A tool combination optimization method, device and medium

Abstract

The application relates to the field of part machining, in particular to a tool combination optimization method, device and medium, the method comprising: in response to monitoring a new machining task, acquiring machining feature parameters; based on the machining feature parameters and a pre-trained cutting tool selection model, determining candidate cutting tools; based on a reinforcement learning model, matching tool accessory components for the candidate cutting tools from a tool accessory library to generate a target combination scheme; and based on the target combination scheme, generating a standardized bill of materials. Through the pre-trained cutting tool selection model, the machining feature parameters are combined with historical data to dynamically recommend optimal cutting tools, reduce manual screening time, and improve cutting tool selection accuracy and efficiency. Based on interface matching rules and total length constraints, various tool accessory combinations are automatically constructed, and rigidity verification is performed to ensure the process feasibility of the combinations, thereby reducing manual trial and error and adjustment time.

Description

Technical Field

[0001] This application relates to the field of parts processing, specifically to a tool combination optimization method, equipment, and medium. Background Technology

[0002] In the field of machining, the configuration of tool sets (cutting tools + tool accessories) directly affects machining efficiency, machining quality, and production costs.

[0003] Currently, tool combination selection mainly relies on the manual experience of process engineers or simple matching based on static rule bases. Traditional tool combination optimization methods typically employ single-objective optimization (such as minimum cost or shortest machining time) or selection logic based on fixed rules, resulting in technical problems such as low efficiency, strong reliance on experience, lack of dynamic optimization capabilities driven by historical data, and reliance on manual intervention for interface matching and overall length control. Summary of the Invention

[0004] To address the aforementioned issues, this application proposes a tool combination optimization method, device, and medium, comprising: in response to detecting a new machining task, acquiring machining characteristic parameters; determining candidate tools based on the machining characteristic parameters and a pre-trained tool selection model; matching tool accessory components from a tool accessory library to the candidate tools based on a reinforcement learning model to generate a target combination scheme; and generating a standardized bill of materials based on the target combination scheme.

[0005] In one example, determining candidate cutting tools based on machining feature parameters and a pre-trained cutting tool selection model specifically includes: acquiring machining feature parameters, which include geometric requirement parameters and material parameters; determining candidate cutting tools based on the geometric requirement parameters and the material parameters; determining the selection probability corresponding to each candidate cutting tool based on the cutting tool selection model; and sorting the candidate cutting tools based on the selection probabilities.

[0006] In one example, before determining the selection probability of each candidate cutting tool based on the cutting tool selection model, the method further includes: cleaning historical processing records and marking historical data of products on the same platform; determining the similarity between the current processing task and historical processing tasks, and weighting the historical data according to whether different weight coefficients are assigned to products on the same platform; training an initial cutting tool selection model using the weighted historical data, wherein the input features of the initial cutting tool selection model include at least processing parameters and cutting tool performance parameters, and the initial cutting tool selection model outputs the selection probability of each candidate cutting tool; and sorting the candidate cutting tools based on the selection probabilities specifically includes: when multiple candidate cutting tools have the same selection probability, performing a secondary sorting according to a preset priority, wherein the priority includes at least one of processing efficiency, overall cost, or quantity in the inventory.

[0007] In one example, determining the similarity between the current processing task and historical processing tasks specifically includes: determining at least one dimension of similarity between the current processing task and historical processing tasks by comparing the historical processing feature parameters of historical processing tasks with the current processing feature parameters of the current processing task; the dimension similarity includes tolerance similarity, depth similarity, and material similarity; and determining the similarity between the current processing task and historical processing tasks based on at least one dimension of similarity.

[0008] In one example, the step of assigning different weight coefficients to historical data based on whether they are products on the same platform specifically includes: in response to the current processing task and the historical processing task being products on the same platform, determining the processing effect evaluation value of the historical processing task and the time difference between the historical processing task and the current time; determining a time similarity coefficient based on the time difference; determining the weight coefficient corresponding to the historical data based on the time similarity coefficient and the processing effect evaluation value; and in response to the current processing task and the historical processing task being products on different platforms, determining the weight coefficient corresponding to the historical data based on the similarity.

