Demand feature extraction method and system based on demand meta-model incidence relation

Through data mining and processing based on the requirements metamodel association relationship, the requirement characteristics are extracted and updated, the problem of insufficient demand coverage caused by complex equipment design changes is solved, and the accuracy of model recommendation is improved.

CN119025882APending Publication Date: 2024-11-26SOUTHWEST JIAOTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310600568.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-25
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The design changes of complex equipment lead to the inability to effectively cover the module requirements, resulting in the inability to learn design changes of the knowledge graph and the reduction in recommendation accuracy.

Method used

By obtaining the data of design changes and demand item sequences based on the requirements metamodel association relationship, extracting data information in the standard data format, filtering product family modules and candidate requirements, eliminating error terms, building a binomial set matrix, and calculating support degree and Confidence, determine the binomial set above the threshold as the requirement feature, and update the requirement metamodel.

Benefits of technology

It realizes accurate extraction of demand characteristics, effectively covers module requirements, and improves the accuracy of model recommendations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025882A_ABST
    Figure CN119025882A_ABST
Patent Text Reader

Abstract

The invention discloses a demand feature extraction method and system based on a demand meta-model association relationship, and the method comprises the steps: obtaining a data source of design change and demand item sequence association mining based on the demand meta-model association relationship, and extracting data information in a standard data format; screening product family modules in the data information and candidate demand items corresponding to the product family modules, and performing data processing on the candidate demand items; removing error items contained in the candidate demand items after data processing to obtain frequent candidate demand items; constructing two-item sets of frequent candidate demand items, and connecting the two-item sets in parallel to generate a frequent two-item set matrix; solving the support degree and the confidence degree of each binomial set in the frequent binomial set matrix; and determining a binomial set of which the support degree and the confidence degree are greater than a threshold value as a demand feature, and updating the demand meta-model by using the demand feature. Accurate extraction of the demand features is realized, so that the demand features of the demand meta-model can effectively cover the demands of the module, and the model recommendation accuracy is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data mining, and in particular to a method and system for extracting demand features based on demand metamodel association relationships. Background Art

[0002] Complex equipment is a typical single-piece small-batch production. The personalized demand for product requirements brings challenges to the availability of metastructure structure and intelligent configuration models. The design change process is a technical decision made by design engineers and technicians based on actual design demand changes or technological development. The decision-making process contains the experience and knowledge of design engineers and technicians. Effectively discovering and extracting this knowledge is of great significance to improving the intelligence of the design process.

[0003] Design changes are mainly reflected in changes in the characteristic parameters of the product design model. There are many factors that cause design changes, including: the demand metamodel is reasonable, the personalized differences in the demand characteristic values ​​are large, and the model cannot infer suitable instances. In this case, there is no need to retrain the model. The demand metamodel is reasonable, the personalized differences in the demand characteristic values ​​are not large, but there are technical updates under the advancement of design technology or process progress. In this case, there is no need to retrain the model.

[0004] With the continuous development of the times, operators and customers will put forward more and more technical requirements for complex equipment, making the demand features of the demand metamodel unable to effectively cover the module requirements. In addition to different requirements for demand feature parameter values, each design project may also propose new demand items (demand features that have not appeared before). Some of these new demand items can be covered by the features in the metamodel, some can be ignored when the technical requirements are not high, and some must be responded to through design changes. At the same time, the existing models of the module may not meet the requirements, and it is necessary to make design changes based on a similar model to generate a new model. If the demand features that lead to design changes are effectively extracted and added to the demand metamodel, the knowledge graph cannot learn the reasons for the design changes, resulting in a decrease in the accuracy of subsequent recommendations. Summary of the invention

[0005] In view of this, the present invention provides a method and system for extracting demand features based on the association relationship of demand metamodel, aiming to solve the above technical problems.

[0006] In order to solve the above technical problems, the technical solution of the present invention is to provide a method for extracting demand features based on the association relationship of demand metamodel, comprising:

[0007] Based on the association relationship of the requirement metamodel, the data source of the association mining between the design changes and the requirement item sequence is obtained, and the data information in the standard data format is extracted;

[0008] Filtering the product family modules in the data information and the corresponding candidate demand items, and performing data processing on the candidate demand items;

[0009] Eliminate the error terms contained in the candidate demand items after data processing to obtain frequent candidate demand items;

[0010] Constructing binomial sets of the frequent candidate demand items, and generating a frequent binomial set matrix by combining the binomial sets;

[0011] Solving the support and confidence of each binomial in the frequent binomial matrix;

[0012] A two-item set whose support and confidence are greater than a threshold is determined as a demand feature, and the demand metamodel is updated using the demand feature.

