Production planning processing method, system, device, and storage medium

Abstract

The present disclosure provides a production plan processing method, system, device and storage medium, relates to the technical field of production plan processing, in particular to the field of production plan processing method. The method comprises the following steps: receiving an adjustment instruction and parsing to obtain a business board to be adjusted, and generating a production plan to be processed based on the business board; wherein the adjustment instruction is used to indicate the business board to be adjusted; based on the prompt word template of the large model and the production plan to be processed, a tool calling sequence is generated by using the large model generation tool; based on the business data and the tool calling sequence, a production plan modification scheme is generated by calling a processing tool.

Description

Technical Field

[0001] This disclosure relates to the field of production planning and processing technology, and more specifically, to a production planning and processing method, system, device, and storage medium. Background Technology

[0002] Production plan adjustments primarily rely on manual intervention combined with Business Intelligence (BI) reports and Excel spreadsheets. The entire process requires manual data retrieval, calculation integration, analysis of upstream and downstream impacts, and manual plan modifications. This comprehensive adjustment process is time-consuming and increases linearly with the number of business segments. Furthermore, manual operation is prone to data errors and cannot quickly respond to urgent adjustment needs. The manual model also leads to strong decision-making subjectivity; different decision-makers may choose completely different adjustment directions, while expert systems and SAP Integrated Business Planning (SAP IBP), with their fixed rules, cannot dynamically handle complex scenarios.

[0003] Furthermore, tools for database querying, data calculation, plan modification, and version archiving are independent of each other. Data transfer requires manual copying and pasting, which is cumbersome and prone to inconsistencies. The manual mode cannot automatically calculate the impact of cross-section linkages, and it is easy to miss key constraints, making the solution unfeasible. Version management relies on manual naming and archiving, which is difficult to trace and cannot be rolled back, resulting in high trial and error costs. The solution generation process lacks user interaction, cannot meet personalized adjustment needs, and users cannot verify the rationality of the solution during the adjustment process, resulting in poor flexibility. Summary of the Invention

[0004] The purpose of this disclosure is to provide a production planning processing method, system, device, and storage medium to solve the aforementioned technical problems.

[0005] To achieve the above objectives, this disclosure provides a production planning processing method, including: The system receives adjustment instructions and parses them to obtain the business segment to be adjusted, and generates a production plan to be processed based on the business segment; wherein, the adjustment instructions are used to indicate the business segment to be adjusted. Based on the prompt word template of the large model and the production plan to be processed, the tool call sequence is generated using the large model; Based on business data and the tool's call sequence, the processing tool is invoked to generate a production plan modification scheme.

[0006] According to another aspect of this disclosure, the process of receiving adjustment instructions and parsing them to obtain the business segment to be adjusted, and generating a production plan to be processed based on the business segment, includes: The system receives an input adjustment instruction and parses it to obtain the identifier of the business segment to be adjusted and the adjustment period in the instruction; wherein the adjustment instruction includes at least one of natural language instructions and structured instructions. Based on the identifier of the business segment and the adjustment cycle, generate the priority result of the business segment; Receive confirmation instructions for the priority result, and generate a production plan to be processed based on the confirmation instructions.

[0007] According to another aspect of this disclosure, the prompt word template based on the large model and the production plan to be processed, using the large model generation tool to call the sequence, includes: Generate prompt word templates based on a large model; Based on the prompt word template and the production plan to be processed, the sequence is invoked through the large model generation tool; The tool call sequence and the prompt word template are used to confirm the tool call.

[0008] According to another aspect of this disclosure, the process of generating a production plan modification scheme based on business data and the tool call sequence includes: The tool calls a sequence to extract the corresponding data table for the business data; the data table includes at least one of the following: basic information table, production plan table, revenue data table, resource dependency table, version management table, and constraint rule table. Based on this business data, calculate the average utilization rate and maximum utilization rate of each business segment, and generate a ranking result for the business segment based on the average utilization rate and maximum utilization rate; Based on the sorting result, the processing tool is invoked to generate the production plan modification proposal.

[0009] According to another aspect of this disclosure, the method further includes: The revenue data for the production plan modification was extracted using this processing tool; Based on this revenue data, the revenue target achievement rate of the modified production plan is verified using this large model, and the verification results are obtained. The version identifier of the production plan modification scheme is updated based on the verification result.

[0010] According to another aspect of this disclosure, based on the revenue data, the revenue target achievement rate of the production plan modification scheme is verified through the large model to obtain the verification results, including: Based on this revenue data, the average and maximum utilization rates of the business sub-segments in the business segment of the production plan modification scheme are calculated using this large model. Based on this revenue data, the resource dependencies of this business sub-segment are extracted using this large model; Based on this revenue data, the resource gap for this business sub-segment is calculated using this large model; the resource gap includes both direct and indirect resource gaps. The revenue target achievement rate is calculated based on the average utilization rate, the maximum utilization rate, the resource dependency, and the resource gap; The revenue target achievement rate of the modified production plan was verified using this large model, and the verification results were obtained.

[0011] According to another aspect of this disclosure, a production planning processing system is provided, the equipment comprising: The interaction layer is used to receive adjustment instructions and parse the business segment to be adjusted, generate a production plan to be processed based on the business segment, and output the production plan to be processed to the orchestration layer; the adjustment instructions are used to indicate the business segment to be adjusted. This orchestration layer communicates and connects with the interaction layer and the tool layer respectively. It is used to generate a tool call sequence based on the prompt word template of the large model and the production plan to be processed, and output it to the tool layer. The data layer, which communicates with the tool layer, is used to output stored business data to the tool layer. The tool layer communicates with the data layer and the orchestration layer respectively. It is used to call the processing tool to generate a production plan modification scheme based on the business data and the tool call sequence, and then return it to the orchestration layer.

[0012] According to another aspect of this disclosure, the interaction layer includes: An input interface is used to receive input adjustment instructions and parse them to obtain the identifier of the business segment to be adjusted and the adjustment period in the adjustment instructions; wherein, the adjustment instructions include at least one of natural language instructions and structured instructions; The analysis interface is used to receive the identifier of the business segment and the adjustment period from the input interface, generate the priority result of the business segment, and send it to the confirmation interface. The confirmation interface is used to receive confirmation instructions for the priority result from the analysis interface, and generate a production plan to be processed based on the confirmation instructions and send it to the orchestration layer.

[0013] According to another aspect of this disclosure, the arrangement layer includes: The prompt word module is used to generate prompt word templates based on the large model; the prompt word templates include process guidance logic templates, tool call rule templates, data parsing rule templates, user interaction templates, and confirmation templates; The tool orchestration module is used to generate a tool call sequence based on the prompt word template and the production plan to be processed, using the large model. The confirmation module is used to confirm the tool call based on the tool call sequence and the prompt word template.