[0009] In one example, the process of matching tool accessory components from a tool accessory library to a candidate cutting tool based on a reinforcement learning model to generate a target combination scheme specifically includes: constructing a state space based on the number, length, and diameter of the connecting rods corresponding to the tool accessory combination scheme, the interface type of the current combination end, the remaining total length error, and the tool accessory inventory status; constructing an action space based on actions such as adding connecting rods, removing connecting rods, replacing connecting rods, adding tool holders, replacing tool holders, adding adapters, termination actions, and backoff actions performed when the tool accessory combination scheme is not feasible; and constructing an action space based on the total number of tool accessories, the remaining total length error, the tool length-to-diameter ratio, and each tool accessory group. The historical usage probability of the combined tool and tool combination scheme on the same platform variant product is used to construct a value function; based on the state space, the action space, and the value function, the reinforcement learning model is constructed; case data of historical tool and tool combination schemes where the total length error is less than the tolerance and the rigidity is compliant are used as training data to train the reinforcement learning model; the tool length, tool output interface type, user-defined recommended tool length, tolerance, and machine tool spindle interface type are input into the reinforcement learning model, and the Pareto optimal solution set is output, sorted by the number of tool and tool and the total length error; based on the Pareto optimal solution set, the target combination scheme is determined.

[0010] In one example, after constructing the action space based on adding connecting rods, removing connecting rods, replacing connecting rods, adding tool holders, replacing tool holders, adding adapters, terminating actions, and reversing actions when the tool accessory combination scheme is not feasible, the method further includes: adding at least one of the following action constraints to the action space: shielding the tool holder addition action in response to an existing tool holder assembly; allowing only termination or reversal actions in response to the total number of tool accessories exceeding a preset threshold; embedding interface compatibility checks when performing the addition or replacement of tool accessories; allowing only the selection of components matching the current combination end interface type according to a predefined compatibility rule base of interface type and diameter matching relationship when adding tool accessories; requiring bidirectional compatibility of front-end interface matching with the previous component and rear-end interface matching with the next component when the interface replacement action is performed; performing the add adapter action when the tool holder interface and machine tool interface do not match after the tool holder addition action is performed; aiming to satisfy the total length constraint of the cutting tool and the tool accessory; determining the length-to-diameter ratio of the combined tool, and adjusting the connecting rod diameter if it exceeds a preset threshold.

[0011] In one example, the standardized bill of materials is a structured table containing tool types, tool accessory sequences, interface parameters, and total length.

[0012] This application also provides a tool combination optimization device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the method described in any of the above examples.

[0013] This application also provides a non-volatile computer storage medium storing computer-executable instructions configured to perform the steps of the method described in any of the above examples.

[0014] The method proposed in this application can bring the following beneficial effects:

[0015] 1. By using a pre-trained tool selection model, combined with machining feature parameters and historical data, the optimal tool is dynamically recommended, reducing manual selection time and improving the accuracy and efficiency of tool selection.

[0016] 2. By learning from historical machining records, especially successful cases of products on the same platform, we assign higher recommendation weights to similar tasks, ensuring that the selected tool combinations have a basis for actual machining verification, thereby improving machining stability and success rate.

[0017] 3. Based on interface matching rules and total length constraints, various tool and tool combinations are automatically constructed, and rigid verification is used to ensure the process feasibility of the combination, reducing manual trial and error and adjustment time.

[0018] 4. Finally, a structured bill of materials is generated, which includes information such as cutting tool model, tool and accessory sequence, interface parameters and total length. It can be directly imported into the process management system to support production planning, tool procurement and inventory management. Attached Figure Description

[0019] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0020] Figure 1 This is a flowchart illustrating a tool combination optimization method in an embodiment of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0022] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0023] Figure 1 This is a flowchart illustrating a tool combination optimization method provided in one or more embodiments of this specification. This method can be applied to automatically select cutting tools and tool accessories during part machining. The tool combination includes both cutting tools and tool accessories; the cutting tool refers to the cutting edge portion at the front of the cutting tool.

