Blade machining parameter determination method and device

Abstract

The invention provides a blade machining parameter determination method and device. According to the method provided by the invention, various data related to complex curved surface processing are integrated and preprocessed to obtain preprocessed data, a basic knowledge graph and a causal relationship graph are constructed on the basis of the preprocessed data, and the preprocessed data and the knowledge graph are utilized to train a large model to obtain a trained large model; performing feature extraction on each piece of data in the processing preparation stage and the actual processing data through the trained large model to obtain a corresponding feature vector, and constructing a feature vector library based on the attribute information and the feature vector of each piece of data; when a processing track of a to-be-processed blade is determined, extracting a target feature vector of the blade by utilizing the large model, searching a matched feature vector matched with the target feature vector from the feature vector library, analyzing attribute information of the target feature vector, extracting a corresponding target processing parameter from the target knowledge graph, and inputting the target processing parameter and the matched feature vector into the large model for optimization; and the optimized processing parameters are obtained.

Description

Technical Field

[0001] This application relates to the technical field of blade processing, and in particular, to a method and device for determining blade processing parameters. Background Art

[0002] In the industrial field, especially in aviation, energy, and mechanical engineering, blades have very wide applications. When processing blades, titanium alloys or superalloys are usually used to manufacture blades. Such materials are known for their excellent mechanical properties, but they also bring challenges in processing. For example, during the cutting process of blades, large cutting forces and rapid tool wear will occur, which are likely to cause machining deformation and vibration, thereby affecting the dimensional accuracy and surface quality of parts.

[0003] To solve this problem, in traditional blade processing methods, phase change materials such as low melting point alloys are used for filling and auxiliary support. In this way, although it helps to improve some processing problems, there are also problems such as decreased positioning accuracy, complex production preparation, difficult workpiece surface residue and cleaning, and the environmental conditions during pouring and melting are relatively harsh, resulting in low precision of the produced blades. Therefore, there is an urgent need for an optimization method for blade processing parameters to ensure that the produced blades meet the requirements of high precision and high quality. Summary of the Invention

[0004] In view of this, this application provides a method and device for determining blade processing parameters to accurately determine the processing parameters of blades and improve the processing precision and quality of blades.

[0005] Specifically, this application is implemented through the following technical solutions: The first aspect of this application provides a method for determining blade processing parameters, and the method for determining blade processing parameters includes: Obtain various types of data related to complex surface processing; wherein, the various types of data reflect the surface data, feature data, and machining data of the blade; Preprocess the various types of data to obtain preprocessed data, and construct a basic knowledge graph related to the blade processing field based on the preprocessed data; the basic knowledge graph takes blades, materials, tools, quality control, processing technology, and processing trajectories as nodes, and the inherent characteristics of blades, materials, tools, quality control, processing technology, and processing trajectories as node attributes; Construct a causal relationship graph related to the blade processing field based on the preprocessed data, and update the basic knowledge graph using the causal relationship graph to obtain a target knowledge graph; Train a large model using the preprocessed data and the target knowledge graph so that the large model learns the knowledge in the blade processing field to obtain a trained large model; For each piece of data in the data during the processing preparation stage and the actual processing data among the various types of data, use the trained large model to extract the features of this piece of data to obtain the feature vector of this piece of data, and construct a feature vector library based on the attribute information and feature vectors of each piece of data; When it is necessary to determine the processing trajectory of the blade to be processed, use the trained large model to extract the target feature vector of the blade to be processed, and search for the matching feature vector that matches the target feature vector from the feature vector library; Search for the target processing parameters corresponding to the attribute information from the target knowledge graph according to the attribute information of the matching feature vector, and input the matching feature vector and the target processing parameters into the trained large model, so that the trained large model optimizes the target processing parameters to obtain the optimized processing parameters.

[0006] The second aspect of the present application provides a device for determining blade processing parameters. The device includes an acquisition module, a processing module, a construction module, a training module, and an optimization module. Among them, The acquisition module is used to acquire various types of data related to complex surface processing; among them, the various types of data reflect the surface data, feature data, and machining data of the blade; The processing module is used to preprocess the various types of data to obtain the preprocessed data, and construct a basic knowledge graph related to the blade processing field according to the preprocessed data; the basic knowledge graph uses blades, materials, tools, quality control, processing technology, and processing trajectory as nodes, and the inherent characteristics of blades, materials, tools, quality control, processing technology, and processing trajectory as node attributes; The construction module is used to construct a causal relationship graph related to the blade processing field according to the preprocessed data, and update the basic knowledge graph with the causal relationship graph to obtain a target knowledge graph; The training module is used to train a large model using the preprocessed data and the target knowledge graph, so that the large model learns the knowledge in the blade processing field to obtain a trained large model; The construction module is used to, for each piece of data in the data during the processing preparation stage and the actual processing data among the various types of data, use the trained large model to extract the features of this piece of data to obtain the feature vector of this piece of data, and construct a feature vector library based on the attribute information and feature vectors of each piece of data; The acquisition module is used to, when it is necessary to determine the processing trajectory of the blade to be processed, use the trained large model to extract the target feature vector of the blade to be processed, and search for the matching feature vector that matches the target feature vector from the feature vector library; The optimization module is used to find target processing parameters corresponding to the attribute information from the target knowledge graph according to the attribute information of the matching feature vector, and input the matching feature vector and the target processing parameters into the trained large model, so that the trained large model optimizes the target processing parameters to obtain optimized processing parameters.

[0007] The method and device for determining blade processing parameters provided by this application integrate various types of data related to complex surface processing, preprocess various types of data to obtain preprocessed data, construct a basic knowledge graph and a causal relationship graph according to the preprocessed data, and then use the preprocessed data and the target knowledge graph to train a large model to obtain a trained large model. After that, the trained large model extracts features from each piece of data in the processing preparation stage and actual processing data to obtain corresponding feature vectors, and constructs a feature vector library based on the attribute information and feature vectors of each piece of data. Further, when it is necessary to determine the processing trajectory of the blade to be processed, the large model is used to extract the target feature vector of the blade, and the matching feature vector is found from the feature vector library. By analyzing the attribute information of the matching feature vector, the corresponding target processing parameters are extracted from the target knowledge graph, and the matching feature vector and the target processing parameters are input into the large model for further optimization to obtain optimized processing parameters. In this way, first, by constructing a knowledge graph for blade processing and training a large model in combination with the causal relationship, the large model can not only remember knowledge, but also reason about the relationship between processing parameters and processing effects, improving the interpretability of knowledge. Further, the trained large model is used to extract features from the processing preparation stage and actual processing data, and a feature vector library is constructed, which enables the model to perform in-depth analysis before and after processing. In addition, by matching the knowledge graph and the feature vector library, effective processing parameters are extracted from similar historical data and further optimized in the large model to ensure that the output parameters are more suitable for the current processing requirements, and the processing parameters can be accurately determined. Brief Description of the Drawings

[0008] Figure 1 It is a flowchart of the first embodiment of the method for determining blade processing parameters provided by this application; Figure 2 It is a flowchart of the second embodiment of the method for determining blade processing parameters provided by this application; Figure 3 It is a schematic diagram of the sub-ontology model of the blade shown in an exemplary embodiment of this application; Figure 4 It is a schematic diagram of the sub-ontology model of the material shown in an exemplary embodiment of this application; Figure 5 It is a schematic diagram of the sub-ontology model of the tool shown in an exemplary embodiment of this application; Figure 6 Schematic diagram of the sub-ontology model for quality control shown in an exemplary embodiment of the present application; Figure 7 Schematic diagram of the sub-ontology model for the machining process shown in an exemplary embodiment of the present application; Figure 8 Schematic diagram of the sub-ontology model for the machining trajectory shown in an exemplary embodiment of the present application; Figure 9 Schematic diagram of the structure of the first embodiment of the blade machining parameter determination device provided by the present application. Detailed implementation manners

[0009] Here, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the present application.

