CN115033800A - Knowledge graph-based numerical control machining cutter recommendation method - Google Patents

Abstract

The invention provides a knowledge graph-based numerical control machining cutter recommendation method, which can be used for accurate recommendation of numerical control machining cutters and comprises the following steps: preprocessing the numerical control processing original data; constructing a body model of numerical control machining; constructing a knowledge graph of numerical control processing; constructing a numerical control machining one-way weighting knowledge map; and obtaining a recommended result of the numerical control machining tool. According to the invention, the knowledge graph of numerical control machining is constructed through the body model of numerical control machining, the numerical control machining one-way weighting knowledge graph constructed through the knowledge graph is used for calculating the correlation degree between the cutter node and the node of the part to be machined by using the PPR algorithm, and then the cutter node with the highest correlation degree is taken as the recommended cutter, so that the influence of the geometric parameters of the cutter in the cutter recommending process is fully considered, and the cutter recommending accuracy is effectively improved.

Description

NC machining tool recommendation method based on knowledge graph

technical field

The invention belongs to the field of numerical control machining, and relates to a method for recommending numerically controlled machining tools, in particular to a method for recommending numerically controlled machining tools based on a knowledge map, which can be used for accurate recommendation of numerically controlled machining tools.

Background technique

Numerical control machining refers to a machining method in which digital information controls the displacement of parts and tools. It is an effective way to solve the problems of variable variety of parts, small batches, complex shapes, and high precision and to achieve efficient and automated machining. Tool is an important part of CNC machining, which has a great influence on the geometry, dimensional accuracy, surface quality and machining cost of the machined surface. With the advancement of machining technology, industrial products continue to develop towards complex structures and high dimensional accuracy, and the types and quantities of tools are becoming more and more numerous, which brings difficulties to mechanical processing and process designers to reasonably recommend tools. The recommendation of CNC machining tools generally relies on the internal relationship between the historical data of past CNC machining, and the recommended process usually relies on manual experience, which has shortcomings such as insufficient accuracy and poor applicability. In the process of tool recommendation, many raw materials and process feature information such as workpiece material, workpiece structure, processing equipment, machining accuracy, heat treatment, etc. are involved. These data are intertwined and complex, and there is rarely a one-to-one correspondence.

In order to take into account the coupling relationship between parameters, Tian Changle et al. published the paper "Feature-based tool selection and cutting parameter integration optimization under carbon emission constraints" in "Computer Integrated Manufacturing Systems", Volume 26, Issue 8 in August 2020 ", a tool selection recommendation method based on machining features is proposed. On the basis of analyzing the influence of tool wear, this method designs a part machining feature information library, tool information library and case library, and builds a tool selection optimization model based on machining features; The optimal solution of the model is the recommended result of the tool. However, this method mainly analyzes the tool recommendation from the machining characteristics of the part, and does not consider the influence of the tool's geometric parameters on the recommended results.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the above-mentioned defects in the prior art, and propose a method for recommending a numerically controlled machining tool based on a knowledge map, which is used to improve the accuracy of tool recommendation.

To achieve the above object, the technical scheme adopted by the present invention comprises the following steps:

(1) Preprocess the raw data of CNC machining:

Use the boxplot method to detect the abnormal values of tools, parts, materials, and machining processes read from the process parameter database, and replace the detected abnormal values with empirical values to obtain a preprocessed data set;

(2) Construct the ontology model of CNC machining:

According to the characteristics and functions of CNC machining tools, parts, materials, and machining processes, determine the respective hierarchical relationships of CNC machining tools, parts, materials, and machining processes; , properties of materials and machining processes, and finally obtain an ontology model that includes the hierarchical relationships and properties of tools, parts, materials, and machining processes;

(3) Build a knowledge map of CNC machining:

(3a) Using the network ontology language, and according to the ontology model of the tools, parts, materials, and machining processes in step (2), construct the names of the tools, parts, and materials in the preprocessed data set, and the processes and machining processes in the process. The feature is the node, and the relationship between the tool, part, material, and processing process in the preprocessed data set is the node relationship, and the inherent feature attribute of the tool, part, and material is the knowledge graph mode layer of the node attribute;