[0024] This process can be executed by computing devices in the relevant field (such as control terminals connected to machining centers), and some input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0025] The analysis method involved in the embodiments of this application can be implemented by a terminal device or a server, and this application does not impose any special limitations on it. For ease of understanding and description, the following embodiments are all described in detail using a server as an example. It should be noted that the server can be a single device or a system composed of multiple devices, that is, a distributed server, and this application does not impose any specific limitations on it.

[0026] like Figure 1 As shown in the figure, this application provides a tool combination optimization method, including:

[0027] S101: In response to the detection of a new processing task, acquire processing characteristic parameters.

[0028] When the server detects a new processing task, it will obtain the processing feature parameters of the workpiece to be processed. These processing feature parameters include parameters such as hole diameter (D), depth (H), tolerance, and material hardness (HRC) required by the processing task.

[0029] S102: Based on machining feature parameters and a pre-trained tool selection model, determine candidate tools.

[0030] After obtaining the machining feature parameters, the server will select cutting tools from the cutting tool library that meet the basic geometric requirements based on the machining feature parameters and a pre-trained cutting tool selection model. Cutting tools are the core components that directly participate in material removal in machining, completing the machining task through the relative movement between the cutting edge and the workpiece.

[0031] S103: Based on a reinforcement learning model, match knife accessory components from the knife accessory library to generate a target combination scheme.

[0032] After identifying candidate cutting tools, a reinforcement learning model can be used to select suitable tool accessories from a tool accessory library to determine the target combination scheme. In the field of machining, tool accessories typically refer to tools or devices used to assist cutting tools, including but not limited to tool holders, extension rods, and adapters. These tools are mainly used to connect the cutting tool to the machine tool spindle, ensuring that the tool can perform machining operations in the correct direction and position. After obtaining candidate cutting tools and tool accessories, the candidate cutting tools and their corresponding tool accessories are combined to form a tool accessory combination scheme. It can be understood that a candidate cutting tool can form multiple tool accessory combination schemes with multiple tool accessories.

[0033] S104: Generate a standardized bill of materials based on the target combination scheme.

[0034] After obtaining the target combination scheme, a structured table can be output based on the target combination scheme list (including component model, interface parameters, total length, rigidity verification results, and confidence level). The output structured table includes cutting tool model, tool and accessory sequence, interface parameters, and total length, and is then imported into the process management system.

[0035] In one embodiment, when determining candidate cutting tools, the server first obtains machining characteristic parameters, including geometric requirement parameters and material parameters; based on the geometric requirement parameters and the material parameters, candidate cutting tools are determined; based on the cutting tool selection model, the selection probability corresponding to each candidate cutting tool is determined; and the candidate cutting tools are sorted based on the selection probability.

[0036] Specifically, after obtaining the processing feature parameters, the server first classifies and extracts these parameters. The processing feature parameters mainly include two categories: geometric requirement parameters and material parameters. Geometric requirement parameters usually cover the dimensional information of the feature to be processed, such as hole diameter (D), depth (H), tolerance, etc.; material parameters describe the physical properties of the workpiece material, such as material hardness (HRC), material type, etc.

[0037] The server first performs an initial screening of the tool library based on geometric requirements, selecting tools that meet the basic geometric conditions. For example, for drilling tasks, the system will select drill bits whose diameter matches the target hole diameter (e.g., with an allowable tolerance of ±0.1mm) and whose cutting length is not less than the machining depth (H). Simultaneously, a secondary filtering is performed based on material parameters. For instance, if the workpiece is made of a high-hardness material, tools unsuitable for such materials (e.g., high-speed steel) will be automatically excluded.

[0038] After completing the initial screening, the server will call the pre-trained knife selection model to evaluate the probability of candidate knives that meet the conditions and calculate a selection probability for each knife. This model is built by analyzing historical machining data and tool performance data, and its output reflects the probability of each tool being successfully used under the current machining task conditions.

[0039] The server then sorts the candidates by their selection probability in descending order, creating a recommended list of cutting tools. In addition to the tool model and probability value, this list can display auxiliary decision-making information based on configuration, such as the predicted cutting speed of the recommended tool, estimated single-piece machining cost, and current inventory levels, providing comprehensive data support for process engineers' final selection.