[0014] Through the above technical solution, the system automatically parses the business segments to be adjusted in response to adjustment commands and generates production plans to be processed. Based on a large-scale model's prompt word template, it automatically generates tool call sequences, eliminating the need for manual arrangement of tool call logic, thus lowering the operational threshold and planning costs. By invoking processing tools based on business data and tool call sequences to generate production plan modification schemes, it achieves automated production plan adjustments driven by business data. This supports precise modifications based on business segment dimensions and utilization indicators, avoiding the biases and inefficiencies caused by manual adjustments, and improving the response speed and accuracy of production plan adjustments.

[0015] Other features and advantages of this disclosure will be described in detail in the following detailed description section. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the following detailed description to explain the present disclosure, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic flowchart of a production planning processing method according to an embodiment of the present disclosure.

[0017] Figure 2 This is a schematic flowchart of a production planning processing method according to another embodiment of the present disclosure.

[0018] Figure 3 This is a schematic flowchart of a production planning processing method according to another embodiment of the present disclosure.

[0019] Figure 4 This is a schematic flowchart of a production planning processing method according to another embodiment of the present disclosure.

[0020] Figure 5 This is a schematic flowchart of a production planning processing method according to another embodiment of the present disclosure.

[0021] Figure 6 This is a schematic diagram of the execution flow of a production planning process.

[0022] Figure 7 This is an overall architecture diagram of a production planning and processing system.

[0023] Figure 8 This is a schematic diagram of the structure of a production planning and processing system according to an embodiment of the present disclosure. Detailed Implementation

[0024] The specific embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit this disclosure.

[0025] In this disclosure, unless otherwise stated, directional terms such as "up," "down," "left," and "right" are generally relative to the normal arrangement of the device; "inner" and "outer" refer to the inner and outer sides relative to the outline boundary of the structure itself; and "far" and "near" refer to the distance relative to a specified reference position.

[0026] In existing technologies, production plan adjustments can be completed manually or in combination with BI and Excel tools. The entire process requires manual data retrieval, calculation integration, and plan modification, resulting in a lengthy process that increases linearly with the number of business segments. Furthermore, manual operation is prone to data errors and cannot quickly respond to urgent adjustments; different decision-makers using manual operation may choose completely different adjustment directions, leading to strong subjectivity in decision-making. Expert systems with fixed rules and SAP IBP cannot dynamically handle complex scenarios.

[0027] Tools for database queries, data calculations, plan modifications, and version archiving operate independently, requiring manual copying and pasting for data transfer. This process is cumbersome and prone to inconsistencies. The manual mode cannot automatically calculate the impact of cross-sectoral interactions, easily overlooking key constraints and preventing the solution from being implemented.

[0028] Version management relies on manual naming and archiving, which is difficult to trace and cannot be rolled back, resulting in high trial and error costs; the solution generation process lacks user interaction, cannot meet personalized adjustment needs, and users cannot verify the rationality of the solution during the adjustment process, resulting in poor flexibility.

[0029] In a production planning adjustment scheme that combines manual guidance with BI system assistance: The first step is to manually receive the group's revenue adjustment target (such as a revenue increase of 20 billion yuan in 2026) and clarify the direction of the adjustment.

[0030] The second step involves manually filtering the current production plan data, historical output data, maximum capacity data, and capacity utilization statistics of business segments such as coal, power, chemicals, and transportation through the preset reports of BI systems (such as Tableau, Power Business Intelligence, and Power BI). The data needs to be integrated and calculated again through Excel.

[0031] The third step involves manually comparing data from different sectors and prioritizing sectors based on personal experience (for example, if one subjectively believes that the power sector has great growth potential, then the power sector will be prioritized for adjustment). There is no unified rule for prioritizing sectors.

[0032] The fourth step involves manual analysis of the potential impact of priority sector adjustments on upstream and downstream businesses (e.g., increased power production requires more coal consumption, necessitating manual querying of coal production capacity), lacking a systematic mechanism for calculating the interconnected impact.

[0033] The fifth step involves manually modifying the production plan data. After modification, the document is manually saved as a version record, without automated verification or draft management functions.

[0034] The scheme lacks a standardized analytical chain in its decision-making logic, relies on subjective human judgment, and results in significant differences in decision-making outcomes among different personnel.

[0035] The solution uses tools such as data analysis systems, Excel, and planning systems, but each tool operates independently without unified arrangement, and data flow still requires manual intervention.

[0036] This solution lacks automated upstream and downstream constraint accounting capabilities and relies solely on manual experience for judgment, which can easily lead to the plan failing to be implemented.

[0037] This solution only achieves basic record keeping through manual archiving and cannot realize draft creation, version backtracking, or multi-version comparison.

[0038] In a business system called SAP IBP (Integrated Business Planning System), the system supports cross-segment production planning collaboration. Its core functions include capacity planning, demand forecasting, and inventory management, and plan optimization is achieved through preset module configurations.

[0039] However, in the collaborative configuration part, cross-segment collaboration requires manual pre-configuration of relationships, which is complex and cannot dynamically adapt to business changes; the optimization solution is executed directly after generation without simulation and step-by-step verification mechanisms, which is risky; it is based on module configuration and fixed algorithms, and lacks the ability to perform natural language interaction and tool orchestration for large models; the operation threshold is high, requiring professional personnel to perform module configuration and maintenance.

[0040] In summary, the disadvantages of the existing technology are: (1) The analysis efficiency is low and the response speed is slow. Current technology (a combination of manual and BI system operation) relies entirely on manual operation: manually querying BI reports takes 1-2 hours, integrating and calculating Excel data takes 30-60 minutes, manually analyzing the impact on upstream and downstream processes takes 1-2 hours, and the complete adjustment process takes at least 3-5 hours, with the time increasing linearly as the number of business segments increases. Furthermore, manual statistical calculations are prone to data entry errors and formula errors. According to industry statistics, the error rate of manually-led production planning adjustments is approximately 8%-12%, making it impossible to quickly respond to urgent adjustments needed to revenue targets.

[0041] (2) Lack of standardized technical logic and strong subjectivity in decision-making The combination of manual and BI models relies on personal experience to determine adjustment priorities, and different decision-makers may choose completely different adjustment directions (for example, decision-maker A prioritizes adjusting section A, while decision-maker B prioritizes adjusting section B), resulting in a lack of consistency in planned adjustments; although expert systems and SAP IBP have fixed rules, the rules are rigid and cannot cope with complex scenarios, resulting in insufficient scientific decision-making.

[0042] (3) Tools are scattered and lack coordination, and the operation process is cumbersome. The core tools (database query, data calculation, plan modification, and version archiving) are independent of each other. For example, users need to first query data through BI reports, then export it to Excel for secondary processing and analysis, then modify the plan in the planning system, and finally manually save the version. The entire process involves 3-4 independent tools, and data transfer requires manual copying and pasting, which is cumbersome and prone to data inconsistency. In addition, the tools have a high learning curve, and ordinary business personnel cannot complete them independently.

