Keyword template matching method and device based on decision tree and computer equipment
By constructing a keyword template matching method based on decision trees, the manufacturing outline is automatically compiled, solving the problem of low efficiency in manual compilation, realizing efficient and accurate process template selection, and improving the efficiency and accuracy of manufacturing outline compilation.
Patent Information
- Application Number
- CN202510984726.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-17
- Publication Date
- 2025-11-21
AI Technical Summary
In the existing manufacturing process, the manufacturing outline preparation stage still generally adopts a manual preparation mode, which leads to low efficiency, easy errors and difficulty in responding quickly to product design changes, becoming a bottleneck restricting the improvement of production efficiency.
By summarizing historical manufacturing outline templates, constructing a vocabulary library of key indicator description languages for numerical models, extracting key technical indicators for processes, and using decision tree algorithms to construct decision trees for key indicators of process knowledge, the system can automatically match process templates.
It improved the efficiency of process engineers in compiling manufacturing outlines, with the accuracy rate of process template selection reaching over 95%, and the efficiency of writing Forms of Order (FO) increased by 50%.
Smart Images
Figure CN120996002A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of information technology, specifically relating to a keyword template matching method based on decision trees. Background Technology
[0002] In the field of modern machining and manufacturing, after long-term technological accumulation and innovative iteration, the current machining process system has become quite mature. From traditional machining methods such as turning, milling, and grinding, to advanced technologies such as CNC machining, EDM, and laser processing, the industry has established standardized process specifications and parameter systems to precisely match various machining needs and achieve a scientific and efficient optimal process configuration for each machining requirement, addressing different material properties, part precision requirements, and structural complexity. However, the manufacturing outline (FO) preparation stage, which occupies a core position in the manufacturing process, still predominantly relies on the traditional manual method. Technicians must spend a significant amount of time consulting documents, analyzing part drawings, planning process flows, and meticulously recording each step, equipment selection, and time estimation. This manual method is not only inefficient and prone to human error, but also has a long preparation cycle, making it difficult to quickly respond to product design changes or production plan adjustments. With the accelerating pace of product updates, the limitations of manually preparing manufacturing outlines (FOs) have become increasingly apparent, becoming a bottleneck restricting machining manufacturing enterprises from improving production efficiency.
[0003] CN114583812A discloses a method for supplying DC power to a welding machine by charging a supercapacitor with a fuel cell and switching between two sets of supercapacitors to provide alternating power. However, this method requires the welding machine to be a DC welding machine, or an additional inverter system to convert the DC power to AC power. This makes the control of the welding machine more complex and increases costs. Summary of the Invention
[0004] This application addresses the aforementioned deficiencies and problems in the prior art by providing a keyword template matching method, apparatus, and computer device based on a decision tree.
[0005] To achieve the above effects, the technical solution of this application is as follows: A keyword template matching method based on decision trees includes the following steps: Step 1. Summarize all the FO templates; Step 2. Construct a vocabulary library for describing key indicators in mathematical modeling; Step 3. Extract the key technical indicators for each process in template FO; Step 4. Construct a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
[0006] Furthermore, the specific plan for step 1 is as follows: Summarize and categorize FO template types from historical FOs, extract existing FOs, and obtain the queue N of all process types in all templates.
[0007] Where m is the number of process types. This represents the process type name and its position in queue N.
[0008] Furthermore, set up a template queue M, which is initially empty, and its elements are as follows:
[0009] Where count is the number of times the template appears (initially 0), and No is the template number.
[0010] Furthermore, the process type configuration information in each FO is extracted and populated into a vector set.
[0011]
[0012] in The process configuration for the k-th FO. For process types In this FO, No is initially empty, where The value can be 0 or 1. If the process exists in the FO, The value is 1 if it does not exist. The value is 0.
[0013] Furthermore, according to Formula 1, calculate any FO ( Euclidean distance from all FO(M) in the template.
[0014] Formula 1 in Let be the Euclidean distance between the k-th template FO and the g-th template FO; if Then the template used by the k-th FO is the template of the g-th FO. The quantity increases by one book, that is:
[0015] if If the Euclidean distance d between the FO and all templates in template M is not 0, then the FO is a new template, and it is added to the template queue M with a quantity of 1, and its number is the maximum NO value in the existing templates plus one, that is: .
[0016] Furthermore, the specific implementation plan for step 2 is as follows: Establish a keyword dictionary and thesaurus to represent different descriptions of the same technical parameter in different environments. The data dictionary should be formatted as follows:
[0017] Where id is the unique identifier of the term, name is the standardized expression of the technical parameter, and key is a concise keyword in natural language used by different designers when describing the standardized term. This keyword is given by process engineers based on experience. Create a dictionary database named dicts for all keywords, where the same term has multiple records.