[0040] In one embodiment, when training the tool selection model, it is necessary to clean historical processing records and label historical data of products on the same platform; determine the similarity between the current processing task and historical processing tasks, and weight the historical data according to whether different weight coefficients are assigned to products on the same platform; train the initial tool selection model using the weighted historical data, wherein the input features of the model include at least processing parameters and tool performance parameters, and the model outputs the selection probability of each candidate tool; the ranking of the candidate tools based on the selection probability specifically includes: when multiple candidate tools have the same selection probability, a secondary ranking is performed according to a preset priority, wherein the priority includes at least one of processing efficiency, overall cost, or quantity in the inventory.

[0041] Specifically, after initially screening out candidate cutting tools that meet basic geometric requirements based on processing feature parameters, in order to further improve the accuracy and practicality of cutting tool selection, the server will call a pre-trained cutting tool selection model to intelligently score and rank the candidate cutting tools.

[0042] First, the server inputs the machining feature parameters of the current machining task (such as hole diameter D, depth H, tolerance, material hardness HRC, feed rate F, etc.) and the performance parameters of each candidate cutting tool (such as cutting speed Vc, coating type, life data, etc.) into the pre-trained cutting tool selection model.

[0043] The tool selection model calculates and outputs the selection probability P for each candidate tool based on the multivariate features of the input. i This probability reflects the likelihood that the tool will be recommended for use in the current machining task, and its value is typically between 0 and 1. After determining the selection probability, the server will sort the candidate tools in descending order based on their selection probabilities to generate a preliminary recommendation order.

[0044] Finally, after initially sorting all candidate cutting tools in descending order based on their selection probabilities, if multiple candidate cutting tools have the same or very close selection probabilities (e.g., overlapping within the confidence interval), a secondary sorting logic is initiated. The secondary sorting follows a preset priority rule, which includes at least one or more of the following: processing efficiency, overall cost, and inventory stability. Processing efficiency can be determined based on the cutting speed of the cutting tool (…). The recommended relative linear velocities between the cutting edge and the workpiece (in m / min) are ranked, with higher velocities ranking higher. The overall cost can be calculated based on the cost per piece. Sort, where For the cost of purchasing a single cutting tool, To estimate tool lifespan, lower costs are ranked higher; for inventory stability, tools with a current inventory of 3 or more pieces can be prioritized to reduce the risk of production interruptions due to stockouts.

[0045] In one embodiment, when determining the similarity between the current processing task and historical processing tasks, the system calculates similarity indices across multiple dimensions by comparing the historical processing feature parameters of historical processing tasks with the current processing feature parameters of the current processing task. Specifically, the system primarily performs quantitative evaluation from three dimensions: tolerance similarity, depth similarity, and material similarity.

[0046] The following formula can be used to calculate tolerance similarity:

[0047]

[0048] in, This represents the width of the overlapping area between the tolerance zones of the current machining task and the tolerance zones of historical tasks. This is the total width after combining the tolerance zones of the current machining task and the tolerance zones of the historical tasks.

[0049] Depth similarity can be calculated using the following formula:

[0050]

[0051] in, The depth corresponding to the current processing task. The depth corresponding to the historical processing tasks.

[0052] Material similarity can be determined by whether the materials' hardness (e.g., HRC value) falls within the same hardness range. A match earns 1 point; otherwise, a score of 0.5 or lower is assigned based on the hardness difference. After obtaining similarity scores in at least one of the above dimensions, the overall similarity can be calculated using the following weighted summation formula:

[0053]

[0054] in, , , Preset weights.

[0055] In one embodiment, when weighting historical data based on whether different weight coefficients are assigned to products on the same platform, in response to the current processing task and the historical processing task being products on the same platform, the processing effect evaluation value of the historical processing task and the time difference between the historical processing task and the current time are determined; based on the time difference, a time similarity coefficient is determined; based on the time similarity coefficient and the processing effect evaluation value, the weight coefficient corresponding to the historical data is determined; in response to the current processing task and the historical processing task being products on different platforms, the weight coefficient corresponding to the historical data is determined based on the similarity.