[0010] The terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms of "a", "the" and "said" used in the present application are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0011] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "when" or "while" or "in response to determining".

[0012] The following specific embodiments are given to introduce the technical solutions of the present application in detail.

[0013] Figure 1 Flowchart of the first embodiment of the blade machining parameter determination method provided by the present application. Please refer to Figure 1 , the blade machining parameter determination method provided in this embodiment includes: S101. Obtain various types of data related to complex surface machining; wherein, the various types of data reflect the surface data, feature data, and machining data of the blade.

[0014] Specifically, various types of data include blade design data, simulation data, data in the machining preparation stage, actual machining data, quality control data, and maintenance and improvement data.

[0015] In one embodiment, various types of data related to complex surface machining can be obtained according to the following steps: Step 1: Determine that the data collection types include structured text, semi-structured text, unstructured text, tables, documents, pictures, audio and video, and / or web pages; Step 2: Determine that the data sources include CAD models and design drawings in the product design stage, blade geometric parameters, material properties, CFD analysis results, FEA stress and strain predictions, thermodynamic performance evaluations in the simulation stage, tool selection parameters, machining machine specifications, NC programming information in the machining accuracy stage, real-time machining parameters, on-line measurement data, temperature monitoring data during the machining process, final product quality inspection reports, defect records, customer feedback in the quality control stage, fault history records, maintenance logs, and improvement measure documents in the maintenance and improvement stage; Step 3: Determine that the data collection methods include direct collection, system docking, manual import, and / or entry; Step 4: Collect various types of data indicated by the data collection types from the data sources according to the data collection methods.

[0016] In summary, through the above steps, the data collection types, data sources, and collection methods of various types of data related to complex surface machining can be clarified, ensuring that various types of data required in the complex surface machining process can be obtained from multiple links (such as design, simulation, machining, and quality control, etc.). In this way, it can optimize the process of machining the blade according to the obtained various types of data, and provide strong data support for selecting suitable machining parameters to obtain blades that meet the requirements.

[0017] It should be noted that various types of data related to complex surface machining include surface data, feature data, and machining data of the blade. The surface data, feature data, and machining data respectively represent technical information in different aspects during the machining process, and each type of data can provide support and decision-making basis at different stages and links of the machining.

[0018] Specifically, the surface data includes information related to the collective shape, surface features, and surface structure of the blade. Among them, the collective shape of the blade includes the blade root, blade tip, blade profile, film cooling holes, cooling channels, damping holes, damping platforms, tenon grooves, tenons, leading edges, trailing edges, blade basins, and blade backs, etc.; further, the surface features of the blade include the surface feature information of each part of the collective shape of the blade. For example, whether the blade tip is sharp or rounded, the length and width information of the blade tip; further, the surface structure of the blade includes the specific structure of each part of the blade. For example, the connection method of the blade root is by insert milling connection, key connection, or bolt connection, and the curvature radius of the blade back, etc.

[0019] Further, the characteristic data of the blade refers to the data describing the functional, structural, or operational characteristics of the blade. Among them, the characteristic data of the blade includes the material properties and thermodynamic properties of the blade, etc. For example, the material for making the blade can be steel, cast iron, or non-ferrous metal. Further, the steel material, cast iron material, and non-ferrous metal material for making the blade can also select specific refined materials according to different requirements. Among them, the steel material is further divided into tool steel, alloy steel, medium alloy steel, and high-quality steel, etc.; the cast iron material includes malleable cast iron, wear-resistant cast iron, and ordinary gray cast iron, etc.; the non-ferrous metal material is further divided into hard metal, single-crystal superalloy, and composite materials, etc.

[0020] The machining data of the blade involves the relevant information in the actual machining process of the blade, including the selection parameters of the cutting tool for cutting the blade, cutting speed, specifications and settings of the machining machine tool, and actual machining parameters, etc. Among them, the selection parameters of the cutting tool include the material classification, usage classification, and structure classification of the cutting tool. For example, the material classification of the cutting tool can include high-speed steel, ceramics, and hard metals, etc.; according to the usage classification of the cutting tool, the cutting tool can be classified into turning tools, hole machining tools, broaches, milling cutters, and gear cutters; according to the structure classification of the cutting tool, the cutting tool can be divided into solid tools, clamped tools, composite tools, and insert tools. It should be noted that in the specific blade machining process, appropriate blade machining data can be set according to the surface data, characteristic data of the blade, and blade requirements to facilitate obtaining blades that meet the preset requirements. In this application, it is not limited.

[0021] S102. Preprocess the various types of data to obtain the preprocessed data, and construct a basic knowledge graph related to the blade processing field according to the preprocessed data; the basic knowledge graph takes blades, materials, tools, quality control, processing technology, and processing trajectories as nodes, and takes the inherent characteristics of blades, materials, tools, quality control, processing technology, and processing trajectories as node attributes.

[0022] It should be noted that the preprocessing methods for various types of data can be selected according to actual needs, and they are not limited in this application. For example, in one embodiment, the preprocessing process may include obtaining key structured information from unstructured text, cleaning noise and interference information in text data, removing garbled characters and illegal characters in structured data, repairing grammar errors and / or logically inconsistent statements, etc.

[0023] Figure 2 The following is a flowchart of the second embodiment of the method for determining blade processing parameters provided by this application. Please refer to Figure 2 , based on the first embodiment, the method provided in this embodiment for constructing a basic knowledge graph related to the blade processing field according to the preprocessed data may include: S201. Based on the preprocessed data, identify the entities in the blade processing field and the attributes corresponding to each entity.

[0024] In this step, after preprocessing various types of data, identify the entities in the blade processing field and the attributes corresponding to each entity.

[0025] Specifically, methods such as rule matching, machine learning-based NER models, or pre-trained language models (such as BERT) can be used to extract key entities in the blade processing field from the preprocessed data. Further, techniques such as statistical analysis combined with TF-IDF and word embedding can be used to identify the attributes of each entity.

[0026] Among them, the entities in the blade processing field include blades, materials, tools, quality control, processing technology, and processing trajectories, which are important components in the blade processing process; further, each entity has multiple attributes to describe the attributes of the entity. For example, the attributes of a blade include geometric shape and blade type, etc., and the attributes of a tool include tool type and tool material, etc.

[0027] For example, in one embodiment, the attributes of a blade include Gaussian curvature, mean curvature, principal curvature, torsion, slope, normal direction, tangent direction, area, volume, connectivity, topological type, continuity, smoothness, boundary, characteristic line, characteristic point, shape description, and processing accuracy, where the topological type includes closed surface, open surface, simply connected surface, and multiply connected surface; the attributes of a processing trajectory can include linear trajectory, curve trajectory, spiral trajectory, random trajectory, and mixed trajectory; the attributes of a tool can include tool name, tool number, tool angle, tool diameter, tool durability, and / or tool life, etc.

[0028] S202. Based on the preprocessed data, analyze and determine the first-level relationships among various entities and the second-level relationships among the attributes of different entities, and for each entity, analyze and determine the hierarchical relationships among the attributes of this entity to obtain the sub-ontology models corresponding to the respective entities.