[0013] Optionally, the screening of the product family modules in the data information and the corresponding candidate demand items and performing data processing on the candidate demand items includes:

[0014] Based on the data of the product group module, data segmentation is performed on the candidate demand items;

[0015] Normalize the candidate demand items after data segmentation;

[0016] The normalized candidate demand items are binary processed into 0 / 1 according to the actual demand item values.

[0017] Optionally, the normalization processing of the candidate demand items after data segmentation includes:

[0018] Determine the candidate attributes of the candidate demand items after data segmentation;

[0019] Data decomposition is performed according to the candidate attributes.

[0020] Optionally, eliminating error items contained in the candidate demand items after data processing to obtain frequent candidate demand items includes:

[0021] When the candidate requirement sub-item first appears, the candidate requirement sub-item without module model change is screened as the first error term, and the candidate requirement sub-item is obtained after the candidate requirement item is processed by data;

[0022] Calculating the support of the candidate requirement sub-items, and selecting the candidate requirement sub-items whose support is less than a minimum support threshold as the second error term;

[0023] The first error term and the second error term are eliminated to generate frequent candidate demand items.

[0024] Optionally, constructing binomial items of the frequent candidate demand items and generating a frequent binomial item set matrix by combining binomial itemsets includes:

[0025] Constructing a binomial set of the frequent candidate demand items according to the frequent candidate demand items and the finite element model corresponding to the product family module;

[0026] The binomial sets constructed by combining different frequent candidate demand items and finite element models corresponding to different product family modules are used to generate a frequent binomial set matrix.

[0027] Optionally, the frequent binomial matrix is ​​specifically:

[0028] Among them, Condreq1-MMSj1 represents the two-item set formed between the first candidate requirement sub-item Condreq and its corresponding finite element model, k is the number of candidate requirement sub-items, and m is the number of finite element models.

[0029] Optionally, before solving the support and confidence of each binomial in the frequent binomial matrix, the method further includes:

[0030] Obtaining the frequently required sub-items of all binomial items in the frequent binomial item set matrix, and querying the earliest appearance time thereof;

[0031] Based on the data information in the standard data format, segment the transaction data after the frequent demand sub-item appears;

[0032] Binary assignment is performed on the transaction data to generate a transaction data set.

[0033] Optionally, solving the support and confidence of each binomial in the frequent binomial matrix includes:

[0034] Based on the transaction data set, the support and confidence of each binomial in the frequent binomial matrix are solved.

[0035] Optionally, determining a binomial set whose support and confidence are greater than a threshold as a demand feature, and updating the demand metamodel using the demand feature, includes:

[0036] Determine a binomial set whose support and confidence are greater than a threshold, and solve the binomial set to obtain a demand sub-item;

[0037] Based on the data mined from the association between the design change and the requirement item sequence and the requirement sub-item, obtaining original candidate requirement items of the requirement sub-item;

[0038] The original candidate requirement items are used as requirement features, and the requirement metamodel is updated using the requirement features.

[0039] Accordingly, the present invention also provides a demand feature extraction system based on demand metamodel association relationship, comprising:

[0040] The data extraction module is used to obtain the data source of the association mining between the design changes and the requirement item sequence based on the association relationship of the requirement metamodel, and extract the data information in the standard data format;

[0041] A data processing module, used to screen the product family modules in the data information and their corresponding candidate demand items, and perform data processing on the candidate demand items;

[0042] A data elimination module is used to eliminate error items contained in candidate demand items after data processing to obtain frequent candidate demand items;

[0043] A binomial construction module, used to construct binomial sets of the frequent candidate demand items, and to generate a frequent binomial matrix by combining the binomial sets;

[0044] A binomial solving module, used for solving the support and confidence of each binomial in the frequent binomial matrix;

[0045] The feature updating module is used to determine a two-item set whose support and confidence are greater than a threshold as a demand feature, and to update the demand meta-model using the demand feature.