[0056] Specifically, the server assigns different weight coefficients to historical data based on whether the historical tasks belong to the same platform as the current task, and performs weighted processing accordingly. The weight of historical data from the same platform can be calculated using the following formula: Wherein, V is the evaluation value of the machining effect of the historical tool (which can be calculated based on indicators such as machining quality and tool life, and can be a positive or negative value). This is a time proximity coefficient; the more recent the data record time, the higher the time proximity coefficient. A larger value reflects the impact of timeliness. For historical data of unrelated products, the weighting is... ,in The preset low weight coefficient (e.g., 0.3) is used to reduce the reference weight of irrelevant data.

[0057] After obtaining the weights, the weighted historical data can be used to train an initial tool selection model. This model can be constructed using machine learning algorithms such as XGBoost and Random Forest. The model's input features include at least: machining parameters, tool performance parameters, and association identifiers. Machining parameters include bore diameter, depth, tolerance, material hardness, and feed rate; tool performance parameters include cutting speed, coating type, and tool life; and association identifiers refer to whether the tool belongs to the same platform as the product. During training, the influence of historical data from products on the same platform can be strengthened by using weighted samples. After training, the model can output the selection probability of each candidate tool in the tool library for the new input machining feature parameters. , and optionally calculate its confidence interval.

[0058] After the candidate cutting tools are determined, the server will perform the matching and combination process of the cutting tools and accessories. The specific steps are as follows:

[0059] In one embodiment, the server constructs a state space based on the number, length, and diameter of connecting rods corresponding to the tool accessory combination scheme, the interface type of the current combination end, the remaining total length error, and the tool accessory inventory status; constructs an action space based on actions such as adding connecting rods, removing connecting rods, replacing connecting rods, adding tool holders, replacing tool holders, adding adapters, termination actions, and backoff actions performed when the tool accessory combination is not feasible; constructs a value function based on the total number of tool accessories, the remaining total length error, the tool length-to-diameter ratio, and the historical usage probability of each tool accessory combination scheme on the same platform variant product; constructs the reinforcement learning model based on the state space, the action space, and the value function; trains the reinforcement learning model using case data of historical tool accessory combination schemes where the total length error is less than the tolerance and the rigidity is compliant; inputs the tool length, tool output interface type, user-defined recommended tool length, tolerance, and machine tool spindle interface type into the reinforcement learning model, and outputs a Pareto optimal solution set sorted by the number of tool accessories and the total length error; and determines the target combination scheme based on the Pareto optimal solution set.

[0060] In one embodiment, after constructing the action space, the method further includes: including at least one of the following action constraints within the action space: shielding the action of adding a tool holder in response to an existing tool holder assembly; allowing only termination or reversal actions in response to the total number of tool accessories exceeding a preset threshold; embedding an interface compatibility check when performing the action of adding or replacing a tool accessory; according to a predefined compatibility rule base of matching relationships between interface types and diameters, allowing only the selection of components that match the current combined end interface type when adding a tool accessory; requiring bidirectional compatibility between the front-end interface and the previous component, and the rear-end interface and the next component when replacing an interface; performing the action of adding a tool holder after the action of adding a tool holder, performing the action of adding an adapter when the tool holder interface and the machine tool interface do not match; aiming to satisfy the total length constraint between the cutting tool and the tool accessory; determining the length-to-diameter ratio of the combined tool, and adjusting the diameter of the connecting rod if it exceeds a preset threshold.

[0061] Specifically, after determining the cutting tools, this implementation uses a reinforcement learning model for intelligent evaluation and ranking to select the optimal target combination. First, the state space required for the reinforcement learning model is constructed. The state space describes the characteristic states of the current tool and accessory combination, including but not limited to: the number of components in the current combination, the length and diameter of each component, the interface type at the end of the current combination, the remaining error value between the current combination and the target total length, and the inventory status information of each tool and accessory component. Second, the action space is defined, corresponding to the set of operations that the system can execute during the optimization process, including: adding a connecting rod, removing a connecting rod, replacing a connecting rod, adding a tool holder, replacing a tool holder, adding an adapter, terminating the action, and performing a backtracking action when the combination is not feasible. The action space executes the following rules: if a tool holder assembly already exists, the "Add Tool Holder" action is disabled; when the tool holder interface and machine tool interface do not match, an adapter is forcibly inserted (this operation does not occupy the tool accessory quantity quota, but is counted in the tool holder length); when the number of tool accessories is greater than 3, only termination or reversal actions are allowed; when performing actions to add or replace a connecting rod / tool ​​holder, an interface compatibility check is embedded. Based on a predefined compatibility rule library of matching relationships between interface types and diameters, when adding a connecting rod / tool ​​holder, only components matching the end interface type of the current combination are allowed to be selected; the interface replacement action must meet the bidirectional compatibility principle of matching the front-end interface with the previous component and the back-end interface with the next component. These actions allow the system to flexibly adjust in the combination space, gradually approaching the optimal solution. Furthermore, a value function is constructed to evaluate the overall advantages and disadvantages of each combination scheme. The value function is based on the principle of multi-objective optimization, comprehensively considering the following factors: the number of tool accessories in the combination should be as small as possible, the total length error should be as small as possible, the rigidity index of the combination should be within a reasonable range, and the historical usage probability of the combination on variant products of the same platform. Specifically, the value function can be expressed as:

