Large model AI integrated intelligent production scheduling system

Abstract

The invention discloses a large model AI integrated intelligent production scheduling system, and relates to the technical field of production scheduling, and the system comprises the steps: obtaining real-time production characteristics, matching a demand module set, and recording the demand module set as a target module set; extracting a first module in the target module set, and matching a first HTML-CSS component package; learning the first HTML-CSS component package, and checking to obtain a scheduling priority prediction model; obtaining first demand feature information of the first module, and performing prediction analysis on the first demand feature information through a scheduling priority prediction model to obtain a first scheduling priority index; performing descending sorting on the target module set to obtain a target module descending list; and performing intelligent production scheduling for the real-time production category according to the descending list of the target modules. The technical problem that code module matching and production scheduling efficiency is low in the prior art is solved, and the technical effect of improving production efficiency and accurate scheduling of task priority is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling, and in particular to a large-model AI-integrated intelligent production scheduling system. Background Art

[0002] As the level of informatization of government services continues to improve, traditional government service processing methods have gradually exposed problems such as inefficiency and waste of resources. Especially in terms of program development and code writing, programmers need to manually write and schedule a lot of code to meet the needs of various government services. At present, the tasks and requirements involved in government services are complex and changeable, and each user's needs have different processing procedures, resulting in the inability of traditional production scheduling methods to effectively respond to dynamic changes, and there are problems such as inaccurate scheduling and unreasonable resource allocation. Summary of the invention

[0003] The present application provides a large-model AI-integrated intelligent production scheduling system, which is used to solve the technical problems of code module matching and low production scheduling efficiency in the prior art.

[0004] In view of the above problems, the present application provides a large-model AI-integrated intelligent production scheduling system.

[0005] The present application provides a large-model AI-integrated intelligent production scheduling system, the system comprising: A real-time production feature acquisition module, the real-time production feature acquisition module acquires the real-time production features of the real-time production category, and matches the demand module set based on the real-time production features, and records the demand module set as the target module set; a matching module, the matching module extracts the first module in the target module set, and matches the first HTML-CSS component package of the first module in a predetermined component knowledge base; a learning and verification module, the learning and verification module introduces a production scheduling integration strategy to learn the first HTML-CSS component package, and verifies to obtain a scheduling priority prediction model; a prediction and analysis module, the prediction and analysis module acquires the first demand feature information of the first module, and predicts and analyzes the first demand feature information through the scheduling priority prediction model to obtain a first scheduling priority index; a descending sorting module, the descending sorting module sorts the target module set in descending order based on the first scheduling priority index to obtain a descending list of target modules; a production scheduling module, the production scheduling module performs intelligent production scheduling for the real-time production category according to the descending list of target modules.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: The present application obtains the real-time production characteristics of the real-time production category, and matches the demand module set based on the real-time production characteristics, and records the demand module set as the target module set; extracts the first module in the target module set, and matches the first HTML-CSS component package of the first module in the predetermined component knowledge base; introduces the production scheduling integration strategy to learn the first HTML-CSS component package, and verifies to obtain the scheduling priority prediction model; obtains the first demand feature information of the first module, and predicts and analyzes the first demand feature information through the scheduling priority prediction model to obtain the first scheduling priority index; sorts the target module set in descending order based on the first scheduling priority index to obtain a descending list of target modules; and performs intelligent production scheduling for the real-time production category according to the descending list of target modules. The present invention solves the technical problems of low code module matching and production scheduling efficiency in the prior art, and accurately matches user needs with edited code modules through automatic extraction and analysis of real-time production characteristics, combined with historical data and rule information, and quickly generates the required production tasks, thereby achieving the technical effect of accurate scheduling that improves production efficiency and task priority. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0008] Figure 1 A schematic diagram of the structure of a large-model AI-integrated intelligent production scheduling system provided in an embodiment of the present application; Figure 2 A flow chart of a large-model AI-integrated intelligent production scheduling method provided in an embodiment of the present application.

