Intelligent manufacturing capability maturity evaluation method and system based on industry characteristics and semantic understanding

Abstract

The invention belongs to the technical field of intelligent manufacturing capability maturity evaluation, and discloses an intelligent manufacturing capability maturity evaluation method and system based on industry characteristics and semantic understanding. The evaluation method comprises the steps of constructing an industry feature knowledge base, generating a personalized evaluation information packet, performing semantic processing on evidence materials, calculating matching degree and similarity, calculating the maturity of the capability subdomains, performing comprehensive rating and the like, and finally integrating maturity scores of all the capability subdomains to form an intelligent manufacturing capability maturity grade result of an evaluated enterprise. Through industry feature matching, semantic vector analysis and dynamic weight calculation, objective, efficient and accurate evaluation of the intelligent manufacturing capability maturity is realized, and the method can be widely applied to manufacturing enterprises and has remarkable economic benefits and popularization values.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing capability maturity assessment technology, specifically to an intelligent manufacturing capability maturity assessment method and system based on industry characteristics and semantic understanding, applicable to the quantitative evaluation of the intelligent level of enterprises in process and discrete manufacturing industries. Background Technology

[0002] As intelligent manufacturing continues to advance, enterprises need to scientifically assess their own level of intelligence. Currently, manual assessment is mainly conducted using GB / T 39116-2020 "Methods for Assessing Capability Maturity of Intelligent Manufacturing". The current intelligent manufacturing capability maturity assessment faces the following core problems: 1. The evaluation standards are fragmented. The use of a unified indicator system by companies in different industries leads to distorted evaluation results (such as the process differences between process manufacturing and discrete manufacturing), and the evaluation indicators are somewhat disconnected from actual business operations.

[0003] 2. Traditional assessment methods rely on static weight allocation, which cannot adapt to the dynamic assessment needs under rapid technological iteration. Existing technologies, such as the Chinese Capability Maturity Model for Intelligent Manufacturing (CMMM), cover 20 process domains across four elements: personnel, resources, and technology, but have not established a dynamic mapping mechanism between industry characteristics and assessment indicators.

[0004] 3. Subjective nature of supporting material analysis: Relying on experts to manually read the text materials, keyword extraction and semantic understanding are easily affected by subjective factors, and the recommended solutions lack objectivity.

[0005] 4. Low evaluation efficiency: The entire evaluation cycle takes 7-10 days, which is difficult to meet the needs of enterprises for rapid evaluation.

[0006] Therefore, it is particularly important to establish a method and system for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding. Summary of the Invention

[0007] The purpose of this invention is to address the aforementioned problems by providing a method and system for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding. This system aims to assess the maturity of intelligent manufacturing capabilities according to industry characteristics, with dynamic weighting and semantic objectivity. It automatically identifies the key characteristics of the specific industry in which the enterprise operates and generates a personalized assessment information package accordingly, thus solving the "one-size-fits-all" problem. The weights automatically evolve with the benchmark pool, addressing the rigidity of "one-time setting, unchanged for many years," and effectively improving assessment efficiency.

[0008] The technical solution of the present invention is as follows: This invention discloses a method and system for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding, comprising the following steps: Construct an industry-specific knowledge base: Based on national industry classification standards, extract key industry features and match recommended sets of capability subdomains, key performance indicators, and typical business processes for each industry. Generate a personalized assessment information package: Receive the industry attribute information of the company being assessed, and generate a personalized assessment information package for the company from the industry feature knowledge base using an industry-capability matching algorithm. The personalized assessment information package includes the capability subdomains to be assessed, key performance indicators, and typical business processes. The personalized assessment information package is generated as follows:

[0009] in, t As keywords, d For document, N For the total number of documents, DF( t (for keywords) t The number of documents; Semantic processing of supporting materials: Obtain supporting materials for the capabilities submitted by the evaluated company, perform natural language processing, extract keywords, and generate semantic vectors; Matching and Similarity Calculation: The semantic vector is matched with the recommendation information set from the industry feature knowledge base to calculate the matching degree; semantic similarity is calculated, outputting the identified key capabilities and recommendation solutions; cosine similarity is calculated to determine the matching degree between the semantic vector and the recommendation information set from the industry feature knowledge base, using the following method:

[0010] By calculating the cosine similarity between the enterprise's industry vector and the industry templates in the knowledge base, the industry template with the highest similarity is selected to generate a personalized evaluation information package.