[0043] (4) Lack of cross-sectoral linkage analysis, resulting in poor feasibility of the proposed solutions. Existing technologies cannot automatically calculate the impact of adjustment plans on upstream and downstream sectors. For example, in a manual-BI combined model, when decision-makers improve power generation plans, they need to consider the capabilities of each link in the chain and manually query data such as coal production capacity, railway transportation capacity, and port loading and unloading capacity. This can easily lead to the omission of key constraints (such as railway capacity gaps), resulting in plans that cannot be implemented. In contrast, SAP IBP can partially consider linkages, but it requires manual configuration of relationships in advance and cannot dynamically identify new constraints (such as transportation restrictions caused by policy impacts), making the feasibility of the plan insufficient.

[0044] (5) Inadequate version management and high risk of adjustments Existing version management technologies rely on manual archiving. For example, in a manual-BI combined model, modified production plans are entered into the planning system and saved, and each version is independently named, which is easy to miss and makes it impossible to quickly trace the adjustment history. There is no draft version mechanism, and once the adjustment plan is implemented, it cannot be rolled back. If an error occurs, it will directly affect production operations, and the cost of trial and error is extremely high.

[0045] (6) Rigid interaction methods and low user engagement The existing solution generation process lacks user participation, the fixed module configuration cannot meet the user's personalized adjustment needs, and users cannot verify the rationality of the solution during the adjustment process. They can only find problems after execution, resulting in poor flexibility.

[0046] This disclosure provides a production planning processing method that adopts a layered architecture design. Through pre-set prompts and logical judgments in a large model, it achieves automated orchestration of tool call order, data transmission, and exception handling, eliminating the need for manual intervention in tool switching. Combined with interactive guidance, it realizes a closed-loop process.

[0047] Figure 1 This is a flowchart illustrating a production planning processing method according to an embodiment of the present disclosure, the method comprising: S110. Receive the adjustment instruction and parse it to obtain the business segment to be adjusted, and generate a production plan to be processed based on the business segment; wherein, the adjustment instruction is used to indicate the business segment to be adjusted; In step S110, a business segment can refer to an enterprise's operating units divided by business type, product line, or region. The business segment to be adjusted can refer to the business unit or operating segment that requires production plan adjustments. For example, business segments include coal, power, chemical, steel, transportation, and sales segments. The adjustment command can be a command issued by the system indicating which business segments need adjustment and how. The adjustment command can be used to instruct the production plan of the business segment to be adjusted. For example, an adjustment command could be input stating that the coal segment's revenue will increase by 15 billion yuan in 2026, and the production plan for the coal segment could be adjusted based on this command.

[0048] In this embodiment of the disclosure, an executable production plan to be adjusted can be generated directly according to the adjustment instruction, or after the production plan to be adjusted is generated, it can be returned to the display terminal for user confirmation. After confirmation, the production plan to be adjusted is modified based on the user's interaction results, and the subsequent processing steps of the production plan processing method are executed.

[0049] S120. Based on the prompt word template of the large model and the production plan to be processed, use the large model to generate the tool call sequence; In step S120, the large model can refer to a deep learning model pre-trained on massive amounts of data based on the Transformer architecture. Large models include, but are not limited to, the GPT (Generative Pre-trained Transformer) series of models and the LLaMA (Large Language Model Meta AI) series of models. The prompt word template can refer to a pre-designed, structured input template used to guide the large model to output in the expected format. The prompt word template can include templates such as process guidance logic, tool call rules, data parsing rules, and user interaction templates. A tool call sequence can be generated using the prompt word template of the large model and the production plan to be processed obtained in step S110. The tool call sequence can refer to a structured list of processing tools called by the large model, such as table structure retrieval tools, query tools, creation tools, modification tools, and execution tools, arranged in a specific order, such as process order or selection condition order.

[0050] S130. Based on business data and the tool's call sequence, the processing tool is invoked to generate a production plan modification scheme.

[0051] In step S130, business data can refer to various types of data generated by the aforementioned business segments in their production and operation activities, and this business data can be stored in the data layer. Based on the business data and the tool call sequence, the corresponding processing tools can be invoked in the tool layer to modify the production plan currently awaiting processing. For example, a query tool can be invoked to extract the data tables corresponding to the business data (including basic information tables, production plan tables, revenue data tables, resource dependency tables, version management tables, constraint rule tables, etc.) from the data layer; another example is invoking a calculation tool to calculate the average and maximum utilization rates of each business segment, and generating a ranking result for each business segment based on the calculation results; yet another example is invoking a modification tool to generate a production plan modification scheme, etc.

[0052] According to embodiments of this disclosure, in response to adjustment instructions, the system automatically parses the business segments to be adjusted and generates production plans to be processed. Based on prompt word templates from a large model, it automatically generates tool call sequences, eliminating the need for manual arrangement of tool call logic, thus reducing the operational threshold and planning costs. By invoking processing tools based on business data and tool call sequences to generate production plan modification schemes, it achieves automated production plan adjustments driven by business data. This supports precise modifications based on business segment dimensions and utilization indicators, avoiding the deviations and inefficiencies caused by manual adjustments, and improving the response speed and accuracy of production plan adjustments.

[0053] Figure 2This is a flowchart illustrating a production planning processing method according to another embodiment of the present disclosure. This method can be used to implement step S110 in the production planning processing method. In one embodiment, the process of receiving an adjustment instruction and parsing it to obtain the business segment to be adjusted, and generating a production plan to be processed based on the business segment, further includes: S210. Receive the input adjustment instruction and parse it to obtain the identifier of the business segment to be adjusted and the adjustment period in the adjustment instruction; wherein, the adjustment instruction includes at least one of natural language instruction and structured instruction.

[0054] In step S210, after receiving the input adjustment instruction, the instruction can be parsed and converted into structured data that the system can recognize. The business segment identifier can refer to a unique code or name that identifies the business segment to be adjusted, for example, using the identifier "C" to represent the coal segment and the identifier "E" to represent the power segment. The adjustment period can refer to the time range to which the adjustment instruction applies; for example, an adjustment period of 2026 could mean that the adjustment will be completed in 2026.

[0055] In this embodiment, the adjustment instruction can include two input methods: natural language instructions and structured instructions. When the adjustment instruction is input in natural language form, such as "coal sector revenue to increase by 15 billion yuan in 2026", the instruction is semantically parsed using Natural Language Processing (NLP) technology. This includes word segmentation and entity recognition of the natural language instruction to extract key entities, including business sector entities such as the coal sector, numerical entities such as 150, unit entities such as 100 million yuan, and time entities such as 2026. According to preset semantic parsing rules, the extracted entities are mapped to fields of structured data. For example, the business sector entity is mapped to the business sector field to be adjusted, the numerical and unit entities are mapped to the adjustment magnitude field, the time entity is mapped to the adjustment period field, and the adjustment target is mapped to the target type field. The output is structured data containing the above fields.

[0056] S220. Based on the identifier of the business segment and the adjustment period, generate the priority result of the business segment.