[0009] Explanation of the accompanying drawings: real-time production feature acquisition module 11, matching module 12, learning and verification module 13, prediction and analysis module 14, descending sorting module 15, production scheduling module 16. DETAILED DESCRIPTION

[0010] This application provides a large-model AI-integrated intelligent production scheduling system to solve the technical problems of code module matching and low production scheduling efficiency in the prior art. Through automatic extraction and analysis of real-time production features, combined with historical data and rule information, it accurately matches user needs with edited code modules and quickly generates the required production tasks, thereby achieving the technical effect of accurate scheduling that improves production efficiency and task priority.

[0011] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0012] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0013] Examples, such as Figure 1 As shown, the embodiment of the present application provides a large-model AI-integrated intelligent production scheduling system for executing Figure 2 A large-model AI-integrated intelligent production scheduling method is shown, and the large-model AI-integrated intelligent production scheduling system includes: The real-time production feature acquisition module 11 acquires the real-time production features of the real-time production category, matches the required module set based on the real-time production features, and records the required module set as the target module set.

[0014] In the embodiment of the present application, real-time production characteristics are first obtained, which include real-time user type characteristics (such as unit users or individual users) and real-time service standard characteristics (such as basic elements of services, processing rules and service cycles).

[0015] Next, according to the preset production feature vectorization plan, the real-time production features are converted into vector format for subsequent processing and comparison. At the same time, the production record logs of any production category are extracted from the historical production records. These historical records include different production features as the basis for comparison with the real-time features. Among them, the production feature vectorization plan is the encoding feature vectorization plan.

[0016] Then, the real-time production features and historical production features are vectorized one by one to generate real-time production vectors and arbitrary production vectors respectively. By comparing these two vectors, the production consistency index is calculated, which is used to measure the matching degree between real-time production features and historical production features. The production consistency index is the coding consistency index.

[0017] When the production consistency index reaches the predetermined threshold, the required module set that matches the real-time characteristics is extracted from the historical production records and added to the candidate module set. Finally, the candidate module set is deduplicated and reorganized to remove duplicate or irrelevant modules, and finally an accurate required module set is obtained, which is the target module set required in the subsequent scheduling process.

[0018] Furthermore, the system provided in the application embodiment also includes: The real-time production features include real-time user type features and real-time service standard features, and the real-time user type features refer to unit users or individual users, and the real-time service standard features at least include basic elements, processing rules, and service cycles.

[0019] In the embodiment of the present application, the real-time production features include real-time user type features and real-time service standard features. These features are important input data in the automatic coding task, through which the system can identify and match the appropriate task module.

[0020] The real-time user type feature is used to distinguish the types of users, which are divided into corporate users and individual users. Corporate users represent a company or organization, and their needs are usually more complex, involving operations of multiple users or departments; while individual users refer to individual consumers, whose needs are relatively simple, usually only involving personal accounts or services.

[0021] The real-time service standard characteristics describe the basic requirements and standards for service provision, including basic elements, processing rules and service cycles. Basic elements refer to the basic components or requirements of services, such as necessary information or preconditions that users need to provide. Processing rules are the specific operating procedures and rules of services, such as how to handle user requests under specific circumstances. For example, after a user submits an application, a decision is made based on the processing rules whether to approve the application. The service cycle is the time requirement or restriction of the service, which usually includes the start and end time of the service, or the number of operations that can be provided within the service cycle. For example, provident fund withdrawal services may only be applied for once a year, and this type of information is part of the service cycle.

[0022] Furthermore, in the system provided in the embodiment of the application, the real-time production feature acquisition module 11 is also used for: Read the production feature vectorization plan; obtain any production record log of any production category in the historical production record, wherein the arbitrary production category corresponds to any production feature; sequentially vectorize the real-time production feature and the arbitrary production feature based on the production feature vectorization plan to obtain a real-time production vector and an arbitrary production vector respectively; compare the real-time production vector with the arbitrary production vector to obtain a production consistency index; when the production consistency index meets a predetermined index threshold, extract any production requirement module set in the arbitrary production record log and add it to the candidate module set; deduplicate and sort the candidate module set to obtain the requirement module set.

