Industrial empowerment path planning method and system based on multi-objective optimization

Abstract

This invention provides a method and system for planning industrial empowerment paths based on multi-objective optimization, belonging to the field of industrial empowerment planning technology. The method includes: constructing an enterprise capability vector for the enterprise to be empowered; performing relational modeling of all-domain technical resources; using multiple single-dimensional weights and priorities as optimization constraints; decomposing the target capability vector into K stage capability vectors; using the K stage capability vectors as stage optimization objectives; performing progressive matching iterations based on multi-objective optimization; and concatenating the K stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path. This addresses the technical problems of existing technologies, such as fragmented resource management and extensive matching, which make it difficult to achieve precise planning and resource allocation during industrial empowerment. It achieves the technical effect of improving the executability of the empowerment path and the accuracy of stage capability matching by constructing a technology resource hypergraph and combining it with multi-objective optimization progressive matching, thus realizing phased collaborative planning of the industrial empowerment path.

Description

Technical Field

[0001] This invention relates to the field of industrial empowerment planning technology, specifically to an industrial empowerment path planning method and system based on multi-objective optimization. Background Technology

[0002] In existing industry empowerment and technology services, enterprise capability enhancement typically relies on the integration and matching of external technological resources, equipment resources, human resources, and scientific research achievements. However, in practical applications, these resources are often distributed across different platforms or systems, lacking a unified data structure and association mechanism, exhibiting obvious characteristics of resource fragmentation and silos, making it difficult to form a global view of the overall resources. Furthermore, existing resource matching methods largely rely on keyword retrieval, manual experience rules, or simple similarity calculations, failing to reflect the complex dependencies and synergies between resources. The matching process is relatively crude, lacking the ability to decompose and plan multi-dimensional capability objectives in stages.

[0003] Furthermore, in the process of designing industrial empowerment paths, the temporal dependencies of technological evolution and the sequential constraints between resources are often not fully considered. As a result, resource allocation is often completed in a static and one-off manner, which is difficult to adapt to the phased capability improvement needs of enterprises. This leads to a lack of continuity and progression in the empowerment process, and low overall planning accuracy. This further results in unclear capability improvement paths for enterprises in the process of technology upgrading and capability building, which in turn leads to a slow overall improvement in enterprise technology capabilities and makes it difficult to quickly achieve the expected development goals.

[0004] In summary, existing technologies suffer from decentralized resource management and extensive matching, which makes it difficult to achieve precise planning and resource allocation during the process of industrial empowerment. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for industrial empowerment path planning based on multi-objective optimization, in order to solve the technical problem that existing technologies suffer from decentralized resource management and extensive matching, making it difficult to achieve accurate planning and resource allocation in the process of industrial empowerment.

[0006] In view of the above problems, this application provides an industry empowerment path planning method and system based on multi-objective optimization.

[0007] The first aspect of this application provides a method for planning an industry empowerment path based on multi-objective optimization. This method includes: constructing an enterprise capability vector by performing a global capability diagnosis on the enterprise to be empowered; receiving the technology plan of the enterprise to be empowered and constructing a target capability vector and multiple single-dimensional weight priorities; performing correlation modeling on all-domain technology resources to construct a technology resource hypergraph; using the multiple single-dimensional weight priorities as optimization constraints, decomposing the capability gap between the enterprise capability vector and the target capability vector into K stage capability vectors based on the technology dependency relationships of the technology resource hypergraph; using the K stage capability vectors as stage optimization objectives, performing progressive matching iterations based on multi-objective optimization on the technology resource hypergraph to generate K stage collaborative resource combinations; and concatenating the K stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path for the enterprise to be empowered.

[0008] Optionally, S1: Compare the first-stage capability vector with the technology resource hypergraph to obtain a first candidate resource group; S2: Use the capability update result of the first candidate resource group as the first state input to perform a phase-by-phase progressive matching iteration to obtain a first-stage collaborative resource combination; S3: Compare the second-stage capability vector with the first-stage collaborative resource combination to obtain a second candidate resource group; Iterate steps S1 to S3 until the Kth-stage collaborative resource combination is output, generating the K-stage collaborative resource combinations.

[0009] Optionally, after crawling the publicly available technical data of the enterprise to be empowered using multi-source heterogeneous technology, redundant information in non-technical dimensions is dehydrated to obtain raw technical data; the raw technical data is then subjected to structured feature aggregation to obtain multiple single-dimensional technical feature values ​​for various technical classification dimensions; an industry technical capability baseline range is obtained through interaction; the multiple single-dimensional technical feature values ​​are then normalized and mapped to obtain multiple standard technical capability feature values; and finally, the enterprise capability vector is output by concatenating these values.

[0010] Optionally, using the multiple technology classification dimensions as an alignment framework, the dimensional mapping of the technology planning is decomposed to obtain multiple single-dimensional technology target values. Normalization mapping based on the industry technology capability baseline interval is then performed to generate the target capability vector. Based on the difference gap between the target capability vector and the industry technology capability baseline interval, the initial uniform allocation weights of the multiple technology classification dimensions are differentially adjusted to generate the multiple single-dimensional weight priorities.

[0011] Optionally, the global technical resources are decomposed to obtain technical resource node groups, equipment resource node groups, talent resource node groups, and technical achievement node groups. Based on the global technical resources, feature extraction is performed on the technical resource node groups, equipment resource node groups, talent resource node groups, and technical achievement node groups based on node type differences to obtain technical feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups. Based on the historical collaborative implementation records of technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features, cross-group hyperedge relationship modeling is performed on the technical feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups to generate the technical resource hypergraph.

[0012] Optionally, the temporal relationship hyperedges of the technology resource hypergraph are extracted to generate a technology dependency strength topology; after prioritizing the multiple technology classification dimensions based on the multiple single-dimensional weight priorities, dependency constraints are injected in combination with the technology dependency strength topology to obtain a dependency constraint dimension sequence; after grouping the dependency constraint dimension sequence by dependency hierarchy according to the technology dependency strength topology, adjacent levels are merged based on the weight priority differences within the group to obtain stage dimension groups; multi-dimensional target values ​​corresponding to the target capability vector are extracted from the stage dimension groups, the capability gap of the enterprise capability vector relative to the multi-dimensional target value is compared and calculated, and after generating stage capability vectors, capability increase verification and division under stage scale balance constraints are performed to obtain the K stage capability vectors.

[0013] Optionally, the implementation prediction of industrial empowerment is performed on the H candidate resources of the first candidate resource group to obtain H prediction results of empowerment implementation potential; the H prediction results of empowerment implementation potential are used as the first state input to perform multi-objective collaborative gain quantification to obtain H multi-objective evaluation scores; F candidate resources are selected based on the descending order of the H multi-objective evaluation scores to expand the actionable actions based on the temporal relationship hyperedge to obtain F candidate extended resource combinations, and industrial empowerment implementation prediction and resource interaction disturbance iteration are performed until the stage progress risk is lower than the threshold, and the first stage collaborative resource combination is selected by descending order.

[0014] Optionally, the technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features are retrieved from the full-domain technical resources, and a parallel feature vectorization transformation based on node type adaptation is performed to obtain the technical resource node group, equipment feature vector group, talent feature vector group, and achievement feature vector group.

[0015] Optionally, the technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features are retrieved from the full-domain technical resources, and a parallel feature vectorization transformation based on node type adaptation is performed to obtain the technical resource node group, equipment feature vector group, talent feature vector group, and achievement feature vector group.