[0057] In step S220, the business segments to be adjusted can be identified based on their identifiers. When only one business segment needs adjustment, it can be executed directly without prioritization. When multiple business segments need adjustment, the growth potential of each segment can be calculated to determine their adjustment priority. The structured data output in step S210 can be used to display the business segment data in chart and / or text format, such as using bar charts to show utilization rates or tables to show revenue contributions. The adjustment period can be used as an additional condition to prioritize business segments according to their adjustment potential within the current adjustment period based on the structured data. The priority results are then output through a large model.

[0058] S230: Receive confirmation instructions for the priority result, and generate a production plan to be processed based on the confirmation instructions.

[0059] In step S230, the priority results output by the large model can be directly used to generate a production plan to be processed. Alternatively, this priority result can be returned to the display terminal for user confirmation. Key nodes on the display terminal, such as priority sections, sub-types, and solutions, can provide users with options for confirmation, rejection, and customization. Based on the input confirmation command, the priority results are adjusted, and a production plan to be processed is generated. For example, accepting the generated priority result by clicking the "Confirm" button, accepting the priority order of the coal sector first, followed by the power sector; or rejecting the generated priority result and requesting regeneration; or adjusting the priority order and confirming by dragging the chemical sector from 3rd to 1st position, then clicking the confirmation option to issue a confirmation command. The production plan to be processed is generated based on the final confirmation command.

[0060] According to the embodiments of this disclosure, the identifier and adjustment cycle of the business segment to be adjusted are obtained by parsing the adjustment instruction, the priority result of each business segment is generated based on preset rules, and a production plan to be processed is generated after confirmation. This realizes the integration of multiple links from instruction parsing and priority generation to production plan generation, without the need to switch between multiple tools, reducing the manpower input for manually analyzing the data of each segment, improving both processing efficiency and controllability of key decision-making links.

[0061] Figure 3 This is a flowchart illustrating a production planning processing method according to another embodiment of the present disclosure. This method can be used to implement step S120 in the production planning processing method. In one embodiment, the large-model-based prompt word template and the production plan to be processed, utilizing the large-model to generate a tool call sequence, further include: S310. Generate prompt word templates based on large models.

[0062] In step S310, the prompt word template can be a preset prompt word template or generated from a large model. Multiple task type prompt word templates can be pre-configured, including process guidance logic templates, tool call rule templates, data parsing rule templates, user interaction templates, and confirmation templates. Each template is stored according to parameters such as business segment type or adjustment cycle. Upon receiving a production plan to be processed, the corresponding template combination is called from the pre-configured template library based on the business segment identifier and adjustment cycle in the adjustment instruction to form a complete prompt word template.

[0063] In this embodiment of the disclosure, current task information (including business segments to be adjusted, adjustment cycles, production plans to be processed, etc.) can also be used as input to generate prompt word templates for the current task through a large model.

[0064] Here is an example of a large model prompt word template: "As the core orchestration engine of the intelligent production planning adjustment system, you need to guide users to complete the adjustment according to the following process:" 1. After receiving the user's revenue target, first call the retrieve_table_schema tool to obtain the structure of the production plan table, basic information table, and revenue data table; 2. Use the execute_query_sql tool to extract the planned and capacity values ​​for each business segment, and calculate unit revenue, utilization rate, revenue increment, and resource gap. Taking the self-produced coal plan as an example, the current_plan value is commercial_coal_sypply in the production plan table, and max_capacity is missing data in design_capacity and average_capacity in the basic information table. At this time, average_capacity is calculated in reverse through the relationship between design_capacity and its related coefficients. Calculate unit_revenue (self-produced coal revenue / commercial_coal_sypply), the average utilization rate of the segment (average_utilization = current_plan / average_capacity), and the maximum utilization rate (capacity_utilization = current_plan / max_capacity). 4. Sorting priority is based on the following rules: sections with average_utilization less than 100% are sorted in ascending order; if none are less than 100%, they are sorted in ascending order of capacity_utilization. 5. Generate user interaction prompts, including data and priority suggestions for each section, to guide users to confirm; 6. Based on the user's confirmation, execute subsequent segmentation analysis or re-sorting logic. S320. Based on the prompt word template and the production plan to be processed, the sequence is called through the large model generation tool.

[0065] In step S320, the prompt word template can be combined with the production plan to be processed into the input of the large model, and the sequence can be called through the large model generation tool. This includes: Sequential invocation mechanism: Following the production planning process flow, such as the entire process from obtaining the structure, executing queries, equivalent calculations, creating drafts, modifying data, to executing modifications. It retrieves the processing tools required for each step and generates a tool invocation sequence based on the invocation instructions of all tools. The order of the production planning process flow can be preset by the business module or customized by the user.

[0066] Conditional Invocation Mechanism: Based on priority results, the system determines whether the steps in the production planning process need to be adjusted using the large model. Based on the adjustment results, it obtains the processing tools that need to be invoked and generates a tool invocation sequence based on the invocation instructions of all tools.

[0067] In this embodiment, each tool in the tool call sequence transmits data through a standardized data interface. The output data of the previous tool is automatically used as the input parameter of the next tool, achieving seamless data flow within the toolchain. After a tool in the tool call sequence executes the corresponding operation step, an exception prompt can be generated and a retry or a large model adjustment instruction can be prompted when a tool call fails, such as when a query tool times out or a verification tool fails data verification.

[0068] S330. Confirm the invocation of the tool based on the tool invocation sequence and the prompt word template.

[0069] In step S330, the tool call sequence output by the large model can be directly used to confirm the tool call. Alternatively, this tool call sequence can be returned to the display terminal for user confirmation. The display terminal can provide users with options such as confirm / reject / custom. Based on the input confirmation command, the tool call sequence can be adjusted before confirming the tool call. The tool call sequence can also be validated based on the prompt word template, such as format validation, tool name validation, parameter integrity validation, execution order validation, and dependency validation, to ensure the completeness and compliance of the tool call sequence as an instruction and the executable capability of each processing tool in the tool call sequence. After successful validation, subsequent processing steps can be executed.

[0070] According to the embodiments of this disclosure, a tool call sequence is generated by using a large model based on the prompt word template and the production plan to be processed, and the generated tool call sequence is verified. This can ensure the compliance and executability of each call instruction in the sequence, avoid data processing interruption or result deviation caused by tool call errors, and improve the accuracy and reliability of tool call sequence generation.

[0071] Figure 4 This is a flowchart illustrating a production planning processing method according to another embodiment of the present disclosure. This method can be used to implement step S130 in the production planning processing method. In one embodiment, the step of generating a production plan modification scheme by invoking a processing tool based on business data and the tool invocation sequence further includes: S410. Extract the data table corresponding to the business data based on the tool's call sequence; wherein, the data table includes at least one of the following: basic information table, production plan table, revenue data table, resource dependency table, version management table, and constraint rule table.