[0011] BERT semantic vector encoding: After Chinese word segmentation and stop word filtering of the supporting materials, the pre-trained BERT model is used to encode the text and generate a fixed-dimensional semantic vector.

[0012]

[0013] Matching score calculation: The semantic vector of the company's supporting materials is matched with the semantic vector of the recommended capability points in the knowledge base, and the cosine similarity is calculated as the matching score.

[0014]

[0015] Computing capability subdomain maturity: Collect or receive actual operational data of enterprises under national standards, calculate the achievement values ​​of each key performance indicator within the capability subdomain based on the actual operational data, determine the weight of each indicator using a dynamic weight model, and calculate the maturity score of the capability subdomain using weighted average. Dynamic weight determination: Collect historical KPI data from benchmark companies in the same industry to construct a data matrix. X ∈R n × m Principal components were extracted using PCA, and the contribution of each indicator to the overall maturity was calculated.

[0016] The initial weights are fine-tuned using expert experience rules to generate the final dynamic weight vector:

[0017] Overall Rating: The maturity scores of all capability subdomains are combined to form the intelligent manufacturing capability maturity level of the evaluated enterprise. The maturity score is calculated as follows:

[0018] in, Si For the first i Subdomain score, w j For the first j Individual indicator weights, KPIs actual For actual values, KPI target The target value is set. The final evaluation score is then obtained:

[0019] in, λi Subdomain weights; Generate interpretable evaluation reports: automatically output scores and levels for each capability subdomain, improvement suggestions for subdomains below the threshold, recommended solutions, and implementation priorities.

[0020] The above methods, as described in this invention, utilize natural language processing and semantic association technologies to deeply understand supporting materials and business data, automatically extracting key capabilities and reducing manual intervention. Based on semantic networks and industry knowledge bases, PEST and embedding models achieve personalized results for each enterprise, avoiding a "one-size-fits-all" approach and making conclusions more credible. The introduction of BERT+GNN performs semantic-level parsing of supporting materials and process data, automatically identifying "hidden" weaknesses and reducing manual workload. It outputs tangible, implementable improvement plans and priorities, achieving a closed loop of evaluation, diagnosis, and improvement.

[0021] This invention improves evaluation efficiency, shortening the evaluation cycle from 7-10 days to 1-2 days and reducing labor costs by more than 70%. It also makes text analysis more objective, achieving an F1 score of 88.4% for keyword extraction and increasing the Top-5 hit rate of solutions by 60%.

[0022] As a preferred approach, in generating personalized evaluation information packages, a retrieval method based on TF-IDF vectors and cosine similarity is adopted, or a trained industry embedding model is used to output the capability subdomains and indicator sets most relevant to the enterprise's industry.

[0023] As a preferred approach, in the semantic processing of supporting materials, natural language processing includes Chinese word segmentation, stop word filtering, keyword extraction, and pre-trained language model encoding.

[0024] As a preferred approach, the dynamic weighting model is obtained in the maturity of the computing power subdomain through the following method: S51: Collect historical indicator data from multiple benchmark companies in the same industry; S52: Using principal component analysis or neural network autoencoders, the contribution of each indicator to the overall maturity is automatically learned, and initial weights are generated. S53: Fine-tune the initial weights by combining expert experience rules to form the final dynamic weight vector.

[0025] The present invention constructs a dynamic weighting model that adaptively adjusts indicator weights based on industry benchmarks, company size, and development stage, thereby improving evaluation sensitivity. The weights evolve dynamically, benchmark data is updated quarterly, and weights are automatically adjusted, ensuring that grade fluctuations within two evaluation intervals are less than 0.5 levels.

[0026] As a preferred approach, in the comprehensive rating, a weighted average, fuzzy comprehensive evaluation, or machine learning regression model is used to map the scores of multiple subdomains to maturity levels 1-5.

[0027] As a preferred approach, the evaluation criteria for each capability subdomain include the following key indicators: Personnel subdomain: personnel skills training coverage rate, digital skills certification pass rate; Resource subdomain: overall equipment efficiency, resource utilization rate; Technology subdomain: number of data analysis models applied, new technology adoption rate; Manufacturing subdomain: production cycle shortening rate, product qualification rate.

[0028] Preferably, the method also includes generating an interpretable evaluation report, which at least includes: S71: Scores and levels for each capability subdomain; S72: Improvement suggestions for subdomains below the threshold; S73: Recommended solutions and implementation priorities.