[0016] Optionally, the cross-group hyperedge relationship modeling covers technology collaboration relationships, equipment sharing relationships, and R&D timing relationships.

[0017] A second aspect of this application provides an industry empowerment path planning system based on multi-objective optimization. This system includes: a capability diagnosis module for constructing an enterprise capability vector by performing a global capability diagnosis on the enterprise to be empowered; a weight construction module for receiving the technology plan of the enterprise to be empowered and constructing a target capability vector and multiple single-dimensional weight priorities; a correlation modeling module for performing correlation modeling on all-domain technology resources and constructing a technology resource hypergraph; a vector decomposition module for decomposing the capability gap between the enterprise capability vector and the target capability vector into K stage capability vectors based on the technology dependency relationships of the technology resource hypergraph, using the multiple single-dimensional weight priorities as optimization constraints; a matching iteration module for performing progressive matching iteration based on multi-objective optimization on the technology resource hypergraph, using the K stage capability vectors as stage optimization objectives, to generate K stage collaborative resource combinations; and a path generation module for concatenating the K stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path for the enterprise to be empowered.

[0018] One or more technical solutions provided in this application have at least the following technical effects or advantages: The method provided in this application embodiment performs a global capability diagnosis on the enterprise to be empowered, constructing an enterprise capability vector; receives the technology plan of the enterprise to be empowered, constructing a target capability vector and multiple single-dimensional weight priorities; performs correlation modeling on all-domain technology resources, constructing a technology resource hypergraph; uses the multiple single-dimensional weight priorities as optimization constraints, and decomposes the capability gap between the enterprise capability vector and the target capability vector into K stage capability vectors based on the technology dependency relationship of the technology resource hypergraph; uses the K stage capability vectors as stage optimization targets, performs progressive matching iteration based on multi-objective optimization on the technology resource hypergraph, generating K stage collaborative resource combinations; and concatenates the K stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path for the enterprise to be empowered. This achieves the technical effect of constructing a technology resource hypergraph and combining it with multi-objective optimization progressive matching to realize phased collaborative planning of the industry empowerment path, improving the executability of the empowerment path and the accuracy of stage capability matching.

[0019] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0021] Figure 1 A flowchart illustrating the industry empowerment path planning method based on multi-objective optimization provided in this application.

[0022] Figure 2 A schematic diagram of the structure of the industry empowerment path planning system based on multi-objective optimization provided in this application.

[0023] Figure labeling: Capability diagnosis module 11, weight construction module 12, association modeling module 13, vector decomposition module 14, matching iteration module 15, path generation module 16. Detailed Implementation

[0024] This application provides a method and system for planning industrial empowerment paths based on multi-objective optimization. It addresses the technical problem of fragmented resource management and inefficient matching in existing technologies, which makes precise planning and resource allocation difficult during industrial empowerment. The method achieves the technical effect of constructing a technology resource hypergraph and combining it with progressive matching through multi-objective optimization to realize phased collaborative planning of industrial empowerment paths, thereby improving the executability of the empowerment paths and the accuracy of phased capability matching.

[0025] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0026] Example 1, as Figure 1 As shown, this application provides an industry empowerment path planning method based on multi-objective optimization, which includes: By conducting a comprehensive capability diagnosis of the enterprises to be empowered, a capability vector of the enterprise is constructed.

[0027] Furthermore, by conducting a global capability diagnosis of the enterprise to be empowered and constructing an enterprise capability vector, the method includes: crawling multi-source public technical data of the enterprise to be empowered using multi-source heterogeneous technology, removing redundant information in non-technical dimensions to obtain technical raw data; performing structured feature aggregation on the technical raw data to obtain multiple single-dimensional technical feature values ​​for various technical classification dimensions; interactively obtaining the industry technical capability baseline range, performing normalization mapping on the multiple single-dimensional technical feature values ​​to obtain multiple standard technical capability feature values, and then concatenating and outputting the enterprise capability vector.

[0028] Specifically, multi-source heterogeneous technology is used to crawl publicly available technical data from publicly available internet data sources for companies seeking empowerment. This multi-source publicly available technical data refers to data related to technical resources, equipment resources, human resources, and technological achievements. Specifically, technical resource data includes company technical solutions, process flows, patent descriptions, and R&D method descriptions; equipment resource data includes production equipment parameters, experimental equipment configurations, and equipment operating indicators; human resource data includes R&D personnel technical resumes, job skill requirements, and research team structure information; and technological achievement data includes patent achievements, academic papers, standard-setting achievements, and technology transfer achievements. The data sources include, but are not limited to, company databases, academic paper databases, company website information, patent disclosure platforms, and recruitment information platforms. For example, through a web crawler program, corresponding crawling rules are set according to the characteristics of different data sources. Data from product introductions, technical documents, and equipment configuration pages on the company's website is crawled according to the page hierarchy. Data related to the company is accurately obtained from the company database based on technology classification tags, equipment category tags, and R&D entity information. Technological achievement information is extracted from patent and paper databases, and human resource information is extracted from recruitment information platforms.

[0029] The multi-source publicly available technical data obtained through crawling includes not only content related to technical resources, equipment resources, human resources, and technological achievements, but also a large amount of non-technical content, such as corporate slogans, corporate news, and marketing descriptions. Natural Language Processing (NLP) technology and data cleaning algorithms are used to remove redundant information from the non-technical dimensions of the crawled multi-source publicly available technical data. Specifically, NLP technology is used to perform semantic analysis on the text data, identifying words and sentence structures unrelated to technical resources, equipment resources, human resources, and technological achievements. For example, through part-of-speech tagging and syntactic analysis, verbs and noun phrases describing corporate activities but not involving technical capabilities are filtered out. The data cleaning algorithm removes text with specific formats, deletes duplicate data, and eliminates invalid fields according to preset rules. These preset rules refer to pre-defined data filtering, format filtering, and validity verification rules, which can be generated based on industry knowledge bases, technical terminology dictionaries, historical annotation samples, and data quality control standards. For example, text paragraphs containing keywords such as company profile, brand promotion, and marketing activities can be classified as non-technical content and removed. Data records with duplicate occurrences and a text similarity exceeding 90% can be deduplicated. Field completeness thresholds can be set; for example, if the missing rate of technical parameter fields exceeds a preset proportion (e.g., 30%), the data can be classified as invalid and removed. Furthermore, only data records containing technical terminology, such as process parameters, equipment models, R&D job titles, and patent classification numbers, can be retained to ensure that the retained data is relevant to the four types of resources: technical resources, equipment resources, human resources, and technological achievements, thus obtaining original technical data. The aforementioned dehydration of redundant non-technical information refers to stripping away non-technical information irrelevant to enterprise capability diagnosis to reduce noise data interference.

[0030] Structured feature aggregation is performed on raw technical data, and a multi-dimensional technology classification system is constructed to uniformly classify and map data related to technical resources, equipment resources, human resources, and technological achievements. For example, technical capabilities are divided into multiple technical classification dimensions, such as manufacturing process dimension, equipment support capability dimension, talent R&D capability dimension, and achievement transformation capability dimension. For each technical classification dimension, information extraction techniques, such as Named Entity Recognition (NER) and dependency parsing, are used to extract key technical indicators from the raw technical data, such as algorithm accuracy, processing latency, equipment load capacity, R&D personnel skill level, patent conversion rate, and system throughput. These indicators are then mapped to numerical or hierarchical indicators, resulting in multiple single-dimensional technical feature values. These single-dimensional technical feature values ​​represent the enterprise's original capability performance in the corresponding technical classification dimension.

