Intelligent task planning method based on large model and operation planning optimization

Abstract

The invention belongs to the technical field of operation planning optimization, and discloses an intelligent task planning method based on a large model and operation planning optimization, and the method comprises the steps: obtaining a user task demand and system available resources; generating a plurality of initial planning schemes by using the large model based on user task demands, and obtaining a corresponding resource demand list; based on the available resources of the system and the resource demand list, obtaining an optimized resource allocation strategy, a sub-task scheduling plan and a corresponding performance index under each initial planning scheme; based on the obtained performance indexes, obtaining a comprehensive score of each initial planning scheme; selecting the initial planning scheme with the highest comprehensive score as a final task planning scheme, and outputting a corresponding resource allocation strategy and a sub-task scheduling plan; according to the method, through integration of a large model and operation planning optimization, efficient planning of tasks and resource optimization configuration are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of operations research and optimization, and more specifically, to an intelligent task planning method based on a large model and operations research and optimization. Background Art

[0002] In today's rapidly developing digital age, intelligent task planning plays an increasingly important role in various industries, such as project management, resource scheduling, supply chain optimization, and robotic path planning. However, with the increasing complexity of tasks, the dynamic changes in available resources, and the increasing diversity of user needs, large models are gradually being used for task planning. Existing methods usually have the following problems in intelligent task planning: First, it is difficult for large models to fully understand the complexity and multidimensionality of user task requirements, especially the user's potential expectations for solution innovation, resulting in the initial generated solutions not being diverse enough or failing to meet the user's deep-seated needs; for example, a logistics company wants to optimize the route planning of its unmanned delivery robots, and the required solution is an innovative strategy that is different from traditional path optimization algorithms and has more forward-looking and self-learning capabilities; however, because its expression is too vague, it is difficult for large models to accurately understand the user's intentions, which may cause the large model to be unable to identify the user's implicit needs for novelty or breakthroughs; second, when effectively combining the cognitive reasoning ability of large models with the precise computing power of operations optimization methods, there is often a lack of a systematic process to ensure that both Collaborative work makes it difficult to generate a solution that is both in line with user intentions and optimal in resource allocation and task scheduling; for example, the large model predicts the order volume and delivery demand trend of each region by analyzing historical data and user order information, but when combined with the operations optimization method, there is no systematic process to convert the prediction results into specific vehicle scheduling, route planning and warehouse allocation plans, resulting in unreasonable vehicle allocation, high empty rates of vehicles on some routes, and serious shortage of transportation capacity in other areas, which not only increases costs but also affects delivery timeliness; thirdly, when faced with multiple potential planning schemes, there is a lack of a comprehensive, objective and multi-index evaluation mechanism, making it difficult to select the optimal solution from numerous schemes; in view of this, the present invention proposes an intelligent task planning method based on large models and operations optimization to solve the above problems. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned objectives, the present invention provides the following technical solution: an intelligent task planning method based on a large model and operations optimization, comprising: S1. Obtain user task requirements and system available resources; S2. Based on user task requirements, use the large model to generate multiple initial planning schemes and obtain the corresponding resource requirement list; S3. Based on the system's available resources and resource requirement lists, obtain the optimal resource allocation strategy, subtask scheduling plan, and corresponding performance indicators under each initial planning scheme; S4. Based on the obtained performance indicators, obtain the comprehensive scores of each initial planning scheme; select the initial planning scheme with the highest comprehensive score as the final task planning scheme, and output the resource allocation strategy and subtask scheduling plan corresponding to the final task planning scheme.

[0004] Furthermore, user task requirements include task objectives, constraints, and innovative descriptions of solutions that need to be achieved.

[0005] Furthermore, the generation of multiple initial planning schemes includes: Decouple the mission objectives into three decision dimensions: process advancement, resource strategy, and technical architecture; Evaluate the innovation of user task requirements based on the preset innovation evaluation dimensions to obtain the innovation coefficient; Preset the characteristics and threshold range of the innovation gradient level, build an innovation coefficient-innovation gradient level mapping table, and convert the innovation coefficient into the innovation gradient level; Input the task objectives, constraints, decision dimensions and innovation gradient levels into the preset large model to generate multiple initial planning schemes.

