A project portfolio optimization method based on heterogeneous networks

Abstract

This invention discloses a project portfolio optimization method based on heterogeneous networks, relating to the fields of project management and intelligent decision-making. The method first acquires project objectives, capability requirements, resource inputs, task dependencies, and risk information, and then standardizes and encodes projects, capabilities, tasks, resources, and strategies. Based on RDF triples, it constructs project association networks, capability association networks, and capability-project fusion heterogeneous networks. It uses grey relational analysis and entropy weighting to determine capability importance, and combines project support, project contribution, data envelopment analysis, CRITIC, critical path method, and TOPSIS to complete project value, core projects, critical path, and risk assessment. Under budget, capability thresholds, project mutual exclusion, and schedule constraints, a multi-objective portfolio optimization model is established, and NSGA-II is used to generate a Pareto optimal solution set. Finally, a recommended solution is output through weighted scoring. This invention can improve the systematicity, accuracy, and risk controllability of resource allocation.

Description

Technical Field

[0001] This invention relates to the field of project management and intelligent decision-making technology, and in particular to a project portfolio optimization method based on heterogeneous networks. Background Technology

[0002] Portfolio decisions typically require coordination among multiple objectives, including capacity building, resource allocation, implementation schedule, and risk control. Existing project management methods often rely on individual project reviews or ranking based on single metrics, making it difficult to uniformly express the supporting, dependent, mutually exclusive, and synergistic relationships between projects, resulting in a lack of systematic approach to portfolio decisions.

[0003] Some existing portfolio decision-making methods draw on financial portfolio theory, simplifying projects into static trade-offs between benefits and costs. However, projects are characterized by non-monetary outputs, path-dependent capacity building, complex inter-project relationships, and the coexistence of multi-source heterogeneous data. Traditional methods struggle to accurately reflect the supporting role of projects in system capabilities and also fail to balance resource efficiency with risk constraints.

[0004] Furthermore, existing methods lack the ability to describe the semantic relationships between project objects, making it difficult to support unified modeling of multiple entities such as projects, capabilities, tasks, resources, and strategies. Consequently, it is challenging to conduct core project mining, critical path identification, project risk quantification, and multi-objective portfolio optimization. Therefore, it is necessary to propose a project portfolio optimization method that can describe heterogeneous project relationships, support multi-criteria comprehensive evaluation, and output feasible portfolio solutions. Summary of the Invention

[0005] The purpose of this invention is to provide a project portfolio optimization method based on heterogeneous networks to solve the problems of insufficient structured description, inadequate multi-criteria evaluation, and difficulty in obtaining high-quality portfolio solutions under complex constraints in existing project portfolio decision-making.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a project portfolio optimization method based on heterogeneous networks, characterized by comprising the following steps:

[0007] Step S1: Obtain basic data for the set of projects to be optimized. The basic data includes at least project target information, capability requirement information, task dependency information, resource input information, and risk information. Entities such as projects, capabilities, tasks, resources, and strategies are then standardized and coded.

[0008] Step S2: Construct a project object relationship description model based on the normalized entity data, use the resource description framework RDF triples to represent the semantic relationships of project objects, and construct a project association network, a capability association network, and a capability-project fusion heterogeneous network accordingly.

[0009] Step S3: Based on the aforementioned capability-project fusion heterogeneous network, conduct a multi-criteria comprehensive evaluation to obtain capability importance, project support, project contribution, project cost-effectiveness ratio, backbone project identification results, critical path identification results, and project risk assessment results.

[0010] Step S4: Establish a project portfolio optimization model based on the comprehensive evaluation results of multiple criteria, with the optimization objectives of maximizing capacity coverage, minimizing portfolio risk, and optimizing the cost-effectiveness ratio of the project portfolio, and introduce budget constraints, capacity lower limit constraints, and project mutual exclusion constraints.