[0031] To eliminate dimensional differences between different enterprises and different technology fields, an industry technical capability baseline range is introduced. This baseline range is derived from historical industry statistical data sets, such as the indicator ranges of top enterprises in the same field regarding technical resource capabilities, equipment resource capabilities, human resource capabilities, and technological achievement capabilities, including industry averages or quantile ranges P10-P90. By normalizing and mapping single-dimensional technical feature values, for example using Min-Max normalization or Z-score standardization, different dimensional indicators are uniformly mapped to comparable scale ranges, forming multiple standard technical capability feature values. These standard technical capability feature values ​​are comparable capability indicators after eliminating industry scale differences.

[0032] Then, all single-dimensional standard technical capability feature values ​​are concatenated into vectors according to a predefined dimensional order to form an enterprise capability vector. The enterprise capability vector not only represents the enterprise's existing technical resource capability level, but also comprehensively reflects the equipment resource support capability, talent resource reserve capability, and technology achievement transformation capability, thereby forming a unified quantitative expression of the overall capability status of the enterprise to be empowered.

[0033] By constructing this enterprise capability vector, we can comprehensively and accurately quantify the technological capabilities of enterprises to be empowered from the perspective of the synergy of four types of resources: technology resources, equipment resources, human resources, and technological achievements. This provides a unified input basis for subsequent resource hypergraph modeling, target capability matching, and empowerment path optimization, and ensures the effectiveness and reliability of empowerment path planning.

[0034] Receive the technology plan of the enterprise to be empowered, and construct the target capability vector and multiple single-dimensional weight priorities.

[0035] Furthermore, the method involves receiving the technology plan of the enterprise to be empowered, constructing a target capability vector and multiple single-dimensional weight priorities, and includes: using the multiple technology classification dimensions as an alignment framework, performing dimensional mapping decomposition of the technology plan to obtain multiple single-dimensional technology target values, performing normalized mapping based on the industry technology capability baseline interval, and generating the target capability vector; and adjusting the initial uniform allocation weights of the multiple technology classification dimensions according to the difference gap between the target capability vector and the industry technology capability baseline interval, thereby generating the multiple single-dimensional weight priorities.

[0036] Specifically, the process involves receiving the technology plans of companies seeking empowerment. These plans refer to a set of technological development goals that the companies intend to achieve within a future period, derived from their strategic planning documents, R&D roadmaps, technology upgrade plans, digital transformation plans, or target parameters input through user interaction. For example, companies might input planning goals such as achieving 90% equipment automation, adding 20 high-level R&D personnel, and increasing patent commercialization rate to 60% through an interactive interface. Using the various technology classification dimensions employed when constructing the company's capability vector as an alignment framework, the technology plans are decomposed through dimensional mapping. This alignment framework unifies the company's current capability space and target capability space under the same dimensional system to ensure comparability between current and target capabilities. For example, a unified classification framework could be used, encompassing manufacturing process dimensions, equipment support capability dimensions, talent R&D capability dimensions, and technology commercialization capability dimensions.

[0037] The semantic descriptions of objectives in the technology planning are semantically analyzed using rule parsing or natural language processing techniques and mapped to corresponding technology classification dimensions to extract multiple single-dimensional technology objective values. For example, by extracting and identifying the equipment automation rate of 90% as an objective parameter, it is mapped to an equipment support capability objective value. These single-dimensional technology objective values ​​represent the target capability level that the enterprise intends to achieve in the corresponding technology classification dimension. To ensure a unified comparative scale for the objective values, the multiple single-dimensional technology objective values ​​are normalized and mapped based on the industry's technical capability baseline range. Min-Max normalization or Z-score standardization methods are used to map each single-dimensional technology objective value to a standard scale range, generating a target capability vector. This target capability vector is a vectorized expression of the enterprise's future target capability state.

[0038] After generating the target capability vector, the initial uniform weights for each technology category dimension are adjusted differentially based on the difference gap between the target capability vector and the industry technical capability baseline range. The difference gap refers to the difference between the target capability value and the industry technical capability baseline range, i.e., difference gap = industry technical capability baseline range - target capability value. The larger the difference gap, the stronger the demand for capability improvement in that dimension, and the higher the priority of resource allocation should be. Initial uniform weight allocation means that before considering the capability gap, each technology category dimension is assigned the same basic weight by default. For example, the initial weight of each dimension is set to 1 / n, where n is the total number of category dimensions and is a positive integer. Based on the size of the difference gap, a differential adjustment algorithm, such as normalized weighted adjustment, is used to generate the final weight values ​​for each dimension, and multiple single-dimensional weight priorities are formed according to the weight size. The higher the weight value, the higher the priority of the corresponding technology dimension in subsequent resource matching and path optimization.

[0039] For example, the technology plan for the enterprise to be empowered proposes: improving manufacturing process efficiency to 95%, equipment automation rate to 90%, high-level R&D talent capability score to 0.80, and achievement transformation rate to 70%. After normalization based on industry baseline, the target capability vector is obtained as [0.90, 0.85, 0.80, 0.75]. The upper limit reference value of the corresponding industry technology capability baseline range is [1.00, 1.00, 1.00, 0.90]. The calculated capability gap is [0.10, 0.15, 0.20, 0.15]. The initial uniform weight is W0=[0.25, 0.25, 0.25, 0.25]. After adjustment based on the difference gap, it is obtained as W=[0.20, 0.28, 0.32, 0.20]. The weight priority ranking result is: talent R&D capability dimension > equipment support capability dimension > manufacturing process dimension = achievement transformation capability dimension. This indicates that in subsequent resource allocation and empowerment path planning, priority should be given to allocating resources related to improving talent R&D capabilities, followed by equipment upgrade resources, and then resources for manufacturing process optimization and achievement transformation. This will form an optimization constraint direction driven by the gap between the target capability and the industry's technical capability baseline.

[0040] By transforming enterprise technology planning from descriptive goals into structured target capability vectors, and generating differentiated weight priorities through capability gaps, the optimization of the empowerment path not only clarifies what goals to achieve, but also identifies which dimensions to prioritize, thereby further improving the pertinence and effectiveness of empowerment.

[0041] Perform relational modeling on all technical resources and construct a technical resource hypergraph.

[0042] Furthermore, a hypergraph of technology resources is constructed by performing relational modeling on the entire domain's technology resources. The method includes: decomposing the entire domain's technology resources to obtain technology resource node groups, equipment resource node groups, talent resource node groups, and technology achievement node groups; extracting features based on node type differences from the technology resource node groups, equipment resource node groups, talent resource node groups, and technology achievement node groups based on the entire domain's technology resources to obtain technology feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups; and modeling cross-group hyperedge relationships among the technology feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups based on historical collaborative implementation records of the technology attribute features, equipment attribute features, talent attribute features, and achievement attribute features to generate the hypergraph of technology resources.

[0043] Furthermore, the technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features are retrieved from the full-domain technical resources, and a parallel feature vectorization transformation based on node type adaptation is performed to obtain the technical resource node group, equipment feature vector group, talent feature vector group, and achievement feature vector group.

[0044] Furthermore, the cross-group hyper-edge relationship modeling covers technology collaboration relationships, equipment sharing relationships, and R&D timing relationships.