[0006] Furthermore, the methods for obtaining the innovation coefficient include: Set corresponding weights for each innovation evaluation dimension based on the domain of user task requirements; Obtain keywords related to each innovation evaluation dimension, establish a correspondence between keywords and score ranges, and form a semantic mapping table; Extract keywords related to innovation from the solution's innovation description; match the extracted keywords with the semantic mapping table to obtain the scores corresponding to each innovation evaluation dimension; if there are multiple relevant descriptions for the same innovation evaluation dimension, calculate the average of the scores of the multiple relevant descriptions as the final score for that innovation evaluation dimension; Based on the weights and final scores of each innovation assessment dimension, the innovation coefficient is calculated through weighted aggregation.

[0007] Furthermore, preferred resource allocation strategies, subtask scheduling plans, and corresponding performance indicator acquisition methods include: Split user task requirements into independent subtasks and obtain the resource requirement data, time consumption data and dependencies between different subtasks for each subtask; based on the dependencies between subtasks, generate the initial scheduling sequence of subtasks through topological sorting method; Based on the total amount of available system resources and the resource demand data of each subtask, a total resource constraint is constructed; based on the time consumption data of the subtask, a time constraint is constructed; based on the dependency relationship between different subtasks, a task dependency constraint is constructed; Taking the shortest total completion time, lowest total cost, highest average resource utilization and least number of key resources used as multi-objective optimization goals, the initial scheduling sequence of subtasks is optimized based on the total resource constraints, time constraints and task dependency constraints using operations research optimization methods to obtain subtask scheduling plans and resource allocation strategies; based on the subtask scheduling plans and resource allocation strategies, the performance indicators of the initial planning scheme are calculated, which include total completion time, total cost, average resource utilization and number of key resources used.

[0008] Furthermore, in the process of optimizing the initial scheduling sequence of subtasks, if multiple subtasks compete for the same resource at the same time, the number of available resources in the current system and the total demand of each subtask for the resource are obtained, and the demand ratio of the total demand to the number of available resources is calculated; A ratio threshold is preset. If the demand ratio is less than or equal to the preset ratio threshold, resources are allocated from high to low based on the comprehensive influence index of the subtask; if the demand ratio is greater than the preset ratio threshold, the set of all subtasks competing for the same type of resources in the current time slice or decision cycle is regarded as an auction market, and each subtask is a bidder in the market; the negative impact of the delay of the subtask due to the failure to obtain resources on the task goal is obtained, and the cost factor is obtained; the subtask is calculated based on its own comprehensive influence index and cost factor in the resource auction, and resources are allocated from high to low based on the size of the comprehensive bid of the subtask.

[0009] Furthermore, the comprehensive influence index of the subtask is obtained in the following ways: Obtain the direct contribution of each subtask to the task goal; based on the dependency relationship between subtasks, construct a task dependency graph, and assign corresponding dependency weights to each directed edge in the task dependency graph, thereby generating a dependency matrix; based on the dependency matrix, calculate the betweenness centrality of each subtask and the sum of the dependency weights of all directed edges received by the node corresponding to each subtask to obtain the weighted in-degree of the subtask; based on the direct contribution, betweenness centrality and weighted in-degree, evaluate the influence of the subtask using fuzzy logic method to obtain the comprehensive influence index of the subtask.

[0010] Furthermore, the methods for evaluating the influence of subtasks using fuzzy logic include: The direct contribution, betweenness centrality and weighted in-degree of the subtask are fuzzified by a predefined fuzzy membership function to obtain the fuzzy membership corresponding to the direct contribution, betweenness centrality and weighted in-degree; the fuzzy membership of the fuzzified direct contribution, betweenness centrality and weighted in-degree of the subtask is inferred using a preset inference rule library to generate the comprehensive fuzzy membership of the subtask; the comprehensive fuzzy membership is defuzzified to obtain the comprehensive influence index of the subtask.

[0011] Furthermore, the comprehensive score can be obtained by: The performance indicators are normalized, and the performance indicators of all the normalized initial planning schemes are subjected to Pareto dominance analysis to identify and eliminate the initial planning schemes dominated by Pareto, and generate a set of candidate schemes; based on the candidate scheme set, an ideal optimal solution consisting of the optimal values ​​of each performance indicator and an ideal worst solution consisting of the worst values ​​of each performance indicator are constructed; combined with the preset weight coefficients of each performance indicator, the weighted optimal solution distance from the initial planning scheme in the candidate scheme set to the ideal optimal solution and the weighted inferior solution distance to the ideal worst solution are calculated; based on the weighted optimal solution distance and the weighted inferior solution distance, the relative proximity of the initial planning schemes in the candidate scheme set is calculated; the relative proximity is the comprehensive score of the initial planning scheme.