[0023] In the embodiment of the present application, the production feature vectorization plan is first read, which is a set of preset rules and methods for converting real-time production features and historical production features into numerical vector representations. Specifically, the real-time production features (such as user type, service standard, etc.) are converted into numerical vectors using feature engineering methods, which are usually achieved through standardization, encoding processing, etc.

[0024] Next, we obtain historical production records from the historical production database. These records include log information of multiple production categories and related features. For example, a historical record may include a request for an individual user to withdraw housing provident fund, including the user's basic information, service standards, processing rules, etc. Each historical record reflects the specific needs and execution status of past production.

[0025] After the historical records are extracted, the real-time production features and historical production features are vectorized according to the feature vectorization plan. At this time, the real-time production features and historical production features are converted into numerical vectors through the vectorization algorithm to form real-time production vectors and arbitrary production vectors.

[0026] Then, the cosine similarity method is used to calculate the production consistency index. Cosine similarity is a common method to measure the similarity between two vectors, with a value range of -1 to 1. When the value is 1, it means that they are completely consistent, and when the value is -1, it means that they are completely opposite. By comparing the real-time production vector and the historical production vector, a numerical value can be calculated to represent the similarity between the two vectors, namely the production consistency index.

[0027] When the production consistency index reaches the predetermined index threshold, the data extraction method is used to extract the required module sets that meet the conditions from the historical production records. These module sets correspond to the service tasks or code modules in the historical records that best match the real-time production characteristics. The extracted modules will be temporarily stored in the candidate module set for further screening. The predetermined index threshold is pre-set.

[0028] Finally, the candidate module set is deduplicated using a hash deduplication algorithm to ensure that the final required module set does not contain duplicate modules. Hash deduplication generates a unique hash value for each module to determine whether it is a duplicate item. If the hash values ​​are the same, it is considered a duplicate module and automatically removed. Finally, the required module set is obtained through the above process.

[0029] A matching module 12 is configured to extract a first module from the target module set and match a first HTML-CSS component package of the first module in a predetermined component knowledge base.

[0030] In the embodiment of the present application, the matching module first randomly extracts a module from the generated demand module set, which is called the first module. The first module represents a specific functional unit that needs to be completed in the current task, such as user data input processing or page layout generation.

[0031] Next, search the pre-defined component knowledge base to find the HTML-CSS component package related to the first module. The pre-defined component knowledge base stores various defined components, which include the HTML and CSS codes required for front-end display, used to generate page layout, style and interactive elements. Each component has a corresponding description and function definition in the knowledge base. For example, the "user form component package" may include multiple form fields and data validation functions. The system will match these components according to the functional requirements of the first module.

[0032] The matching process is usually based on a content matching algorithm, that is, comparing the functional requirements of the first module with the functions in the HTML-CSS component package. For example, if the first module is about the coding function of the user input interface, the system will search and match a form component package suitable for users to fill in information in the knowledge base. In this way, the system can quickly find component packages that meet the requirements according to the module requirements, including "user form component package" or "multi-field data validation component package". This matching process is based on the module's requirement description and the functional attributes of the component package, and intelligent recommendations are made through semantic similarity or keyword matching. Through this process, the first HTML-CSS component package of the first module is obtained. This component package not only meets the requirements of the current task, but also is consistent with the automated coding logic of the digital innovation engine. For example, if the goal is to generate a page for users to fill in information, the matching component package may contain elements such as "text box" and "drop-down list", which are directly applied to page generation.

[0033] The learning and verification module 13 introduces a production scheduling integration strategy to learn the first HTML-CSS component package and verifies and obtains a scheduling priority prediction model.

[0034] In the embodiment of the present application, the workflow of the learning and verification module aims to optimize the scheduling priority of production tasks by introducing a production scheduling integration strategy, wherein the production scheduling integration strategy is an automatic encoding task scheduling integration strategy.

[0035] Specifically, the first HTML-CSS component package is first divided into data to obtain the first division result. On this basis, supervised learning is performed on each data group to train multiple learners. Subsequently, these learners are integrated and fused in combination with the production scheduling integration strategy, and finally a scheduling priority prediction model is constructed to predict the scheduling priorities of different tasks.