(3b) Using the network ontology language, the data information of the tools, parts, materials, and processing processes in the preprocessed data set is used as the content, and is filled into the node attributes of the corresponding type of the knowledge map mode layer to obtain the CNC machining knowledge map. data layer;

(4) Constructing a one-way weighted knowledge map of CNC machining:

(4a) According to the node attributes and the relationship between parts and tools in the CNC machining knowledge map, measure the similarity between parts and parts, tools and tools for nodes with the same attributes in the knowledge map, and establish parts and parts and tools. The similarity measurement criterion with the tool, and the corresponding weight set is set according to the importance of each attribute of the part and the tool; the similarity relationship matrix corresponding to each node attribute of the part and the tool is calculated according to the established similarity measurement criterion; Calculate the multivariate similarity relationship matrix between nodes by using the similarity relationship matrix of each node attribute and the corresponding weight;

(4b) The value in the multivariate similarity relationship matrix is regarded as the relationship weight between the nodes of the CNC machining knowledge map, and then the relationship weight between the nodes with the relationship in the CNC machining knowledge map is regarded as a vector. For the relationship between nodes, the size of the vector is directly used as the weight of the relationship between nodes; for the two-way relationship between nodes, the sum of the vectors is used as the weight of the relationship between nodes, and finally the one-way weighted knowledge map of CNC machining for tool recommendation is obtained;

(5) Obtain the recommended results of CNC machining tools:

Use the PPR algorithm and build a CNC machining tool recommendation model according to the CNC machining one-way weighted knowledge graph, then use the part node to be processed as the input of the recommended model, calculate the PR value of the tool node, and then use the tool node with the highest PR value as the recommended tool node. knives.

Compared with the prior art, the present invention has the following advantages:

The invention constructs the knowledge graph of numerical control machining through the ontology model of numerical control machining, and uses the PPR algorithm to calculate the correlation degree between the tool node and the part node to be processed through the unidirectional weighted knowledge graph of numerical control machining constructed by the knowledge graph, and then calculates the correlation degree by using the PPR algorithm. The highest tool node is used as the recommended tool, which fully considers the influence of tool geometric parameters in the tool recommendation process, which effectively improves the accuracy of tool recommendation.

Description of drawings

Fig. 1 is the realization flow chart of the present invention;

Fig. 2 is the general block diagram of the numerical control machining knowledge graph construction in the present invention;

Fig. 3 is a schematic diagram of the hierarchical relationship of cutters in the present invention;

Fig. 4 is a schematic diagram of the hierarchical relationship of parts in the present invention;

Fig. 5 is a schematic diagram of the hierarchical relationship of materials in the present invention;

Fig. 6 is a schematic diagram of the hierarchical relationship of the processing process in the present invention;

Fig. 7 is the numerical control machining knowledge map pattern view in the present invention;

8 is a schematic diagram of a one-way weighted knowledge graph constructed in the present invention;

FIG. 9 is a simulation comparison diagram of the recommended effect of the present invention and the prior art.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

1, the present invention includes the following steps:

Step 1) Preprocess the raw data of CNC machining:

Use the boxplot method to read the abnormal values of tools, parts, materials, and machining processes from the process parameter database for detection, and use the inter _- quartile IQR to represent the lower quartile Q1 and the upper quartile _Q3 . Spacing, values between Q ₃ +1.5×IQR and Q ₁ -1.5×IQR are normal values, and the rest are outliers; outliers are replaced by empirical values determined according to historical processing data and expert experience, and the preprocessed data set;

Step 2) Construct the ontology model of CNC machining:

According to the characteristics and functions of CNC machining tools, parts, materials, and machining processes, determine the respective hierarchical relationships of CNC machining tools, parts, materials, and machining processes; , properties of materials and machining processes, and finally obtain an ontology model that includes the hierarchical relationships and properties of tools, parts, materials, and machining processes;

Step 2a) Classify the CNC machining tools according to the processing technology. The tools can be divided into turning tools, milling tools, hole processing tools, reciprocating processing tools, gear processing tools, thread tools, integral tools, machine-clamped indexable tools, machine tools. Clamp regrindable tools, welding tools, welding machine clip-type tools, and then classify the functions of each tool after the classification of the processing technology to obtain the hierarchical relationship of the CNC machining tools, as shown in Figure 2; at the same time, according to the tool name, including the tool name , tool number, tool angle, tool diameter, tool durability, tool life attribute information, establish the attribute relationship of each tool after functional classification, and obtain the ontology model including the tool hierarchical relationship and attribute relationship;