[0045] Specifically, comprehensive technical resources refer to the set of resources acquired through industry resource databases, industry-academia-research collaboration platforms, equipment asset management platforms, talent resume databases, patent achievement databases, and historical project implementation record databases, participating in the enterprise empowerment process. These resources include four categories: technical resources, equipment resources, talent resources, and technological achievements. The comprehensive technical resources undergo resource decomposition, breaking down the entire resource set into technical resource node groups, equipment resource node groups, talent resource node groups, and technological achievement node groups according to resource attribute categories. A node group refers to a set of resource entities with the same type of attribute in a graph structure, where each node corresponds to an independent resource object. For example, technical resource nodes include algorithm modules, process flows, or software tools; equipment resource nodes include, but are not limited to, production equipment, experimental instruments, or computing power platforms; talent resource nodes include, but are not limited to, R&D personnel, expert teams, or technical service providers; and technological achievement nodes include, but are not limited to, patent achievements, technical standards, or R&D outputs. The node decomposition process is achieved through resource classification tag matching, entity recognition, or rule mapping, uniformly mapping heterogeneous resources into node representations in a graph structure.

[0046] Based on the full domain of technical resources, feature extraction is performed on technical resource node groups, equipment resource node groups, talent resource node groups, and technical achievement node groups based on node type differences. These node type differences refer to the different attribute structures of different node categories. An adaptive feature extraction mechanism is used to retrieve corresponding node attribute features from the full domain of technical resources, including technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features. Technical attribute features include, but are not limited to, Technology Maturity Level (TRL), suitable scenario category, and process complexity parameters. Equipment attribute features include, but are not limited to, equipment performance parameters, load capacity, and shared availability. Talent attribute features include, but are not limited to, professional skill tags, project experience values, and R&D capability scores. Achievement attribute features include, but are not limited to, patent conversion rate, achievement maturity, and industry applicability. Then, based on node type adaptive parallel feature vectorization transformation, that is, adopting an appropriate vector encoding method for different node types, the technical attribute features, equipment attribute features, talent attribute features and achievement attribute features are converted into unified computable vectors. For example, normalized encoding is used for continuous parameters, One-Hot encoding or embedding encoding is used for categorical labels, and semantic vector encoding is used for text descriptions, thereby obtaining technical resource node groups, equipment feature vector groups, talent feature vector groups and achievement feature vector groups.

[0047] After completing the node vectorization representation, cross-group hyperedge relationship modeling is performed based on historical collaborative implementation records corresponding to technical attribute characteristics, equipment attribute characteristics, talent attribute characteristics, and achievement attribute characteristics. Here, a hyperedge differs from a point-to-point edge in a regular structural graph; it represents a relationship that can simultaneously connect multiple heterogeneous nodes. That is, a hyperedge can simultaneously connect technical nodes, equipment nodes, talent nodes, and achievement nodes, thus reflecting complex multi-entity collaborative relationships. For example, an industrial vision inspection technology node may simultaneously depend on GPU computing power equipment nodes, algorithm R&D talent nodes, and related patent achievement nodes. These four types of nodes can be associated through the same hyperedge. Hyperedge relationship modeling is completed based on historical collaborative implementation records, which can originate from project execution logs, resource joint call records, or R&D collaboration chain data. Hyperedge connection relationships are constructed by statistically analyzing resource co-occurrence frequency, call dependency strength, and implementation order.

[0048] Among them, cross-group hyperedge relationship modeling covers at least three types of relationships: technology collaboration relationship, equipment sharing relationship, and R&D timing relationship. Technology collaboration relationship indicates that there is a joint application dependency between different technical resources, such as the matching relationship between algorithm modules and process flow. Equipment sharing relationship indicates that multiple technical tasks share the same equipment resource, which is reflected by equipment call frequency and resource conflict coefficient. R&D timing relationship indicates that there is a sequential dependency in the implementation of resources, such as the formation of a certain patent result needs to be triggered after the completion of specific technology R&D. Furthermore, edge weights are determined for technical collaboration relationships, equipment sharing relationships, and R&D timing relationships. The edge weights of technical collaboration relationships are used to characterize the synergy strength between different technical resources. They are calculated based on the frequency of historical joint applications and the synergy gain effect. The number of times two technical resources co-occur in historical projects or R&D tasks is counted to form a co-occurrence frequency index. At the same time, a synergy gain coefficient is introduced, which is the performance improvement ratio brought by the joint use of two technical resources compared to their individual use, such as the improvement in algorithm accuracy, efficiency, or cost reduction. The edge weights of technical collaboration relationships are then expressed as a weighted combination of the normalized frequency value and the synergy gain, for example: W=α×f+β×g, where f represents the normalized co-occurrence frequency, g represents the synergy performance improvement rate, and α and β are weight coefficients used to balance the influence between the frequency of use and the actual effect. Thus, the technical collaboration relationship reflects both the historical dependence strength and the actual value of technical integration. The weight coefficients can be set based on actual needs and expert experience, such as α=0.6 and β=0.4.

[0049] The edge weights of device sharing relationships are used to measure the degree of dependence of multiple resources on the same device, and are determined based on the intensity of device usage and the degree of resource competition. By statistically analyzing the number of times different tasks call the same device, the duration of use, and the queuing time within a certain time window, the device resource utilization rate is calculated, such as usage time / total available time. At the same time, the degree of resource competition when multiple tasks are used concurrently is introduced, for example, the longer the device queuing time, the higher the conflict. Similarly, the edge weights of device sharing relationships are determined by weighted combination, thereby reflecting the load pressure and scheduling difficulty of the device in a multi-resource sharing scenario.

[0050] The edge weights of R&D sequence relationships reflect the strength of sequential dependencies between different resources in the R&D process. Their determination is based on a comprehensive calculation of process dependency frequency and time interval sensitivity. By analyzing task execution logs from historical R&D projects, the sequential call relationships between resources are extracted, the frequency of a resource appearing as a prerequisite is statistically analyzed, and the average time interval between preceding and following resources is calculated. A temporal criticality coefficient is introduced to measure the constraint strength of a prerequisite resource on subsequent R&D results. For example, if the absence of a prerequisite resource will prevent subsequent tasks from executing, its weight is higher. The edge weights of R&D sequence relationships are determined through weighting, thus accurately reflecting the structural sequential constraints in the R&D process. Furthermore, a hyperedge association matrix is ​​used to represent the relationship structure, forming a technology resource hypergraph G=(V,H), where V represents the set of nodes consisting of technology, equipment, talent, and achievement nodes, and H represents the set of cross-group hyperedges containing temporal relationships to reflect the sequential dependencies between resources.

[0051] By representing the originally scattered technical resources, equipment resources, human resources, and technological achievements as nodes, and further constructing a technical resource hypergraph through hyperedge association modeling, the problem of various resources being isolated from each other in traditional resource management can be effectively avoided. This allows complex resource collaboration relationships, resource sharing relationships, and R&D dependency relationships to be structurally expressed and uniformly characterized in a unified graph structure, thereby improving the global collaboration and overall executability in the enterprise empowerment path planning process.

[0052] Using the multiple single-dimensional weight priorities as optimization constraints, the capability gap between the enterprise capability vector and the target capability vector is decomposed into K stage capability vectors based on the technology dependency relationship of the technology resource hypergraph.