[0012] The technical effects and advantages of the present invention are as follows: At the level of understanding user needs, the present invention breaks down user task requirements into task objectives, constraints, and descriptions of solution innovation, and constructs an innovation evaluation system. This converts users' potential expectations for solution innovation into quantifiable indicators, and inputs the three decision-making dimensions of task goal decoupling into a large model to generate diverse initial planning solutions that meet deep needs. In terms of solution optimization, a system process is established that collaborates with the large model and operations optimization. Operations optimization optimizes subtask scheduling and resource allocation based on resource constraints and multi-objective optimization, and handles resource competition through a comprehensive influence index and auction mechanism to ensure reasonable resource allocation. In the solution evaluation link, based on performance indicators, a scientific and objective comprehensive scoring system is constructed through Pareto dominance analysis and relative proximity calculation of ideal solutions to select the global optimal solution. In addition, the comprehensive influence evaluation of subtasks and multi-objective optimization settings enable this method to flexibly adapt to complex scenarios in multiple fields, effectively addressing problems such as high task complexity, dynamic resource changes, and diverse user needs, significantly improving the intelligence level of task planning and resource allocation efficiency, and possessing broad application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic diagram of an intelligent task planning method based on a large model and operations optimization according to the present invention; Figure 2Schematic diagram of resource allocation of the present invention. DETAILED DESCRIPTION

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0015] Example 1 See also Figure 1 As shown, the intelligent task planning method based on a large model and operations optimization described in this embodiment includes: S1. Obtain user task requirements and system available resources; S2. Based on user task requirements, use the large model to generate multiple initial planning schemes and obtain the corresponding resource requirement list; S3. Based on the system's available resources and resource requirement lists, obtain the optimal resource allocation strategy, subtask scheduling plan, and corresponding performance indicators under each initial planning scheme; S4. Based on the obtained performance indicators, obtain the comprehensive scores of each initial planning scheme; select the initial planning scheme with the highest comprehensive score as the final task planning scheme, and output the resource allocation strategy and subtask scheduling plan corresponding to the final task planning scheme.

[0016] Furthermore, user task requirements include task objectives, constraints, and a description of the innovativeness of the solution. Constraints include the time frame for achieving the task objectives, the budget ceiling, whether it is restricted to a specific solution or implemented through a specific process, etc. The description of the innovativeness of the solution is a textual description or requirement for the novelty, uniqueness, or breakthrough of the technology, method, or strategy of the solution provided by the large model when the user submits the task requirements. The number of available system resources includes the resources that the current system can invest in completing the task, including the number of equipment, the number of personnel, the funds that can be invested, etc.

[0017] Furthermore, the generation of multiple initial planning schemes includes: Decouple the task objectives in user task requirements into three decision-making dimensions: process advancement, resource strategy, and technical architecture; Taking software project development as an example, the process advancement can be divided into different methods such as serial development and parallel development. Serial development is to advance step by step according to the established development process, while parallel development is to divide the project into different modules, carry out multiple modules simultaneously, and finally splice them together to complete the project. Resource strategy refers to the amount and method of resource investment, such as in-house development, partial outsourcing, full outsourcing, and the number of personnel and funds invested. Technical architecture refers to the technology used to implement the software, such as using mature industry technologies for development, using brand-new technologies for development, or using a mixture of old and new technologies. Taking engineering project implementation as an example, process advancement includes the implementation or construction sequence of each sub-project of the project, which may include certain dependencies, that is, the implementation of one sub-project must be dependent on the completion of another sub-project; resource strategy indicates the amount and method of resource input, such as the amount of engineering equipment, the number of construction personnel, and the number of engineering equipment and construction personnel allocated to each sub-project; technical architecture can be divided into the construction plan used to complete the project, such as using a traditional construction plan, a construction plan with process innovation, or a completely new construction plan; Based on the innovative description of the solution in the user's task requirements, the innovation of the task requirements is evaluated through five innovation evaluation dimensions: novelty, breakthrough, practicality, uniqueness, and technical difficulty, and the innovation coefficient is obtained; Preset the characteristics and threshold range of the innovation gradient level, and construct a mapping table between the innovation coefficient and the innovation gradient level, so as to convert the innovation coefficient into a specific innovation gradient level; The task objectives, constraints, decision dimensions and innovation gradient levels are submitted as input parameters to the big model (such as GPT, Deep seek, etc.); the big model adopts a generative architecture based on deep learning, and through unsupervised and supervised mixed training on massive historical project cases, industry knowledge graphs and technical documents, establishes a multi-dimensional mapping relationship between goal setting, constraints, decision dimensions, innovation gradient levels and planning schemes; when the model receives the input information, it will first pre-process the data, use the semantic parsing algorithm to extract key elements and perform structured encoding; then, based on the knowledge system and logical reasoning rules formed by the training, the big model combines natural language processing and generative adversarial network (GAN) technology to generate multiple candidate planning schemes under the three decision dimensions of process advancement, resource strategy and technical architecture; each scheme strictly adheres to the established innovation gradient level standards, and dynamically verifies the feasibility and compliance of the scheme through the constraint verification module, and finally outputs multiple initial planning schemes that are both targeted and practical.