[0029] In this step, based on the preprocessed data, analyze and determine the first-level relationships among various entities and the second-level relationships among the attributes of different entities. The first-level relationship is the hierarchical relationship or association between entities. For example, there is a first-level relationship between the processing technology and the tool selection. Among them, the processing technology is one of the core entities, and the tool selection is closely associated with the processing technology, and there is a clear dependence relationship between them. A specific processing technology will require the use of a specific type of tool. Therefore, the relationship between the processing technology and the tool selection is a first-level relationship. Further, the second-level relationship among the attributes of different entities is the interaction relationship among the attributes of different entities. For example, the tool material is an attribute of the entity tool, and the cutting speed is an attribute of the entity processing technology. Between them, the tool material will directly affect the applicable cutting speed range, that is, different tool materials will affect the selection of the cutting speed in the processing technology, and there is a second-level relationship of dependence between them.

[0030] Specifically, when implementing, methods based on rules (such as dependency syntactic analysis), methods based on machine learning, and methods based on knowledge reasoning can be used to extract the first-level relationships among entities and the second-level relationships among the attributes of different entities.

[0031] Further, in a possible implementation manner, the step of analyzing and determining the hierarchical relationships among the attributes of each entity to obtain the sub-ontology models corresponding to the respective entities may include: S2021. For the blade, according to the geometric shape, physical properties, processing method, optimization objective, and application field of the blade, classify the various features of the blade in detail, determine the relationships among the attributes under each category, establish a hierarchical relationship based on the classification results and the attribute relationships, and formulate inference rules according to the above classification and attribute relationships to obtain the sub-ontology model of the blade.

[0032] In this step, the blades are classified according to their characteristics. Among them, the geometric shape of the blade includes the size and shape of the blade, the physical properties of the blade include strength and density, etc., and the processing method of the blade is determined by how the blade is processed. Then, by analyzing the interactions between these characteristics, the dependency relationships between each attribute are determined, and these dependency relationships are connected through certain inference rules. For example, the strength of the material of the blade will affect the design of the geometric shape of the blade. In this way, the attributes and relationships of the blade can be clearly presented through their hierarchical structure, thus forming a sub-ontology model of the blade and providing effective data support for subsequent decision-making and optimization.

[0033] Figure 3 It is a schematic diagram of the sub-ontology model of the blade shown in an exemplary embodiment of the present application. Please refer to Figure 3 , from Figure 3 , it can be seen that in the sub-ontology model of this blade, the subordinate attributes of the geometric shape of the blade are shape and size, the subordinate attributes of the physical properties are strength and density, the subordinate attributes of the processing method are milling and laser cutting, the subordinate attributes of the optimization goal are strength and corrosion resistance, and the subordinate attributes of the application field are aviation and wind power generation; further, in this sub-ontology model, characteristics such as the geometric shape, physical properties, processing method, optimization goal, and application field of the blade are used as top-level nodes, and they form sub-nodes at different levels through dependency relationships, reflecting the interactions and dependencies in the blade design and optimization process. For example, the geometric shape of the blade (such as size and shape) will affect the selection of its physical properties (such as strength and density), and at the same time these physical properties also determine the processing method of the blade.

[0034] S2022. For materials, classify the tool materials and workpiece materials according to their types, classify the manufacturing processes for each material after type classification, determine the relationships between the various attributes under each category after manufacturing process classification, and finally detect the hierarchical relationships based on the classification results and attribute relationships to obtain the sub-ontology model of the materials.

[0035] It should be noted that the types of tool materials and workpiece materials include metal materials and alloy materials, and each material also has corresponding specific manufacturing processes, such as casting, forging, and welding. After that, for each manufacturing process, by analyzing the physical and chemical properties of the tool material, the changes in the specific properties of the material under different processes can be determined. Further, analyze the relationship between the properties and attributes of the material under different manufacturing processes to form a hierarchical structure to ensure that each material and its related attributes are clearly established. Finally, a sub-ontology model that can reflect the relationship between material properties and processes is obtained. For example, when using aluminum alloy as the manufacturing material, the manufacturing processes that can be used during manufacturing include casting, forging, and welding. Different manufacturing processes will have different degrees of influence on the properties of aluminum alloy material products. Among them, the manufacturing process will affect the properties such as plasticity, strength, hardness, and corrosion resistance of aluminum alloy material products. Specifically, when using the casting process to process aluminum alloy materials, the influence on the strength property of the obtained aluminum alloy material product is relatively low, the hardness property is relatively low, and the plasticity property is relatively high. That is, there is no significant difference between the strength property and hardness property of the aluminum alloy material product obtained through the casting process and the strength property and hardness property of the aluminum alloy material before the casting process. However, the aluminum alloy material product obtained after casting has higher plasticity than the aluminum alloy material before casting; similarly, when using the forging process to process aluminum alloy materials, the influence on the strength property of the obtained aluminum alloy material product is relatively high, the influence on the hardness property is relatively high, and the influence on the plasticity property is medium. That is, the strength of the aluminum alloy material product obtained through the forging process has a significant increase compared with the strength property and hardness property of the aluminum alloy material before the forging process, and the plasticity of the aluminum alloy material product after forging has a certain degree of improvement compared with the aluminum alloy material before forging; when using the welding process to process aluminum alloy materials, the influence on the strength property of the welding area in the obtained aluminum alloy material product is relatively high, the influence on the hardness property is medium, and the influence on the corrosion resistance property is relatively low. That is, the welding area of the aluminum alloy material product obtained through the welding process has lower strength, a certain degree of increased hardness, and basically the same corrosion resistance compared with the aluminum alloy material before the welding process; in this way, an ontology model of aluminum alloy materials is obtained, which can clearly understand the influence of different processes on the performance of aluminum alloy, so as to select a more scientific manufacturing strategy in practical applications.

[0036] Figure 4 Schematic diagram of the sub-ontology model of the material shown in an exemplary embodiment of the present application. Please refer to Figure 4 , Figure 4Taking the aluminum alloy material in the material type as an example of the sub-ontology model, after selecting the aluminum alloy material as the tool and workpiece materials, manufacturing processes such as casting, forging, and welding can be selected for processing. Among them, when selecting the casting process for processing, the material properties that affect the aluminum alloy material include strength property, hardness property, and corrosion resistance property. Specifically, the strength, hardness, and corrosion resistance of the manufactured tool and process are relatively low.

[0037] S2023. For the tool, classify the CNC machining tools according to the processing technology, classify the functions of each classified tool, determine the relationships between the various attributes under each category after function classification, and finally establish a hierarchical relationship based on the classification results and attribute relationships to obtain the sub-ontology model of the tool.

[0038] Specifically, CNC machining tools can be classified into milling cutters, turning tools, drills, etc. according to the processing technology. Taking the milling cutter as an example, for the function classification of the milling cutter in different situations, the milling cutter can be divided into cutting milling cutters and feed milling cutters. Further, the attributes of the cutting milling cutter include cutting force, surface quality, tool life, and cutting depth, and the attributes of the feed milling cutter include cutting force, surface quality, tool life, and feed rate. In this way, by establishing a hierarchical relationship based on the classification results and attribute relationships of the milling cutter, the sub-ontology model of the milling cutter can be obtained, thereby providing a basis for the selection of tools.

[0039] Figure 5 The figure shows a schematic diagram of the sub-ontology model of the tool shown in an exemplary embodiment of the present application. Please refer to Figure 5 , Figure 5 Taking the milling cutter in the CNC machining tools as an example, the milling cutter is classified into cutting milling cutters and feed milling cutters according to the processing technology. Under the classification of cutting milling cutters, cutting force, surface quality, tool life, and cutting depth are defined, and under the classification of feed milling cutters, cutting force, surface quality, tool life, and feed rate are defined, clarifying the hierarchical structure relationship between the classification results of the tools and the attributes defined under each tool.

[0040] S2024. For quality control, taking quality control as the core entity, taking detection methods, quality standards, error ranges, and detection frequencies as the attributes of this entity, and taking the detection methods including detection means, detection tools, detection procedures, and detection environments, the quality standards including international standards, national standards, industry standards, enterprise standards, and technical specifications, the error ranges including absolute error, relative error, quality uncertainty, tolerance range, process error, and detection error, and the detection frequencies including time frequency, sampling ratio, dynamic adjustment, and key point detection, establish a sub-ontology model with quality control as the core entity.