[0053] Furthermore, using the multiple single-dimensional weight priorities as optimization constraints, the capability gap between the enterprise capability vector and the target capability vector is decomposed into K stage capability vectors based on the technology dependency relationship of the technology resource hypergraph. The method includes: extracting the temporal relationship hyperedges of the technology resource hypergraph to generate a technology dependency intensity topology; prioritizing the multiple technology classification dimensions based on the multiple single-dimensional weight priorities, and then injecting dependency constraints in combination with the technology dependency intensity topology to obtain a dependency constraint dimension sequence; grouping the dependency constraint dimension sequence by dependency hierarchy according to the technology dependency intensity topology, and then merging adjacent levels based on the weight priority differences within the group to obtain stage dimension groups; extracting the multi-dimensional target values ​​corresponding to the target capability vector from the stage dimension groups, comparing and calculating the capability gap between the enterprise capability vector and the multi-dimensional target values, generating stage capability vectors, and then performing capability increase verification and division under stage scale balance constraints to obtain the K stage capability vectors.

[0054] Specifically, temporal relationship hyperedges are extracted from the technology resource hypergraph. These temporal relationship hyperedges refer to resource combinations with clear sequential execution constraints, such as a chain dependency structure from algorithm development to computing power deployment, and then to model verification. By statistically analyzing the task start time, completion time, and dependency call relationships in historical collaborative implementation records, a technology dependency strength topology is constructed. Nodes represent technology classification dimensions or resource sets, and edge weights represent temporal dependency strength, determined by a weighted calculation function of dependency occurrence frequency and average time interval, thus forming a technology dependency strength topology reflecting the technology evolution path. Multiple technology classification dimensions are sorted based on multiple single-dimensional weight priorities, that is, dimensions such as manufacturing process, equipment capability, talent capability, and achievement transformation are sorted from high to low according to their weight values, forming an initial priority sequence.

[0055] This paper integrates the priority sequence with the technology dependency strength topology. Through a dependency constraint injection mechanism, the sequential dependencies in the technology dependency strength topology are embedded into the priority sequence. Specifically, when a low-weight dimension belongs to a higher-order dependency precursor node in the technology dependency strength topology, its order is adjusted upwards or a mandatory precursor constraint is applied. This results in a dependency constraint dimension sequence that balances business importance ranking and technology dependency constraints. The dependency constraint dimension sequence is then grouped hierarchically according to the technology dependency strength topology. This hierarchical grouping refers to dividing each technology dimension into layers based on the topological hierarchy of nodes in the technology dependency strength topology, such as topological depth or the number of layers from the central node, thus forming a multi-layered structure with a clear dependency depth.

[0056] Building upon this, a weight priority difference judgment mechanism within groups is introduced to merge the dimension sets between adjacent levels. When the weight difference between two adjacent levels is less than a preset threshold, such as 0.1, they are merged into the same stage, avoiding excessive fragmentation of stage division, thus obtaining stage dimension grouping. After obtaining the stage dimension grouping, a stage-dimensional index mapping table is established. The stage-dimensional index mapping table is used to clarify the set of technical classification dimensions included in each stage. For example, the dimension set corresponding to stage k is Dk={di,dj,...}, where di represents the i-th technical classification dimension, dj represents the j-th technical classification dimension, such as manufacturing process dimension, equipment support capability dimension, talent R&D capability dimension, and achievement transformation capability dimension, etc.

[0057] The target value ti is read sequentially from the target capability vector according to the dimension index, and sub-vectors are extracted based on the mapping relationship. For example, this can be achieved through a vector index mask or a sparse selection matrix. For instance, a sparse selection matrix Mk is constructed, where a value of 1 is taken when the dimension belongs to that stage, and 0 otherwise. Then, the stage capability vector can be represented as: Vk = Mk · V target , where V target Mk is used to filter the target values ​​of the dimensions involved in this stage, forming the overall target capability vector. Dimensions not included in this stage are assigned a value of 0 or null to ensure vector dimension consistency. This method extracts the multi-dimensional target values ​​corresponding to each stage from the global target capability vector. After extracting the multi-dimensional target values ​​corresponding to the stage dimension groups, the standard technical capability feature values ​​of the corresponding dimensions in the enterprise capability vector are extracted simultaneously and their differences are calculated dimension by dimension with the multi-dimensional target values ​​of this stage to quantify the capability gap that the enterprise to be empowered needs to fill in the current stage, obtaining the capability gap value for each dimension. When the calculation result of a certain dimension is less than or equal to zero, it indicates that the current capability of that dimension has met or exceeded the target requirements, and the corresponding gap value is recorded as 0. After obtaining the capability gap values ​​for each dimension, the gap values ​​are further concatenated according to the stage dimension groups to form the stage capability vector.

[0058] After generating the initial stage capability vector, a stage scale balancing constraint mechanism is introduced to constrain and control the capability improvement and resource load of each stage. First, the capability increase index of each stage is calculated, defined as the sum or weighted sum of the differences between the target value of each dimension of the stage and the corresponding baseline value of the previous stage. Statistical analysis is performed by selecting stage capability improvement data from historical empowerment projects in similar industries. For example, the average capability increase and standard deviation of each stage in the past N projects are extracted. Combined with the industry technology evolution speed coefficient, which reflects the speed of the industry's average technology upgrade cycle, the maximum carrying capacity threshold of the stage is constructed: Maximum carrying capacity threshold of the stage = Industry technology evolution speed coefficient × (Average capability increase + Standard deviation). The industry technology evolution speed coefficient usually ranges from 0.8 to 1.2 and is used to adjust the impact of the speed of industry development on the stage capacity.

[0059] When the capability growth rate of a certain stage exceeds the maximum carrying capacity threshold of that stage, a stage splitting mechanism is triggered. This involves re-dividing the high-growth dimensions within that stage according to dependency order or weight, and allocating them to adjacent stages. When the capability growth rate of a certain stage is less than or equal to the maximum carrying capacity threshold of that stage, a stage merging mechanism is triggered, merging it with adjacent stages to avoid excessively fine-grained stages leading to excessively high execution costs. Through this two-way adjustment strategy of splitting for exceeding growth limits and merging for insufficient growth, the capability growth rate of each stage is kept within a reasonable range, ultimately converging to obtain K stage capability vectors that satisfy equilibrium constraints. By structurally decomposing the overall target capability vector under multiple single-dimensional weight and priority constraints, combined with the temporal dependencies in the technology resource hypergraph, K stage capability vectors with dependency order and stage constraints are formed. This ensures that the capability objectives of each stage not only meet the overall technological development needs of the enterprise but also consider the dependencies between technologies, guaranteeing the rationality and operability of the stage objectives and further improving the efficiency and success rate of achieving the target capabilities.

[0060] Using the K stage capability vectors as stage optimization objectives, a progressive matching iteration based on multi-objective optimization is performed on the technology resource hypergraph to generate K stage collaborative resource combinations.

[0061] Furthermore, using the K stage capability vectors as stage optimization objectives, a progressive matching iteration based on multi-objective optimization is performed on the technology resource hypergraph to generate K stage collaborative resource combinations. The method includes: S1: comparing the first stage capability vector with the technology resource hypergraph to obtain a first candidate resource group; S2: using the capability update result of the first candidate resource group as a first state input, performing a stage progressive matching iteration to obtain a first stage collaborative resource combination; S3: comparing the second stage capability vector with the first stage collaborative resource combination to obtain a second candidate resource group; iterating steps S1 to S3 until the Kth stage collaborative resource combination is output, generating the K stage collaborative resource combinations.