[0029] The above methods provide highly interpretable results: the weights and membership curves throughout the entire process are archived, meeting the requirements for audit review.

[0030] This invention also discloses a smart manufacturing capability maturity assessment system based on industry characteristics and semantic understanding, comprising: The industry feature knowledge base construction module, based on the national industry classification standard, uses the PEST method to extract key features in four dimensions: policy, economy, society, and technology. It establishes, stores, and manages the associated data of industry, capability subdomain, indicator, and process to construct the industry feature knowledge base. Information recommendation module: used to output personalized evaluation information packages based on the enterprise's industry attributes; Semantic processing module: used to extract keywords from supporting materials and calculate semantic vectors; The solution matching module is used to retrieve recommended solutions based on semantic vectors; Capability Subdomain Maturity Calculation Module: Used to acquire actual operational data and calculate key performance indicator achievement values, determine the weight of each indicator using dynamic weights, and calculate the maturity score of each capability subdomain using weighted averages. The comprehensive rating and report output module is used to summarize maturity scores and output maturity levels and improvement suggestions, and to display the assessment progress, subdomain scores and historical comparison trends in real time.

[0031] Preferably, the semantic processing module has a built-in pre-trained Chinese language model and supports users to upload domain-specific custom dictionaries.

[0032] As a preferred option, each module is deployed as a microservice on a containerized platform, supporting horizontal scaling and multi-tenant concurrent evaluation.

[0033] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention improves industry adaptability by using PEST and embedded models to achieve "personalized evaluation for each enterprise". Under the same national standard, different industries can automatically map out differentiated evaluation items. The difference in evaluation dimensions between different industries is >40%, avoiding "one-size-fits-all" templates and making the conclusions more credible.

[0034] 2. This invention utilizes deep semantic understanding, introducing BERT+GNN to perform semantic-level parsing of supporting materials and process data, automatically discovering "hidden" shortcomings and reducing manual reading workload by more than 70%.

[0035] 3. This invention adopts dynamic weights, which automatically evolve with the benchmark pool, solving the rigidity problem of "setting once and keeping it unchanged for many years" and making the evaluation results sensitive to the technological evolution of the industry.

[0036] 4. This invention improves the closed loop. The solution generated based on semantic network can directly locate equipment, process, and personnel skill level, and give ROI ranking, shortening the enterprise implementation cycle by 40%.

[0037] 5. This invention adopts a high-concurrency service and a containerized microservice architecture. A single cluster can support 500+ enterprises to conduct online assessments simultaneously, shortening the assessment cycle from the traditional 4-6 weeks to 5-7 working days.

[0038] 6. This invention is standard compatible and fully compatible with GB / T 39116-2020, and can be directly used in scenarios such as standard implementation, benchmark selection, and government subsidy review.

[0039] 7. This invention achieves objective text analysis: the keyword extraction F1 score reaches 88.4%, and the Top-5 hit rate of the solution is improved by 60%.

[0040] 8. The results of this invention are highly interpretable: the weights and membership curves of the entire process are archived, which meets the requirements for audit review. Attached Figure Description

[0041] The present invention will be described by way of example and with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart of an intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding, according to the present invention.

[0042] Figure 2 This is a schematic diagram of the structure of an intelligent manufacturing capability maturity assessment system based on industry characteristics and semantic understanding, according to the present invention.

[0043] Figure 3 This is an overall technical flowchart of an embodiment of the present invention.

[0044] Figure 4 This is a flowchart illustrating the construction of an industry feature knowledge base in an embodiment of the present invention.

[0045] Figure 5 This is a flowchart illustrating the generation process of the personalized assessment information package in an embodiment of the present invention.

[0046] Figure 6 This is a flowchart of semantic processing of supporting materials in an embodiment of the present invention.

[0047] Figure 7 This is a flowchart of the dynamic weight model construction process in an embodiment of the present invention.

[0048] Figure 8 This is a flowchart of the maturity assessment and rating process in an embodiment of the present invention. Detailed Implementation

[0049] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0050] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0051] The features and performance of the present invention will be further described in detail below with reference to embodiments.