[0072] In step S410, the data table corresponding to the business data can refer to a structured dataset stored in tabular form after standardization. The data table corresponding to the business data in the corresponding step can be extracted based on the execution steps of each processing tool in the tool call sequence. The specific fields and relationships of one such data table are shown below: Taking coal mine basic information as an example, the fields of the basic information table may include: plan identifier ID, version identifier ID, coal mine code, coal mine name, code of the subordinate secondary unit, name of the subordinate secondary unit, designed capacity, raw coal supply, region, equity attributable to the parent company, loading station code, dedicated line code, etc. Taking the self-produced coal plan as an example, the fields of the production plan table may include: plan identifier ID, version identifier ID, coal mine code, coal type code, and commercial coal supply quantity, etc. Taking the electricity revenue table as an example, the revenue data table fields may include: plan identifier ID, version identifier ID, power plant code, electricity sales price, heat sales price, etc. Taking the power plant coal consumption table as an example, the resource dependency table fields may include: plan identifier ID, version identifier ID, power plant code, power supply coal consumption, power plant power consumption rate, heating coal consumption, furnace calorific value, etc. The version management table fields may include: Version ID, Version Name, Version Number, Activation ID, Deletion ID, Creator, Creation Time, Updater, Update Time, Creator's Department, Sort, etc. Taking coal mine supply as an example, the fields in the constraint rule table may include: rule ID, version ID, constraint level, region, second-level company code, coal mine code, coal type code, supply quantity, fluctuation coefficient, etc.

[0073] Data tables can be stored using distributed relational databases such as PostgreSQL and Structured Query Language (SQL); standardized data access is provided through the Java Database Connectivity (JDBC) interface, which supports structured query language queries, data updates, and management.

[0074] In this embodiment of the disclosure, after receiving the tool call sequence, the tool for obtaining the table structure in the tool call sequence can parse the structure, field types, and relationships of the data table. Based on the database metadata query interface such as INFORMATION_SCHEMA, standardized query statements can be written to extract information such as the data table name, field names, data types, primary keys, and foreign keys, and generate a structured table structure document (e.g., JSON format). Calling an interface such as retrieve_table_schema can obtain the target table name as the input parameter and output JSON data containing field information and relationships.

[0075] In this embodiment, query instructions can be generated based on the query tools in the tool call sequence to extract corresponding data from the data table. By encapsulating the JDBC query interface, dynamic SQL statement input can be supported, with built-in parameter validation and data format conversion. The query results can be converted to JSON format, and batch and paginated queries are also supported. By calling an interface such as execute_query_sql, the input parameters are SQL statements and query parameters, and the output is a list of structured query results. For example, the large model automatically generates standardized SQL query statements based on table structure information, calls the SQL query execution tool, and extracts data: the current production plan (current_plan) of each business segment (coal / electricity / chemical / transportation capacity) and its sub-segments; the average capacity (average_capacity) and maximum capacity (max_capacity) of each business segment; the unit product revenue (unit_revenue) and historical revenue contribution of each business segment; the upstream and downstream resource dependencies and dependency coefficients of each business segment; the structured data returned by the SQL query execution tool, the orchestration layer cleans the data (removing null values ​​and outliers) and performs format conversion to generate a unified dataset.

[0076] S420. Based on the business data, calculate the average utilization rate and maximum utilization rate of each business segment, and generate the ranking result of the business segment based on the average utilization rate and maximum utilization rate.

[0077] In step S420, utilization rate calculation can be performed based on a large model. The formula for calculating the average utilization rate of a business segment is shown below: Average utilization rate = Current production plan / Average capacity; The formula for calculating the maximum utilization rate of a business segment is as follows: Maximum utilization rate = Current production plan / Maximum capacity; The large model uses ranking rules to generate ranking results for business segments based on average and maximum utilization rates. For example, the first priority is given to business segments with remaining capacity that can be prioritized for improvement, i.e., those with an average utilization rate of less than 100%. The second priority is given to business segments with even greater remaining capacity, i.e., those with an average utilization rate of not less than 100%, ranked in ascending order of maximum utilization rate. The ranking rules for the large model can be preset or customized.

[0078] S430. Based on the sorting result, call the processing tool to generate the production plan modification scheme.

[0079] In step S430, the sorting results of the business segments output by the large model can be directly used to generate a production plan modification scheme. Alternatively, the sorting results of this business segment can be used to generate a visual analysis report and interactive prompts through the large model, and then returned to the display terminal for user confirmation. The display terminal can provide users with options such as confirm / reject / custom. Based on the input confirmation command, the sorting results of the business segments output by the large model can be used to generate a production plan modification scheme.

[0080] The following is a visualization analysis report generated from a large model: "Analysis Report on the 2026 Revenue Growth Target of 20 Billion Yuan" I. Core data for each sector: 1. Coal sector: Average utilization rate 92%, maximum utilization rate 48%, unit revenue 120 yuan / ton, current revenue contribution 35 billion yuan; 2. Power Sector: Average utilization rate 70%, maximum utilization rate 35%, unit revenue 0.3 yuan / kWh, current revenue contribution 42 billion yuan; 3. Chemical sector: average utilization rate 98%, maximum utilization rate 55%, unit revenue 800 yuan / ton, current revenue contribution 28 billion yuan; 4. Transportation capacity segment: average utilization rate of 85%, maximum utilization rate of 40%, unit revenue of 50 yuan / ton-kilometer, current revenue contribution of 15 billion yuan.

[0081] II. Priority Recommendation: First priority: Power sector (lowest average utilization rate, greatest growth potential). Second priority: Transportation capacity sector; Third priority: Coal sector; Fourth priority: Chemical sector.

[0082] Please confirm whether you want to prioritize adjusting the power sector? (Options: Confirm / Reject / Custom Recommendation)

[0083] According to the embodiments of this disclosure, the data tables corresponding to business data are extracted according to the tool call sequence, and the average utilization rate and maximum utilization rate of each business segment are automatically calculated based on the data tables. This eliminates the need for manual consultation of multiple data sources and manual calculation of utilization indicators for each segment, reducing the human resource input for data querying and calculation. Based on the sorting results, a quantifiable adjustment basis is provided for subsequent production plan modifications, avoiding the subjectivity and randomness of manual judgment of the adjustment order based on experience, and improving the accuracy and consistency of production plan adjustments.

[0084] Figure 5 This is a flowchart illustrating a production planning processing method according to another embodiment of the present disclosure, the method further comprising: S510. Based on this processing tool, extract the revenue data of the production plan modification scheme.

[0085] In step S510, the processing tool can refer to various functional tools that operate on business data, and can be in the form of software modules or components. The processing tool is a general term for functional modules used to perform business data operations. The specific type of processing tool can be configured according to the actual application scenario. For example, it can be one or more of the following: data query tool, calculation tool, sorting tool, modification tool, verification tool, version management tool. It can also be other tool modules with data processing functions that can implement the production planning processing method of this application; this application does not limit this.