[0018] It should be noted that the innovation gradient levels are divided into three categories, including L1 innovation: mature technology reorganization, that is, using existing and widely verified technologies to achieve goals through recombination and optimized configuration; L2 innovation: industry innovation technology application, which refers to the introduction of emerging and innovative technologies in the industry to solve problems; L3 innovation: cross-domain technology migration, introducing advanced technologies from other fields into the current field to create new solutions; the threshold range of each innovation gradient level is: L1 innovation: [0, 0.4]; L2 innovation: [035, 0.75]; L3 innovation: [0.6, 1.0]; if the innovation coefficient belongs to both L1 and L2 or L2 and L3 innovation gradient levels, the solutions corresponding to the two gradient levels will be given respectively to better meet user needs.

[0019] Furthermore, the methods for obtaining the innovation coefficient include: Set corresponding weights for each innovation evaluation dimension based on the field of user needs (such as technology, business, engineering, medical, etc.), and the sum of the weights of each dimension is 1; Obtain keywords related to each innovation evaluation dimension, combine industry terminology with semantic features, establish a correspondence between keywords and scoring intervals (0-1), and form a semantic mapping table. This semantic mapping table clarifies the scoring criteria for different keywords under the corresponding dimensions, providing a quantitative basis for subsequent text scoring. Perform word and sentence segmentation on the text describing the solution's innovation to eliminate redundant information irrelevant to the innovation assessment. Use natural language processing tools (such as NLP algorithms) or manual annotation to accurately identify and extract keywords and key descriptions related to innovation in the text, laying the foundation for dimension scoring. The extracted keywords and key descriptions are matched with the semantic mapping table to obtain the corresponding score for each innovation evaluation dimension, with a score range of 0-1. If there are multiple relevant descriptions for the same dimension, the average of the multiple description scores is calculated as the final score for that dimension. For example, if a technical solution has two descriptions in the novelty dimension, and the matching scores of the mapping table are 0.8 and 0.9 respectively, the score for that dimension is the average of the two descriptions, 0.85. Based on the preset weights of each innovation evaluation dimension and the corresponding dimension scoring results, the innovation coefficient of the user task requirement (with a value of 0-1) is calculated through weighted summary to achieve a quantitative evaluation of the text innovation.

[0020] It should be noted that the resource requirement list is generated by the large model using prompt words, and only includes the resources required to implement the initial planning scheme generated by the large model, and does not include the quantity of required resources; the prompt words are the types of resources required for the given scheme.

[0021] Furthermore, preferred resource allocation strategies, subtask scheduling plans, and corresponding performance indicator acquisition methods include: Based on user needs, use a work breakdown structure (WBS) to hierarchically break down user tasks into independently executable subtasks. Each subtask includes several steps to achieve the task, preventing over-segmentation that leads to task fragmentation (e.g., breaking down report writing into meaningless tasks like opening a document and entering a title). For example, if a user's goal is to design an app, it can be broken down into five sub-goals: product research, product planning, product development, product testing, and product deployment. Each sub-goal includes several steps to achieve the goal, without further breaking down the steps. Use large models or expert systems to obtain resource requirement data and time consumption data for each subtask, as well as the dependencies between different subtasks; Construct resource constraints based on the system's available resources to ensure that the total amount of resources allocated to subtasks does not exceed the total capacity of the system's available resources; Based on the time consumption data of subtasks, time constraints are constructed; time constraints are used to ensure that each subtask is fully executed and the completion time of all subtasks does not exceed the time constraints in the constraints of the user's task requirements; Based on the dependencies between different subtasks, task dependency constraints are constructed; task dependency constraints are used to ensure that the execution order of subtasks is logical; Based on the dependencies between subtasks, the initial scheduling sequence of subtasks is generated by topological sorting method; With the shortest total completion time, lowest total cost, highest average resource utilization and least number of key resources used as optimization goals, based on the total resource constraints, time constraints and task dependency constraints, the initial scheduling sequence of subtasks is optimized using operations optimization methods (such as particle swarm optimization algorithm, ant colony algorithm, etc.) to obtain the subtask scheduling plan and resource allocation strategy; based on the subtask scheduling plan and resource allocation strategy, the performance indicators of the initial planning scheme are calculated, including total completion time, total cost, average resource utilization and number of key resources used.