[0052] Example 1: like Figure 1 As shown, the present invention provides a method for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding, characterized by comprising the following steps: Construct an industry-specific knowledge base: Based on national industry classification standards, extract key industry features and match recommended sets of capability subdomains, key performance indicators, and typical business processes for each industry. Generate personalized assessment information package: Receive industry attribute information of the company being assessed, and generate a personalized assessment information package for the company being assessed in the industry feature knowledge base through industry-capability matching algorithm. The personalized assessment information package includes the capability subdomain to be assessed, key performance indicators and typical business processes. Semantic processing of supporting materials: Obtain the supporting materials of the capabilities submitted by the evaluated company, perform natural language processing, extract keywords, and generate semantic vectors; natural language processing includes Chinese word segmentation, stop word filtering, keyword extraction, and BERT semantic vector encoding; Matching and similarity calculation: Match the semantic vector with the recommendation information set of the industry feature knowledge base and calculate the matching degree; calculate semantic similarity and output the identified key capabilities and recommendation solutions; Computing capability subdomain maturity: Collect or receive actual operational data of enterprises under national standards, calculate the achievement values ​​of each key performance indicator within the capability subdomain based on the actual operational data, determine the weight of each indicator using a dynamic weight model, and calculate the maturity score of the capability subdomain using weighted average. Overall rating: The rating combines the maturity scores of all capability subdomains to form the smart manufacturing capability maturity level of the evaluated enterprise.

[0053] Example 2: Compared to Example 1, the intelligent manufacturing capability maturity assessment method in this example generates a personalized assessment information package: the personalized assessment information package is generated as follows:

[0054] in, t As keywords, d For document, N For the total number of documents, DF( t(for keywords) t The number of documents; Matching Degree and Similarity Calculation: Cosine similarity is used to calculate the matching degree between semantic vectors and the recommendation information set from the industry feature knowledge base. The method is as follows:

[0055] in , A is a semantic vector. i B i For elements in the vector; By calculating the cosine similarity between the enterprise's industry vector and the industry templates in the knowledge base, the industry template with the highest similarity is selected to generate a personalized evaluation information package.

[0056] BERT semantic vector encoding: After Chinese word segmentation and stop word filtering of the supporting materials, the pre-trained BERT model is used to encode the text and generate a fixed-dimensional semantic vector.

[0057]

[0058] Matching score calculation: The semantic vector of the company's supporting materials is matched with the semantic vector of the recommended capability points in the knowledge base, and the cosine similarity is calculated as the matching score.

[0059]

[0060] Dynamic weight determination: Collect historical KPI data from benchmark companies in the same industry to construct a data matrix. X ∈R n × m Principal components were extracted using PCA, and the contribution of each indicator to the overall maturity was calculated.

[0061] The initial weights are fine-tuned using expert experience rules to generate the final dynamic weight vector:

[0062] Overall Rating: The maturity scores of all capability subdomains are combined to form the intelligent manufacturing capability maturity level of the evaluated enterprise. The maturity score is calculated as follows:

[0063] in, Si For the first i Subdomain score, w j For the first j Individual indicator weights, KPIs actual For actual values, KPItarget The target value is set. The final evaluation score is then obtained:

[0064] in, λi This represents the subdomain weight.

[0065] Example 3: like Figure 2 As shown, this invention also discloses a smart manufacturing capability maturity assessment system based on industry characteristics and semantic understanding, comprising: The industry feature knowledge base construction module extracts key features in four dimensions—policy, economy, society, and technology—based on national industry classification standards and using the PEST method. It establishes, stores, and manages the associated data of industry, capability subdomains, indicators, and processes. Information recommendation module: used to output personalized evaluation information packages based on the enterprise's industry attributes; Semantic processing module: used to extract keywords from supporting materials and calculate semantic vectors; The solution matching module is used to retrieve recommended solutions based on semantic vectors; Capability Subdomain Maturity Calculation Module: Used to acquire actual operational data and calculate key performance indicator achievement values, determine the weight of each indicator using dynamic weights, and calculate the maturity score of each capability subdomain using weighted averages. The comprehensive rating and report output module is used to summarize maturity scores and output maturity levels and improvement suggestions, and to display the assessment progress, subdomain scores and historical comparison trends in real time.

[0066] Example 4: In one embodiment, the evaluation system includes: an industry feature knowledge base module, an information recommendation module, a semantic processing module (including a pre-trained Chinese language model and a user-defined dictionary interface), a solution matching module, a data collection and indicator calculation module, a dynamic weighting and scoring module, and a comprehensive rating and report output module (including a visualization dashboard). Each module is containerized and deployed as a microservice, supporting horizontal scaling and multi-tenant concurrency.