[0018] Furthermore, step 3 specifically includes: Extract keywords from each template in template queue X, and extract the actual values of keywords from each FO, in the following format:
[0019] The template number is the template's number in queue X, and its value is the index position in the template queue. The name is the term, the value is the parameter value corresponding to the term, and n is the number of keywords appearing in the template. Different templates have different numbers of keywords.
[0020] Furthermore, step 4 specifically includes: Step 4.1. Keyword information entropy filtering; Step 4.2. Construct the process model decision tree T based on the ID3 decision tree algorithm.
[0021] Furthermore, step 4.1 specifically involves: The empirical entropy of the template is calculated using the empirical entropy calculation formula (Formula 5). ,Right now:
[0022] Where K is the number of templates. For vectors The total number of all Fos in the system. template The quantity.
[0023] Furthermore, based on the empirical conditional entropy of the keyword for the template queue... The calculation formula is shown in Formula 6:
[0024] Where n represents the possible values of keyword A in the template. The count value in template i The empirical entropy of keyword A, The value is the sum of the counts corresponding to the templates when keyword A takes different values. This represents the total number of templates associated with keyword A when it takes different values.
[0025] Furthermore, the information entropy gain of each keyword is calculated. The calculation formula is shown in Formula 7:
[0026] Calculate the information entropy of all keywords using formulas 2, 3, 4, and 5.
[0027] Furthermore, step 4.2 specifically involves: using keywords as the feature set, setting the threshold to 0.01, and considering that all FOs using the template number contain the keyword, and constructing a decision tree based on the ID3 decision tree construction algorithm.
[0028] A keyword template matching device based on decision trees, comprising: Summary module: Summarizes all FO templates; Vocabulary building module: Constructs a vocabulary library for describing key indicators in mathematical modeling; Extraction module: Extracts key technical indicators for each process in template FO; Decision tree construction module: Constructs a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
[0029] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of a decision tree-based keyword template matching method.
[0030] The beneficial effects of this application are as follows: 1. This method summarizes and generalizes historical File of Order (FO) instances to obtain templates for all FOs, and then constructs a process template decision tree based on the keyword distribution in the FO templates. Using this process template decision tree, when creating a new FO, a matching template can be quickly provided based on the designer's processing requirements for the part, thereby improving the efficiency of process engineers in creating FOs.
[0031] 2. This method, based on the process keyword decision tree constructed by the previous method, extracts natural language descriptions of technical indicators from the numerical model, and then matches them with key technical indicators in the keyword thesaurus to form a relatively standardized description of process technical indicators. After decision-making through the process decision tree, the accuracy rate of process template selection reaches over 95%. The work efficiency of process engineers in writing FOs is improved by 50%. Attached Figure Description
[0032] Figure 1 This is a flowchart of a keyword template matching method based on decision trees.
[0033] Figure 2 This is a schematic diagram of the process decision tree in the embodiment. Detailed Implementation
[0034] Example 1 like Figure 1 As shown, a keyword template matching method based on decision trees includes the following steps: Step 1. Summarize all the FO templates; Step 2. Construct a vocabulary library for describing key indicators in mathematical modeling; Step 3. Extract the key technical indicators for each process in template FO; Step 4. Construct a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
[0035] This method summarizes and generalizes historical File Entities (FO) instances to obtain templates for all FOs. Then, based on the keyword distribution within these FO templates, it constructs a process template and builds a decision tree. Using this process template decision tree, when creating a new FO, a matching template can be quickly provided based on the designer's processing requirements for the part, thereby improving the efficiency of process engineers in creating FOs.
[0036] Example 2 like Figure 1 As shown, a keyword template matching method based on decision trees includes the following steps: Step 1. Summarize all the FO templates; Step 2. Construct a vocabulary library for describing key indicators in mathematical modeling; Step 3. Extract the key technical indicators for each process in template FO; Step 4. Construct a decision tree for key process knowledge indicators based on the decision tree construction algorithm; The specific plan for step 1 is as follows: Summarize and categorize FO template types from historical FOs, extract existing FOs, and obtain the queue N of all process types in all templates.
[0037] Where m is the number of process types. This represents the process type name and its position in queue N.
[0038] Set up a template queue M, which is initially empty. The elements of M are as follows:
[0039] Where count is the number of times the template appears (initially 0), and No is the template number.
[0040] Extract the process type configuration information from each FO and populate it into a vector set.
[0041]
[0042] in The process configuration for the k-th FO. For process types In this FO, No is initially empty, where The value can be 0 or 1. If the process exists in the FO, The value is 1 if it does not exist. The value is 0.