[0062]

[0063] in, These are the weighting coefficients for each item; The total number of knife accessories (including extension rods and knife handles); This represents the remaining total length error. The length-to-diameter ratio of the assembled tool. , This is the actual total length of the cutting tool. For the first The effective length of each tool assembly For the first The diameter of each tool assembly; This represents the historical usage probability of the combination on variant products on the same platform (the system automatically removes records of combinations that are too early or have poor performance; if there are no records, this value is 0). Next, the reinforcement learning model is trained. Training data comes from successful combination cases in historical machining records where the total length error does not exceed the tolerance range and is rigidly compliant. By repeatedly learning the state-action correspondence of these high-quality combinations, the model gradually establishes a mapping strategy from state to optimal action. After training, the tool length, tool output interface type (e.g., cylindrical-D8), user-defined recommended tool total length (L), tolerance (ΔL, default ±10mm), and machine tool spindle interface type (e.g., HSK-A100) are input into the reinforcement learning model. The model will output a Pareto optimal solution set sorted by the number of tool accessories and the total length error. Each solution in this set achieves a balance among multiple optimization objectives, and the user can select the final target combination scheme from them according to actual process requirements.

[0064] Through the above methods, the system can automatically evaluate and select tool combinations that are structurally reasonable, technologically feasible, and have a historical verification basis without relying on human experience, significantly improving the intelligence level and decision-making efficiency of combination optimization.

[0065] This application embodiment also provides a tool combination optimization device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: acquire machining feature parameters in response to detecting a new machining task; determine candidate cutting tools based on the machining feature parameters and a pre-trained tool selection model; match tool accessory components from a tool accessory library for the candidate cutting tools based on a reinforcement learning model to generate a target combination scheme; and generate a standardized bill of materials based on the target combination scheme.

[0066] This application embodiment also provides a non-volatile computer storage medium storing computer-executable instructions, the computer-executable instructions being configured to: in response to detecting a new machining task, acquire machining feature parameters; determine candidate cutting tools based on the machining feature parameters and a pre-trained cutting tool selection model; match cutting tool components from a tool library for the candidate cutting tools based on a reinforcement learning model to generate a target combination scheme; and generate a standardized bill of materials based on the target combination scheme.

[0067] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0068] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0069] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0071] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0073] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0074] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0075] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0076] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0077] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A tool combination optimization method, characterized in that, include: In response to the detection of a new processing task, processing characteristic parameters are acquired; Candidate cutting tools are determined based on machining feature parameters and a pre-trained cutting tool selection model. Obtain processing characteristic parameters, including geometric requirement parameters and material parameters; Based on the geometric requirements parameters and the material parameters, candidate cutting tools are determined; Based on a reinforcement learning model, tool accessory components are matched from a tool accessory library for the candidate cutting tools to generate a target combination scheme; specifically including: Based on the number, length, diameter, interface type at the end of the current combination, remaining total length error, and inventory status of the tool and accessory combination scheme, a state space is constructed. The action space is constructed based on adding connecting rods, removing connecting rods, replacing connecting rods, adding tool holders, replacing tool holders, adding adapters, terminating actions, and retraction actions performed when the tool and accessory combination scheme is not feasible; A value function is constructed based on the total number of tool accessories, the remaining total length error, the tool length-to-diameter ratio, and the historical usage probability of each tool accessory combination scheme on the same platform variant products. The reinforcement learning model is constructed based on the state space, the action space, and the value function. The reinforcement learning model is trained using case data from historical knife and tool combination schemes where the total length error is less than the tolerance and the rigidity complies with regulations. Input the tool length, tool output interface type, user-defined recommended total tool length, tolerance, and machine tool spindle interface type into the reinforcement learning model, and output the Pareto optimal solution set sorted by the number of tool accessories and the total length error; Based on the Pareto optimal solution set, determine the target combination scheme; Based on the target combination scheme, a standardized bill of materials is generated.