[0067] Example 5: like Figure 3 As shown in this embodiment, a method for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding is disclosed, including: Collect the following data and information: basic industry information (industry / scale / type), supporting materials (documents / reports / process data), enterprise operation data (actual KPI values), and industry benchmark data (≥10 companies in the same industry). Data input is followed by knowledge base construction and matching, semantic processing and solution matching, weight calculation and maturity assessment, and then combined with the comprehensive maturity level (levels 1-5), assessment report (shortcomings + improvement suggestions), and solutions for weaknesses, and finally the results are output.

[0068] Among them, such as Figure 4 As shown, the process of constructing the industry characteristic knowledge base is as follows: (1) Classify the industry according to GB / T 4575; (2) Divide the manufacturing type (process type / discrete type); (3) Analyze the four-dimensional characteristics (policy + economy + society + technology); (4) Establish the relationship (industry - capability subdomain - KPI - business process); (5) Enter industry benchmark data; (6) Update the knowledge base regularly.

[0069] like Figure 5 As shown, the personalized assessment information package generation process is as follows: receive enterprise industry attributes; select matching method; generate industry feature vector; calculate vector matching degree; filter high matching information; output personalized assessment package (subdomain + KPI + business activities).

[0070] like Figure 6 As shown, the semantic processing flow for supporting materials is as follows: receiving supporting materials, standardizing the format, segmenting Chinese words to remove useless words, extracting core keywords, generating semantic vectors, supplementing business data semantic vectors, and forming a complete semantic vector set.

[0071] like Figure 7 As shown, the dynamic weight model construction process is as follows: collect industry benchmark data, generate initial weights, propose principal components, calculate loading weights, learn mapping relationships, output weight vectors, fine-tune with expert experience, form the final dynamic weights, and periodically update benchmark values ​​- repeat the process.

[0072] like Figure 8 As shown, the maturity assessment and rating process is as follows: collect actual operational data of the enterprise, calculate the actual KPI achievement value, normalize the KPI (actual value / industry standard value), adjust the commission dynamic weight - calculate the subdomain score, the calculator has a comprehensive score, match the maturity level (level 1-5), and generate an assessment package - solution.

[0073] Example 5: In another embodiment, a method for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding is disclosed, including: S1: Construct an industry feature knowledge base: Based on the national industry classification standards, use the PEST method to extract key features in four dimensions: policy, economy, society, and technology, and establish a four-tuple of "industry - capability subdomain - KPI - typical business process".

[0074] S2: Generate personalized assessment information package: Receive enterprise industry attributes, and output the subdomains of capabilities to be assessed, recommended KPIs, and business processes through TF-IDF cosine retrieval or industry embedding model.

[0075] S3: Supporting Material Semantic Processing: Perform Chinese word segmentation, stop word filtering, Sentence-LDA+TF-IDF keyword extraction, and BERT semantic vector encoding on uploaded documents; calculate word weights and construct semantic vectors for process data exported from business systems.

[0076] S4: Matching and Similarity Calculation: Match semantic vectors with the knowledge base recommendation set, calculate semantic similarity, and identify key capability shortcomings; use graph neural networks (GNN) to construct keyword semantic networks and generate targeted technical improvement suggestions and process optimization schemes.

[0077] S5: Subdomain Maturity Measurement: a) Collect actual operational data of the enterprise according to the standard process defined in GB / T 39116; b) Calculate the achievement value of each KPI and normalize it with national standards / industry best practices; c) Learn benchmark data through principal component analysis or autoencoder to obtain initial weights, then introduce expert rules for fine-tuning to form a dynamic weight vector, and calculate the subdomain score by weighting.

[0078] S6: Overall Rating: Using AHP-fuzzy comprehensive evaluation or machine learning regression model, the scores of 20 subdomains are mapped to maturity levels 1-5; S7: Interpretable Reports: Automatically outputs scores for each subdomain, weakness analysis, improvement suggestions, and implementation priorities, supporting both PDF and online dashboard formats.

[0079] Example 6: In this embodiment, an application to a cigarette manufacturing company is disclosed, and the specific evaluation process is as follows: 1. Constructing an Industry Characteristic Knowledge Base: The PEST analysis method was used to analyze the cigarette industry, including: P dimension: national tobacco monopoly system, strict quality supervision, and high tax policies; E dimension: high-profit industry, low raw material price fluctuations, and stable market demand; S dimension: increased consumer health awareness and rising demand for low-tar products; T dimension: widespread adoption of digital twins, machine vision quality inspection, and intelligent logistics traceability systems. The analysis identified key evaluation points for the cigarette industry.