[0086] In one implementation, based on the revenue data, the revenue target achievement rate of the production plan modification scheme is verified through the large model to obtain the verification result, including: Based on this revenue data, the average and maximum utilization rates of the business sub-segments in the business segment of the production plan modification scheme are calculated using this large model. Based on this revenue data, the resource dependencies of this business sub-segment are extracted using this large model; Based on this revenue data, the resource gap for this business sub-segment is calculated using this large model; the resource gap includes both direct and indirect resource gaps. The revenue target achievement rate is calculated based on the average utilization rate, the maximum utilization rate, the resource dependency, and the resource gap; The revenue target achievement rate of the modified production plan was verified using this large model, and the verification results were obtained.

[0087] In this embodiment, the SQL query tool in the sequence can be invoked to extract data from the business sub-segments of the business segment according to priority results. A business sub-segment can refer to a further subdivided business unit within the business segment. The classification of business sub-segments can be preset by the larger model and can also be customized. For example, the business sub-segments of the power segment include coal-fired power generation, new energy power generation (including wind power and photovoltaic power generation), and hydropower. The business sub-segments of the coal segment include open-pit mining, underground mining, and coal washing. The business sub-segments of the chemical segment include coal chemical industry, petrochemical industry, and fine chemical industry.

[0088] Based on the priority section confirmed by the user, business sub-segment data of that section type can be extracted. For example, according to the priority result, the business section is extracted as the power section, and then the extracted business sub-segments are: thermal power section, hydropower section, and new energy sub-segment (including wind power section and / or photovoltaic sub-segment). The data extracted for each type of business sub-segment are: production plan, average capacity, maximum capacity, unit revenue, and resource dependence coefficient (for example, a thermal power business segment that produces 100 million kWh of electricity requires 200,000 tons of coal).

[0089] In this embodiment of the disclosure, the average utilization rate and maximum utilization rate of each type of business sub-segment can also be recalculated based on the extracted revenue data through a large model. The calculation formula is the same as the calculation formula for the average utilization rate and maximum utilization rate of the business sub-segment, and will not be repeated here.

[0090] In this embodiment of the disclosure, the priority results of each type of business sub-segment can be calculated based on the average and maximum utilization rates of each type of business sub-segment. After generating a visual analysis report and interactive prompts through the large model, the results are returned to the display terminal for user confirmation. The display terminal can provide users with options such as confirmation / rejection / customization. Based on the input confirmation command, the sorting results of each type of sub-segment output by the large model are used to generate the adjustment direction of each type of sub-segment determined by the production plan modification scheme.

[0091] In this embodiment, based on the adjustment direction, the large model can also call and execute SQL query tools to extract resource dependency data for each business sub-segment. This resource dependency data is divided into direct dependency data and indirect dependency data. For example, direct dependency data may include the requirement of 200,000 tons of coal for producing 100 million kWh of electricity, the requirement of 500 ton-kilometers of railway transport capacity for transporting 10,000 tons of coal, and the requirement of 300 × 20 = 60 million tons of coal for an increase of 30 billion kWh in thermal power production. Indirect dependencies may include port loading and unloading capacity, coal mine production capacity (e.g., 60 million tons of coal requires 6000 × 500 = 3 million ton-kilometers of railway transport capacity), etc.

[0092] The large model can be used to analyze the interrelationships between various business sub-segments. It can calculate the direct and indirect resource gaps of business sub-segments through recursive calculations, generate the interrelationship analysis results, and return them to the display terminal for user confirmation. The display terminal can provide users with options such as confirmation, rejection, and customization. Based on the input confirmation command, an adjustment plan can be generated, and the adjustment plan can be used to create a draft version of the production plan modification plan.

[0093] An example of a linkage impact analysis report is shown below: Report on the Linkage and Impact of an Increase of 30 Billion kWh in Thermal Power Production I. Direct resource requirements: Coal demand: 60 million tons (30 billion kWh × 200,000 tons / 100 million kWh). Railway transport demand: 3 million ton-kilometers (60 million tons × 500 ton-kilometers / 10,000 tons).

[0094] II. Resource Gap Accounting: 1. Coal sector: Current spare capacity is 10 million tons, with a shortage of 50 million tons; 2. Railway transport capacity: Currently, there is a remaining capacity of 5 million ton-kilometers, with no shortage; 3. Port loading and unloading: Current spare capacity is 80 million tons, with no shortage.

[0095] III. Recommended solutions for the gap: 1. Increase internal coal mine production by 10 million tons (utilizing remaining capacity at full capacity); 2. Purchase 40 million tons of coal from external sources (the cost of external purchases and the supplier's capacity need to be confirmed).

[0096] Please select a gap solution? (Options: Solution 1 / Solution 2 / Custom Solution)

[0097] If the user selects a solution such as Solution 2: procuring 40 million tons of coal from external sources, additional constraints, such as those related to external coal procurement, can be further calculated using a large model. These additional constraints may include the impact of procurement costs on revenue, transportation capacity constraints of external procurement routes, etc. Once the large model confirms that the solution will not introduce new resource gaps, subsequent operations can be executed. If the large model confirms that the solution has new gaps, subsequent operations will not be executed, and repeated linkage will affect the calculation.

[0098] In this embodiment of the disclosure, a draft version creation tool can be called through a large model to generate a draft version of the current production plan adjustment scheme. By inputting the version ID, creation time and creator, a draft version is generated. For example, the version ID is Draft_20260417_001, the creator is A, and the creation time is April 17, 2026.

[0099] Draft versions are inactive by default. Larger models can use version data modification tools to update the draft version according to the adjustment plan described above. The updated draft version data is then returned to the display terminal for user confirmation. The display terminal can provide users with options such as confirm / reject / custom. Based on the confirmation command, the current draft version status can be changed to active. Taking the power sector as an example, a report on updating draft version data according to the adjustment plan is shown below: Power sector: The current production plan has been revised from 50 billion kWh to 80 billion kWh; Revenue target: The revenue target for the power sector has been revised from 42 billion yuan to 42 billion + 30 billion × 0.32 = 51.6 billion yuan; Coal procurement plan: 40 million tons of additional coal will be procured from external sources, and the procurement cost budget will be updated; The version data modification tool performed data verification (the thermal power sector's capacity of 80 billion kWh is no greater than the maximum capacity of 156.2 billion kWh, and the verification passed). Modification was allowed, and the modification log is recorded as follows: "Preview of the draft version of the production plan adjustment (Draft_20260417_001)" I. Core Adjustments: 1. Thermal power production plan: adjusted from 50 billion kWh to 80 billion kWh (requiring an additional 30 billion kWh); 2. Coal procurement plan: 40 million tons more; 3. Revenue Target: The Group’s total revenue target needs to be increased by RMB 19.6 billion (RMB 9.6 billion for the power sector and RMB 10 billion for coal procurement revenue).

[0100] II. Resource Constraint Verification: Thermal power capacity: 80 billion kWh or no more than 156.2 billion kWh (maximum capacity), which meets the requirements; Coal supply: 60 million tons (10 million tons internally and 40 million tons externally), sufficient to meet demand; Railway transport capacity: 3 million ton-kilometers, not exceeding the remaining 5 million ton-kilometers, is sufficient.