[0041] In this step, for the specific sub - attributes of each attribute of entity quality control, selection can be made according to actual needs, and in this application, no limitation is imposed on them. For example, in the detection method, the detection means in the detection method include electrical testing and functional testing, the detection tools include a multimeter and an oscilloscope, the detection procedure includes first performing electrical testing and then performing functional testing, and the detection environment is under the conditions of room temperature (between 20° and 25°) and humidity less than 70%.

[0042] Figure 6 The following is a schematic diagram of the sub - ontology model of quality control shown in an exemplary embodiment of this application. Please refer to Figure 6 , Figure 6 In, taking the detection method attribute in the core entity as an example, the detection method further includes detection means, detection tools, detection procedures, and detection environment. By refining these contents and through the hierarchical relationship between the attribute detection method and sub - attributes, a systematic and specific detection method can be provided for actual quality control work to determine the detection quality.

[0043] S2025: Regarding the processing technology, taking the processing technology as the core entity, and dividing the processing technology into work - piece materials to be processed, processes, working steps, cutting speed, depth of cut, and feed rate as the attributes of this entity, and taking the type of work - piece materials to be processed as the attribute of the work - piece materials to be processed, a sub - ontology model with the processing technology as the core entity is established.

[0044] In this step, the work - piece materials to be processed include metal materials, non - metal materials, composite materials, and special materials; the processes include rough machining, finish machining, and ultra - finish machining; the working steps include cutting, grinding, and electrical machining; the cutting speed includes low cutting speed, medium cutting speed, and high cutting speed; the depth of cut includes small depth of cut, medium depth of cut, and large depth of cut; the feed rate includes low feed rate, medium feed rate, and high feed rate. Suppose milling is being carried out, and the work - piece material to be processed is aluminum alloy. Through the above - mentioned attributes, a sub - ontology model can be constructed for this processing technology. For example, the work - piece material to be processed in the processing technology is the metal material aluminum alloy, the process is rough machining, the working step is cutting, the cutting speed is medium cutting speed (100 m / min to 150 m / min), the depth of cut is medium depth of cut (2 mm to 4 mm), and the feed rate is medium feed rate (0.1 mm / rev to 0.2 mm / rev).

[0045] It should be noted that a feed rate of 0.1 mm / rev means that whenever the tool rotates one circle, the tool will advance 0.1 mm along the cutting direction of the work - piece, and a feed rate of 0.2 mm / rev means that whenever the tool rotates one circle, the tool will advance 0.2 mm along the cutting direction of the work - piece.

[0046] Figure 7The schematic diagram of the sub-ontology model of the machining process shown in an exemplary embodiment of the present application is as follows. Please refer to Figure 7 , Figure 7 which shows that starting from the core entity of the machining process, it is refined layer by layer to each specific machining link. Through this hierarchical structure, the relationship between different machining processes and their attributes can be clearly understood, and the connection between each hierarchical structure can be intuitively displayed.

[0047] S2026. For the machining trajectory, classify the machining trajectory according to the machining trajectory type to obtain the hierarchical relationship of the machining trajectory, and for each type of machining trajectory, establish a sub-ontology model with the machining trajectory as the core entity for the attributes of the scene suitable for the machining trajectory and the tool adapted to the machining trajectory.

[0048] Specifically, classifying the machining trajectory according to the machining trajectory type, the machining trajectory can be divided into a straight trajectory, an arc trajectory, a spiral trajectory, and an arbitrary curve trajectory. Among them, the scenes suitable for the straight trajectory are machining planes, grooves, and holes, and the adapted tools are milling cutters and turning tools; the scenes suitable for the arc trajectory are machining curved surfaces and inner and outer circles, and the adapted tools are milling cutters, turning tools, and tool radius compensation tools; the scenes suitable for the spiral trajectory are machining threads and internal spiral structures in holes, and the adapted tools are milling cutters, thread tools, and drills; the scenes suitable for the arbitrary curved surface trajectory are machining free-form surfaces, complex ink molds, and special surfaces, and the adapted tools are 5-axis milling cutters, electric machining tools, and special tools.

[0049] Figure 8 The schematic diagram of the sub-ontology model of the machining trajectory shown in an exemplary embodiment of the present application is as follows. Please refer to Figure 8 , from Figure 8 it can be seen that classifying the core main body machining trajectory into a straight trajectory, an arc trajectory, a spiral trajectory, and an arbitrary curve trajectory, and the attributes of each machining trajectory include sub-attributes of the suitable scene and the adapted tool. Figure 8 Taking the straight trajectory as an example in , the scenes suitable for the straight trajectory include machining planes, grooves, and holes, and the adapted tools are milling cutters and turning tools. In this way, different machining trajectories can be adopted according to different scenes and tools, with a clear hierarchical structure, which is convenient for selecting a suitable machining trajectory according to the sub-ontology model of the machining trajectory.

[0050] S203. According to the first hierarchical relationship, the sub-ontology models corresponding to each entity, and the second hierarchical relationship, obtain the ontology model of the blade machining field; the ontology model is used to characterize the hierarchical relationship and mutual relationship between entities and attributes.

[0051] In this step, the first hierarchical relationship, the sub-ontology models corresponding to each entity, and the second hierarchical relationship can be fused to obtain the ontology model of the blade machining field, and this ontology model can accurately characterize the hierarchical structure of the blade machining field.

[0052] S204. Identify triple data of entities, attributes, and values from the preprocessed data based on the ontology model, and construct a basic knowledge graph for the blade processing field using the identified triple data; wherein, the basic knowledge graph represents blade field knowledge through the relationships between entities, and at the same time forms fine-grained relationship and attribute descriptions through triple data, and can accurately represent various entities, attributes, and the relationships between them in the blade processing field.

[0053] Specifically, in the ontology model, the hierarchical relationships and interactions between various entities and attributes have been clearly characterized. After that, entities, attributes, and values are extracted from the preprocessed data to form specific triple data. Among them, entities include processing technology, tool selection, tool material, etc., attributes include cutting speed, feed rate, tool material and hardness, etc., and values include cutting speed of 150 m / min, feed rate of 150 m / rev, etc. The formed triples can be (processing technology, cutting speed, 150 m / min), (tool selection, tool material, cemented carbide), (processing technology, feed rate, 0.05 mm / rev), etc.

[0054] Furthermore, after obtaining the triple data, a basic knowledge graph for the blade processing field can be constructed through the triple data, forming fine-grained relationship and attribute descriptions. In this way, it can accurately represent various entities, attributes, and the relationships between them in the blade processing field.

[0055] It should be noted that based on the knowledge graph, processing parameters can be determined. For example, when a workpiece that needs to process a complex curved surface requires a suitable processing trajectory, the workpiece may include complex free-form surfaces and requires precise three-dimensional cutting. Then, an arbitrary curve trajectory is selected as the processing trajectory. Due to the complexity of the trajectory, a 5-axis milling cutter is selected as the tool.

[0056] It should be noted that by constructing an ontology model and then obtaining a basic knowledge graph for the blade processing field based on the ontology model, through this basic knowledge graph, the hierarchical relationships and mutual relationships between various entities and their attributes in the blade processing process can be effectively represented and organized. Among them, through the fine-grained triple data in the basic knowledge graph, the dependency relationships between entities such as processing technology, tool selection, and material characteristics can be clearly described, providing more accurate information for subsequent determination of processing parameters.

[0057] S103. Construct a causal relationship graph related to the blade processing field according to the preprocessed data, and update the basic knowledge graph using the causal relationship graph to obtain a target knowledge graph.