[0080] Select the "Tobacco Products Industry - Cigarette Manufacturing" sub-sector, and the system will match characteristic sub-domains such as tobacco blending ratio, cigarette quality, and logistics traceability, and recommend industry-specific indicators such as tobacco consumption rate, packaging machine OEE, and energy consumption per 10,000 cigarettes.

[0081] 2. Text semantic processing: Upload the intelligent transformation report of the tobacco processing workshop, extract industry keywords such as "tobacco blending model, infrared moisture meter, small package imaging detection", and recommend solutions such as digital twin of tobacco and machine vision quality inspection.

[0082] 3. Weights are dynamically determined: Based on data from 10 benchmark tobacco factories, the initial weighting of PCA emphasizes the "ratio accuracy" indicator, while the expert AHP emphasizes the importance of "moisture control". The combined weighting reflects the characteristics of the tobacco industry.

[0083] 4. Evaluation Results: The silk-making subdomain score S=0.912, and the maturity level is judged to be level 4.

[0084] Example 7: This invention also discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following method steps: (1) Construct an industry characteristic knowledge base, extract key industry characteristics based on the national industry classification standard, and match recommended capability subdomain set, key performance indicator set and typical business process set for each industry; (2) Receive industry attribute information of the enterprise being evaluated, and generate a personalized evaluation information package for the enterprise in the knowledge base through the industry-capability matching algorithm. The information package includes the capability subdomain to be evaluated, recommended key performance indicators and recommended business processes. (3) Obtain the supporting documentation of the capabilities submitted by the evaluated enterprise, perform natural language processing on the materials, extract keywords and generate semantic vectors. Perform word segmentation and stop word processing on the enterprise's business process data, extract keywords and calculate weights; (4) Match the preprocessed enterprise business process data with the recommendation information set and calculate the matching degree. Calculate semantic similarity and output the identified key capabilities and recommended solutions; evaluate the effectiveness of the business process and capability effectiveness based on the matching degree, and generate an effectiveness evaluation report. Use the Sentence-LDA topic model combined with the TF-IDF algorithm to extract keywords from the material; analyze the relationships between keywords through semantic association technology to construct a semantic network; generate targeted solutions based on the semantic network and the recommendation information set, including technical improvement suggestions and process optimization schemes.

[0085] (5) For each subdomain of capability being evaluated, perform the following operations: a) Define the standard business processes for this subdomain in accordance with national standards, and collect or receive actual operational data of the enterprise under these processes; b) Calculate the achievement values ​​of each key performance indicator within the subdomain based on actual operational data; c) Use a dynamic weighting model to determine the weights of each indicator and calculate the maturity score of the subdomain using a weighted average. (6) The maturity scores of all capability subdomains are combined to form the intelligent manufacturing capability maturity level of the evaluated enterprise. According to the "Intelligent Manufacturing Capability Maturity Model" (GB / T 39116-2020), intelligent manufacturing capability is divided into four basic elements and 20 core capability domains: personnel, resources, technology and manufacturing. Evaluation standards are formulated for each capability subdomain, including data collection real-time performance, analysis model application depth and equipment interconnection capability indicators. Actual data of enterprises are collected through interviews, evidence, operation demonstrations and on-site inspections, and performance is evaluated against the evaluation standards. The Analytic Hierarchy Process (AHP) is used to determine the indicator weights of each capability subdomain and to comprehensively calculate the enterprise's intelligent manufacturing capability level score. The enterprise's intelligent manufacturing capability level is determined based on the score and is divided into five levels: planning level, standardization level, integration level, optimization level and leading level.

[0086] The embodiments described above merely illustrate specific implementation methods of this application, and while the descriptions are detailed and specific, they should not be construed as limiting the scope of protection of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the technical solution of this application, and these modifications and improvements all fall within the scope of protection of this application.