[0011] Step S5: The non-dominated sorting genetic algorithm NSGA-II is used to solve the item combination optimization model to obtain the Pareto optimal solution set, and a recommended item combination scheme is output by combining the weighted scoring mechanism.

[0012] Furthermore, in step S1, entity identification, synonym merging, invalid character cleaning, attribute completion, and identifier mapping are performed on the original project data to form a unified metadata structure that can be used for heterogeneous network modeling.

[0013] Furthermore, in step S2, the RDF triple adopts the expression method of "subject-predicate-object". The subject and object include at least project entity, capability entity, task entity, resource entity and strategy entity. The predicate is used to describe support relationship, dependency relationship, containment relationship, constraint relationship and temporal relationship.

[0014] Furthermore, in step S2, the node set of the capability-project fusion heterogeneous network includes at least a set of project nodes and a set of capability nodes, and the edge set is used to characterize the support relationship between projects and capabilities, the dependency relationship between projects, and the correlation relationship between capabilities.

[0015] Furthermore, in step S2, network structure features are extracted from the capability-project fusion heterogeneous network. These network structure features include at least node degree, betweenness centrality, clustering coefficient, and path relationship, which are used to identify important nodes and key association structures in the network.

[0016] Furthermore, in step S3, the importance of capability is determined by combining grey relational analysis and entropy weight method. Grey relational analysis is used to calculate the degree of closeness between capability indicators and ideal reference sequences, while entropy weight method is used to determine objective weights based on the dispersion of indicators, and thereby form the comprehensive importance of capability.

[0017] Furthermore, in step S3, by constructing a capability-project support matrix, the support strength of each project for different capability dimensions is quantified, and the support results are normalized to obtain the project support degree.

[0018] Furthermore, in step S3, the project contribution is obtained by aggregating the project support level and capability importance, and the project input-output efficiency is evaluated by data envelopment analysis (DEA) to obtain the project cost-effectiveness ratio.

[0019] Furthermore, in step S3, the CRITIC method is used to jointly analyze the project support, project contribution, project cost-effectiveness ratio, and project efficiency in order to identify key projects.

[0020] Furthermore, in step S3, the Critical Path Method (CPM) is used to identify the critical paths that affect the overall progress of the project portfolio based on the project task dependencies.

[0021] Furthermore, in step S3, a risk assessment index system is constructed for project technical risks, resource risks, schedule risks and external environmental risks, and the TOPSIS (Topological Solution Ranking) method is used to obtain the comprehensive risk ranking result of the project.

[0022] Furthermore, in step S4, the project portfolio optimization model uses binary decision variables to represent whether a project is selected, and uses a constraint handling mechanism to penalize over-budget schemes, schemes that fail to meet capacity standards, and mutually exclusive conflict schemes.

[0023] Furthermore, in step S5, the NSGA-II solution process includes population initialization, crossover mutation, non-dominated sorting, crowding calculation, elite retention, and constraint penalty update steps to output a Pareto optimal solution set that satisfies the multi-objective trade-off requirements.

[0024] Furthermore, in step S5, each alternative solution in the Pareto optimal solution set is scored according to the comprehensive weights of capability coverage, project contribution, resource efficiency and risk control, and the final recommended solution is output.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] (1) By introducing RDF triples and capability-item fusion heterogeneous networks, the standardized expression of project object relationships is realized, which improves the structured level of project portfolio modeling.

[0027] (2) By integrating grey relational analysis, entropy weight method, DEA, CRITIC, CPM and TOPSIS, a multi-criteria evaluation system covering capability, value, schedule and risk has been formed.

[0028] (3) By outputting the Pareto optimal solution set under multiple constraints through NSGA-II, the scientific nature, flexibility and risk controllability of resource allocation can be significantly improved. Attached Figure Description

[0029] Figure 1This is a schematic diagram of the project portfolio optimization model architecture of the present invention.

[0030] Figure 2 This is a schematic diagram of the system framework of the present invention.

