CN-121998041-A - Operation planning control method, device, equipment and medium based on data analysis

CN121998041ACN 121998041 ACN121998041 ACN 121998041ACN-121998041-A

Abstract

The invention relates to the technical field of machine learning, and discloses an operation planning control method, device, equipment and medium based on data analysis, which comprise the following steps: extracting user intention information in an operation preparation optimization consultation text input by a user, filling content into a preset model frame according to the user intention information to obtain a large language model of the operation preparation optimization consultation text, converting the large language model into executable codes, executing the executable codes, correcting the large language model according to an execution result if the executable codes fail to execute, returning to the step of converting the large language model into the executable codes, and generating an optimization strategy of the operation preparation optimization problem according to the execution result and a verification result if the executable codes succeed in executing. The invention improves the universality of the operation planning control method based on data analysis.

Inventors

ZHANG YIFAN
SHAN JINXIAO
CHEN XIANLI

Assignees

招商局金融科技有限公司

Dates

Publication Date: 20260508
Application Date: 20251230

Claims (10)

1. An operational scenario control method based on data analysis, comprising: Acquiring a task plan scheme preset by a user, acquiring operation planning optimization consultation text corresponding to the task plan scheme, and extracting user intention information in the operation planning optimization consultation text, wherein the user intention information comprises a target optimization direction of the task plan scheme, constraint conditions during optimization of the task plan scheme and decision variables of the task plan scheme; converting the target optimization direction, the constraint condition and the decision variable into model parameters, and carrying out parameter assignment on a preset large language model frame by utilizing the model parameters to obtain a large language model; converting the large language model into executable code and executing the executable code; if the executable code fails to execute, acquiring error reporting information in the executing process, carrying out parameter adjustment on the large language model according to the error reporting information and the target optimization direction, and returning to the step of converting the large language model into the executable code; if the executable code is successfully executed, outputting a constraint condition and a decision variable which are currently corresponding to the large language model, and generating an operation planning optimization strategy of the task planning scheme according to the output constraint condition and the decision variable; And carrying out operation planning optimization adjustment on the task plan scheme according to the operation planning optimization strategy to obtain an optimized task scheme.
2. The operation planning control method based on data analysis according to claim 1, wherein the extracting the user intention information in the operation planning optimization consultation text includes: word segmentation processing is carried out on the operation research optimization consultation text to obtain a word segmentation text set; Labeling the quantity words, units and entity nouns in the word segmentation text set to obtain a standardized text fragment set; Identifying phrases representing variables to be determined in the standardized text segment set by using a preset formatting template to obtain decision variables; Identifying phrases representing maximum and minimum intentions in the standardized text fragment set by using the formatting template to obtain a target optimization direction; identifying phrases representing the constraint conditions in the standardized text segment set by using the formatting template to obtain constraint conditions; and summarizing the decision variables, the target optimization direction and the constraint conditions to obtain user intention information.
3. The method for controlling an operation planning scheme based on data analysis according to claim 1, wherein the converting the target optimization direction, the constraint condition and the decision variable into model parameters, and performing parameter assignment on a preset large language model frame by using the model parameters to obtain a large language model comprises: extracting decision variables contained in the user intention information, and defining variable types based on keywords contained in the decision variables and context content associated with the keywords to obtain a defined decision variable set; Combining a target optimization direction contained in the user intention information and a measurement index associated with the target optimization direction into a complete mathematical expression to obtain a target function; Converting constraint conditions contained in the user intention information into mathematical conditional expressions to obtain constraint condition sets in mathematical forms; And converting the decision variable set, the objective function and the constraint condition set in the mathematical form into a data format of preset model parameters to obtain model parameters, and assigning the converted model parameters to a preset model frame to obtain a large language model.
4. The operational scenario control method based on data analysis of claim 1, wherein said converting said large language model into executable code comprises: identifying variables, coefficients, operators and function invokers in the large language model to obtain a symbol set; Matching the symbol set with a preset structure contained in a preset solver, and identifying and marking fixed parameters, variable parameters and decision variable type parameters in the large language model according to a matching result to obtain a structured model description object; identifying the model type of the large language model according to the structured model description object, and taking a solver with highest matching score between the feature of the solver in a preset solver library and the model type as a target solver according to the model type; And filling the structured model description object into a preset code template by using the target solver to obtain an executable code.
5. The method for controlling an operational scenario based on data analysis according to claim 1, wherein before said parameter adjustment is performed on said large language model according to said error reporting information and said target optimization direction, said method further comprises: acquiring an execution result of the executable code; extracting key information in the execution result and structuring and outputting the key information to obtain structured information; Performing constraint feasibility verification on the structured information based on a constraint condition set of a mathematical form in the large language model to obtain a constraint verification result; carrying out service logic rationality verification on the structured information by using a preset service rule to obtain a service logic verification result; and when the structural information is judged to pass the constraint feasibility verification and the business logic rationality verification according to the constraint verification result and the business logic verification result, the execution result is judged to pass the feasibility verification, the execution feedback of the executable code is obtained, and whether the executable code is successfully executed is judged according to the execution feedback. And when the structural information is judged to not pass the constraint feasibility verification and the business logic rationality verification according to the constraint verification result and the business logic verification result, judging that the execution result does not pass the feasibility verification.
6. The operational scenario control method based on data analysis of claim 5, wherein said determining whether said executable code is executed successfully based on said execution feedback comprises: Extracting parameter fields in the execution feedback to obtain a parameter field set; Identifying a boolean type field in the set of parameter fields; According to the same Boolean type field as a preset successful execution feedback field; If the executable codes are the same, judging that the executable codes are successfully executed; if not, judging that the executable codes fail to execute.
7. The method for controlling an operational scenario based on data analysis according to claim 1, wherein said generating an operational scenario optimization strategy of said mission planning scenario according to output constraints and decision variables comprises: Converting the constraint condition and the decision variable into structured data; Mapping the structured data into service terms according to a preset parameter mapping table; Mapping the numerical results in the structured data into business meanings; Carrying out business semantic analysis on the execution result according to the business terms and the business meaning analysis based on a preset business rule base to obtain a business interpretation result; And calling a preset service template to generate a structured optimization strategy according to the business interpretation result.
8. An operational scenario control device based on data analysis, comprising: The data acquisition module is used for acquiring a task plan scheme preset by a user, acquiring an operation planning optimization consultation text corresponding to the task plan scheme and extracting user intention information in the operation planning optimization consultation text, wherein the user intention information comprises a target optimization direction of the task plan scheme, constraint conditions during optimization of the task plan scheme and decision variables of the task plan scheme; The model construction module is used for converting the target optimization direction, the constraint condition and the decision variable into model parameters, and carrying out parameter assignment on a preset large language model frame by utilizing the model parameters to obtain a large language model; the execution feedback module is used for converting the large language model into an executable code and executing the executable code, acquiring error reporting information in an execution process if the executable code fails to execute, carrying out parameter adjustment on the large language model according to the error reporting information and the target optimization direction, and returning to the step of converting the large language model into the executable code; The strategy generation module is used for outputting constraint conditions and decision variables corresponding to the large language model currently and generating an operation planning optimization strategy of the task planning scheme according to the output constraint conditions and decision variables; and the operation planning optimization module is used for carrying out operation planning optimization adjustment on the task planning scheme according to the operation planning optimization strategy to obtain an optimized task scheme.
9. Computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the data analysis based operational scenario control method according to any one of claims 1 to 7 when the computer program is executed.
10. A computer-readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the data analysis-based operation planning control method according to any one of claims 1 to 7.