[0043] Calculate any FO (according to Formula 1) Euclidean distance from all FO(M) in the template.
[0044] Formula 1 in Let be the Euclidean distance between the k-th template FO and the g-th template FO; if Then the template used by the k-th FO is the template of the g-th FO. The quantity increases by one book, that is:
[0045] if If the Euclidean distance d between the FO and all templates in template M is not 0, then the FO is a new template, and it is added to the template queue M with a quantity of 1, and its number is the maximum NO value in the existing templates plus one, that is: .
[0046] The specific implementation plan for step 2 is as follows: Establish a keyword dictionary and thesaurus to represent different descriptions of the same technical parameter in different environments. The data dictionary should be formatted as follows:
[0047] Where id is the unique identifier of the term, name is the standardized expression of the technical parameter, and key is a concise keyword in natural language used by different designers when describing the standardized term. This keyword is given by process engineers based on experience. Create a dictionary database named dicts for all keywords, where the same term has multiple records.
[0048] Step 3 specifically involves: Extract keywords from each template in template queue X, and extract the actual values of keywords from each FO, in the following format:
[0049] The template number is the template's number in queue X, and its value is the index position in the template queue. The name is the term, the value is the parameter value corresponding to the term, and n is the number of keywords appearing in the template. Different templates have different numbers of keywords.
[0050] Step 4 specifically includes: Step 4.1. Keyword information entropy filtering; Step 4.2. Construct the process model decision tree T based on the ID3 decision tree algorithm.
[0051] Step 4.1 specifically involves: The empirical entropy of the template is calculated using the empirical entropy calculation formula (Formula 5). ,Right now:
[0052] Where K is the number of templates. For vectors The total number of all Fos in the system. template The quantity.
[0053] Empirical conditional entropy of template queues based on keywords The calculation formula is shown in Formula 6:
[0054] Where n represents the possible values of keyword A in the template. The count value in template i The empirical entropy of keyword A, The value is the sum of the counts corresponding to the templates when keyword A takes different values. This represents the total number of templates associated with keyword A when it takes different values.
[0055] Calculate the information entropy gain of each keyword. The calculation formula is shown in Formula 7:
[0056] Calculate the information entropy of all keywords using formulas 2, 3, 4, and 5.
[0057] Step 4.2 specifically involves: using keywords as the feature set and a threshold of 0.01. Keywords contained in the template are considered to be included in all FOs using the template number. A decision tree is constructed based on the ID3 decision tree construction algorithm.
[0058] Example 3 Based on Embodiments 1 and 2, a keyword template matching device based on a decision tree includes: Summary module: Summarizes all FO templates; Vocabulary building module: Constructs a vocabulary library for describing key indicators in mathematical modeling; Extraction module: Extracts key technical indicators for each process in template FO; Decision tree construction module: Constructs a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
[0059] Example 4 Based on Embodiments 1 and 2, a computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the decision tree-based keyword template matching method as in Embodiment 1 or Embodiment 2.
[0060] Example 5 Based on Example 1 or Example 2, this example assumes that there are 5 different types of processes in the historical FO (ultrasonic inspection, penetrant testing and surface treatment).
[0061] In this embodiment, there are a total of 10 FOs. As described in step 1, 1 indicates that the process exists and 0 indicates that the process does not exist. The process configuration is as follows:
[0062] Based on step 1, the template M is calculated using Euclidean distance as follows:
[0063] Constructing a vocabulary of key indicator description languages for mathematical modeling The following knowledge base was constructed based on process knowledge. The keywords in the knowledge base come from digital model design documents and process content.
[0064]
[0065] Extract key technical indicators for each process in template FO The keywords and values of the template are extracted by matching keywords from the word library.
[0066]
[0067] Constructing a decision tree of key indicators of process knowledge Based on the keywords and values in each template shown in step 3, the empirical entropy H(D) for this example is calculated.
[0068] Calculate the information entropy gain of the keywords based on step 3:
[0069]
[0070] =0.878 in This is a sample subset of different templates under the configuration of ultrasonic waves in the process. This is a sample subset of different templates under the condition of no ultrasonic wave configuration in the process.
[0071] Similarly, the following can be calculated:
[0072]
[0073] Since the template ultrasonic and surface treatment information entropy gains are the same and the value is greater than 0.01, either one can be chosen as the root node to build a decision tree.
[0074] In this study, the ultrasonic node was chosen as the root node. Since the node without ultrasonic nodes only contains template 2, the child node without ultrasonic nodes is template 2.