[0022] Methods for obtaining performance indicators of the initial planning scheme include: Build a scheduling Gantt chart and resource allocation table based on the subtask scheduling plan and resource allocation strategy; the scheduling Gantt chart intuitively displays the start and end time of subtasks and the parallel execution interval; the resource allocation table details the resource allocation of each subtask in each time slice; In the scheduling Gantt chart, iterate over the end times of all subtasks. If there are task execution constraints caused by dependencies, ensure that the maximum value is selected as the total completion time of the entire task while satisfying the task dependency logic. If all subtasks have no dependencies, the maximum end time can be directly selected as the total completion time. This indicator reflects the total time taken for a task from start to finish and is one of the key indicators for measuring task execution efficiency. Based on the resource allocation table, the resource usage of each subtask in each time slice is calculated based on the pre-set resource cost unit price. The cost of each subtask is accumulated in chronological order and the total cost is finally summed up to measure the economic investment of task execution. Traverse each time slice in the resource allocation table, and for each resource type, summarize the amount of resource occupied by all subtasks in the time slice to obtain the actual usage of each resource in each time slice; for each resource, in each time slice, divide the actual usage of the resource by its pre-set total capacity to obtain the utilization rate of the resource in the time slice; for example, if the total CPU capacity is 8 cores and the actual CPU usage in the TS1 time slice is 5 cores, then the CPU utilization rate of the TS1 time slice is 62.5%; for each resource, take the arithmetic average of its utilization rate in all time slices, that is, add up the utilization rate of each time slice and divide it by the total number of time slices to obtain the average utilization rate of the resource in the entire task cycle; summarize the average utilization rate of all resources to form a resource utilization data set; correct the data through weighted averaging, standardization, etc., and finally obtain the average resource utilization rate used to measure the utilization rate of all resources; In the resource allocation table, each time slice is traversed in chronological order. For each resource type, the total demand for the resource by all subtasks in the time slice is compared with the pre-set available capacity. If the total demand for a resource in a time slice exceeds its available capacity, the resource corresponding to the time slice is determined to be a critical resource, and the time slice and resource type are recorded. If the total demand does not exceed the available capacity, the resource is not a critical resource in the time slice. After traversing all time slices and resource types, the recorded critical resource entries are deduplicated and counted, and the final count result is the number of critical resources used. This indicator can accurately locate the links where resource supply and demand are unbalanced and there is tension and pressure during task execution, providing a key basis for subsequent optimization of resource allocation.

[0023] Furthermore, if Figure 2 As shown in the figure, during the process of optimizing the initial scheduling sequence of subtasks, if multiple subtasks compete for the same resource at the same time, the resource allocation method is as follows: Obtain the number of available resources R in the current system and the total resource demand D of each subtask; calculate the demand ratio C of the total demand D to the number of available resources R; A preset ratio threshold (between 1.1 and 1.3, reflecting the intensity of competition for resources) is set. If the demand ratio C is less than or equal to the preset ratio threshold, resources are allocated based on the comprehensive influence index of the subtasks. If the demand ratio C is greater than the preset ratio threshold, the resources are allocated as follows: The set of all subtasks competing for the same type of resource R in the current time slice (or decision cycle) is regarded as a bidding market, and each subtask is a bidder in the market; Obtain the negative impact of the delay on the task goal caused by the failure to obtain resources, and obtain the cost factor; the function expression of the cost factor is: ;in, Indicates the The cost factor paid by each subtask to obtain resources; Indicates the The total float time of each subtask is obtained from the result of the above topological sorting algorithm; Indicates the The estimated time required to complete each subtask; Indicates the The direct contribution of each subtask to the task goal; The subtask gives a comprehensive bid in the resource auction based on its own comprehensive influence index and cost factor ; ; and is the preset weight coefficient, Indicates the The comprehensive influence index of each subtask; Resources are allocated from high to low based on the comprehensive bids of subtasks. Each subtask will obtain all the resources it needs until the resources are exhausted. If the comprehensive bids of multiple subtasks are the same, resources will be allocated first to the subtask with a higher comprehensive influence index.