[0046] The present invention discloses a method and system for extracting demand features based on demand metamodel association relationships. The method comprises the following steps: obtaining a data source for association mining between design changes and demand item sequences based on demand metamodel association relationships, and extracting data information in a standard data format; screening product family modules and their corresponding candidate demand items in the data information, and performing data processing on the candidate demand items; eliminating error items contained in the candidate demand items after data processing to obtain frequent candidate demand items; constructing binaries of the frequent candidate demand items, and generating a frequent binary matrix by combining the binaries; solving the support and confidence of each binary in the frequent binary matrix; determining binaries with support and confidence greater than a threshold as demand features, and using the demand features to update the demand metamodel.

[0047] The benefit of the present invention lies in that the accurate extraction of demand features enables the demand features of the demand metamodel to effectively cover the requirements of the module. At the same time, the demand features that cause design changes are screened out and the model is updated using the demand features, so that the accuracy of model recommendation is effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 It is a schematic diagram of the steps of a method for extracting demand features based on demand metamodel association relationships provided by one embodiment of the present invention;

[0050] Figure 2 It is a structural diagram of a demand feature extraction system based on demand metamodel association relationship provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to enable those skilled in the art to better understand the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0052] Reference Figure 1 , is a schematic diagram of the steps of a method for extracting demand features based on demand metamodel association relationships provided by an embodiment of the present invention.

[0053] S11. Based on the association relationship of the requirement metamodel, obtain the data source of the association mining between the design changes and the requirement item sequence, and extract the data information in the standard data format.

[0054] The data sources for mining the association between design changes and requirement item sequences include two: (1) candidate requirement item records generated in requirement instance extraction; (2) module instances, process instances, and change logs generated in configuration design and design / process design. The data from the two data sources are aligned based on project codes and component codes, and the extracted data format is as follows:

[0055] α = (module code, product family code, project code, project establishment time, {candidate requirement item}, module model, whether to change, parent model).

[0056] Taking the bogie as an example, the product configuration design and change data format are shown in Table 1.

[0057]

[0058] Table 1-Product configuration design and change data format

[0059] S12. Filter the product family modules in the data information and their corresponding candidate demand items, and perform data processing on the candidate demand items.

[0060] According to the requirements and design records in Table 1, data segmentation is performed based on the product family DTi and module Modelj. 候选需求项 (σ 产品族编码=DTi^零部件编码=Modelj (Data)), where π is attribute projection, σ is conditional screening, ∧ means that two conditions need to be met at the same time, and the query result is stored in CondReqMij.

[0061] Furthermore, the candidate demand item data is normalized, and for each candidate demand item, its data type and data value are analyzed and normalized so that it meets the conditions for binary analysis, including binary value type, nominal value type, and continuous value type.

[0062] Among them, the binary value type has only two values. For this type of candidate attribute, it is decomposed into two sub-items. For example, the wheelset type is divided into powered wheelset and non-powered wheelset, that is, {wheelset type-powered wheelset, wheelset type-non-powered wheelset}.

[0063] The nominal value type has a finite number of values. For this type of candidate attribute, there are two cases: (1) The value type and the number of samples in each category are suitable for analysis: directly use the decomposition method to decompose the candidate demand items into multiple items according to the value. For example: traction motor type: DC traction motor, AC asynchronous traction motor, AC synchronous traction motor, decomposed into 3, namely {traction motor type-DC, traction motor type-AC asynchronous, traction motor type-AC synchronous}. (2) There are many value types and few samples in each category: use concept stratification. Concept stratification is a concept induction and evaluation process. Incorporating concept stratification into association analysis can solve the problem that the bottom-level items do not have sufficient support and therefore do not appear in the frequent item set.

[0064] Continuous value type ③ refers to the requirement item whose value is not finite, such as the inner distance of the wheelset. Continuous value requirement items can be processed by discretization-based methods and statistical methods. For complex equipment requirement items, most of these continuous values ​​are digital types, so the discretization method is used. This method groups continuous attributes according to adjacent values ​​to form a finite number of intervals. The intervals can be divided by equal width, equal frequency, based on entropy or clustering, and the method of dividing intervals by manual analysis can be used. The transformation of continuous values ​​into nominal values ​​involves two subtasks: determining how many classification values ​​n are required, and determining how to map continuous values ​​to these classification values. The result can be represented by an interval set {(x0,x1],(x1,x2],…,(xn-1,xn)}.