Step 2b) Classify the parts according to the shape features. The parts can be divided into shaft parts, disc sleeve parts, bracket box parts, hexahedron parts, and fuselage base parts. The parts are classified according to the part composition, and the hierarchical relationship of the parts is obtained, as shown in Figure 3; at the same time, according to the attribute information of the parts including size, machining accuracy and machined surface quality, the attribute relationship of each part after classification according to the part composition is established. , get ontology model including part hierarchy and attribute relationship;

Step 2c) Classify the tool material and the part material according to the type, and classify the manufacturing process of each material after the type classification. The tool material can be divided into coating material, cemented carbide, superhard tool material, ceramics; parts The material can be divided into steel, cast iron, non-ferrous alloy, and finally the hierarchical relationship of the material is obtained, as shown in Figure 4; at the same time, according to the attribute information of the material including strength, hardness, stiffness, plasticity, and impact toughness, the manufacturing process according to the material is established. After classifying the attribute relationship of each material, an ontology model including material hierarchy and attribute relationship is obtained;

Step 2d) Divide the machining process into processes, work steps, machining elements, cutting speed, back-cut amount, and feed amount, and divide the machining elements into inner circle, outer circle, plane, forming surface, groove, thread, Cone surface and tooth shape, the hierarchical structure of the machining process is obtained, as shown in Figure 5; at the same time, the values of cutting speed, back cut amount, and feed amount are used as their respective attributes, and the shape feature parameters of each machining element are used as machining elements. Attributes, get the process hierarchy and attribute ontology model;

Step 3) Build a knowledge map of CNC machining:

The construction of the CNC machining knowledge map is divided into the construction of the model layer and the construction of the data layer, and the process is shown in Figure 6;

Step 3a) Using the network ontology language, and according to the ontology model of the tools, parts, materials, and machining processes in step (2), construct the names of the tools, parts, and materials in the preprocessed data set, and the processes and machining processes in the process. The feature is the node, and the relationship between the tool, part, material, and machining process in the preprocessed data set is the node relationship, and the inherent feature attribute of the tool, part, and material is the knowledge graph mode layer of the node attribute, as shown in Figure 7. Show;

Step 3b) Using the network ontology language, the data information of the tools, parts, materials, and processing processes in the preprocessed data set is used as the content, and is filled into the node attributes of the corresponding type of the knowledge map mode layer to obtain the CNC machining knowledge map. data layer;

Step 4) Constructing a one-way weighted knowledge graph of CNC machining:

Step 4a) Calculate the multivariate similarity relationship matrix between nodes:

Step 4a1) The knowledge graph schema layer contains nodes P={p ₁ ,...,p _d ,...,p _D } and M node attributes S={s ₁ ,...,s _m ,...,s _M }, according to P and S determine the similarity measurement criterion a={a ₁ ,..., _{am ,...,a M }} of the similarity relationship between nodes, and then use the function P _m ( _{pi ,p j )} corresponding to each criterion a _m Determine the similarity relationship between the i-th node p _i and the j-th node p _j with respect to the attribute s _m , and form a similarity measurement function set P={p ₁ ,p ₂ ,p ₃ ,p ₄ ,p ₅ }, where:

Among them, 1 and 0 respectively indicate that there is a similarity relationship between p _i and p _j , and there is no similarity relationship;

Step 4a2) The similarity relationship P _m (pi ,p _j ) is represented by the similarity adjacency matrix A _m (i,j): A _m ( _i ,j)=P _m (p _i ,p _j ), and a corresponds to The similarity adjacency matrix set A={A ₁ ,...,A _m ,...,A _M }, when A _m (i,j) and A _m (j,i) are both 1, it means that the nodes p _i , p _j The similarity relationship of is mutual;