[0024] Furthermore, the comprehensive influence index information of the subtask is obtained by: Decompose OKR or KPI to obtain the direct contribution of each subtask to the task goal (0-1); Based on the dependencies between subtasks, a task dependency graph is constructed. The task dependency graph is a directed acyclic graph (DAG). Nodes in the task dependency graph represent subtasks and carry information about the subtask's comprehensive influence index. Subtasks with dependencies are connected by directed edges to show the execution order and parallelization possibility between subtasks. The large model assigns a dependency weight of 1-10 to each directed edge in the task dependency graph (1 for a weak dependency and 10 for a strong dependency). Based on the dependency weights of the directed edges, a corresponding dependency matrix is ​​generated. The element values ​​in the matrix are the dependency weights of the edges in the task dependency graph; an element value of 0 indicates that there is no dependency relationship. Based on the dependency matrix, we use tools such as Gephi to obtain the betweenness centrality of each subtask (with a value range of 0-1). We also calculate the sum of the weights of all edges received by the node corresponding to each subtask to obtain the weighted in-degree of the subtask. Based on the direct contribution, betweenness centrality, and weighted in-degree, we use fuzzy logic to evaluate the influence of the subtask and obtain the comprehensive influence index of the subtask. The methods for evaluating the influence of subtasks using fuzzy logic include: Membership functions are defined for direct contribution, betweenness centrality, and weighted in-degree respectively. The membership function maps a specific numerical indicator (such as direct contribution of 0.7) to one or more fuzzy memberships (between 0 and 1). For example, for the direct contribution indicator, it is defined as follows: Fuzzy set: {low, medium, high}; in the membership function, low is defined as 1 when the direct contribution is less than or equal to 0.3, and then decreases to 0; medium is defined as 0 when the membership is 0 at 0.2, 1 at 0.5, and 0 at 0.8; high is defined as 0 when the contribution is greater than or equal to 0.7, and then increases to 1; The direct contribution, betweenness centrality and weighted in-degree are fuzzified by the predefined fuzzy membership function to obtain the fuzzy membership corresponding to the direct contribution, betweenness centrality and weighted in-degree. For example, if the direct contribution of subtask A is 0.8, its membership to "high" contribution may be 0.9, its membership to "medium" contribution may be 0.1, and its membership to "low" contribution may be 0. Establish rules in the IF-THEN format and store them in the inference rule library. The IF-THEN format rules are used to describe how different indicator combinations affect the comprehensive influence of subtasks. These rules are determined by domain experts or through data-driven methods, for example: IF (direct contribution IS high) AND (betweenness centrality IS medium) THEN (comprehensive influence IS strong); IF (direct contribution IS medium) AND (weighted in-degree IS high) THEN (comprehensive influence IS medium); IF (betweenness centrality IS is low) THEN (comprehensive influence IS is weak); IF (direct contribution IS low) OR (weighted in-degree IS low) THEN (comprehensive influence IS weak); Based on the preset inference rule library, the inference operation is performed on the membership of the fuzzified direct contribution, betweenness centrality, and weighted in-degree. Specifically, for each rule, the minimum value of the input indicator membership is taken as the activation strength. For example, in the rule "IF (direct contribution IS high) AND (betweenness centrality IS medium) THEN (comprehensive influence IS strong)", if the membership of direct contribution is "high" is 0.9 and the membership of betweenness centrality is "medium" is 0.6, then the activation strength of this rule is 0.6. For all activation rules with the same output term (such as "strong"), the maximum strength is selected as the final membership of this term. After completing the reasoning operation, the fuzzy outputs generated by all rule reasoning (multiple subsets representing strong, medium, and weak comprehensive influence) are merged to generate a comprehensive fuzzy membership; a defuzzification process (such as the center of gravity method, the maximum membership average method, etc.) is used to convert the aggregated fuzzy output set into an accurate value, which is the comprehensive influence index of the subtask; the higher the comprehensive influence index of all subtasks, the more important and critical the subtask is in the task process.

