoupling of operating conditions and calculation of cutting edge fillet:
[0159] First, calculate the resultant force of the measured cutting forces: N;
[0160] Substitute the theoretical cutting force into the datum mapping model: N (the coefficient here is an example value, and the actual value is obtained by experimental fitting);
[0161] Calculate the wear force difference: N;
[0162] Dimensionless transformation yields the relative wear characteristic quantity: ;
[0163] Substitute the fitting function to calculate the corner radius of the cutting edge: For carbide cutting tools... , , , μm, calculated μm, combined with initial μm indicates that the tool is currently in a state of slight wear.
[0164] 3. Lifecycle and Dressing Determination: Cumulative Use Time of the Tool If the current r-value is below the dressing threshold of 30μm, dressing will not be triggered. If the r-value wears down to 35μm during subsequent machining, falling within the dressing threshold range, and the tool is brand new and un-dressed, then the first dressing determination will be triggered. After dressing, the measured r-value is 18μm, the accuracy verification meets the standard, the baseline r-value is updated to 18μm, the dressing log is recorded, and the status changes to "dressing available".
[0165] 4. Life Prediction and Condition Classification: Substituting the modified Taylor life equation, carbide cutting tools... min, machining 45# steel After the first repair ,index , , Since the current operating conditions are consistent with the baseline operating conditions, the ratio of dimensionless terms is 1 for all of them, and the total life is calculated accordingly. min, a total of 120 minutes have been used, therefore the remaining lifespan min. The current r value corresponds to a slight wear state, and there is still sufficient margin before reaching the ultimate wear threshold of 65μm.
[0166] 5. Inventory Scheduling Plan Generation: The current task is routine batch processing with an urgency level of 2. The warehouse currently has 8 brand-new cutting tools, 12 tools that can be repaired, and 3 tools nearing the end of their lifespan, with a safety stock of 5 tools. The system obtains the Pareto solution set using NSGA-II and, based on current scenario preferences, prioritizes scheduling 10 repaired tools with sufficient remaining lifespan for the routine batch processing task. The brand-new tools are reserved for subsequent high-precision processes. Simultaneously, 2 tools nearing the end of their lifespan are scheduled for use in low-load processes until they are scrapped, generating an early scrap warning. After calculating the remaining inventory, a procurement plan for 2 new tools is generated to ensure the safety stock level. This plan maximizes the reuse of repaired tools, reduces the cost of purchasing new tools, and ensures production continuity.
[0167] 6. Closed-loop update: After 8 hours of cumulative processing, the update node is triggered. The measured wear data within the cycle is collected and compared with the prediction results. The deviation is 2.3μm. The gradient descent method is used to correct the fitting function coefficient and the life model decay coefficient. The learning rate is 0.03, and one iteration update is completed, gradually improving the accuracy of subsequent detection and prediction.
[0168] Taking the machining of aluminum alloy with an 8mm diameter high-speed steel twist drill as an example, with a batch size of 500 pieces, the system adaptation parameters are as follows: initial cutting edge radius. μm, dressing threshold range 45μm~65μm, ultimate wear threshold 90μm, maximum dressing times 3; high-speed steel cutting tool min, processing aluminum alloys After the first repair The system can complete the entire process control according to the same logic, verifying the versatility of the solution for different types of cutting tools.
[0169] The embodiments of this example have been described above. However, this example is not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms based on the guidance of this example, and all of them are within the protection scope of this example.
Claims
1. A tooling full lifecycle inventory intelligent scheduling system for cost optimization, characterized in that, include: The data acquisition module is used to acquire cutting force signals Fx, Fy, and Fz, as well as three types of machining parameters: feed rate f, depth of cut ap, and cutting speed v, during the tool machining process. The working condition decoupling and tool state extraction module is used to eliminate the interference of machining conditions on cutting forces, extract tool wear characteristics, and calculate and output the tool cutting edge fillet radius r. The tool lifecycle modeling module is used to statistically analyze the tool's usage time and state change patterns based on the cutting edge fillet radius r, generate wear state data and cutting edge dressing records, and realize traceable management of tool life. At the same time, it determines whether the tool meets the cutting edge dressing conditions based on the initial cutting edge fillet radius r, plans the number of cutting edge dressing times, tracks the state wear after dressing, and performs accuracy verification on the dressed cutting edge fillet radius. The tool condition prediction and life assessment module is used to combine tool usage time, wear data and dressing records to predict the life of tools in stock and in use, output remaining life information, classify health status levels, and define the critical state of tool scrapping. The inventory optimization and scheduling decision module is used to manage inventory access operations based on remaining life, critical scrap status, combined with the number of times the cutting edge can be repaired, production task requirements, and inventory constraints including the maximum inventory capacity of cutting tools and the minimum safety stock, and to generate scheduling, repair and scrapping schemes through comprehensive cost optimization. The closed-loop update module is used to collect real-time cutting force and working condition data, and combine the cutting edge radius r, remaining life and scrap status data. Based on the deviation between the actual wear status and the prediction results, it updates the wear judgment criteria and life assessment basis, dynamically corrects the judgment and prediction results, and realizes adaptive updates.
2. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The data acquisition module adopts a high-frequency synchronous acquisition mechanism, and the sampling frequency is set according to the processing scenario: 1000Hz for fine processing scenario, 500Hz for rough processing scenario, and 200Hz for regular batch processing scenario; the module marks the corresponding timestamps of the two types of data to complete the time sequence alignment, uses the 3σ criterion to remove abnormal data, uses linear interpolation to complete missing data, and outputs standardized multi-source state data.
3. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The working condition decoupling and tool state extraction module calculates the resultant force amplitude by vector synthesis of the synchronously acquired three-axis cutting force components Fx, Fy, and Fz, obtaining the measured scalar value of the cutting force; it establishes a benchmark mapping model between the cutting force and working condition parameters, calculates the theoretical cutting force without wear interference under the current working condition, and obtains the pure tool wear characteristic quantity by subtracting the measured cutting force from the theoretical cutting force, thus filtering out the mechanical influence caused by working condition changes; it has a built-in fitting function for tool wear characteristics and cutting edge fillet radius, and calculates the tool cutting edge fillet radius r by substituting the wear characteristic quantity into the fitting function.
4. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The tool lifecycle modeling module sets a tool dressing threshold range and compares the real-time cutting edge radius r with the initial standard radius. When the wear of r exceeds the standard allowable range but does not reach the scrap threshold, the cutting edge dressing condition is determined to be met. The cutting edge dressing condition includes the first dressing condition and the second dressing condition. The first dressing condition is when a brand-new tool wears to the dressing threshold range, and the second dressing condition is when the tool wears to the dressing threshold range again after the first dressing. The maximum number of dressing times is set in combination with the tool material. For carbide tools, it is 1-2 times, and for high-speed steel tools, it is 2-3 times. After each dressing, the tool reference radius and wear traceability data are updated synchronously.
5. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The tool condition prediction and life assessment module sets four levels of tool health thresholds, corresponding to four health status levels: brand new and intact, slightly worn, requiring repair, and nearing scrap. Based on the cumulative tool usage time, wear from previous repairs, and the real-time attenuation rate of the cutting edge radius r, the remaining tool life is calculated through a multi-parameter fusion life calculation model. The state where the cutting edge radius r reaches the limit wear threshold is defined as the critical scrap state, and the limit wear threshold is set according to the tool material and machining accuracy requirements.
6. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The inventory optimization and scheduling decision module constructs a multi-objective optimization model with four independent optimization objectives: inventory cost, equipment downtime cost, tool replacement cost, and cutting edge trimming cost. The inventory constraints also include procurement cycle and procurement cost. Based on the urgency of production tasks, tool inventory, remaining lifespan, and number of trimming cycles, the module determines the scheme selection preference. A non-dominated sorting genetic algorithm is used to obtain the Pareto optimal solution set. From the optimal solution set, the optimal scheduling combination that fits the current production scenario is selected to avoid production losses caused by premature tool scrapping or processing with defects.
7. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 6, characterized in that, The inventory optimization and scheduling decision module adopts a classification, partitioning, and hierarchical inventory storage and retrieval strategy; it classifies and stores brand-new unused tools, repaired and usable tools, and tools nearing the end of their service life into different partitions, and manages them hierarchically according to their remaining lifespan; it prioritizes scheduling repaired tools with sufficient remaining lifespan for regular machining tasks, and prioritizes scheduling brand-new tools for high-precision, high-load machining tasks.
8. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 1, characterized in that, The closed-loop update module is equipped with a periodic iterative update mechanism. The update node is determined according to the processing scenario: for regular continuous batch production scenarios, the node is the cumulative processing time of 8 hours, and for small-part high-frequency batch processing scenarios, the node is the cumulative processing time of 500 workpieces. Real-time processing data is collected within each update node cycle, and the deviation between the actual wear state and the predicted state is compared. Based on the deviation, the wear judgment algorithm parameters and life prediction model coefficients are adaptively corrected using the gradient descent method to gradually improve the detection and prediction accuracy.
9. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 4, characterized in that, After the tool edge trimming is completed, the tool lifecycle modeling module performs accuracy verification on the trimmed edge fillet radius r. If the verification meets the standard, the tool status is updated to trimmed and usable. If the verification fails, the tool is directly determined to be scrapped, and the tool's subsequent use cycle is terminated.
10. The intelligent scheduling system for the entire lifecycle inventory of cutting tools oriented towards cost optimization according to claim 7, characterized in that, The inventory optimization and scheduling decision module is linked to the production task scheduling to dynamically match tool resources with processing procedures. For tools with no remaining repair cycles and a remaining lifespan less than the time required for a single batch of tasks, a scrap warning and replacement plan are generated in advance. The time required for a single batch of tasks is calculated and determined based on the processing time of a single piece in the corresponding process and the number of workpieces processed. At the same time, a tool procurement plan is generated to ensure that the inventory meets production needs.