[0065] Furthermore, after the candidate requirement items are decomposed, the data is binary processed into 0 / 1 according to the actual project requirement item values. Taking the candidate requirement items in Table 1 as an example, the results are shown in Table 2.

[0066]

[0067]

[0068] Table 2 - Decomposition and binary processing of candidate requirements

[0069] S13. Eliminate error terms contained in the candidate demand items after data processing to obtain frequent candidate demand items.

[0070] Assume that the lowest level demand item set after decomposition transformation is ReqMij, and the transformed data set is RegData, then the error items are eliminated according to Table 2.

[0071] First, check whether the module model has changed when the candidate requirement sub-item appears for the first time. If not, the unchanged candidate requirement sub-item is used as the first error term. The corresponding relational algebraic expression is: nonreq1=π req (σ 项目时间=min(项目时间) ∧ 变更标识=‘否’ (Regdata)), where π is the attribute projection, σ is the conditional screening, and min(project time) is the earliest time when the requirement item appears.

[0072] Furthermore, the support of the candidate requirement sub-items is calculated, and the items whose support is less than the minimum support threshold are taken as the second error term. The corresponding algebraic formula is:

[0073] nonreq2=π req (σ reg=1∧count(*)≥minsup (Regdata)).

[0074] Furthermore, the first error term and the second error term are eliminated to obtain the eliminated frequent candidate demand items candreqs.

[0075] S14. Construct binomial sets of frequent candidate demand items, and generate a frequent binomial set matrix by combining binomial sets.

[0076] After obtaining the frequent candidate demand items, construct a binomial set of frequent candidate demand items. Assuming that there are k candidate demand items candreqs and m MMSj models, we get a binomial set: Condreq-MMSj, where Condreq is a candidate demand sub-item of the candidate demand item candreqs.

[0077] Further, the frequent binomial matrix is ​​generated by combining the binomial sets, and we get

[0078]

[0079] S15. Calculate the support and confidence of each binomial in the frequent binomial matrix.

[0080] After obtaining the frequent binomial matrix, it is necessary to obtain the transaction data set, and then calculate the support and confidence of each binomial in the frequent binomial matrix. Specifically, obtain the earliest appearance time of the demand sub-item. For any item Condreqp-MMSjq, use Condreqp to query the transformed data set RegData, and obtain its earliest appearance time: Time =(min(π 项目时间 (σ Condreqp =1(RegData))).

[0081] Furthermore, based on the data information in the standard data format, using the relationship: Dataofcansp=π 序号,候选需求项,模块模型,父模型 (σ 项目时间≥Timei (RegData)), split out the transaction data after the frequent demand sub-items appear, where the serial numbers are automatically generated in ascending order of time. Taking the "inside distance - (1200,1400]" demand sub-item in Table 2 as an example, the transaction data corresponding to the candidate sub-item is obtained after selecting by time and projecting by column, as shown in Table 3.

[0082]

[0083] Table 3 - Transaction data of requirement sub-item "Inside distance - (1200,1400]"

[0084] Further, take the module model set and convert it into columns, with the column name being the model name. Initialize the column data to 0, and update the column value of the model name in the row corresponding to the TID of the model change data set to 1; take the module model corresponding to the first transaction, scan from the second transaction, and if it is equal to the parent model of the transaction row, update its corresponding value to 1; take the module model corresponding to the second transaction, scan from the third transaction, and if it is equal to the parent model of the transaction row, update its corresponding value to 1, and so on, to obtain the transaction set of candreqsp. Taking Table 3 as an example, the transaction set obtained after conversion is shown in Table 4.

[0085]

[0086] Table 4 - Requirement sub-item "Inside distance - (1200,1400]" transaction set

[0087] It should be noted that there are no more than 2 item sets involved here, so no complex pruning algorithm is required, but only 2 item sets with support and confidence greater than the threshold are selected from the combination of each model in candreqp and MMsp. Specifically, assuming that the support threshold is 0.5 and the confidence threshold is 0.7, taking the first 4 records of the transaction set of the requirement sub-item "Inside distance - (1200, 1400]" in Table 4 as an example, the calculation results shown in Table 5 can be obtained.