Description

Operation planning control method, device, equipment and medium based on data analysis Technical Field The present invention relates to the field of machine learning technologies, and in particular, to a method, an apparatus, a device, and a medium for controlling an operation planning scheme based on data analysis. Background The operation planning optimization is an important branch of application mathematics and system science, is widely applied to the fields of resource allocation, production planning, logistics scheduling and the like, and has the core aim of modeling and solving a complex system on the premise of meeting a series of constraint conditions so as to realize optimal decisions such as income maximization, cost minimization and the like. Traditionally, an operation preparation optimization problem is solved with high dependency on field experts and programmers, wherein the operation preparation experts firstly analyze the real problem described by natural language manually, abstract and construct a formal mathematical large language model (comprising decision variables, objective functions and constraint conditions), and then the programmers manually write calling codes for specific solvers according to the model. The manual driving process has the inherent defects of long development period, high labor cost, high technical threshold and the like, severely limits the operation optimization technology, and particularly ensures that the rapid construction and application of complex models are difficult for vast middle and small enterprises to effectively utilize the advanced optimization technology to improve the operation efficiency. In recent years, a large language model shows strong natural language understanding and code generating capability, and brings new opportunities for automatic solution of operation planning optimization problems. Prior art approaches have attempted to use large language models to directly translate a user's natural language problem descriptions into a mathematical model or even executable code. Such methods achieve automated linking from natural language to solution processes to some extent. However, the existing automation schemes based on large language models have the core limitation of insufficient universality, namely, the existing automation schemes generally depend on specific templates, preset structures or limited problem patterns, and are difficult to flexibly adapt to various business scenes, morphological constraint conditions and complicated and changeable optimization targets of various industries. Once the practical industrial problems of nonstandard expression, logic interleaving or deep domain knowledge are faced, the generated model often lacks necessary service laminating property and expansibility, so that the migration and multiplexing cost of the solution is high, and the general requirements of wide users in diversified and personalized scenes cannot be really met. Disclosure of Invention The invention provides an operation planning control method, device, computer equipment and medium based on data analysis, which are used for solving the problem of low universality of the existing operation planning optimization method in the current market. In a first aspect, an operation planning control method based on data analysis is provided, including: Acquiring a task plan scheme preset by a user, acquiring operation planning optimization consultation text corresponding to the task plan scheme, and extracting user intention information in the operation planning optimization consultation text, wherein the user intention information comprises a target optimization direction of the task plan scheme, constraint conditions during optimization of the task plan scheme and decision variables of the task plan scheme; converting the target optimization direction, the constraint condition and the decision variable into model parameters, and carrying out parameter assignment on a preset large language model frame by utilizing the model parameters to obtain a large language model; converting the large language model into executable code and executing the executable code; if the executable code fails to execute, acquiring error reporting information in the executing process, carrying out parameter adjustment on the large language model according to the error reporting information and the target optimization direction, and returning to the step of converting the large language model into the executable code; if the executable code is successfully executed, outputting a constraint condition and a decision variable which are currently corresponding to the large language model, and generating an operation planning optimization strategy of the task planning scheme according to the output constraint condition and the decision variable; And carrying out operation planning optimization adjustment on the task plan scheme according to the operation planning optimization