Step 4a3) Determine the weight value w _m of each attribute s _m in the attribute set S={s ₁ ,...,s _m ,...,s _M } according to the mechanism knowledge and the numerical control machining process, and obtain the weight corresponding to S and the sum is 1 The value set W={w ₁ ,...,w _m ,...,w _M }, and the similarity relationship matrix A _p between nodes under the multi-criteria constraint is calculated by the similarity adjacency matrix _{Am and its corresponding weight value w m :}

Step 4a4) The feature attribute S ₁ = {s ₁₁ , s ₁₂ , s ₁₃ , s ₁₄ , s ₁₅ } of the part is represented by the material, feature size, machining accuracy, surface roughness, and shape position accuracy of the part, and similar from the name Similarity in structure, similarity in structure (similarity in shape, size, accuracy), material similarity (similarity in material type, blank form like idea, similarity in heat treatment), and similarity in process (similarity in processing method, use of equipment) similarity, process sequence similarity, fixture similarity, gage similarity) to determine the similarity measure criterion a ₁ ={a ₁₁ ,a ₁₂ ,a ₁₃ ,a ₁₄ ,a ₁₅ } for each attribute respectively, and then according to a ₁ A similarity measurement function set P ₁ ={p ₁₁ ,p ₁₂ ,p ₁₃ ,p ₁₄ ,p ₁₅ } is formed, and a weight value set corresponding to a ₁ is set according to the importance of each attribute W ₁ ={w ₁₁ ,w ₁₂ ,w ₁₃ ,w ₁₄ ,w ₁₅ };

Step 4a5) Represent the tool's diameter by tool material, tool diameter, tool length, rake angle γ _o , clearance angle α _o , entering angle κ _r , edge inclination angle λ _s , secondary deflection angle κ' _r , secondary clearance angle α' _o Feature property S ₂ ={s ₂₁ ,s ₂₂ ,s ₂₃ ,s ₂₄ ,s ₂₅ ,s ₂₆ ,s ₂₇ ,s ₂₈ ,s ₂₉ }, and from tool material similarity (strength, hardness, stiffness, plasticity, impact Toughness) and tool cutting machining mechanism (the influence of tool diameter, tool length, tool angle on CNC machining of parts) to determine the similarity measure criterion a ₂ ={a ₂₁ ,a ₂₂ ,a ₂₃ ,a ₂₄ ,a respectively for each attribute ₂₅ ,a ₂₆ ,a ₂₇ ,a ₂₈ ,a ₂₉ }, and then form a similarity measure function set P ₂ ={p ₂₁ ,p ₂₂ ,p ₂₃ ,p ₂₄ ,p ₂₅ ,p ₂₆ , _{p 27 ,} p ₂₈ , p ₂₉ }, according to the importance of each attribute, set the weight value set W ₂ corresponding to a ₂ ={w ₂₁ ,w ₂₂ ,w ₂₃ ,w ₂₄ ,w ₂₅ ,w ₂₆ ,w ₂₇ ,w ₂₈ , w ₂₉ };

Step 4a6) According to the similarity measurement function set P ₁ of the part and the similarity measurement function set P ₂ of the tool, calculate the similarity relationship matrix corresponding to the node attributes of the part and the tool respectively, and then according to the corresponding weight of each attribute, according to In step (4a3), a multivariate similarity relationship matrix between nodes is calculated, and the value in the multivariate similarity relationship matrix is regarded as the relationship weight between the nodes of the CNC machining knowledge graph.

Step 4b) A schematic diagram of constructing a one-way weighted knowledge graph is shown in Figure 8. The numerical value in the multivariate similarity relationship matrix is regarded as the relationship weight between the nodes of the CNC machining knowledge map, and then the relationship weight between the nodes with the relationship in the CNC machining knowledge map is regarded as a vector. For the one-way relationship between nodes, directly The size of the vector is used as the weight of the relationship between nodes; for the two-way relationship between nodes, the sum of the vectors is used as the weight of the relationship between nodes, and finally a unidirectional weighted knowledge map of CNC machining for tool recommendation is obtained. Feature information, the geometric parameters of the tool are also fully considered in the construction process, including tool diameter, tool length, rake angle γ _o , clearance angle α _o , leading angle κ _r , edge inclination angle λ _s , secondary angle κ' _r , secondary relief angle α'_o;