[0036] Furthermore, in the system provided in the embodiment of the application, the learning verification module 13 is also used for: The first HTML-CSS component package is divided into data to obtain a first division result; supervised learning is performed on any data group in the first division result, and any learner is obtained by verification; and the any learner is integrated and fused according to the production scheduling integration strategy to obtain the scheduling priority prediction model.

[0037] In an embodiment of the present application, data is first divided for the first HTML-CSS component package to obtain a first division result. The first HTML-CSS component package contains various front-end component information related to the production task, such as page layout, style elements, interactive controls, etc. These components have different functions in the coding task, such as user form input, button interaction, page display, etc. Among them, the production task refers to the automatic coding task. Through data division, the information in the component package is decomposed into multiple sub-data groups, such as user forms, button controls, display areas, etc. This division process can be carried out according to the function and usage scenario of the component to ensure that each sub-data group represents a different functional module or attribute in the component package.

[0038] Next, supervised learning is performed on each data group extracted from the first partition result. Supervised learning refers to the use of labeled historical data to train the model so that it can predict the output results based on the input features. In this process, historical production data (such as previous page loading time, user interaction records, page response data, etc.) will be used as training data, and a series of learners are trained through these historical data. Each learner can predict the corresponding production scheduling output, such as page rendering time, priority, etc., based on the input component features, such as layout type, interaction complexity, style configuration, etc.

[0039] After completing supervised learning, multiple learners are integrated and fused using the production scheduling integration strategy. The production scheduling integration strategy aims to combine the prediction results of multiple learners to improve the accuracy and robustness of the final prediction. Specifically, the module can select the prediction results of multiple learners and fuse their outputs according to the strategy through integration methods such as weighted average and voting. For example, if multiple learners predict different priority values, the priority with the highest frequency is selected as the final scheduling result. Through this integration strategy, a more accurate scheduling priority prediction model is obtained, thereby improving the accuracy and efficiency of production scheduling.

[0040] Finally, by integrating the outputs of multiple learners, a unified scheduling priority prediction model is obtained.

[0041] Furthermore, in the system provided in the embodiment of the application, the learning verification module 13 is also used for: Analyze the arbitrary data group to obtain an arbitrary priority coefficient; train the training data group composed of the arbitrary data group and the arbitrary priority coefficient to obtain an arbitrary priority learner; take the mean of the arbitrary priority coefficient as the target scheduling priority index; use the arbitrary priority coefficient as the input data of the meta-learner, and use the target scheduling priority index as the output data of the meta-learner to train the meta-learner; build the arbitrary priority learner and the meta-learner according to the production scheduling integration strategy to obtain the scheduling priority prediction model.

[0042] In the embodiment of the present application, firstly, an arbitrary data group is analyzed to obtain an arbitrary priority coefficient. Wherein, the arbitrary data group includes arbitrary production data and arbitrary rule information. The arbitrary production data includes data and rule information related to the automatic encoding task. When analyzing these data groups, the module extracts a first constraint from the arbitrary rule information, and the constraint corresponds to the first rule in the encoding task rule.

[0043] Then, it is determined whether the current real-time production category satisfies the first constraint. If this constraint is satisfied, the first rule is added to the priority list; if this constraint is not satisfied, the first rule is added to the non-priority list. Next, the ratio of the number of rules in the priority list to the total number of rules (i.e., the sum of the number of rules in the priority list and the non-priority list) is calculated, and this ratio is recorded as the arbitrary priority coefficient.

[0044] Next, feature data related to the auto-encoding task is extracted from any data group (including any auto-encoding task data and any rule information), and combined with the calculated arbitrary priority coefficient to construct a training data group. The training data group contains multiple samples, each of which includes relevant information about the production task (such as task type, resource requirements, etc.) and the calculated arbitrary priority coefficient. These data will be used as training sets for the learning model. Then, supervised learning methods are used to train these training data to generate an arbitrary priority learner. In this process, labeled data (such as the priority of historical tasks, actual resource scheduling, etc.) are used to train the model, and the priority of the task is predicted through methods such as regression analysis or classification models. Ultimately, the arbitrary priority learner will be able to predict the priority of the task based on new input data (such as new production task information).