[0058] Optionally, in a possible implementation manner, the process of constructing a causal relationship graph related to the field of blade processing according to the preprocessed data includes: (1) Mining causal relationships from the preprocessed data based on a causal discovery algorithm to obtain a first causal relationship set.

[0059] It should be noted that the first causal relationship set contains potential causal relationships in the preprocessed data, and these causal relationships can reveal the direct influence relationships between different processing parameters; further, unsupervised causal algorithms include the PC algorithm, the LiNGAM algorithm, the GES algorithm, etc. Taking the PC algorithm as an example, the principle of this unsupervised causal algorithm is to infer causal relationships by constructing a conditional independence graph. First, construct a fully connected undirected graph, and then gradually delete the edges that do not conform to the causal hypothesis according to the conditional independence test in the preprocessed data. Finally, obtain the causal relationship represented by a directed acyclic graph to get the first causal relationship graph.

[0060] (2) Identifying correlation relationships from the preprocessed data based on a statistical method to obtain a second causal relationship set.

[0061] Specifically, first, select a suitable statistical method according to the type of preprocessed data. Taking the statistical method of Pearson correlation coefficient as an example, first, it is necessary to ensure that the preprocessed data is continuous and conforms to the linear relationship hypothesis; then, calculate the Pearson correlation coefficient of each pair of variables, and then conduct a hypothesis test. Set the null hypothesis as "no correlation" (H 0 :r = 0), use the t-test to calculate the p-value, and judge whether the correlation is significant; for example, if the p-value is less than the significance level (such as 0.05), it is considered that there is a significant linear correlation. If the p-value is greater than the significance level (such as 0.05), it is considered that there is no significant linear correlation.

[0062] Further, after calculating the correlation coefficients between variables in the preprocessed data, it is further possible to determine the relevant variable pairs based on the calculated correlation coefficients, and determine the set composed of the relevant variable pairs as the second causal relationship set with causal relationships. Among them, the second causal relationship set identifies possible correlation relationships in the data through statistical methods, and these relationships can provide statistical association information between different entities and attributes.

[0063] (3) Fusing the first causal relationship set and the second causal relationship set, and using prior causal relationships and intervention learning to verify whether each causal relationship in the fused causal relationship base holds, so as to find the target causal relationships that hold from the causal relationship set.

[0064] In specific implementation, the first causal relationship set and the second causal relationship can be combined (taking the union or intersection) to obtain the fused causal relationship.

[0065] It should be noted that the prior causal relationship is the causal relationship between entities and attributes preset based on the knowledge, experience or known theoretical models in the field before data analysis; intervention learning is a causal reasoning method based on "intervention" or "operation". In intervention learning, by artificially intervening in a certain variable in the system (for example, adjusting the value of a parameter in a certain process step of the processing), observing the reaction in the whole system, the causal influence of this variable on other variables can be inferred, and the causal relationship can be obtained.

[0066] In specific implementation, when verifying the causal relationship based on intervention learning, the causal relationship can be verified based on A / B experiments (control variable method), causal inference models, data-driven intervention experiments (using reinforcement learning RL, adjusting processing parameters in the simulation environment and observing the results), etc.

[0067] In this step, by fusing the first causal relationship set and the second causal relationship set, combining the prior causal relationship and the verification method of intervention learning, and then screening out the established target causal relationship, in this way, the effectiveness and feasibility of the causal relationship in the causal relationship set can be ensured.

[0068] (4) Construct a causal relationship diagram related to the blade processing field according to the target causal relationship.

[0069] It should be noted that in this embodiment, potential causal relationships are extracted from the data by using unsupervised causal discovery algorithms (such as the PC algorithm) and statistical methods (such as Pearson correlation coefficient). These methods can reveal the true correlations in the data. Secondly, the preprocessed data has been strictly preprocessed to remove noise and missing values, ensuring the quality of the analysis results; further, the prior knowledge of the field and intervention learning are also combined to verify the causal relationship, ensuring that the identified causal relationship is not only valid in the data but also verified in actual operations. In this way, through multi-level verification and fusion, the finally obtained causal relationship diagram can accurately reflect the causal mechanism in the blade processing field. Therefore, the finally obtained causal relationship diagram has a scientific basis and can provide effective decision support in practical applications.

[0070] Furthermore, after obtaining the causal relationship, the basic knowledge graph is updated by using the causal relationship diagram to obtain the target knowledge graph.

[0071] Specifically, by mapping the entities and relationships in the causal relationship diagram to the nodes and edges of the basic knowledge graph, the content of the existing graph is extended, and the validity of the new relationships is verified through domain knowledge. At the same time, based on the statistical support and experimental verification in the causal relationship diagram, the weights and credibility of the relationships in the basic knowledge graph are updated to obtain the target knowledge graph. In this way, the obtained target knowledge graph will be more comprehensive and accurate, and can reflect the causal mechanism in the blade processing field in real time, providing more effective support for decision-making and optimization.

[0072] It should be noted that the target knowledge graph can also clearly display the causal action chain among entities such as tool materials, processing technologies, and tool selection. In this way, it can provide more comprehensive data support and decision-making basis for subsequent processing optimization and parameter adjustment.

[0073] S104. Use the preprocessed data and the target knowledge graph to train the large model so that the large model learns the knowledge in the blade processing field and obtains a trained large model.

[0074] The goal of training the large model is to enable the model to fully learn the knowledge in the blade processing field and be used for feature extraction, processing parameter prediction, etc. Specifically, the causal reasoning ability of the model can be enhanced by combining with the knowledge graph. In specific implementation, knowledge can be effectively captured from the preprocessed data and the target knowledge graph and used as pre-training data for the large model for data annotation and training. Further, knowledge injection can be carried out so that the large model can also obtain the association information of entities when learning word vectors. At the same time, the triple relationships in the target knowledge graph are introduced to effectively make up for the knowledge gaps in the large model. In addition, auxiliary tasks such as edge prediction or triple prediction can be added to enable the large model to learn to better understand the entities and relationships in the blade processing field. Further, in the pre-training process, the large model learns the statistical characteristics and patterns of the blade feature attributes and the data related to blade processing, extracts the features related to blade quality, and finally, in the fine-tuning and verification, the pre-trained model is fine-tuned according to the specific task. In the fine-tuning training data, the annotations in the target knowledge graph are introduced to provide additional supervision information; and clear evaluation indicators are formulated according to different downstream tasks.

[0075] Optionally, in a possible implementation manner, the specific implementation process of this step may include: (1). Construct training samples based on the preprocessed data to train the large model using the training samples.

[0076] In this step, the preprocessed data includes the key parameters and relevant information in the blade processing field, which can provide necessary data support for the large model. Training samples for training the large model are constructed according to the preprocessed data. In this way, it can be ensured that the large model obtains rich and representative data during the training process.

[0077] (2) Extract triples from the target knowledge graph, and introduce the triples as additional context information during the word vector training of the large model.

[0078] In this step, extract triple data from the target knowledge graph, such as (processing technology, cutting speed, 150 m / min). During the word vector training of the large model, introduce this triple as additional context information. In this way, when the large model learns word vectors, it can not only consider the context of words, but also understand the interaction and background information between words based on the relationships in the knowledge graph, thereby enhancing the large model's understanding of the knowledge in the blade processing field.

[0079] (3) Add edge prediction and triple prediction as auxiliary tasks to the training process, and through joint optimization, enable the large model to understand and infer the relationships between domain entities while learning word vectors.

[0080] Specifically, edge prediction refers to predicting the potential relationships between entities in the knowledge graph. For example, for two entities "tool type" and "processing technology", the model needs to predict whether there is a certain interaction relationship between them; triple prediction involves predicting the integrity or existence of triples. For example, in a given triple including entity 1, entity 2, and an attribute, given partial information in the triple, such as the relationship between entity 1 and the attribute, predict the other entity 2.