[0101] Please confirm whether to execute this draft version? (Options: Confirm Execution / Return to Modifications / Discard)

[0102] If you choose "Confirm Execution", the large model can call the modification execution tool to officially make the draft version effective and synchronize it to the production system; if you choose "Return to Modification", the above draft version modification process will be repeated to add a new draft version; if you choose "Abandon", the adjustment process will be terminated.

[0103] S520. Based on the revenue data, the revenue target achievement rate of the production plan modification scheme is verified through the large model, and the verification result is obtained.

[0104] In step S520, after the draft version takes effect, the SQL query tool can be executed through the large model to extract the adjusted revenue data and the revenue contribution of each segment. Based on the revenue data, the revenue target achievement rate of the draft version of the production plan modification scheme is calculated. A formula for calculating the revenue target achievement rate of the production plan modification scheme is as follows: Revenue target achievement rate = (Actual revenue increase / Target increase) × 100% The actual revenue increment can be calculated based on the difference between the expected revenue and the benchmark revenue. The expected revenue and the benchmark revenue are extracted from the business segments through query tools; the target revenue increment can be obtained through adjustment instructions. If the achievement rate is not less than 95%, for example, if the actual increase is 19.6 billion yuan and the target is 20 billion yuan, the achievement rate is 98%, then the verification is deemed to have passed, and the output draft version can be used as the final production plan modification scheme, and the process ends. If the achievement rate is less than 95%, for example, if the actual increase is 18 billion yuan and the target is 20 billion yuan, the achievement rate is 90%, then the verification is deemed to have failed and the draft version of the current production plan needs to be optimized. The draft version iteration optimization prompts are generated by using the prompt word template in the large model.

[0105] A revenue target verification report is shown below: Revenue Target Verification Report I. Goal Achievement Status: Target revenue increase: 20 billion yuan; Actual revenue increase: 18 billion yuan; Achievement rate: 90% is less than 95%, the goal has not been fully achieved.

[0106] II. Recommendations for Iterative Optimization: It is recommended to further adjust the transportation capacity sector. Currently, the average utilization rate of the transportation capacity sector is 85%, which has room for growth. It is estimated that an increase of 2 million ton-kilometers in production could generate an additional 1 billion yuan in revenue.

[0107] Please confirm whether iterative optimization should be performed? The revenue target verification report is returned to the display terminal for user confirmation. The display terminal can provide users with options such as confirm / reject / custom. Based on the input confirmation command, the revenue target verification report output by the large model is used to update the version identifier of the production plan modification scheme.

[0108] S530. Update the version identifier of the production plan modification scheme based on the verification result.

[0109] In step S530, the version identifier can refer to the version number of the production plan modification scheme. The update method of the version identifier can be determined based on the passed verification result: for example, when the verification result is passed, the current draft version is taken as the final production plan modification scheme, and the version identifier of the current production plan modification scheme is updated to the latest version number of the draft version. Alternatively, a new version number can be generated based on the latest version number. For example, if the latest version number of the current draft version is V1.0, then the new version number is V2.0. Conversely, when the verification result fails, the orchestration layer does not update the version identifier, retains the current version number of the draft version, marks the current version status as "verification failed," and generates a prompt message to notify the user that the modification scheme failed the verification and needs to be readjusted.

[0110] In this embodiment, once any version of the production plan modification is confirmed to be effective, that version is marked as "effective" and automatically archived in the historical version record area of ​​the version management table, including information such as version number, effective time, person who made the change, and verification result. Previously effective draft versions can be used as historical versions. The currently effective production plan modification is used as the baseline version, which cannot be directly modified.

[0111] When users need to view the evolution of a solution, the large model can retrieve historical versions through a query tool, searching a list of historical version records to view the adjustments and effective dates of each version. When users need to revert to a specific historical version, they can select the corresponding historical version and trigger a rollback operation. The rollback operation includes: copying the historical version's production plan modification scheme to the current draft version, adding a version number, and allowing the user to confirm before it takes effect again.

[0112] According to embodiments of this disclosure, by extracting revenue data from the modified plan and verifying the revenue target achievement rate through a large model, and updating the version identifier after successful verification, automatic verification and version management of the production plan modification plan are achieved. The version identifier is automatically updated after successful verification and remains unchanged when verification fails, facilitating subsequent traceability, rollback, and comparison of the production plan modification plan. This avoids version confusion and omissions that may occur with manual archiving, and improves the standardization and traceability of the production plan modification plan.

[0113] This disclosure proposes a production planning processing method that automates the orchestration of tool invocation order, data transfer, and exception handling through pre-set prompts and logical judgments in a large model. A schematic diagram of the execution flow of this production planning processing method is shown below. Figure 6 As shown, it includes: S601, Revenue Target Input: Supports users to input revenue adjustment targets in natural language or structured forms (e.g., inputting a revenue adjustment target of 15 billion yuan in revenue growth by 2026), and analyzes the target value and time period.

[0114] S602. Data initialization: Display data for each segment in the form of charts and / or text (e.g., bar charts to show utilization rate, tables to show revenue contribution).

[0115] S603. Module priority determination: Calculate the priority of the business module based on the input data to obtain the priority result. If the user confirms yes, execute step S640; if the user confirms no, repeat step S630 to calculate the priority result.

[0116] S604, User Confirmation: Provides users with "Confirm / Reject / Custom" options, supporting users to input personalized commands.

[0117] S605, segment analysis: Generate a revenue analysis report based on priority results and revenue data of business segments.

[0118] S606, Linkage Impact Calculation: Extract resource dependency data for each business segment.

[0119] S607. Resource Gap Resolution: Calculate direct and indirect resource gaps, and generate results for the linkage impact analysis.

[0120] S608, Version Management: Generate draft versions from large models and update draft version data according to adjustment plans.

[0121] S609. Target verification: Calculate the revenue target achievement rate of the draft version based on the revenue data. If the verification passes (e.g., the achievement rate is not less than 95%), proceed to step S610. If the verification fails, proceed to step S603.

[0122] S610. Use the output draft version as the final production plan modification scheme and end the process.

[0123] 1. High degree of standardization in decision-making, significantly reducing reliance on manual intervention. This disclosure utilizes a large-scale model tool to orchestrate fixed technical processes, incorporating unified priority determination rules, linked accounting logic, and data verification standards to replace subjective human judgment. Regardless of the operator's expertise, all processes and rules are executed according to a unified framework, eliminating decision-making biases caused by differences in human experience. Non-professionals can complete production plan adjustments through interactive guidance, without needing specialized skills such as SQL writing or ERP configuration. Automated data extraction and calculation replace manual operations, reducing the data error rate to below 1%.