[0031] Figure 3 This is a schematic diagram of the heterogeneous network structure for the capability-project integration of the present invention.

[0032] Figure 4 This is a schematic diagram of the support calculation process for the project of this invention.

[0033] Figure 5 This is a schematic diagram of the cost-effectiveness calculation process for the present invention. Detailed Implementation

[0034] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0035] Example: Taking a project portfolio in a certain field as the implementation target, there are 20 candidate projects, covering areas such as intelligent management, data analysis, and security monitoring. The total project budget is 98.8 million yuan. Project portfolio optimization analysis is carried out around 15 core capabilities. A basic dataset is formed by collecting information on the objectives, capability requirements, task dependencies, funding, human resource allocation, and risks of each candidate project.

[0036] In this embodiment, a project object relationship description model is constructed based on entities such as projects, capabilities, tasks, and strategies. RDF triples are used to structure semantic relationships such as "project supporting capabilities," "project dependent tasks," and "project constraints." Furthermore, a project association network, a capability association network, and a capability-project fusion heterogeneous network are constructed; the structure of these networks can be found in [reference needed]. Figure 3 .

[0037] Structural features such as node connectivity, critical links, and path relationships are extracted from the aforementioned capability-project fusion heterogeneous network to identify key project nodes with high connectivity and high controllability in the project portfolio, and to provide a foundation for subsequent backbone project identification and critical path analysis.

[0038] In this embodiment, for the 15 core capabilities, grey relational analysis is used to calculate the degree of correlation between each capability indicator and the ideal reference sequence, and the entropy weight method is used to determine the objective weight of each capability indicator, thereby obtaining the overall importance ranking of capabilities. The results are shown in Table 1.

[0039] Table 1. Results of Comprehensive Weighting of Core Competencies

[0040] As shown in Table 1, capability C6 has the highest overall weight of 0.3806735, therefore it should be prioritized in the project portfolio optimization process. In addition, C9, C11, and C1 also have relatively high overall weights and can be considered as key capability dimensions in subsequent project selection, project support analysis, and project portfolio optimization. Based on the capability overall weight results shown in Table 1, a capability-project support matrix can be further constructed, and project support and project contribution can be calculated.

[0041] A capability-project support matrix is ​​constructed based on the correspondence between projects and capabilities. The support strength of each project across different capability dimensions is quantified and normalized to obtain the project support degree. Furthermore, the project contribution degree is calculated based on the support degree to achieve a comprehensive evaluation of the project's technical value and system contribution. The calculation process for the project support degree can be found in [link to relevant documentation]. Figure 4 .

[0042] In this embodiment, project funding and human resource input are used as input indicators, while application benefits, target completion rate, and technological breakthroughs are used as output indicators. Data Envelopment Analysis (DEA) is employed to calculate the input-output efficiency of each candidate project, yielding the project cost-effectiveness ratio. The calculation results show significant differences in resource utilization efficiency among different projects. The integrated business application system has the highest cost-effectiveness ratio at 3.67, indicating that this project generates high comprehensive benefits per unit of resource input. The cost-effectiveness ratios for the big data modeling and analysis training platform and the industry application software operation and maintenance system are 0.38 and 0.23, respectively. In contrast, the cost-effectiveness ratios for the signal intelligent optimization platform, the multi-source data fusion analysis platform, and the collaborative communication management system are 0.10, 0.09, and 0.04, respectively. Therefore, the project cost-effectiveness ratio can be used as an evaluation criterion for subsequent project combination selection.

[0043] Based on this, the CRITIC method is used to comprehensively analyze indicators such as project support, project contribution, project cost-effectiveness, and project efficiency to identify key projects; at the same time, the CPM method is used to identify critical paths that affect the overall duration of the project portfolio based on project task dependencies.