[0081] In this step, add edge prediction and triple prediction as auxiliary tasks to the training process. Through joint optimization, the large model can not only learn how to represent the vectors of words, but also understand and infer the relationships and influence mechanisms between domain entities.

[0082] Specifically, during joint optimization, a joint loss function can be set based on the training tasks, and joint optimization can be carried out based on this joint loss function.

[0083] It should be noted that joint optimization means simultaneously optimizing the tasks of word vector learning, triples, and edge prediction.

[0084] In this embodiment, a large model is trained by combining the preprocessed data and the target knowledge graph. Among them, the preprocessed data is used to construct training samples to ensure that the large model can receive rich feature information related to the blade processing field during the training process, thereby improving the accuracy of its feature extraction and processing parameter prediction. Further, by introducing the triples in the target knowledge graph as context information into word vector training, the model can learn the context of words while also understanding the relationships and background knowledge between different entities, enhancing its causal reasoning ability. After that, edge prediction and triple prediction are used as auxiliary tasks. Through joint optimization, the model can infer the potential relationships and interactions between entities in the blade processing field while learning word vectors, thereby improving the understanding of domain knowledge and decision-making support capabilities. In this way, when the large model optimizes the blade processing process, it can perform more accurate feature extraction, processing parameter prediction, and optimization suggestions based on domain knowledge.

[0085] S105. For each piece of data in the processing preparation stage data and the actual processing data among the various types of data, use the trained large model to extract features from this piece of data to obtain the feature vector of this piece of data, and construct a feature vector library based on the attribute information and feature vector of each piece of data.

[0086] Specifically, use the trained large model to process each piece of data and convert each piece of data into a feature vector. Among them, this data can include records of processing parameters, processing techniques, tool types, etc. during the processing preparation stage and the actual processing stage. After that, through the trained large model, this data can be converted into a high-dimensional vector representation to capture the potential information in the data.

[0087] Further, based on the attribute information of each piece of data and the generated feature vector, construct a feature vector library. Among them, multiple records are stored in the feature vector library, and each record is used to record the corresponding relationship between the attribute information of a piece of data and the feature vector. The attribute information of a piece of data can be information such as the blade type, material, and processing status carried by this piece of data.

[0088] In a possible implementation manner, after constructing the feature vector library, the method may further include: Step 1: According to the entities in the target knowledge graph, map each feature vector in the feature vector library to the corresponding entity in the target knowledge graph to ensure that each feature vector is correctly associated with the corresponding entity.

[0089] Specifically, during the mapping process, first determine the corresponding items of each entity in the target knowledge graph in the feature vector library. For example, match the corresponding feature vectors in the feature vector library according to the entity names or attribute information in the target knowledge graph, and associate each feature vector in the feature vector library with the corresponding entity. Among them, if an entity has multiple attributes or relationships, select the most relevant feature vector for mapping; further, if a feature vector cannot directly match an entity, it can match the upper-level concept.

[0090] In this step, according to the entity information in the target knowledge graph, map each feature vector in the feature vector library to the corresponding entity in the knowledge graph. In this way, it can be ensured that each feature vector can correctly establish a connection with its corresponding domain entity, so that the feature vector is not only an abstract representation of data, but also can accurately reflect the semantic information of the entity.

[0091] Step 2: For each feature vector, use the relationships and attributes of the entity corresponding to the feature vector in the target knowledge graph to adjust the expression of the feature vector.

[0092] When specifically implemented, the entity attributes in the target knowledge graph (such as the hardness and material type of the blade, etc.) can be fused with the feature vector to form an enhanced feature representation. In addition, the relationships and attributes of the entity in the target knowledge graph can be combined into the feature vector by means of splicing, weighting, etc.

[0093] Optionally, the relationship information in the target knowledge graph (such as "the processing relationship between the blade and the material") can also be used to adjust or enhance the feature vector. For example, through the semantic relationship between entities in the target knowledge graph, adjust the position of the feature vector in the vector space to make the relevant feature vectors closer, thereby improving the accuracy of feature matching.

[0094] In this step, for each feature vector, according to the relationships and attributes of the entity corresponding to the feature vector in the knowledge graph, adjust the expression of the feature vector. It can be understood that the influence relationships and attributes in the knowledge graph provide in-depth background information in the field of blade processing. These information help to adjust the expression of the feature vector, so that the feature vector can more accurately reflect the interaction and influencing factors between entities, thereby optimizing the representation effect of the feature vector.

[0095] In summary, after the feature vector library is constructed, through the above two steps, it can be ensured that each feature vector in the feature vector library not only has a high-quality semantic representation, but also can closely combine each feature vector with the entities and influence relationships in the knowledge graph, which can improve the accuracy of feature matching.

[0096] S106. When it is necessary to determine the machining trajectory of the blade to be machined, use the trained large model to extract the target feature vector of the blade to be machined, and search for the matching feature vector that matches the target feature vector in the feature vector library.

[0097] In this step, first use the trained large model to extract features from the relevant data of the blade to be machined, such as the geometric shape of the blade, the blade material, etc., to obtain the target feature vector containing important feature information in the leaf processing process; then, in the feature vector library, search for the matching feature vector that matches the extracted target feature vector.

[0098] It should be noted that searching for the matching feature vector that matches the extracted target feature vector in the feature vector library can be based on the similarity measure of the feature vector. For example, according to the cosine similarity or Euclidean distance similarity, search for the feature vector most similar to the target feature vector in the feature vector library, or search for the feature vector whose similarity is greater than the preset threshold.

[0099] S107. Search for the target machining parameters corresponding to the attribute information in the target knowledge graph according to the attribute information of the matching feature vector, and input the matching feature vector and the target machining parameters into the trained large model, so that the trained large model optimizes the target machining parameters to obtain the optimized machining parameters.

[0100] In this step, first, according to the attribute information of the matching feature vector, search for the target machining parameters corresponding to the attribute information of the matching feature vector in the target knowledge graph. For example, the attribute information in the matching feature vector may include information about the blade material, tool type, and cutting speed, etc. From the target knowledge graph, machining parameters related to the above information, such as the optimal cutting depth and the rotation speed of the tool, can be searched.

[0101] Furthermore, after the machining parameters are found, the trained large model can be further used to optimize the machining parameters. Specifically, when implementing, input the matching feature vector and the target machining parameters into the trained large model. Based on the understanding of the historical machining data and the knowledge of the blade machining field, the large model can further optimize the target machining parameters considering factors such as the geometric shape of the current blade, the machining environment, and the machining process requirements, and then obtain the optimized machining parameters.

[0102] Furthermore, in a possible implementation manner, before inputting the matching feature vector and the target machining parameters into the trained large model, the method may further include: Step 1: According to the target knowledge graph and the attribute information of the blade to be processed, verify the rationality of the matching feature vector to exclude the matching feature vectors that do not conform to the target knowledge graph and screen out the target matching feature vectors that conform to the target knowledge graph.

[0103] Step 2: Input the target matching feature vector and the target processing parameters corresponding to the target matching feature vector into the trained large model.

[0104] In the above steps, by verifying the rationality of the matching feature vector according to the target knowledge graph and the attribute information of the blade to be processed, the matching feature vectors that do not conform to the target knowledge graph are excluded, and the target matching feature vectors that conform to the target knowledge graph are screened out. In this way, it can be ensured that the matching feature vectors before inputting data into the large model are not only technically effective but also conform to the actual processing rules in the field of blade processing, thereby improving the accuracy of the large model in optimizing processing parameters.