[0025] Furthermore, the comprehensive score can be obtained by: Determine the weight coefficient of each performance indicator through the Delphi method or AHP; The total completion time, total cost, average resource utilization and number of key resources are processed using a normalization method to obtain corresponding normalized indicators, thereby ensuring that each indicator is measured on a unified scale; Perform Pareto dominance analysis on the normalized performance indicators: If there exists an initial planning scheme A whose performance indicators are not inferior to those of the initial planning scheme B, and at least one indicator is strictly superior to that of scheme B, then scheme A dominates scheme B; all dominated initial planning schemes are eliminated to generate a set of candidate schemes; based on the set of candidate schemes, an ideal optimal solution and an ideal worst solution for each performance indicator are constructed, where the ideal optimal solution is composed of the optimal value of each performance indicator in the candidate scheme set, and the ideal worst solution is composed of the worst value of each performance indicator in the candidate scheme set; Based on the weight coefficients of each performance indicator, the weighted optimal solution distance and the weighted inferior solution distance of the initial planning scheme in the candidate solution set to the ideal optimal solution and the ideal worst solution are calculated; based on the weighted optimal solution distance and the weighted inferior solution distance, the relative proximity of the initial planning scheme in the candidate solution set is calculated; the relative proximity is the comprehensive score of the initial planning scheme; The initial planning scheme with the highest comprehensive score is determined as the final task planning scheme, and the resource allocation strategy and task scheduling plan corresponding to the final task planning scheme are output.

[0026] At the level of understanding user needs, the present invention breaks down user task requirements into task objectives, constraints, and descriptions of solution innovation, and constructs an innovation evaluation system. This converts users' potential expectations for solution innovation into quantifiable indicators, and inputs the three decision-making dimensions of task goal decoupling into a large model to generate diverse initial planning solutions that meet deep needs. In terms of solution optimization, a system process is established that collaborates with the large model and operations optimization. Operations optimization optimizes subtask scheduling and resource allocation based on resource constraints and multi-objective optimization, and handles resource competition through a comprehensive influence index and auction mechanism to ensure reasonable resource allocation. In the solution evaluation link, based on performance indicators, a scientific and objective comprehensive scoring system is constructed through Pareto dominance analysis and relative proximity calculation of ideal solutions to select the global optimal solution. In addition, the comprehensive influence evaluation of subtasks and multi-objective optimization settings enable this method to flexibly adapt to complex scenarios in multiple fields, effectively addressing problems such as high task complexity, dynamic resource changes, and diverse user needs, significantly improving the intelligence level of task planning and resource allocation efficiency, and possessing broad application value.

[0027] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will be able to modify the technical solutions described in the foregoing embodiments or to substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

[0028] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0029] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0030] In the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0031] In the description of the present invention, “several” means one or more, and “a large number” means two or more.

[0032] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0033] The formulas in this manual are all dimensionless and calculated using numerical values. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field based on actual conditions.

[0034] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. An intelligent task planning method based on large models and operations optimization, characterized in that: include: S1. Obtain user task requirements and system available resources; S2. Based on user task requirements, use the large model to generate multiple initial planning schemes and obtain the corresponding resource requirement list; S3. Based on the system's available resources and resource requirement lists, obtain the optimal resource allocation strategy, subtask scheduling plan, and corresponding performance indicators under each initial planning scheme; S4. Obtaining a comprehensive score for each initial planning scheme based on the obtained performance indicators; The initial planning scheme with the highest comprehensive score is selected as the final task planning scheme, and the resource allocation strategy and subtask scheduling plan corresponding to the final task planning scheme are output.

2. The intelligent task planning method based on large model and operations optimization according to claim 1 is characterized in that: User task requirements include task objectives, constraints, and innovative descriptions of solutions.

3. The intelligent task planning method based on large model and operations optimization according to claim 2 is characterized in that: Methods for generating multiple initial planning schemes include: Decouple the mission objectives into three decision dimensions: process advancement, resource strategy, and technical architecture; Evaluate the innovation of user task requirements based on the preset innovation evaluation dimensions to obtain the innovation coefficient; Preset the characteristics and threshold range of the innovation gradient level, build an innovation coefficient-innovation gradient level mapping table, and convert the innovation coefficient into the innovation gradient level; Input the task objectives, constraints, decision dimensions and innovation gradient levels into the preset large model to generate multiple initial planning schemes.

4. The intelligent task planning method based on large model and operations optimization according to claim 3 is characterized in that: The methods for obtaining the innovation coefficient include: Set corresponding weights for each innovation evaluation dimension based on the domain of user task requirements; Obtain keywords related to each innovation evaluation dimension, establish a correspondence between keywords and score ranges, and form a semantic mapping table; Extract keywords related to innovation from the solution's innovation description; match the extracted keywords with the semantic mapping table to obtain the scores corresponding to each innovation evaluation dimension; if there are multiple relevant descriptions for the same innovation evaluation dimension, calculate the average of the scores of the multiple relevant descriptions as the final score for that innovation evaluation dimension; Based on the weights and final scores of each innovation assessment dimension, the innovation coefficient is calculated through weighted aggregation.