[0045] After obtaining multiple arbitrary priority coefficients, these coefficients are averaged. Specifically, the priority coefficients of all tasks are first calculated, which reflect the degree to which the tasks meet the priority rules. By averaging all priority coefficients, the target scheduling priority index is obtained.

[0046] Next, the calculated arbitrary priority coefficient is used as the input data of the meta-learner, and the target scheduling priority index is used as the output data of the meta-learner. The meta-learner is a high-level learning model that can learn from the prediction results of multiple low-level learners and finally output a comprehensive prediction result. At this time, the arbitrary priority coefficient (as input) and the target scheduling priority index (as output) constitute the training data, and the prediction performance of the meta-learner is optimized by training. Usually, the training process uses ensemble learning methods, such as random forests, gradient boosting trees, etc., to integrate the results of multiple low-level learners. After training, the meta-learner can generate more accurate scheduling priority predictions based on the priority coefficients.

[0047] Finally, according to the production scheduling integration strategy, the arbitrary priority learner and the meta-learner are integrated and fused. The integration process adopts weighted average to synthesize the prediction results of different learners to improve the accuracy and robustness of the prediction. By integrating multiple learners, the final scheduling priority prediction model is generated.

[0048] Furthermore, the system provided in the application embodiment also includes: The arbitrary data set includes arbitrary production data and arbitrary rule information.

[0049] In the embodiment of the present application, any data group includes any production data and any rule information. Any production data refers to task data related to the automatic coding task, which includes program codes for implementing specific government functions. For example, a code module may be used to extract a user's provident fund payment record, or to process services such as user identity verification. These code modules are pre-written program codes for implementing a certain government service, involving functions such as data extraction, verification, and processing.

[0050] Arbitrary rule information refers to rules related to government services, which stipulate specific conditions or requirements that must be met for tasks, including service conditions, service restrictions, and execution order. For example, a service may require rules such as "individual users can only withdraw housing provident funds once a year" or "users must make continuous deposits for one year before they can withdraw."

[0051] Furthermore, the system provided in the application embodiment is also used for: Extract the first constraint in the arbitrary rule information, where the first constraint corresponds to the first rule; determine whether the real-time production category has the first constraint; if so, add the first rule to the priority list; if not, add the first rule to the non-priority list; take the ratio of the number of priority rules in the priority list to the total number of rules, recorded as the arbitrary priority coefficient, wherein the total number of rules refers to the sum of the number of rules in the priority list and the number of rules in the non-priority list.

[0052] In the embodiment of the present application, a rule in any rule information is first randomly extracted, which is called the first rule. The first rule contains a specific first constraint, that is, a condition or requirement related to the current task. For example, a service may have specific rules, such as users must meet the requirement of "continuously paying provident fund for more than one year" before applying for withdrawal. This extraction process uses a random algorithm to randomly select a rule from the predefined rule set for processing.

[0053] Next, it is determined whether the current real-time production category meets the extracted first constraint. The real-time production category refers to the current automatic coding task type, such as urgent tasks, ordinary tasks, etc. The determination of whether the first constraint is met is usually performed by comparing the task attributes with the first constraint conditions, such as checking whether the user's deposit record meets a specific time limit or whether it meets other constraints.

[0054] If the judgment result is "compliant", the first rule is added to the priority list, indicating that the rule is applicable to the current task and needs to be executed first. If the judgment result is "non-compliant", the first rule is added to the non-priority list, indicating that the rule is not applicable to the current task and can be processed later.

[0055] Then count the number of rules in the priority list (i.e., the number of priority rules), and calculate the total number of rules, which is the sum of all rules in the priority list and non-priority list. By calculating the ratio of the number of priority rules to the total number of rules, we get the arbitrary priority coefficient. The arbitrary priority coefficient is used to quantify the degree to which the task complies with the priority rule. It is a value between 0 and 1. The larger the value, the more the task complies with the priority rule.