[0105] It should be noted that in this embodiment, by constructing a knowledge graph and a causal relationship graph and combining the training of the large model, the optimization of processing parameters no longer depends on a single experience. Instead, by combining data-driven, knowledge graph, causal reasoning, and large model optimization to optimize processing parameters, the scientificity and accuracy of parameter optimization can be improved. In addition, through the feature vector matching method, the historical processing parameters most suitable for the current blade processing can be found and further optimized to make the processing parameters more accurate.

[0106] The method for determining blade processing parameters provided in this embodiment integrates various types of data related to complex surface machining, preprocesses various types of data to obtain preprocessed data, constructs a basic knowledge graph and a causal relationship graph based on the preprocessed data, and then trains a large model using the preprocessed data and the target knowledge graph to obtain a trained large model. After that, the trained large model extracts features from each piece of data in the machining preparation stage and actual machining data to obtain corresponding feature vectors, and constructs a feature vector library based on the attribute information and feature vectors of each piece of data. Further, when it is necessary to determine the machining trajectory of the blade to be machined, the large model is used to extract the target feature vector of the blade, and the matching feature vector is searched for in the feature vector library. By analyzing the attribute information of the matching feature vector, the corresponding target machining parameters are extracted from the target knowledge graph, and the matching feature vector and the target machining parameters are input into the large model for further optimization to obtain optimized machining parameters. In this way, first, by constructing a knowledge graph for blade machining and training a large model in combination with causal relationships, the large model can not only remember knowledge but also infer the relationship between machining parameters and machining effects, improving the interpretability of knowledge. Further, the trained large model is used to extract features from the machining preparation stage and actual machining data and construct a feature vector library, which enables the model to perform in-depth analysis before and after machining. In addition, by combining the matching of the knowledge graph and the feature vector library, effective machining parameters are extracted from similar historical data and further optimized in the large model to ensure that the output parameters better meet the current machining requirements and can accurately determine the machining parameters.

[0107] Optionally, in a possible implementation manner, the method further includes: Using the trained large model to analyze the processed data according to the machining process flow to realize the extraction of causal association knowledge among blades, tools, machining processes, machining trajectories, quality problems, and influencing factors, and correcting the causal relationship graph based on the extracted causal association indications.

[0108] In specific implementation, the preprocessed data is input into the trained large model to enable the large model to identify causal associations. Further, the causal relationship graph can be corrected based on the causal associations.

[0109] Specifically, in this method, by layer-by-layer analyzing the preprocessed data according to the machining process flow (such as design, machining, quality inspection, etc.), the extraction of causal association knowledge among various entities is realized. In this way, the large model can understand the influence of each stage and step on the final machining result. After that, correcting the causal relationship graph based on the extracted causal association indications can identify and prevent possible problems in the machining process before the machining process starts, thereby improving machining efficiency and machining quality.

[0110] Optionally, in another possible implementation, using the natural language and multi-modal understanding capabilities of the trained large model, entities and relationships related to the blade processing quality are identified from the preprocessed data, and the identified entities and relationships are used to supplement or correct the target knowledge graph.

[0111] Specifically, in this method, the trained large model is used to identify entities and relationships related to the blade processing quality from the preprocessed data, and the target knowledge graph is supplemented and corrected according to the identified entities and relationships. In this way, the target knowledge graph is continuously iteratively updated, so that the data and logic in the large model can be continuously updated and improved, enhancing the intelligence and decision-making accuracy of the large model.

[0112] Corresponding to the foregoing embodiment of a method for determining blade processing parameters, the present application also provides an embodiment of a device for determining blade processing parameters.

[0113] Figure 9 The following is a schematic structural diagram of Embodiment 1 of a device for determining blade processing parameters shown in an exemplary embodiment of the present application. Please refer to Figure 9 The device includes an acquisition module 910, a processing module 920, a construction module 930, a training module 940, and an optimization module 950, where The acquisition module 910 is configured to acquire various types of data related to complex surface processing; among them, the various types of data reflect the surface data, feature data, and machining data of the blade; The processing module 920 is configured to preprocess the various types of data to obtain preprocessed data, and construct a basic knowledge graph related to the blade processing field according to the preprocessed data; the basic knowledge graph uses blades, materials, tools, quality control, processing technology, and processing trajectories as nodes, and the inherent characteristics of blades, materials, tools, quality control, processing technology, and processing trajectories as node attributes; The construction module 930 is configured to construct a causal relationship graph related to the blade processing field according to the preprocessed data, and update the basic knowledge graph using the causal relationship graph to obtain a target knowledge graph; The training module 940 is configured to train a large model using the preprocessed data and the target knowledge graph, so that the large model learns the knowledge in the blade processing field to obtain a trained large model; The construction module 930 is configured to, for each piece of data in the processing preparation stage data and the actual processing data among the various types of data, use the trained large model to extract features of the piece of data to obtain a feature vector of the piece of data, and construct a feature vector library based on the attribute information and feature vectors of each piece of data; The obtaining module 910 is configured to, when it is necessary to determine the processing trajectory of the blade to be processed, extract the target feature vector of the blade to be processed by using the trained large model, and search for a matching feature vector that matches the target feature vector in the feature vector library; The optimization module 950 is configured to search for the target processing parameters corresponding to the attribute information in the target knowledge graph according to the attribute information of the matching feature vector, and input the matching feature vector and the target processing parameters into the trained large model, so that the trained large model optimizes the target processing parameters to obtain optimized processing parameters.

[0114] The device in this embodiment can be used to execute Figure 1 the steps of the method embodiment shown. The specific implementation principle and process are similar and will not be elaborated here.

[0115] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. Those of ordinary skill in the art can understand and implement it without creative work.

[0116] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining blade processing parameters, characterized in that: The blade processing parameter determination method comprises: Acquire various data related to complex surface processing; wherein the various data reflect the surface data, feature data and machining data of the blade; Preprocessing the various types of data to obtain preprocessed data, and constructing a basic knowledge map related to the blade processing field based on the preprocessed data; the basic knowledge map uses blades, materials, tools, quality control, processing technology and processing trajectories as nodes, and uses inherent characteristics of blades, materials, tools, quality control, processing technology and processing trajectories as node attributes; According to the preprocessed data, a cause-and-effect relationship graph related to the blade processing field is constructed, and the cause-and-effect relationship graph is used to update the basic knowledge graph to obtain a target knowledge graph; Using the preprocessed data and the target knowledge graph to train a large model, so that the large model learns knowledge in the field of blade processing, and obtains a trained large model; For each piece of data in the processing preparation stage data and the actual processing data in the various types of data, the trained large model is used to extract features of the data to obtain a feature vector of the data, and a feature vector library is constructed based on the attribute information and feature vectors of each piece of data; When it is necessary to determine the machining trajectory of the blade to be machined, the target feature vector of the blade to be machined is extracted using the trained large model, and a matching feature vector matching the target feature vector is searched from the feature vector library; According to the attribute information of the matching feature vector, the target processing parameters corresponding to the attribute information are searched from the target knowledge graph, and the matching feature vector and the target processing parameters are input into the trained large model so that the trained large model can optimize the target processing parameters to obtain the optimized processing parameters.

2. The method according to claim 1, characterized in that The method of using the preprocessed data and the target knowledge graph to train the large model so that the large model learns knowledge in the field of blade processing to obtain a trained large model includes: Constructing training samples based on the preprocessed data to train the large model using the training samples; Extracting triples from the target knowledge graph, and introducing the triples as additional context information into the word vector training when training the word vector of the large model; Edge prediction and triple prediction are added as auxiliary tasks to the training process. Through joint optimization, the large model can understand and infer the relationship between entities in the leaf field while learning word embeddings.