[0088]

[0089] Table 5 - Frequent binomial sets of "Inside distance - (1200,1400]"

[0090] S16. Determine a binomial set whose support and confidence are greater than a threshold as a demand feature, and use the demand feature to update the demand metamodel.

[0091] Further, according to the above calculation results, if the associated model set is not empty, it means that candreqp is associated with the model, that is, Condreqp→MMSjq holds, indicating that the change is related to the Condreqp sub-item. Traverse all binomial sets to obtain all related demand sub-items.

[0092] Furthermore, according to the data and requirement sub-items of the association mining between the design change and the requirement item sequence, the original candidate requirement items of the requirement sub-items are obtained, and the original candidate requirement items are used as requirement features, and the requirement meta-model is updated using the requirement features. Taking Table 5 as an example, based on the requirement sub-items, we get: inner distance -(1200,1400]→DCZXJLD-M02, which means that the candidate requirement sub-item "inner distance -(1200,1400]" is related to the change of the module model from DCZXJLD-M01→DCZXJLD-M02, that is, the original candidate requirement item "inner distance" is used as the requirement feature to update the requirement meta-model.

[0093] Furthermore, after the demand metamodel is updated, the demand instances in the knowledge graph can be updated, and the model can be retrained after the update to be updated according to the changes in demand.

[0094] This embodiment provides a method for extracting demand features based on the association relationship of demand metamodels, which obtains the data source of the association mining between design changes and demand item sequences based on the association relationship of demand metamodels, and extracts data information in a standard data format; screens the product family modules and their corresponding candidate demand items in the data information, and performs data processing on the candidate demand items; removes the error items contained in the candidate demand items after data processing to obtain frequent candidate demand items; constructs binomial sets of frequent candidate demand items, and generates a frequent binomial set matrix by combining each binomial set; solves the support and confidence of each binomial set in the frequent binomial set matrix; determines the binomial set with support and confidence greater than a threshold as the demand feature, and uses the demand feature to update the demand metamodel. Accurate extraction of demand features is achieved, so that the demand features of the demand metamodel can effectively cover the requirements of the module. At the same time, the demand features that cause design changes are screened out, and the model is updated using the demand features, so that the accuracy of model recommendation is effectively improved.

[0095] Correspondingly, the present invention also provides a demand feature extraction system based on demand metamodel association relationship, including a data extraction module, a data processing module, a data elimination module, a binomial set construction module, a binomial set solution module and a feature update module.

[0096] Furthermore, a data extraction module is used to obtain the data source for mining the association between design changes and requirement item sequences based on the association relationship of the requirement metamodel, and to extract data information in a standard data format; a data processing module is used to screen the product family modules and their corresponding candidate requirement items in the data information, and to perform data processing on the candidate requirement items; a data elimination module is used to eliminate the error terms contained in the candidate requirement items after data processing, and obtain frequent candidate requirement items; a binomial construction module is used to construct binomials of frequent candidate requirement items, and to generate a frequent binomial matrix by combining each binomial; a binomial solution module is used to solve the support and confidence of each binomial in the frequent binomial matrix; a feature update module is used to determine binomials whose support and confidence are greater than a threshold as requirement features, and to use the requirement features to update the requirement metamodel.

[0097] The above is a method and system for extracting demand features based on the association relationship of demand metamodels provided in an embodiment of the present invention. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the embodiments can be referenced to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principle of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

[0098] Professionals may further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention. The steps of the method or algorithm described in conjunction with the embodiments disclosed herein can be directly implemented by hardware, a software module executed by a processor, or a combination of the two. The software module can be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the technical field. 。

Claims

1. A method for extracting demand features based on the association relationship of demand metamodel, characterized in that: include: Based on the association relationship of the requirement metamodel, the data source of the association mining between the design changes and the requirement item sequence is obtained, and the data information in the standard data format is extracted; Filtering the product family modules in the data information and the corresponding candidate demand items, and performing data processing on the candidate demand items; Eliminate the error terms contained in the candidate demand items after data processing to obtain frequent candidate demand items; Constructing binomial sets of the frequent candidate demand items, and generating a frequent binomial set matrix by combining the binomial sets; Solving the support and confidence of each binomial in the frequent binomial matrix; A two-item set whose support and confidence are greater than a threshold is determined as a demand feature, and the demand metamodel is updated using the demand feature.