[0056] For example, suppose 10 rules are randomly extracted, 6 of which are judged to meet the priority conditions of the task, and 4 are not. Then, the number of priority rules is 6, the total number of rules is 10, and the final arbitrary priority coefficient is 0.6.

[0057] The prediction and analysis module 14 obtains the first demand feature information of the first module, and performs prediction and analysis on the first demand feature information through the scheduling priority prediction model to obtain a first scheduling priority index.

[0058] In an embodiment of the present application, the prediction and analysis module first obtains the first demand characteristic information from the first module. The first demand characteristic information refers to specific data or parameters related to the current task or service, such as the user's personal information, the type of service, the urgency of the request, etc. This information is automatically extracted through user requests or task descriptions. The first module plays a role in connecting user needs with specific tasks in this process, and its content includes the basic requirements of the task and related resource requirements. Next, the first demand characteristic information of the first module is input into the scheduling priority prediction model for prediction analysis to obtain the first scheduling priority index.

[0059] The descending order sorting module 15 sorts the target module set in descending order based on the first scheduling priority index to obtain a descending list of target modules.

[0060] In the embodiment of the present application, the descending sorting module first uses the first scheduling priority index as the basis for sorting. The first scheduling priority index is the task priority score calculated by the prediction and analysis module in the previous stage based on the demand feature information of the task or service (such as user demand, task urgency, etc.). A higher scheduling priority index indicates that the task has a higher priority and needs to be executed earlier.

[0061] Next, the descending sorting module sorts the target module set according to the first scheduling priority index, and sorts the modules from high to low priority, that is, the modules with high priority are arranged in front, and the modules with low priority are arranged in the back.

[0062] Finally, the target module set is rearranged in descending order to generate a descending list of target modules. This list represents the execution order of all tasks or services.

[0063] The production scheduling module 16 performs intelligent production scheduling for the real-time production category according to the descending list of target modules.

[0064] In the embodiment of the present application, the production scheduling module starts scheduling for the real-time production category based on the priority order in the descending list of target modules. The real-time production category represents the type of task that needs to be processed currently, such as "housing fund withdrawal coding task", and selects the appropriate task module to execute according to the characteristics of the task.

[0065] During this process, the production scheduling module intelligently allocates processing resources according to the priority of each module. Tasks with higher priorities are processed quickly, while tasks with lower priorities are postponed. For example, if the real-time production category is to handle "housing fund withdrawal business" for a certain user, the production scheduling module will first schedule task modules related to housing fund withdrawal, such as user identity authentication module and account information query module. Since these modules have higher priorities, they will be ranked at the top of the descending list of target modules, and computing resources will be allocated to these tasks first to ensure that the tasks can be completed quickly and efficiently.

[0066] Through this intelligent scheduling method, the production scheduling module not only ensures that high-priority tasks can be processed in a timely manner, but also reasonably adjusts the execution order of other low-priority tasks according to resource demand and allocation, thereby optimizing resource usage. This scheduling method ensures the smooth completion of key tasks such as provident fund withdrawal business, improves the processing efficiency of the overall system, and enables automatic coding tasks to be processed efficiently and orderly according to priority.

[0067] Furthermore, in the system provided in the embodiment of the application, the production scheduling module 16 is also used for: Extract the top module in the descending list of the target modules; obtain a test user, wherein the test user has an identification of a test production category; generate a test production scheduling result of the test production category through the top module; provide services for the test user according to the test code obtained based on the test production scheduling result to obtain a test service record; manually verify the test service record to obtain a test verification result; when the test verification result meets the predetermined service constraints of the test production category, determine the scheduling of the top module.

[0068] In the embodiment of the present application, the top module is first extracted from the target module descending list. The target module descending list has been sorted according to the scheduling priority prediction model through the previous steps, and the modules with high priority are arranged in front. Therefore, the top module represents the most urgent and priority task module at present.

[0069] Next, obtain a trial user with the "trial production category" identifier. Trial users are used to test the effect of production scheduling. The production tasks of these users will be used as experimental data to verify the actual effect of the scheduling plan. By screening out qualified trial users from the user database, the applicability and accuracy of scheduling are ensured.