[0062] Specifically, taking K stage capability vectors as stage optimization objectives, a progressive matching iteration based on multi-objective optimization is performed on the technology resource hypergraph. First, step S1 is executed, comparing and matching the first stage capability vector with the technology resource hypergraph. This is achieved by calculating the matching degree function between the stage capability vector and each resource node or resource combination subgraph, such as using weighted cosine similarity or Euclidean distance minimization methods. After decomposing the first stage capability vector into multi-dimensional objective constraints, the candidate resource subset that can maximize the satisfaction of the capability requirement is searched in the technology resource hypergraph, thereby obtaining the first candidate resource group. The first candidate resource group refers to the initial resource combination set that can satisfy the capability objective constraints in the current stage.

[0063] In step S2, the capability update result of the first candidate resource group is used as the first state input for a phased progressive matching iteration. Capability aggregation calculation is performed on various resource nodes in the first candidate resource group, that is, the corresponding attribute vectors are weighted and fused to form the actual achievable capability state vector at the current stage, reflecting the comprehensive capability output of the current resource combination in terms of technology, equipment, talent, and results. Then, the capability state vector is used as feedback input to perform residual calculation with the first stage target capability vector to form a capability gap vector, which drives the next round of resource adjustment and optimization search. Through operations such as local replacement, hyperedge expansion, or neighborhood reconnection, the resource combination is iteratively optimized to obtain the first stage collaborative resource combination, that is, the optimal resource subset that meets the convergence conditions, such as the capability gap being less than a threshold, such as 0.05, or the number of iterations reaching the upper limit, such as 50 to 100 iterations.

[0064] After completing the first stage of optimization, proceed to step S3, where the second-stage capability vector is used as the new optimization objective, and the state of the first-stage collaborative resource combination generated in the previous stage is used as the initial constraint input. Matching calculations are then performed again in the technical resource hypergraph. This involves performing a difference analysis between the second-stage capability vector and the current resource state to determine the incremental capability requirements. Candidate resource nodes with compensation or enhancement capabilities are then prioritized in the technical resource hypergraph to generate a second candidate resource group. This progressive iterative mechanism is repeated, gradually optimizing and updating the candidate resource combinations until convergence is achieved, generating the second-stage collaborative resource combination.

[0065] The above process repeats steps S1 to S3 for each stage, that is, using the stage capability vector as the driving target and the output state of the previous stage as the input constraint of the next stage, a progressive dynamic optimization search is performed in the technology resource hypergraph, thereby generating K stage collaborative resource combinations step by step, and finally forming a complete multi-stage resource configuration sequence.

[0066] By mapping phased capability objectives to a progressive multi-objective optimization problem on a technology resource hypergraph, and through phased state feedback and iterative optimization of resource combinations, a dynamic matching process from capability requirements to resource allocation is achieved. This generates an optimal collaborative resource combination sequence that corresponds one-to-one with the capability vectors of the K phases, ensuring that the industry empowerment path has an executable resource support foundation at each stage. This improves the reliability and execution stability of the industry empowerment path planning, and ensures that the enterprises to be empowered can achieve phased and orderly improvement of their technological capabilities and steadily achieve their overall capability goals under resource constraints.

[0067] Furthermore, the method further includes: performing industry empowerment implementation prediction on H candidate resources of the first candidate resource group to obtain H empowerment implementation potential prediction results; using the H empowerment implementation potential prediction results as the first state input, performing multi-objective collaborative gain quantification to obtain H multi-objective evaluation scores; selecting F candidate resources based on the descending order of the H multi-objective evaluation scores to expand the actionable actions based on temporal relationship hyperedges to obtain F candidate extended resource combinations, performing industry empowerment implementation prediction and resource interaction disturbance iteration until the stage progress risk is lower than the threshold, and selecting the first stage collaborative resource combination by descending order.

[0068] Specifically, the method for obtaining the first-stage collaborative resource combination also includes predicting the implementation of industry empowerment for the H candidate resources in the first candidate resource group. The H candidate resources refer to the resource set obtained through the initial matching stage in the technology resource hypergraph, and their sources include equipment resource nodes, talent resource nodes, and technology resource nodes. Each candidate resource corresponds to an attribute feature vector, obtained based on a resource feature vector library and a historical collaborative implementation record database. The attribute feature vector includes parameters such as capability strength, collaborative adaptability, resource availability, and historical participation frequency. The industry empowerment implementation prediction refers to estimating the actual contribution effect of a single resource in the target industry scenario using a rule-based scoring function and historical statistical mapping method, outputting the prediction result of empowerment implementation potential, forming a set of H predicted values ​​to represent the potential empowerment capability of each resource. For example, for each candidate resource, its basic attributes are extracted from the resource feature vector library and the historical collaborative implementation record database, including indicators such as capability strength, resource availability, historical participation frequency, and average contribution efficiency, and then a standardized evaluation function is established. For example, capability strength is defined as the industry normalized value Ai, resource availability is defined as the time occupancy rate Ui, historical participation frequency is defined as the stability index Fi, and the historical average contribution coefficient Hi, which is obtained by manual statistics or rule induction of past projects, is introduced. The predicted value of industry empowerment implementation is obtained by calculating through a weighted function.

[0069] Using the aforementioned H potential implementation predictions for empowerment as the first state input, multi-objective synergistic gain quantification is performed. This multi-objective synergistic gain refers to the comprehensive improvement effect of a single resource when combined with other resources. Its evaluation dimensions include capability enhancement contribution, cross-resource synergistic matching degree, and resource conflict reduction degree. A multi-objective evaluation function is constructed to calculate a comprehensive score for each candidate resource, for example, using a weighted multi-index evaluation model: Score. i =w1 × Industry Empowerment Prediction + w2 × Collaborative Matching Degree + w3 × Structural Collaborative Gain, where w1, w2, and w3 are weighting coefficients, thus obtaining H multi-objective evaluation scores. The H multi-objective evaluation scores are sorted in descending order, and the top F candidate resources with the highest scores are selected based on resource size and hypergraph density. F is generally set to a value between 0.2H and 0.5H to balance the search space and combinatorial quality.

[0070] Based on the selected F candidate resources, actionable expansion is performed using temporal relationship hyperedges in the technology resource hypergraph. Specifically, by traversing the F candidate resources, all adjacent hyperedges satisfying temporal dependency constraints are matched in the technology resource hypergraph. Resource nodes satisfying pre-dependency and complementary relationships are combined and expanded to form F candidate expanded resource combinations. For each candidate expanded resource combination, industry empowerment implementation prediction is executed again, and iterative optimization is performed using a resource interaction perturbation mechanism. This resource interaction perturbation refers to considering resource conflicts, substitutions, and synergistic enhancement effects during resource combination, adjusting the resource combination structure to optimize the overall empowerment effect. Through iterative execution of prediction, evaluation, and adjustment mechanisms, the quality of the resource combination is continuously optimized until the stage progress risk indicator falls below a preset threshold. The stage progress risk measures the potential delay risk, capability deviation risk, and resource conflict risk during the implementation of the current resource combination. The preset threshold is determined based on historical project risk distribution statistics, for example, set to 0.1. After the convergence condition is met, the candidate combinations are sorted in descending order, and the resource combination with the highest score is selected as the first-stage collaborative resource combination, thereby realizing the whole-process optimization and screening from candidate resources to the optimal collaborative resource combination.