[0044] To assess project uncertainty, a risk evaluation index system was constructed, encompassing technical risk, resource risk, schedule risk, and external environmental risk. After standardizing the relevant indicators, TOPSIS was used to calculate the closeness of each project to the positive and negative ideal solutions, obtaining a ranking of the projects' overall risk levels. The evaluation results show that among the candidate projects, P3, P4, and P5 are classified as Level II high-risk projects, P1 and P6 as Level III medium-risk projects, and P2 and P7 as Level IV low-risk projects. Based on this risk level ranking, constraints can be imposed on high-risk projects or their priority can be reduced during project portfolio optimization.

[0045] Based on the aforementioned factors such as capability importance, project support, project contribution, project cost-effectiveness, key project identification results, critical path identification results, and project risk assessment results, a project portfolio optimization model is established. The optimization model aims to maximize capability coverage, minimize portfolio risk, and optimize the project portfolio cost-effectiveness, while also setting budget constraints, capability lower limit constraints, and project mutual exclusion constraints.

[0046] In this embodiment, binary encoding is used to represent whether a candidate item is selected, and the encoding result is used as the population individual for the NSGA-II algorithm. The algorithm performs a global search of the item combination space through population initialization, crossover, mutation, non-dominated sorting, crowding calculation, and elite retention, and uses a dynamic penalty function to constrain over-budget, under-capacity, and mutually exclusive conflict schemes.

[0047] After obtaining the Pareto optimal solution set, candidate project combinations that satisfy budget constraints, lower capacity constraints, and mutual exclusion constraints are screened, and three representative schemes are selected for comprehensive comparison. The results are shown in Table 2. For each representative scheme, the sum of project combination cost-effectiveness, sum of contribution, sum of importance, sum of combination risk, and sum of completion time are statistically analyzed. After normalizing each indicator, a comprehensive score is calculated based on preset weights to determine the recommended implementation scheme.

[0048] Table 2. Comprehensive Comparison Results of Candidate Project Combination Schemes

[0049] Table 2 shows that Scheme 1 is superior in terms of the total cost-effectiveness, total contribution, and total importance of the project portfolio; Scheme 3 is superior in terms of the total portfolio risk and total completion time; Scheme 2's indicators fall between those of Schemes 1 and 3. After comprehensive comparison, Scheme 1 is selected as the recommended implementation scheme. This scheme ultimately retains 11 core projects, with a total project portfolio budget of 88.78 million yuan, saving 10.02 million yuan compared to the original scheme. Therefore, the method of this invention can ensure coverage of key capabilities while balancing risk control and resource allocation efficiency.

[0050] The method proposed in this invention can be applied not only to portfolio decision-making in management projects, but also to technology research and development plans, configuration of innovation task groups, and other complex resource allocation scenarios with "capability-project" relationships.

[0051] The above description is only a preferred embodiment of the present invention. Those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and all such modifications and improvements should fall within the protection scope of the present invention.

Claims

1. A project portfolio optimization method based on heterogeneous networks, characterized in that, The process includes the following steps: S1, obtaining basic data for a candidate project set, including at least project objective information, capability requirement information, task dependency information, resource input information, and risk information, and standardizing project entities, capability entities, task entities, resource entities, and strategy entities; S2, constructing a project object relationship description model based on the standardized entity data, and building a project association network, a capability association network, and a capability-project fusion heterogeneous network based on the project object relationship description model; S3, conducting a multi-criteria comprehensive evaluation of capabilities and projects based on the capability-project fusion heterogeneous network to obtain capability importance, project support, project contribution, project cost-effectiveness ratio, backbone project identification results, critical path identification results, and project risk assessment results. S4. Based on the comprehensive evaluation results of the multi-criteria system, establish a project portfolio optimization model. The project portfolio optimization model aims at at least to maximize capacity coverage, minimize portfolio risk, and optimize the cost-effectiveness ratio of the project portfolio, and sets budget constraints, lower capacity constraints, and mutual exclusion constraints for projects. S5. Use the non-dominated sorting genetic algorithm NSGA-II to solve the project portfolio optimization model, obtain the Pareto optimal solution set, and perform a weighted score on the Pareto optimal solution set based on decision preferences to output a recommended project portfolio scheme.