[0070] Then, the top module generates the test production scheduling results for the test production category. This scheduling result will be generated through an intelligent scheduling algorithm based on the needs and task requirements of the test users, covering task priorities, required resources, time arrangements, etc. The scheduling process will ensure that tasks are scheduled according to the predetermined priority and resources are optimally allocated.

[0071] Then, according to the generated test production scheduling result, corresponding services are provided to the test user, specifically, corresponding services or tasks are provided to the test user based on the generated scheduling plan. At this time, a test service record is recorded and generated as the actual situation of the execution of the scheduling plan.

[0072] The trial service records are then manually verified. The purpose of manual verification is to verify whether the scheduling results meet the predetermined service constraints and requirements. Through manual verification, possible scheduling problems or services that do not meet the regulations are identified to ensure the correctness and effectiveness of the scheduling results.

[0073] Finally, if the test verification results meet the predetermined service constraints of the test production category, the top module is scheduled. This means that through actual testing, the scheduling results have achieved the expected effect and meet the service standards, so the scheduling plan for this module can be finally confirmed and applied to the formal production scheduling.

[0074] Furthermore, in the system provided in the embodiment of the application, the production scheduling module 16 is also used for: When the test verification result does not meet the predetermined service constraint of the test production category, extract the next top module in the descending list of target modules; and perform a definite scheduling analysis on the next top module.

[0075] In the embodiment of the present application, when the test verification result does not meet the predetermined service constraint, the adjustment mechanism is activated to extract the next top module in the descending list of target modules, wherein the next top module refers to the module ranked second only to the top module.

[0076] After obtaining the sub-top module, a scheduling analysis is performed on it. The scheduling analysis process includes detailed analysis and processing based on the task characteristics of the sub-top module. A new scheduling plan is generated based on the specific requirements of each task, such as required resources, time limits, priorities, etc. At the same time, the feasibility of the sub-top module under different constraints is evaluated, and the optimal execution order and resource allocation method are determined by simulating different scheduling scenarios. The goal of this stage is to solve the reasons for the failure of the top module scheduling as much as possible to ensure that the task can be successfully completed under the new scheduling plan.

[0077] After completing the scheduling analysis of the sub-top module, a new scheduling plan is generated based on the analysis results and verified again. If the new scheduling plan meets the predetermined service constraints and the requirements of the trial production category, this plan will be adopted and the subsequent scheduling will be executed; if the conditions are still not met, adjustments will continue until a scheduling plan that meets the requirements is found.

[0078] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects: The present application obtains the real-time production characteristics of the real-time production category, and matches the demand module set based on the real-time production characteristics, and records the demand module set as the target module set; extracts the first module in the target module set, and matches the first HTML-CSS component package of the first module in the predetermined component knowledge base; introduces the production scheduling integration strategy to learn the first HTML-CSS component package, and verifies to obtain the scheduling priority prediction model; obtains the first demand feature information of the first module, and predicts and analyzes the first demand feature information through the scheduling priority prediction model to obtain the first scheduling priority index; sorts the target module set in descending order based on the first scheduling priority index to obtain a descending list of target modules; and performs intelligent production scheduling for the real-time production category according to the descending list of target modules. The present invention solves the technical problems of low code module matching and production scheduling efficiency in the prior art, and accurately matches user needs with edited code modules through automatic extraction and analysis of real-time production characteristics, combined with historical data and rule information, and quickly generates the required production tasks, thereby achieving the technical effect of accurate scheduling that improves production efficiency and task priority.

[0079] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0080] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0081] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A large-model AI-integrated intelligent production scheduling system, characterized in that: include: A real-time production feature acquisition module, wherein the real-time production feature acquisition module acquires the real-time production features of the real-time production category, matches the required module set based on the real-time production features, and records the required module set as the target module set; A matching module, wherein the matching module extracts a first module from the target module set and matches a first HTML-CSS component package of the first module in a predetermined component knowledge base; A learning and verification module, wherein the learning and verification module introduces a production scheduling integration strategy to learn the first HTML-CSS component package and verifies and obtains a scheduling priority prediction model; A prediction and analysis module, wherein the prediction and analysis module obtains the first demand feature information of the first module, and performs prediction and analysis on the first demand feature information through the scheduling priority prediction model to obtain a first scheduling priority index; A descending order sorting module, wherein the descending order sorting module sorts the target module set in descending order based on the first scheduling priority index to obtain a descending list of target modules; A production scheduling module performs intelligent production scheduling for the real-time production category according to the descending list of target modules.