Claims

1. A method for assessing the maturity of intelligent manufacturing capabilities based on industry characteristics and semantic understanding, characterized in that, Includes the following steps: Construct an industry-specific knowledge base: Based on national industry classification standards, extract key industry features and match recommended sets of capability subdomains, key performance indicators, and typical business processes for each industry. Generate personalized assessment information package: Receive industry attribute information of the company being assessed, and generate a personalized assessment information package for the company being assessed in the industry feature knowledge base through industry-capability matching algorithm. The personalized assessment information package includes the capability subdomain to be assessed, key performance indicators and typical business processes. Semantic processing of supporting materials: Obtain supporting materials for the capabilities submitted by the evaluated company, perform natural language processing, extract keywords, and generate semantic vectors; Matching and similarity calculation: Match the semantic vector with the recommendation information set of the industry feature knowledge base, and calculate the matching degree; Semantic similarity calculation, outputting the identified key capabilities and recommendation solutions; Computing capability subdomain maturity: Collect or receive actual operational data of enterprises under national standards, calculate the achievement values ​​of each key performance indicator within the capability subdomain based on the actual operational data, determine the weight of each indicator using a dynamic weight model, and calculate the maturity score of the capability subdomain using weighted average. Overall rating: The rating combines the maturity scores of all capability subdomains to form the smart manufacturing capability maturity level of the evaluated enterprise.

2. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 1, characterized in that, In generating personalized assessment information packages, a retrieval method based on TF-IDF vectors and cosine similarity is used, or a trained industry embedding model is employed to output the capability subdomains and indicator sets most relevant to the enterprise's industry. The personalized assessment information package generation method is as follows: in, t As keywords, d For document, N For the total number of documents, DF( t (for keywords) t The number of documents.

3. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 1, characterized in that, In the semantic processing of supporting materials, natural language processing includes Chinese word segmentation, stop word filtering, keyword extraction, and BERT semantic vector encoding.

4. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 1, characterized in that, In the maturity of the computing power subdomain, the dynamic weight model is obtained through the following method: S51: Collect historical indicator data from multiple benchmark companies in the same industry; S52: Using principal component analysis or neural network autoencoders, the contribution of each indicator to the overall maturity is automatically learned, and initial weights are generated. S53: Fine-tune the initial weights by combining expert experience rules to form the final dynamic weight vector.

5. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 1, characterized in that, In the comprehensive rating, weighted average, fuzzy comprehensive evaluation, or machine learning regression models are used to map the scores of multiple subdomains to maturity levels 1-5.

6. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 5, characterized in that, The evaluation criteria for each capability subdomain include the following key indicators: Personnel subdomain: personnel skills training coverage rate, digital skills certification pass rate; Resource subdomain: overall equipment efficiency, resource utilization rate; Technology subdomain: number of data analysis models applied, new technology adoption rate; Manufacturing subdomain: production cycle shortening rate, product qualification rate.

7. The intelligent manufacturing capability maturity assessment method based on industry characteristics and semantic understanding according to claim 1, characterized in that, This also includes generating an interpretable evaluation report, which at least contains: S71: Scores and levels for each capability subdomain; S72: Improvement suggestions for subdomains below the threshold; S73: Recommended solutions and implementation priorities.

8. A smart manufacturing capability maturity assessment system based on industry characteristics and semantic understanding, characterized in that, include: The industry feature knowledge base construction module extracts key features in four dimensions—policy, economy, society, and technology—based on national industry classification standards and using the PEST method. It establishes, stores, and manages the associated data of industry, capability subdomains, indicators, and processes. Information recommendation module: used to output personalized evaluation information packages based on the enterprise's industry attributes; Semantic processing module: used to extract keywords from supporting materials and calculate semantic vectors; The solution matching module is used to retrieve recommended solutions based on semantic vectors; Capability Subdomain Maturity Calculation Module: Used to acquire actual operational data and calculate key performance indicator achievement values, determine the weight of each indicator using dynamic weights, and calculate the maturity score of each capability subdomain using weighted averages. The comprehensive rating and report output module is used to summarize maturity scores and output maturity levels and improvement suggestions, and to display the assessment progress, subdomain scores and historical comparison trends in real time.

9. The intelligent manufacturing capability maturity assessment system based on industry characteristics and semantic understanding according to claim 8, characterized in that, The semantic processing module has a built-in pre-trained Chinese language model and supports users to upload domain-specific custom dictionaries.

10. The intelligent manufacturing capability maturity assessment system based on industry characteristics and semantic understanding according to claim 8, characterized in that, Each module is deployed as a microservice on a containerized platform, supporting horizontal scaling and multi-tenant concurrent evaluation.

Images