[0071] By introducing industrial empowerment implementation prediction, multi-objective synergistic gain assessment, and a combination expansion mechanism under time constraints, the initial candidate resources are screened and dynamically optimized layer by layer. This ensures the feasibility and efficiency of resources and determines the optimal combination of collaborative resources in the first stage, thereby further improving the accuracy and stability of resource allocation and the feasibility and reliability of industrial empowerment path planning.

[0072] By combining the K stages of collaborative resources according to the technology dependence time sequence, a multi-stage collaborative empowerment path is generated for the enterprise to be empowered.

[0073] Specifically, based on the technology dependency temporal relationship hyperedges extracted from the technology resource hypergraph, the generated K-stage collaborative resource combinations are sequentially reconstructed and their paths are concatenated. Each stage collaborative resource combination is regarded as a stage subgraph with an internal resource collaboration structure. The stages are then topologically ordered according to the technology dependency temporal relationships defined in the technology resource hypergraph. This ordering process is constrained by the dependency strength topology formed by the temporal relationship hyperedges, ensuring that the key technology resources, equipment resources, or human resources included in the previous stage provide the necessary capability support conditions for the next stage. For example, when the equipment deployment resources or basic process resources included in a certain stage are prerequisites for the algorithm optimization or results transformation in the next stage, their priority execution order is forcibly maintained during path concatenation.

[0074] Then, the collaborative resources of each stage are sequentially linked according to the topological sorting results to form a stage chain structure with a clear temporal sequence. A state transfer relationship is established between adjacent stages, that is, the resource output capability state of the previous stage serves as the input constraint condition for the next stage, thereby forming a multi-stage collaborative empowerment path for the enterprise to be empowered.

[0075] By introducing time constraints on technology dependence, the originally independent K-stage collaborative resource combinations are structurally spliced ​​together, transforming them from a discrete set of stages into a continuous empowerment path with sequential logic and capability transfer relationships. This ensures that the industrial empowerment process conforms to the laws of technological evolution and resource dependence in the time dimension, improves the executability, continuity, and stage capability matching accuracy of the overall path, provides effective industrial empowerment guidance for enterprises to be empowered, and helps them gradually improve their technological capabilities and achieve their technological development goals.

[0076] Example 2, based on the same inventive concept as the multi-objective optimization-based industry empowerment path planning method in the previous examples, such as... Figure 2 As shown, this application provides an industry empowerment path planning system based on multi-objective optimization, wherein the industry empowerment path planning system based on multi-objective optimization includes: The capability diagnosis module 11 is used to construct an enterprise capability vector by performing a global capability diagnosis on the enterprise to be empowered; the weight construction module 12 is used to receive the technology plan of the enterprise to be empowered, construct a target capability vector and multiple single-dimensional weight priorities; the association modeling module 13 is used to perform association modeling on the full-domain technology resources and construct a technology resource hypergraph; the vector decomposition module 14 is used to decompose the capability gap of the enterprise capability vector relative to the target capability vector into K stage capability vectors based on the technology dependency relationship of the technology resource hypergraph, using the multiple single-dimensional weight priorities as optimization constraints; the matching iteration module 15 is used to perform progressive matching iteration based on multi-objective optimization on the technology resource hypergraph with the K stage capability vectors as stage optimization targets, and generate K stage collaborative resource combinations; the path generation module 16 is used to splice the K stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path for the enterprise to be empowered.

[0077] Furthermore, the matching iteration module 15 is also used for: S1: comparing the first-stage capability vector and the technology resource hypergraph to obtain a first candidate resource group; S2: using the capability update result of the first candidate resource group as the first state input, performing a phase-progressive matching iteration to obtain a first-stage collaborative resource combination; S3: comparing the second-stage capability vector and the first-stage collaborative resource combination to obtain a second candidate resource group; iterating steps S1 to S3 until the Kth-stage collaborative resource combination is output, generating the K-stage collaborative resource combinations.

[0078] Furthermore, the capability diagnosis module 11 is also used for: crawling multi-source public technical data of the enterprise to be empowered using multi-source heterogeneous technology, dehydrating non-technical redundant information to obtain technical raw data; performing structured feature aggregation on the technical raw data to obtain multiple single-dimensional technical feature values ​​of various technical classification dimensions; interactively obtaining the industry technical capability baseline range, performing normalization mapping on the multiple single-dimensional technical feature values ​​to obtain multiple standard technical capability feature values, and then splicing and outputting the enterprise capability vector.

[0079] Furthermore, the weight construction module 12 is also used to: use the multiple technology classification dimensions as an alignment framework to perform dimensional mapping decomposition of the technology planning, obtain multiple single-dimensional technology target values, perform normalization mapping based on the industry technology capability baseline interval, and generate the target capability vector; and adjust the initial uniform distribution weights of the multiple technology classification dimensions differently according to the difference gap between the target capability vector and the industry technology capability baseline interval, and generate the multiple single-dimensional weight priorities.

[0080] Furthermore, the association modeling module 13 is also used to: decompose the global technical resources to obtain technical resource node groups, equipment resource node groups, talent resource node groups, and technical achievement node groups; based on the global technical resources, perform feature extraction on the technical resource node groups, equipment resource node groups, talent resource node groups, and technical achievement node groups based on node type differences to obtain technical feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups; and, based on the historical collaborative implementation records of technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features, perform cross-group hyperedge relationship modeling for the technical feature vector groups, equipment feature vector groups, talent feature vector groups, and achievement feature vector groups to generate the technical resource hypergraph.

[0081] Furthermore, the vector decomposition module 14 is also used to: extract the temporal relationship hyperedges of the technology resource hypergraph to generate a technology dependency strength topology; prioritize the multiple technology classification dimensions based on the multiple single-dimensional weight priorities, and inject dependency constraints in combination with the technology dependency strength topology to obtain a dependency constraint dimension sequence; group the dependency constraint dimension sequence according to the technology dependency strength topology, and merge adjacent levels based on the weight priority differences within the group to obtain a stage dimension group; extract the multi-dimensional target values ​​corresponding to the target capability vector from the stage dimension group, compare and calculate the capability gap of the enterprise capability vector relative to the multi-dimensional target values, generate stage capability vectors, and perform capability increase verification and division under stage scale balance constraints to obtain the K stage capability vectors.

[0082] Furthermore, the matching iteration module 15 is also used to: perform industry empowerment implementation prediction on the H candidate resources of the first candidate resource group to obtain H empowerment implementation potential prediction results; use the H empowerment implementation potential prediction results as the first state input to perform multi-objective collaborative gain quantification to obtain H multi-objective evaluation scores; select F candidate resources based on the descending order of the H multi-objective evaluation scores to expand the actionable actions based on the temporal relationship hyperedge to obtain F candidate extended resource combinations, perform industry empowerment implementation prediction and resource interaction disturbance iteration until the stage progress risk is lower than the threshold, and select the first stage collaborative resource combination by descending order.

[0083] Furthermore, the association modeling module 13 is also used to: retrieve the technical attribute features, equipment attribute features, talent attribute features and achievement attribute features from the global technical resources, perform parallel feature vectorization transformation based on node type adaptation, and obtain the technical resource node group, equipment feature vector group, talent feature vector group and achievement feature vector group.

[0084] Furthermore, the association modeling module 13 is also used to: cover the technology collaboration relationship, equipment sharing relationship and R&D sequence relationship in the cross-group hyper-edge relationship modeling.