2. According to the large-model AI integrated intelligent production scheduling system of claim 1, it is characterized in that: The real-time production features include real-time user type features and real-time service standard features, and the real-time user type features refer to unit users or individual users, and the real-time service standard features at least include basic elements, processing rules, and service cycles.

3. According to the large-model AI integrated intelligent production scheduling system of claim 2, it is characterized in that: The real-time production feature acquisition module includes: Read the production feature vectorization plan; Obtain any production record log of any production category in the historical production records, wherein the any production category corresponds to any production feature; Sequentially performing vector processing on the real-time production feature and the arbitrary production feature based on the production feature vectorization plan to obtain a real-time production vector and an arbitrary production vector respectively; Comparing the real-time production vector with the arbitrary production vector to obtain a production consistency index; When the production consistency index meets the predetermined index threshold, extracting any production requirement module set in the any production record log and adding it to the candidate module set; The candidate module set is deduplicated and reorganized to obtain the required module set.

4. According to the large-model AI integrated intelligent production scheduling system of claim 1, it is characterized in that: The learning verification module comprises: Performing data division on the first HTML-CSS component package to obtain a first division result; Performing supervised learning on any data group in the first division result, and verifying and obtaining any learner; According to the production scheduling integration strategy, the arbitrary learners are integrated, fused and constructed to obtain the scheduling priority prediction model.

5. According to the large-model AI integrated intelligent production scheduling system of claim 4, it is characterized in that: The learning verification module comprises: Analyzing the arbitrary data set to obtain an arbitrary priority coefficient; Training a training data group formed by the arbitrary data group and the arbitrary priority coefficient to obtain an arbitrary priority learner; The average of the arbitrary priority coefficients is taken as the target scheduling priority index; Taking the arbitrary priority coefficient as input data of a meta-learner, taking the target scheduling priority index as output data of the meta-learner, and training to obtain the meta-learner; The arbitrary priority learner and the meta learner are constructed according to the production scheduling integration strategy to obtain the scheduling priority prediction model.

6. According to the large-model AI integrated intelligent production scheduling system of claim 5, it is characterized in that: The arbitrary data set includes arbitrary production data and arbitrary rule information.

7. According to claim 6, a large-model AI integrated intelligent production scheduling system is characterized in that: Analyzing the arbitrary data set to obtain an arbitrary priority coefficient includes: Extracting a first constraint from the arbitrary rule information, where the first constraint corresponds to a first rule; Determining whether the real-time production category meets the first constraint; If it exists, add the first rule to the priority list; if it does not exist, add the first rule to the non-priority list; The ratio of the number of priority rules in the priority list to the total number of rules is taken as the arbitrary priority coefficient, wherein the total number of rules refers to the sum of the number of rules in the priority list and the non-priority list.

8. According to the large-model AI integrated intelligent production scheduling system of claim 1, it is characterized in that: The production scheduling module further includes: Extracting the top module in the descending list of target modules; Acquire a test user, wherein the test user has an identification of a test production category; Generate a test production scheduling result of the test production category through the top module; Providing services to the test user according to the test code obtained based on the test production scheduling result, and obtaining a test service record; Manually verify the test service record to obtain a test verification result; When the test verification result meets the predetermined service constraint of the test production category, the top module is scheduled for determination.

9. According to the large-model AI integrated intelligent production scheduling system of claim 8, it is characterized in that: Manually verifying the test service record to obtain a test verification result, further comprising: When the test verification result does not meet the predetermined service constraint of the test production category, extracting the next top module in the descending list of target modules; A determination scheduling analysis is performed on the sub-top module.

Images