Step 5) Get the recommended results for CNC machining tools:

Step 5a) According to the one-way weighted knowledge map of CNC machining established in step (4), the weight of the relationship between nodes is used as the probability of jumping between nodes, and a PPR algorithm-based CNC machining tool recommendation model is constructed;

Step 5b) The PPR algorithm has three hyperparameters: the damping coefficient ζ, the convergence accuracy α, and the maximum number of iterations Max. In order to determine the value of the hyperparameters, the processed part nodes are used as input, the PR value of each tool node is calculated, and the tool node with the highest PR value is used as the recommended tool, and the recommended result is compared with the actual tool and the accuracy is calculated. The function of the damping coefficient ζ is to ensure the stability of the iteration of the PPR algorithm and avoid the interruption or divergence of the results. Its empirical value is 0.85; set ζ = 0.85, first do not set the upper limit of the maximum number of iterations Max, when the convergence accuracy α reaches 0.000001 The recommended result of the tool node and the recommended accuracy of the tool will not change; then set the convergence accuracy α = 0.000001, when the maximum number of iterations Max reaches 100, the recommended result of the tool node and the recommended accuracy of the tool will not change; therefore, initialize the PPR algorithm ζ=0.85, α=0.000001, Max=100;

Step 5c) Use the part node to be processed as the input of the CNC machining tool recommendation model, and calculate the PR value of each tool node; after 100 iterations, the absolute difference of the PR value calculated by the two iterations of each tool node is less than 0.000001. The tool node with the highest PR value is the recommended tool.

The traditional machine learning algorithm model is characterized by part material, size, machining accuracy, etc. After reaching a certain threshold, the accuracy is difficult to improve again, even if the amount of data is increased, there will be no greater improvement. The construction of the tool recommendation model of the present invention is based on the knowledge map, and increasing the amount of data can effectively enrich the data information, which is beneficial to the improvement of the recommendation accuracy.

The technical effects of the present invention are described below in conjunction with simulation experiments:

1. Simulation conditions and content:

The simulation uses AMD Ryzen 7 5800H with Radeon Graphics 3.20GHz processor, the compiler uses PyCharm (Community Edition) version: 11.0.11+9-b1504.13 amd64, the compilation language uses Python 3.8.5, and the compilation environment is anaconda version:conda 4.10.1;

The recommended accuracy of the present invention and the existing tool selection recommendation method based on machining characteristics is compared and simulated, and the results are shown in Figure 9.

2. Analysis of simulation results:

Referring to Figure 9, the ordinate in the figure is the recommended precision of the tool. It can be seen from the figure that the recommended precision of the present invention is 93.23%, and the recommended precision of the prior art is 85.76%, and the difference between the two is 7.47%. Compared with the prior art, the recommendation accuracy is effectively improved.

To sum up, the present invention takes the construction and application of a knowledge map in the field of CNC machining as a research goal, and proposes a method for recommending CNC machining tools based on the knowledge map, which includes preprocessing the original data of CNC machining, constructing an ontology model of CNC machining, and constructing The knowledge map of CNC machining, the construction of one-way weighted knowledge map of CNC machining, and the recommendation results of CNC machining tools. The design of the whole scheme is rigorous and complete, and the recommended parameters are highly accurate, so as to realize the accurate recommendation of CNC machining tools.

Claims (4)