[0075] Excluding template 2, the decision tree for ultrasonic nodes will continue to be calculated iteratively using the ID3 algorithm, as shown in the attached figure. Figure 2 As shown.
[0076] The decision tree is constructed based on the decision tree algorithm as shown in the attached figure. Figure 2 As shown.
[0077] According to the appendix Figure 2 After extracting the digital model file, the corresponding template can be matched according to the keywords in the digital model and the process decision tree.
Claims
1. A keyword template matching method based on decision trees, characterized in that, Includes the following steps: Step 1. Summarize all the FO templates; Step 2. Construct a vocabulary library for describing key indicators in mathematical modeling; Step 3. Extract the key technical indicators for each process in template FO; Step 4. Construct a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
2. The keyword template matching method based on decision trees according to claim 1, characterized in that, The specific plan for step 1 is as follows: Summarize and categorize FO template types from historical FOs, extract existing FOs, and obtain the queue N of all process types in all templates. Where m is the number of process types. This represents the process type name and its position in queue N.
3. The keyword template matching method based on decision trees according to claim 2, characterized in that, Set up a template queue M, which is initially empty. The elements of M are as follows: Where count is the number of times the template appears (initially 0), and No is the template number.
4. The keyword template matching method based on decision trees according to claim 3, characterized in that, Extract the process type configuration information from each FO and populate it into a vector set. in The process configuration for the k-th FO. For process types In this FO, No is initially empty, where The value can be 0 or 1. If the process exists in the FO, The value is 1 if it does not exist. The value is 0.
5. The keyword template matching method based on decision trees according to claim 4, characterized in that, Calculate any FO (according to Formula 1) Euclidean distance from all FO(M) in the template: Formula 1 in Let be the Euclidean distance between the k-th template FO and the g-th template FO; if Then the template used by the k-th FO is the template of the g-th FO. The quantity increases by one book, that is: if If the Euclidean distance d between the FO and all templates in template M is not 0, then the FO is a new template, and it is added to the template queue M with a quantity of 1, and its number is the maximum NO value in the existing templates plus one, that is: 。 6. The keyword template matching method based on decision trees according to claim 1, characterized in that, The specific implementation plan for step 2 is as follows: Establish a keyword dictionary and thesaurus to represent different descriptions of the same technical parameter in different environments. The data dictionary should be formatted as follows: Where id is the unique identifier of the term, name is the standardized expression of the technical parameter, and key is a concise keyword in natural language used by different designers when describing the standardized term. This keyword is given by process engineers based on experience. Create a dictionary database named dicts for all keywords, where the same term has multiple records.
7. The keyword template matching method based on decision trees according to claim 6, characterized in that, Step 3 specifically involves: Extract keywords from each template in template queue X, and extract the actual values of keywords from each FO, in the following format: The template number is the template's number in queue X, and its value is the index position in the template queue. The name is the term, the value is the parameter value corresponding to the term, and n is the number of keywords that appear in the template.
8. The keyword template matching method based on decision trees according to claim 7, characterized in that, Step 4 specifically includes: Step 4.
1. Keyword information entropy filtering; Step 4.
2. Construct the process model decision tree T based on the ID3 decision tree algorithm.
9. The keyword template matching method based on decision tree according to claim 8, characterized in that, Step 4.1 specifically involves: Calculate the empirical entropy of the template using the empirical entropy calculation formula. ,Right now: Where K is the number of templates. For vectors The total number of all Fos in the middle. template The quantity.
10. The keyword template matching method based on decision trees according to claim 9, characterized in that, Empirical conditional entropy of template queues based on keywords : Where n represents the possible values of keyword A in the template. The count value in template i The empirical entropy of keyword A, The value is the sum of the counts corresponding to the templates when keyword A takes different values. This represents the total number of templates associated with keyword A when the keyword takes different values.
11. The keyword template matching method based on decision trees according to claim 10, characterized in that, Calculate the information entropy gain of each keyword. : Calculate the information entropy of all keywords using formulas 2, 3, 4, and 5.
12. The keyword template matching method based on decision trees according to claim 11, characterized in that, Step 4.2 specifically involves: using keywords as the feature set and a threshold of 0.01; considering all FOs using the template number as containing the keyword; and constructing a decision tree based on the ID3 decision tree construction algorithm.
13. A keyword template matching device based on decision trees, characterized in that, include: Summary module: Summarizes all FO templates; Vocabulary building module: Constructs a vocabulary library for describing key indicators in mathematical modeling; Extraction module: Extracts key technical indicators for each process in template FO; Decision tree construction module: Constructs a decision tree for key indicators of process knowledge based on the decision tree construction algorithm.
14. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the keyword template matching method based on decision tree as described in any one of claims 1-12.