3. The method according to claim 1, characterized in that: After constructing the feature vector library, the method further includes: According to the entities in the target knowledge graph, mapping each feature vector in the feature vector library with the corresponding entity in the target knowledge graph to ensure that each feature vector is correctly associated with the corresponding entity; For each feature vector, the expression of the feature vector is adjusted using the relationship and attributes of the entity corresponding to the feature vector in the target knowledge graph.

4. The method according to claim 1, characterized in that The process of constructing a cause-effect relationship diagram related to the blade processing field according to the preprocessed data includes: Mining causal relationships from the preprocessed data based on a causal discovery algorithm to obtain a first causal relationship set; Identifying correlations from the preprocessed data based on a statistical method to obtain a second causal relationship set; Fusing the first causal relationship set and the second causal relationship set, and using prior causal relationships and intervention learning to verify whether each causal relationship in the fused causal relationship set is established, so as to find a valid target causal relationship from the causal relationship set; According to the target causal relationship, a causal relationship diagram related to the blade processing field is constructed.

5. The method according to claim 1, characterized in that According to the pre-processed data, a basic knowledge graph related to the blade processing field is constructed, including: Based on the preprocessed data, identifying entities in the blade processing field and attributes corresponding to each entity; Based on the preprocessed data, the first-level relationships between entities and the second-level relationships between attributes of different entities are analyzed and determined, and for each entity, the hierarchical relationships between the attributes of the entity are analyzed and determined to obtain a sub-ontology model corresponding to each entity; According to the first hierarchical relationship, the sub-ontology models corresponding to each entity and the second hierarchical relationship, an ontology model of the blade processing field is obtained; the ontology model is used to characterize the hierarchical relationship and mutual relationship between entities and attributes; Based on the ontology model, triple data of entities, attributes and values ​​are identified from the preprocessed data, and the identified triple data are used to construct a basic knowledge graph in the field of blade processing; wherein the basic knowledge graph represents blade domain knowledge through entities and relationships between entities, and at the same time forms fine-grained relationship and attribute descriptions through triple data, which can accurately characterize various entities, attributes and the relationships between them in the field of blade processing.

6. The method according to claim 5, characterized in that The entities include blades, materials, tools, quality control, processing technology and processing trajectory; for each entity, the hierarchical relationship between the attributes of the entity is analyzed and determined to obtain the sub-ontology model corresponding to each entity, including: For blades, the various characteristics of the blades are carefully classified according to the blade's geometric shape, physical properties, processing methods, optimization goals and application fields, and the relationship between the various attributes under each category is determined. A hierarchical relationship is established based on the classification results and attribute relationships, and inference rules are formulated based on the above classification and attribute relationships to obtain the sub-ontology model of the blade; For materials, tool materials and workpiece materials are classified according to their types, and the manufacturing process of each material after type classification is classified, and the relationship between the various attributes under each category after the manufacturing process classification is determined. Finally, the hierarchical relationship is detected based on the classification results and the attribute relationship to obtain the sub-ontology model of the material; For tools, CNC machining tools are classified according to the machining process, and each tool is functionally classified after classification, and the relationship between the attributes under each category after functional classification is determined. Finally, a hierarchical relationship is established based on the classification results and the attribute relationship to obtain the sub-ontology model of the tool; For quality control, with quality control as the core entity, detection methods, quality standards, error range, and detection frequency as the attributes of the entity, and detection methods including detection means, detection tools, detection procedures, and detection environment, quality standards including international standards, national standards, industry standards, enterprise standards, and technical specifications, error ranges including absolute error, relative error, quality uncertainty, tolerance range, process error, and detection error, and detection frequency including time frequency, sampling ratio, dynamic adjustment, and key point detection, a sub-ontology model with quality control as the core entity is established; For the processing technology, the processing technology is taken as the core entity, and the material to be processed, process, step, cutting speed, back cutting amount and feed rate are taken as the attributes of the entity, and the type of the material to be processed is taken as the attribute of the material to be processed to establish a sub-ontology model with the processing technology as the core entity; For the processing trajectory, the processing trajectory is classified according to the processing trajectory type to obtain the hierarchical relationship of the processing trajectory. For each type of processing trajectory, a sub-ontology model with the processing trajectory as the core entity is established based on the attributes of the processing trajectory, including the scene suitable for the processing trajectory and the tool suitable for the processing trajectory.

7. The method according to claim 1, characterized in that Before inputting the matching feature vector and the target processing parameter into the trained large model, the method further includes: According to the target knowledge graph and the attribute information of the blade to be processed, verify the rationality of the matching feature vector to exclude the matching feature vector that does not conform to the target knowledge graph, and screen out the target matching feature vector that conforms to the target knowledge graph; The step of inputting the matching feature vector and the target processing parameter into the trained large model comprises: The target matching feature vector and the target processing parameters corresponding to the target matching feature vector are input into the trained large model.

8. The method according to claim 1, characterized in that: The method further comprises: Analyze the processed data according to the machining process flow using the trained large model to extract causal relationship knowledge among blades, tools, machining processes, machining trajectories, quality issues and influencing factors, and modify the causal relationship diagram based on the extracted causal relationship indications; or, The natural language and multimodal understanding capabilities of the trained large model are used to identify entities and relationships related to blade processing quality from the preprocessed data, and the identified entities and relationships are used to supplement or correct the target knowledge graph.

9. The method according to claim 1, characterized in that: The acquisition of various data related to complex surface processing includes: Determine the data collection type as structured text, semi-structured text, unstructured text, tables, documents, pictures, audio and video, and / or web pages; The data sources include CAD models and design drawings, blade geometry parameters, material properties in the product design phase, CFD analysis results, FEA stress-strain prediction, thermodynamic performance evaluation in the simulation phase, tool selection parameters, machine tool specifications, NC programming information in the precision machining phase, real-time machining parameters, online measurement data, temperature monitoring data in the machining process, final product quality inspection reports, defect records, customer feedback in the quality control phase, and fault history records, maintenance logs, and improvement measures documents in the maintenance and improvement phase; Determine the data collection method including direct collection, system connection, manual import and / or input; Various types of data indicated by the data collection type are collected from the data source according to the data collection method.

10. A blade processing parameter determination device, characterized in that: The device comprises an acquisition module, a processing module, a construction module, a training module and an optimization module, wherein: The acquisition module is used to acquire various data related to complex surface processing; wherein the various data reflect the surface data, feature data and machining data of the blade; The processing module is used to preprocess the various types of data to obtain preprocessed data, and to construct a basic knowledge map related to the blade processing field based on the preprocessed data; the basic knowledge map uses blades, materials, tools, quality control, processing technology and processing trajectories as nodes, and uses inherent characteristics of blades, materials, tools, quality control, processing technology and processing trajectories as node attributes; The construction module is used to construct a cause-effect relationship graph related to the blade processing field according to the preprocessed data, and use the cause-effect relationship graph to update the basic knowledge graph to obtain a target knowledge graph; The training module is used to train the large model using the preprocessed data and the target knowledge graph, so that the large model learns the knowledge in the field of blade processing to obtain a trained large model; The construction module is used to extract features of each piece of data in the processing preparation stage data and the actual processing data in the various types of data using the trained large model to obtain a feature vector of the data, and to construct a feature vector library based on the attribute information and feature vector of each piece of data; The acquisition module is used to extract the target feature vector of the blade to be processed by using the trained large model when it is necessary to determine the processing trajectory of the blade to be processed, and to search for a matching feature vector that matches the target feature vector from the feature vector library; The optimization module is used to search for target processing parameters corresponding to the attribute information from the target knowledge graph according to the attribute information of the matching feature vector, and input the matching feature vector and the target processing parameters into the trained large model so that the trained large model can optimize the target processing parameters to obtain optimized processing parameters.

Images