1. a kind of numerical control machining tool recommendation method based on knowledge map, is characterized in that, comprises the steps: (1) Preprocess the raw data of CNC machining: Use the boxplot method to detect the abnormal values of tools, parts, materials, and machining processes read from the process parameter database, and replace the detected abnormal values with empirical values to obtain a preprocessed data set; (2) Construct the ontology model of CNC machining: According to the characteristics and functions of CNC machining tools, parts, materials, and machining processes, determine the respective hierarchical relationships of CNC machining tools, parts, materials, and machining processes; , properties of materials and machining processes, and finally obtain an ontology model that includes the hierarchical relationships and properties of tools, parts, materials, and machining processes; (3) Build a knowledge map of CNC machining: (3a) Using the network ontology language, and according to the ontology model of the tools, parts, materials, and machining processes in step (2), construct the names of the tools, parts, and materials in the preprocessed data set, and the processes and machining processes in the process. The feature is the node, and the relationship between the tool, part, material, and processing process in the preprocessed data set is the node relationship, and the inherent feature attribute of the tool, part, and material is the knowledge graph mode layer of the node attribute; (3b) Using the network ontology language, the data information of the tools, parts, materials, and processing processes in the preprocessed data set is used as the content, and is filled into the node attributes of the corresponding type of the knowledge map mode layer to obtain the CNC machining knowledge map. data layer; (4) Constructing a one-way weighted knowledge map of CNC machining: (4a) According to the node attributes and the relationship between parts and tools in the CNC machining knowledge map, measure the similarity between parts and parts, tools and tools for nodes with the same attributes in the knowledge map, and establish parts and parts and tools. The similarity measurement criterion with the tool, and the corresponding weight set is set according to the importance of each attribute of the part and the tool; the similarity relationship matrix corresponding to each node attribute of the part and the tool is calculated according to the established similarity measurement criterion; Calculate the multivariate similarity relationship matrix between nodes by using the similarity relationship matrix of each node attribute and the corresponding weight; (4b) The value in the multivariate similarity relationship matrix is regarded as the relationship weight between the nodes of the CNC machining knowledge map, and then the relationship weight between the nodes with the relationship in the CNC machining knowledge map is regarded as a vector. For the relationship between nodes, the size of the vector is directly used as the weight of the relationship between nodes; for the two-way relationship between nodes, the sum of the vectors is used as the weight of the relationship between nodes, and finally the one-way weighted knowledge map of CNC machining for tool recommendation is obtained; (5) Obtain the recommended results of CNC machining tools: Use the PPR algorithm and build a CNC machining tool recommendation model according to the CNC machining one-way weighted knowledge graph, then use the part node to be processed as the input of the recommended model, calculate the PR value of the tool node, and then use the tool node with the highest PR value as the recommended tool node. knives. 2. The method for recommending a CNC machining tool based on a knowledge map according to claim 1, the construction of the ontology model of the tool, part, material, and machining process described in step (2), the implementation steps are: (2a) Classify the CNC machining tools according to the processing technology, and then classify the functions of each tool after the processing technology classification to obtain the hierarchical relationship of the CNC machining tools; at the same time, according to the tool name, tool number, tool angle, tool Attribute information of diameter, tool durability and tool life, establish the attribute relationship of each tool after functional classification, and obtain the ontology model including the tool hierarchy relationship and attribute relationship; (2b) Classify the parts according to the shape features, and then classify each type of parts after the shape features are classified according to the components of the parts to obtain the hierarchical relationship of the parts; Establish the attribute relationship of each part after classification according to the part composition, and obtain the ontology model including the part hierarchy and attribute relationship; (2c) Classify the tool material and the part material according to the type, and classify the manufacturing process of each material after the type classification, so as to obtain the hierarchical relationship of the material; The attribute information is established, the attribute relationship of each material is established after classification according to the manufacturing process of the material, and the ontology model including the material hierarchy and attribute relationship is obtained; (2d) Divide the machining process into processes, steps, machining elements, cutting speed, back-cut amount, and feed amount, and divide the machining elements into inner circle, outer circle, plane, forming surface, groove, thread, Taper surface and tooth shape to obtain the hierarchical structure of the machining process; at the same time, the values of cutting speed, back-cut amount, and feed amount are used as their respective attributes, and the shape feature parameters of each machining element are used as the attributes of the machining elements, and the machining process is included. Process hierarchy and attribute ontology model. 3. The method for recommending CNC machining tools based on knowledge map according to claim 1, the multi-element similarity relationship matrix between the calculation nodes described in the step (4a), the implementation step is: (4a1) The knowledge graph schema layer contains D nodes. P={p ₁ ,...,p _d ,...,p _D }. and M node attributes S={s ₁ ,...,s _m ,...,s _{M }} , and determine the _similarity measurement _criterion a _{= {} a ₁ , . , p _j ) determine the similarity relationship between the i-th node p _i and the j-th node p _j with respect to the attribute s _m , and form a similarity measurement function set P={p ₁ , p ₂ , p ₃ , p ₄ , p ₅ },in:

Among them, 1 and 0 respectively indicate that there is a similarity relationship between p _i and p _j , and there is no similarity relationship; (4a2) The similarity relationship P _m (pi, p _j ) is represented by the similarity adjacency matrix A _m (i, _j ): A _m (i, _j )=P _m (pi, p _j ), and a corresponds to The similarity adjacency matrix set A={A ₁ ,...,A _m ,...,A _M }, when A _m (i,j) and _Am (j, i) are both 1, it means that the nodes p _i , p _j The similarity relationship of is mutual; (4a3) Determine the weight value w _{m of each attribute s m in the} attribute set _S ={s ₁ , . The value set W={w ₁ ,...,w _m ,...,w _M }, and the similarity relationship matrix A _p between nodes under the multi-criteria constraint is calculated by the similarity adjacency matrix _{Am and its corresponding weight value w m :}

(4a4) The feature attribute S ₁ = {s ₁₁ , s ₁₂ , s ₁₃ , s ₁₄ , s ₁₅ } of the part is represented by the material, feature size, machining accuracy, surface roughness, and shape position accuracy of the part, and from the name feature , structural characteristics, material characteristics and manufacturing process characteristics, respectively determine the similarity measurement criterion a ₁ ={a ₁₁ , a ₁₂ , a ₁₃ , a ₁₄ , a ₁₅ } of each attribute, and then form a similarity measurement function set according to a ₁ P ₁ ={p ₁₁ , p ₁₂ , p ₁₃ , p ₁₄ , p ₁₅ }, set the weight value set corresponding to a ₁ according to the importance of each attribute W ₁ ={w ₁₁ , w ₁₂ , w ₁₃ , w ₁₄ , w ₁₅ }; (4a5) By tool material, tool diameter, tool length, rake angle γ _o , clearance angle α _o , main deflection angle κ _r , edge inclination angle λ _s , secondary deflection angle κ′ _r , secondary clearance angle α′ _o indicate the The characteristic property S ₂ = {s ₂₁ , s ₂₂ , s ₂₃ , s ₂₄ , s ₂₅ , s ₂₆ , s ₂₇ , s ₂₈ , s ₂₉ }, and from the strength, hardness, stiffness, plasticity, impact toughness and The tool cutting machining mechanism determines the similarity measurement criterion a ₂ ={a ₂₁ , a ₂₂ , a ₂₃ , a ₂₄ , a ₂₅ , a ₂₆ , a ₂₇ , a ₂₈ , a ₂₉ } for each attribute respectively, and then according to a ₂ A similarity measurement function set P ₂ ={p ₂₁ , p ₂₂ , p ₂₃ , p ₂₄ , p ₂₅ , p ₂₆ , p ₂₇ , p ₂₈ , p ₂₉ } is formed, and the corresponding value of a ₂ is set according to the importance of each attribute Weight value set W ₂ ={w ₂₁ , w ₂₂ , w ₂₃ , w ₂₄ , w ₂₅ , w ₂₆ , w ₂₇ , w ₂₈ , w ₂₉ }; (4a6) According to the similarity measurement function set P ₁ of the part and the tool similarity measurement function set P ₂ , calculate the similarity relationship matrix corresponding to the node attributes of the part and the tool respectively, and then according to the weight corresponding to each attribute, according to In step (4a3), a multivariate similarity relationship matrix between nodes is calculated. 4. The method for recommending CNC machining tools based on a knowledge map according to claim 1, wherein the PPR algorithm described in step (5) is used, and a CNC machining tool recommendation model is built according to the one-way weighted knowledge map of CNC machining , to obtain the recommended results of CNC machining tools. The implementation steps are: (5a) According to the one-way weighted relationship knowledge graph established in step (4), the weight of the relationship between nodes is used as the probability of jumping between nodes, and a PPR algorithm-based CNC machining tool recommendation model is constructed; (5b) The convergence accuracy of the initialization PPR algorithm is α, and the maximum number of iterations is Max; (5c) Use the node of the part to be processed as the input of the recommended model of the CNC machining tool, and calculate the PR value of each tool node; after Max iterations, the absolute difference of the PR value calculated by the two iterations of each tool node is less than α, and the The tool node with the highest PR value is the recommended tool.

Images