2. The project portfolio optimization method according to claim 1, characterized in that, The normalization process in step S1 includes: data cleaning, invalid character removal, text segmentation and formatting, entity recognition, entity relationship recognition, node merging and attribute standardization of the original project data, and constructing a semantic entity library based on Uniform Resource Identifier (URI).

3. The project portfolio optimization method according to claim 1, characterized in that, In step S2, the Resource Description Framework (RDF) is used to describe the relationships between project objects in the form of subject, predicate, and object triples. The subject and object include at least project entities, capability entities, task entities, resource entities, and strategy entities. The predicate is used to describe support relationships, dependency relationships, containment relationships, and constraint relationships.

4. The project portfolio optimization method according to claim 3, characterized in that, The nodes of the capability-project fusion heterogeneous network include a set of project nodes and a set of capability nodes, which are used to characterize the support relationship between projects and capabilities, the dependency relationship between projects, and the correlation relationship between capabilities. Network structure features such as node connectivity, critical links, and path relationships are extracted from the project association network, capability association network, and capability-project fusion heterogeneous network to identify important nodes and critical paths.

5. The project portfolio optimization method according to claim 1, characterized in that, In step S3, the importance of capability is calculated by combining grey relational analysis and entropy weighting. Grey relational analysis is used to calculate the correlation coefficient between each capability factor and the ideal reference sequence, while entropy weighting is used to determine the weight based on the dispersion of capability index data and to rank each capability dimension according to the weighted grey relational degree.

6. The project portfolio optimization method according to claim 1, characterized in that, In step S3, the support strength of each project to each capability dimension is quantified by constructing a capability-project support matrix, and the project support degree is obtained after normalization; and the project contribution degree is obtained by aggregation based on the project support degree. At the same time, Data Envelopment Analysis (DEA) is used to calculate the input-output efficiency between project funding, human resources input and project application benefits, and technological breakthrough output, so as to obtain the project cost-effectiveness ratio.

7. The project portfolio optimization method according to claim 1, characterized in that, In step S3, the CRITIC index correlation weighting method is used to comprehensively analyze the project support, project contribution, project cost-effectiveness ratio and project efficiency to identify key projects; based on the project task dependency relationship, the Critical Path Method (CPM) is used to identify critical paths.

8. The project portfolio optimization method according to claim 1, characterized in that, In step S3, a project risk indicator system is constructed, including technical risks, resource risks, schedule risks, and external environmental risks. The Top-Ideal Solution Ranking Method (TOPSIS) is used to quantify and rank the project risk indicators to obtain the comprehensive risk assessment results of the project.

9. The project portfolio optimization method according to claim 1, characterized in that, In step S4, a binary decision variable is used to represent whether a project is selected, where a value of 1 indicates that the corresponding project is selected, and a value of 0 indicates that the corresponding project is not selected; the budget constraint is used to limit the total cost of the project portfolio to not exceed the available funds, the capability lower limit constraint is used to limit the key capabilities to reach a preset threshold, and the project mutual exclusion constraint is used to limit projects with resource conflicts or category conflicts from being selected at the same time.

10. The project portfolio optimization method according to claim 9, characterized in that, The solution process of NSGA-II described in step S5 includes: binary encoding of the project portfolio and initializing the population; performing crossover and mutation operations; performing non-dominated sorting and crowding calculation; using an elite retention strategy to retain superior individuals; and using a dynamic penalty function to constrain schemes that exceed the budget, fail to meet the capability requirements, or do not satisfy mutual exclusion constraints. After obtaining the Pareto optimal solution set, it is normalized according to capability coverage, risk level, and cost-effectiveness ratio indicators, and the comprehensive score of each Pareto solution is calculated based on the preset decision preference weights. The project portfolio with the best score is selected as the recommended scheme.

Images