[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The industry empowerment path planning method and specific examples based on multi-objective optimization in the aforementioned embodiment 1 are also applicable to the industry empowerment path planning system based on multi-objective optimization in this embodiment. Through the foregoing detailed description of the industry empowerment path planning method based on multi-objective optimization, those skilled in the art can clearly understand the industry empowerment path planning system based on multi-objective optimization in this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0087] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. An industry empowerment path planning method based on multi-objective optimization, characterized in that, The method includes: By conducting a comprehensive capability diagnosis of the enterprises to be empowered, a capability vector of the enterprise is constructed; Receive the technology plan of the enterprise to be empowered, and construct a target capability vector and multiple single-dimensional weight priorities; Perform relational modeling on all domain technology resources to construct a technology resource hypergraph; Using the multiple single-dimensional weight priorities as optimization constraints, the capability gap between the enterprise capability vector and the target capability vector is decomposed into K stage capability vectors based on the technology dependency relationship of the technology resource hypergraph. Using the K stage capability vectors as stage optimization objectives, a progressive matching iteration based on multi-objective optimization is performed on the technology resource hypergraph to generate K stage collaborative resource combinations; By combining the K stages of collaborative resources according to the technology dependence time sequence, a multi-stage collaborative empowerment path is generated for the enterprise to be empowered.

2. The multi-objective optimization-based industry enablement pathway planning method of claim 1, wherein, Using the K stage capability vectors as stage optimization objectives, a progressive matching iteration based on multi-objective optimization is performed on the technology resource hypergraph to generate K stage collaborative resource combinations. The method includes: S1: Compare the first-stage capability vector with the technical resource hypergraph to obtain the first candidate resource group; S2: Take the capability update result of the first candidate resource group as the first state input, perform phased progressive matching iteration, and obtain the first phase collaborative resource combination; S3: Compare the second-stage capability vector with the first-stage collaborative resource combination to obtain the second candidate resource group; Iterate through steps S1 to S3 until the Kth stage collaborative resource combination is output, generating the K stage collaborative resource combinations.

3. The multi-objective optimization-based industry enablement pathway planning method of claim 1, wherein, The method involves conducting a comprehensive capability diagnosis of the enterprises to be empowered and constructing an enterprise capability vector, including: After crawling the publicly available technical data of the enterprise to be empowered using multi-source heterogeneous technology, redundant information in non-technical dimensions is dehydrated to obtain the original technical data. The raw technical data is subjected to structured feature aggregation to obtain multiple single-dimensional technical feature values ​​for various technical classification dimensions; The system interactively obtains the industry's technical capability baseline range, performs normalization mapping on the multiple single-dimensional technical feature values, obtains multiple standard technical capability feature values, and then concatenates them to output the enterprise capability vector.

4. The multi-objective optimization-based industry enablement pathway planning method of claim 3, wherein, The method includes receiving the technology plan of the enterprise to be empowered, constructing a target capability vector and multiple single-dimensional weight priorities, and comprising: Using the aforementioned multiple technology classification dimensions as an alignment framework, the dimensional mapping of the technology planning is decomposed to obtain multiple single-dimensional technology target values. Then, a normalized mapping based on the industry technology capability baseline range is performed to generate the target capability vector. Based on the difference gap between the target capability vector and the industry technical capability baseline range, the initial uniform weight allocation of the multiple technical classification dimensions is adjusted differentially to generate the multiple single-dimensional weight priorities.

5. The multi-objective optimization-based industry enablement pathway planning method of claim 3, wherein, The method for constructing a technology resource hypergraph by performing relational modeling on all technology resources includes: By disassembling the entire domain of technical resources, we obtain technical resource node groups, equipment resource node groups, human resource node groups, and technical achievement node groups. Based on the global technical resources, feature extraction is performed on the technical resource node group, equipment resource node group, talent resource node group and technical achievement node group based on the differences in node type to obtain technical feature vector group, equipment feature vector group, talent feature vector group and achievement feature vector group; Based on the historical collaborative implementation records of technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features, cross-group hyperedge relationship modeling is performed for the technical feature vector group, equipment feature vector group, talent feature vector group, and achievement feature vector group to generate the technical resource hypergraph.

6. The multi-objective optimization-based industry enablement pathway planning method of claim 5, wherein, Using the multiple single-dimensional weight priorities as optimization constraints, and based on the technology dependency relationships in the technology resource hypergraph, the method decomposes the capability gap between the enterprise capability vector and the target capability vector into K-stage capability vectors. Extract the temporal relationship hyperedges of the technology resource hypergraph to generate a technology dependency strength topology; After prioritizing the multiple technology classification dimensions based on the multiple single-dimensional weight priorities, dependency constraints are injected in combination with the technology dependency strength topology to obtain a dependency constraint dimension sequence. After grouping the dependency constraint dimension sequence according to the technology dependency strength topology, adjacent levels are merged based on the weight priority difference within the group to obtain stage dimension grouping. Extract the multi-dimensional target values ​​corresponding to the target capability vector from the stage dimension group, compare and calculate the capability gap of the enterprise capability vector relative to the multi-dimensional target value, generate stage capability vectors, and then perform capability increase verification and division under stage scale balance constraint to obtain the K stage capability vectors.

7. The multi-objective optimization-based industry enablement pathway planning method of claim 2, wherein, The method further includes: The industrial empowerment implementation prediction is performed on the H candidate resources of the first candidate resource group to obtain the prediction results of the H empowerment implementation potential; Using the H predicted results of the empowerment implementation potential as the first state input, multi-objective collaborative gain quantization is performed to obtain H multi-objective evaluation scores; Based on the descending order of the H multi-objective evaluation scores, F candidate resources are selected for actionable expansion based on temporal relationship hyperedges, resulting in F candidate extended resource combinations. Industrial empowerment implementation prediction and resource interaction disturbance iteration are performed until the stage progress risk is lower than the threshold. The first stage collaborative resource combination is then selected by descending order.

8. The multi-objective optimization-based industry enablement pathway planning method of claim 5, wherein, The technical attribute features, equipment attribute features, talent attribute features, and achievement attribute features are retrieved from the full-domain technical resources, and a parallel feature vectorization transformation based on node type adaptation is performed to obtain the technical resource node group, equipment feature vector group, talent feature vector group, and achievement feature vector group.

9. The multi-objective optimization-based industry enablement pathway planning method of claim 5, wherein, The cross-group hyperedge relationship modeling covers technology collaboration relationships, equipment sharing relationships, and R&D timing relationships.

10. An industry enabling path planning system based on multi-objective optimization, characterized by, The step of implementing the industry empowerment path planning method based on multi-objective optimization according to any one of claims 1 to 9, wherein the industry empowerment path planning system based on multi-objective optimization comprises: The capability diagnosis module is used to perform a global capability diagnosis on the enterprise to be empowered and construct the enterprise capability vector. The weight construction module is used to receive the technology plan of the enterprise to be empowered, and construct the target capability vector and multiple single-dimensional weight priorities. The association modeling module is used to model the associations of all technical resources in the domain and construct a technical resource hypergraph. The vector decomposition module is used to decompose the enterprise capability vector relative to the target capability vector into K stage capability vectors based on the technical dependency relationship of the technology resource hypergraph, with the multiple single-dimensional weight priorities as the optimization constraint direction; The matching iteration module is used to perform progressive matching iteration based on multi-objective optimization on the technical resource hypergraph, with the K stage capability vectors as stage optimization objectives, to generate K stage collaborative resource combinations; The path generation module is used to concatenate the K-stage collaborative resource combinations according to the technology dependency time sequence to generate a multi-stage collaborative empowerment path for the enterprise to be empowered.

Images