2. The method for extracting demand features based on demand metamodel association relationships according to claim 1 is characterized in that: The screening of the product family modules in the data information and the corresponding candidate demand items and performing data processing on the candidate demand items includes: Based on the data of the product group module, data segmentation is performed on the candidate demand items; Normalize the candidate demand items after data segmentation; The normalized candidate demand items are binary processed into 0 / 1 according to the actual demand item values.

3. The method for extracting demand features based on demand metamodel association relationships according to claim 2 is characterized in that: The normalization process of the candidate demand items after data segmentation includes: Determine the candidate attributes of the candidate demand items after data segmentation; Data decomposition is performed according to the candidate attributes.

4. The method for extracting demand features based on demand metamodel association relationships according to claim 1 is characterized in that: The step of eliminating error items contained in the candidate demand items after data processing to obtain frequent candidate demand items includes: When the candidate requirement sub-item first appears, the candidate requirement sub-item without module model change is screened as the first error term, and the candidate requirement sub-item is obtained after the candidate requirement item is processed by data; Calculating the support of the candidate requirement sub-items, and selecting the candidate requirement sub-items whose support is less than a minimum support threshold as the second error term; The first error term and the second error term are eliminated to generate frequent candidate demand items.

5. The method for extracting demand features based on demand metamodel association relationships according to claim 1 is characterized in that: The step of constructing binomial sets of the frequent candidate demand items and generating a frequent binomial set matrix by combining binomial sets includes: Constructing a binomial set of the frequent candidate demand items according to the frequent candidate demand items and the finite element model corresponding to the product family module; The binomial sets constructed by combining different frequent candidate demand items and finite element models corresponding to different product family modules are used to generate a frequent binomial set matrix.

6. The method for extracting demand features based on demand metamodel association relationships according to claim 5 is characterized in that: The frequent binomial matrix is ​​specifically: Among them, Condreq1-MMSj1 represents the two-item set formed between the first candidate requirement sub-item Condreq and its corresponding finite element model, k is the number of candidate requirement sub-items, and m is the number of finite element models.

7. The method for extracting demand features based on demand metamodel association relationships according to claim 5 is characterized in that: Before solving the support and confidence of each binomial in the frequent binomial matrix, the method further includes: Obtaining the frequently required sub-items of all binomial items in the frequent binomial item set matrix, and querying the earliest appearance time thereof; Based on the data information in the standard data format, segment the transaction data after the frequent demand sub-item appears; Binary assignment is performed on the transaction data to generate a transaction data set.

8. The method for extracting demand features based on demand metamodel association relationships according to claim 7 is characterized in that: The step of solving the support and confidence of each binomial in the frequent binomial matrix includes: Based on the transaction data set, the support and confidence of each binomial in the frequent binomial matrix are solved.

9. The method for extracting demand features based on demand metamodel association relationships according to claim 1, characterized in that: The step of determining a binomial set whose support and confidence are greater than a threshold as a demand feature, and updating the demand metamodel using the demand feature, includes: Determine a binomial set whose support and confidence are greater than a threshold, and solve the binomial set to obtain a demand sub-item; Based on the data mined from the association between the design change and the requirement item sequence and the requirement sub-item, obtaining original candidate requirement items of the requirement sub-item; The original candidate requirement items are used as requirement features, and the requirement metamodel is updated using the requirement features.

10. A demand feature extraction system based on demand metamodel association relationship, characterized in that: include: The data extraction module is used to obtain the data source of the association mining between the design changes and the requirement item sequence based on the association relationship of the requirement metamodel, and extract the data information in the standard data format; A data processing module, used to screen the product family modules in the data information and their corresponding candidate demand items, and perform data processing on the candidate demand items; A data elimination module is used to eliminate error items contained in candidate demand items after data processing to obtain frequent candidate demand items; A binomial construction module, used to construct binomial sets of the frequent candidate demand items, and to generate a frequent binomial matrix by combining the binomial sets; A binomial solving module, used for solving the support and confidence of each binomial in the frequent binomial matrix; The feature updating module is used to determine a two-item set whose support and confidence are greater than a threshold as a demand feature, and to update the demand meta-model using the demand feature.