5. The intelligent task planning method based on large model and operations optimization according to claim 4 is characterized in that: The preferred resource allocation strategy, subtask scheduling plan, and corresponding performance indicator acquisition methods include: Split user task requirements into independent subtasks and obtain the resource requirement data, time consumption data and dependencies between different subtasks for each subtask; based on the dependencies between subtasks, generate the initial scheduling sequence of subtasks through topological sorting method; Based on the total amount of available system resources and the resource demand data of each subtask, a total resource constraint is constructed; based on the time consumption data of the subtask, a time constraint is constructed; based on the dependency relationship between different subtasks, a task dependency constraint is constructed; Taking the shortest total completion time, lowest total cost, highest average resource utilization and least number of key resources used as multi-objective optimization goals, the initial scheduling sequence of subtasks is optimized based on the total resource constraints, time constraints and task dependency constraints using operations research optimization methods to obtain subtask scheduling plans and resource allocation strategies; based on the subtask scheduling plans and resource allocation strategies, the performance indicators of the initial planning scheme are calculated, which include total completion time, total cost, average resource utilization and number of key resources used.

6. The intelligent task planning method based on large model and operations optimization according to claim 5 is characterized in that: During the process of optimizing the initial scheduling sequence of subtasks, if multiple subtasks compete for the same resource at the same time, the number of available resources in the current system and the total demand of each subtask for the resource are obtained, and the ratio of the total demand to the number of available resources is calculated; A preset ratio threshold is set. If the demand ratio is less than or equal to the preset ratio threshold, resources are allocated from high to low based on the comprehensive influence index of the subtasks. If the demand ratio is greater than the preset ratio threshold, the set of all subtasks competing for the same type of resources in the current time slice or decision cycle is regarded as a bidding market, and each subtask is a bidder in the market; Obtain the negative impact of the delay in obtaining subtasks due to failure to obtain resources on the task goal and obtain the cost factor; the subtask is calculated based on its own comprehensive influence index and cost factor to obtain the comprehensive bid in the resource auction, and resources are allocated from high to low based on the size of the comprehensive bid of the subtask.

7. The intelligent task planning method based on large model and operations optimization according to claim 6 is characterized in that: The methods for obtaining the comprehensive influence index of a subtask include: Obtain the direct contribution of each subtask to the task goal; based on the dependency relationship between subtasks, construct a task dependency graph, and assign corresponding dependency weights to each directed edge in the task dependency graph, thereby generating a dependency matrix; based on the dependency matrix, calculate the betweenness centrality of each subtask and the sum of the dependency weights of all directed edges received by the node corresponding to each subtask to obtain the weighted in-degree of the subtask; based on the direct contribution, betweenness centrality and weighted in-degree, evaluate the influence of the subtask using fuzzy logic method to obtain the comprehensive influence index of the subtask.

8. The intelligent task planning method based on large model and operations optimization according to claim 7 is characterized in that: The methods for evaluating the influence of subtasks using fuzzy logic include: The direct contribution, betweenness centrality and weighted in-degree of the subtask are fuzzified by a predefined fuzzy membership function to obtain the fuzzy membership corresponding to the direct contribution, betweenness centrality and weighted in-degree; the fuzzy membership of the fuzzified direct contribution, betweenness centrality and weighted in-degree of the subtask is inferred using a preset inference rule library to generate the comprehensive fuzzy membership of the subtask; the comprehensive fuzzy membership is defuzzified to obtain the comprehensive influence index of the subtask.

9. The intelligent task planning method based on large model and operations optimization according to claim 8 is characterized in that: The comprehensive score can be obtained by: The performance indicators are normalized, and the performance indicators of all the normalized initial planning schemes are subjected to Pareto dominance analysis to identify and eliminate the initial planning schemes dominated by Pareto, and generate a set of candidate schemes; based on the candidate scheme set, an ideal optimal solution consisting of the optimal values ​​of each performance indicator and an ideal worst solution consisting of the worst values ​​of each performance indicator are constructed; combined with the preset weight coefficients of each performance indicator, the weighted optimal solution distance from the initial planning scheme in the candidate scheme set to the ideal optimal solution and the weighted inferior solution distance to the ideal worst solution are calculated; based on the weighted optimal solution distance and the weighted inferior solution distance, the relative proximity of the initial planning schemes in the candidate scheme set is calculated; the relative proximity is the comprehensive score of the initial planning scheme.

Images