2. The method according to claim 1, characterized in that, The process of determining candidate cutting tools based on machining feature parameters and a pre-trained cutting tool selection model specifically includes: Based on the aforementioned blade selection model, the selection probability of each candidate blade is determined; The candidate blades are sorted based on the selected probability.

3. The method according to claim 2, characterized in that, Before determining the selection probability of each candidate blade based on the blade selection model, the method further includes: Clean historical processing records and mark historical data of products on the same platform; Determine the similarity between the current processing task and historical processing tasks, and weight the historical data according to whether different weight coefficients are assigned to products on the same platform; The initial tool selection model is trained using weighted historical data. The input features of the initial tool selection model include at least processing parameters and tool performance parameters. The initial tool selection model outputs the selection probability of each candidate tool. The process of sorting the candidate blades based on the selection probability specifically includes: When multiple candidate cutting tools have the same probability of being selected, they are sorted a second time according to a preset priority, which includes at least one of the following: processing efficiency, overall cost, or quantity in the inventory.

4. The method according to claim 3, characterized in that, Determining the similarity between the current processing task and historical processing tasks specifically includes: By comparing the historical processing feature parameters of historical processing tasks with the current processing feature parameters of the current processing task, the similarity between the current processing task and the historical processing tasks in at least one dimension is determined. The dimensional similarity includes tolerance similarity, depth similarity, and material similarity; The similarity between the current processing task and the historical processing tasks is determined based on similarity in at least one dimension.

5. The method according to claim 3, characterized in that, The weighting of historical data based on whether it is a product on the same platform with different weight coefficients specifically includes: In response to the fact that the current processing task and the historical processing task are products of the same platform, the processing effect evaluation value of the historical processing task and the time difference between the historical processing task and the current moment are determined. Based on the time difference, a time similarity coefficient is determined; Based on the time similarity coefficient and the processing effect evaluation value, the weight coefficient corresponding to the historical data is determined; In response to the fact that the current processing task and the historical processing task are products from different platforms, the weight coefficient corresponding to the historical data is determined based on the similarity.

6. The method according to claim 1, characterized in that, After constructing the action space based on adding, removing, replacing, adding, or replacing tool holders, adding adapters, terminating actions, and retraction actions performed when the tool accessory combination scheme is not feasible, the method further includes: adding at least one of the following action constraints to the action space: In response to an existing toolholder assembly, the addition of toolholder movement is disabled; If the total number of knife accessories exceeds a preset threshold, only termination or reversal actions are allowed. When performing actions to add or replace tool accessories, an embedded interface compatibility check is performed. Based on a compatibility rule base that predefines the matching relationship between interface type and diameter, the tool accessory is made to only allow the selection of components that match the current combined end interface type; The action of changing the interface must meet the bidirectional compatibility requirement of the front-end interface matching the previous component and the back-end interface matching the next component; After the tool holder action is executed, if the tool holder interface and the machine tool interface are not compatible, the adapter action is executed. The goal is to satisfy the total length constraint between the cutting tool and the tool accessory. Determine the length-to-diameter ratio of the combined cutter. If it exceeds the preset threshold, adjust the diameter of the connecting rod.

7. The method according to claim 1, characterized in that, The standardized bill of materials is a structured table that includes the cutting tool model, tool accessory sequence, interface parameters, and total length.

8. A tool combination optimization device, characterized in that, include: At least one processor; And, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the steps of the method as claimed in any one of claims 1-7.

9. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are configured to perform the steps of the method as claimed in any one of claims 1-7.

Images