[0124] 2. Automated and collaborative processes significantly improve adjustment efficiency. This disclosure automates the entire process—from target input, data extraction, analysis and calculation, plan modification, and version archiving—through a tool call sequence and integrated processing tools, and dynamically orchestrates a large model, eliminating the need for manual tool switching and data transfer. Existing technologies require 3-5 hours to adjust the process, while this proposal takes less than 90 minutes, improving efficiency by over 80%. It integrates the original 3-4 independent tool operations into a one-stop process, reducing the number of steps from 15-20 to 5-6. It supports urgent adjustments to revenue targets, providing an executable solution within 1.5 hours to meet the rapid decision-making needs of enterprises.

[0125] This disclosure, based on resource dependency graphs and recursive algorithms, automatically calculates the constraint impact of adjustment plans on multiple upstream and downstream sectors, quantifies resource gaps, and identifies potential implementation risks in advance. It covers constraints across the entire supply chain, including production, transportation, and procurement. All adjustment plans undergo resource gap verification, increasing the implementation success rate to over 95%. Through interconnected impact calculations, it identifies potential constraints (such as transportation gaps) in advance and provides solutions to avoid production interruptions after implementation.

[0126] 3. Version control standards reduce adjustment risks. This design employs a three-tiered system: baseline version, draft version, and historical version. It supports draft creation, multiple rounds of modification, version comparison, and rollback traceability. Adjustments are verified within the draft version, ensuring no impact on the currently effective production plan and mitigating production risks caused by erroneous adjustments. Each modification's content, executor, and time are fully recorded for auditing and issue tracing. Parallel modification and comparison of multiple versions are supported, allowing selection of the optimal solution for different scenarios.

[0127] 4. Real-time and accurate data supports the achievement of revenue targets. This disclosure utilizes SQL tools to extract business database data in real time, replacing the reliance on historical data or manual input in existing technologies, and combines it with equivalent calculation tools to ensure accurate data calculation. Analysis based on real-time business data ensures that production plans match current capacity and revenue status; the revenue increment prediction error rate is no greater than 3%, and the preset revenue target can be stably achieved or approached; it supports real-time changes in business data (such as sudden capacity reductions), and adjustment plans can be quickly iterated and optimized.

[0128] 5. Flexible interactive guidance, adaptable to personalized needs. This disclosure employs a step-by-step interactive guidance system, providing user confirmation and customization options at key nodes such as priority determination, sub-type selection, and gap solution. It supports users in adjusting priorities based on corporate strategic considerations (e.g., prioritizing the development of new energy sectors rather than thermal power as recommended by rules), balancing automation and flexibility. Users can clearly view the analysis results and logic at each step, understand the solution generation process, and improve acceptance. Furthermore, by modifying the large model's prompts and rule parameters, it can adapt to the production plan adjustment needs of different industries (energy, manufacturing, chemicals) and companies of different sizes.

[0129] 6. The architecture is scalable, supporting business iteration and upgrades. This disclosure adopts a layered architecture design. The tool layer supports the addition of new tools (such as AI prediction tools and cost optimization tools), the orchestration layer supports dynamic updates of prompts and rules, and the data layer supports the addition of data for new business segments. New functional modules can be added without refactoring the overall architecture, adapting to enterprise business development (such as adding an energy storage segment); priority rules and linkage calculation logic can be adjusted and updated according to market changes and enterprise strategies; it is compatible with future large-scale model technology upgrades and business system iterations, extending the system's lifespan.

[0130] This disclosure adopts a layered architecture design for the production planning and processing system. An overall architecture diagram of the production planning and processing system is shown below. Figure 7 As shown, it includes: (1) Interaction layer: Human-computer interaction interface and guidance mechanism, providing users with a visual interaction interface, outputting analysis results, guiding users to confirm, receiving user instructions, and realizing human-computer collaborative decision-making.

[0131] Core interaction flow design: Target input interface: Supports users to input revenue adjustment targets in natural language or structured forms, and automatically parses target values ​​and time periods; Analysis results display interface: The data of each section is displayed in the form of charts and text (such as bar charts to show utilization rate and tables to show revenue contribution), highlighting priority recommendations; Decision Confirmation Interface: Provides "Confirm / Reject / Custom" options for nodes (priority sections, sub-types, gap solutions), and supports users to input personalized instructions; The solution preview interface displays the adjusted production plan, revenue forecast, capacity utilization rate, etc., and supports comparison of multiple versions; The operation log interface records the entire process of adjustment, including operation steps, tool call records, and data modification history, and supports traceability queries.

[0132] (2) Data layer: core data storage and management, which can store core data of the entire business segments of coal, electricity, railway, port, shipping, chemical and sales, provide data support for upper-level tools, and ensure the integrity and real-time performance of data.

[0133] This includes a set of business data tables, with specific fields and relationships as follows: Taking coal mine basic information as an example, the fields of the basic information table may include: plan ID, version ID, coal mine code, coal mine name, code of the subordinate secondary unit, name of the subordinate secondary unit, designed capacity, raw coal supply, region, equity attributable to the parent company, loading station code, and dedicated line code; Taking the self-produced coal plan as an example, the fields in the production plan table can include: plan ID, version ID, coal mine code, coal type code, and commodity coal supply. Taking the electricity revenue table as an example, the revenue data table fields may include: plan ID, version ID, power plant code, electricity sales price, and heat sales price; Taking the power plant coal consumption table as an example, the resource dependency table fields may include: plan ID, version ID, power plant code, power supply coal consumption, power plant power consumption rate, heating coal consumption, and furnace calorific value; The version management table fields may include: version ID, version name, version number, effective flag, deletion flag, creator, creation time, updater, update time, creator's department, sorting, etc. Taking coal mine supply as an example, the fields in the constraint rule table can include: rule ID, version ID, constraint level, region, second-level company code, coal mine code, coal type code, supply quantity, and fluctuation coefficient.

[0134] It employs a distributed relational database (such as PostgreSQL) to support massive data storage and high-concurrency queries; it provides standardized data access through a JDBC interface, supporting SQL queries, data updates, and transaction management to ensure the atomicity and consistency of data operations.

[0135] (3) Arrangement layer: including the large model intelligent arrangement engine. The large model is based on preset prompt words and standardized technical processes to realize prompt word engineering, tool call logic and logical judgment instructions. It is the core decision layer of the system.

[0136] (4) Tool Layer: Standardized tool encapsulation and implementation, encapsulating the core tools required for the entire production plan adjustment process, providing standardized interfaces for the orchestration layer, and ensuring the reusability and callability of the tools. The tool layer includes processing tools such as tools for obtaining table structures, executing SQL queries, creating draft versions, modifying version data, and executing modifications.

[0137] Figure 8 This is a schematic diagram of the structure of a production planning processing system 800 according to an embodiment of the present disclosure. The system includes an interaction layer 810, an orchestration layer 820, a data layer 830, and a tool layer 840, and further comprises: The interaction layer 810 is used to receive adjustment instructions and parse the business segment to be adjusted, generate a production plan to be processed based on the business segment, and output the production plan to be processed to the orchestration layer; the adjustment instructions are used to indicate the business segment to be adjusted. The orchestration layer 820 is communicatively connected to the interaction layer 810 and the tool layer 840, respectively, and is used to generate a tool call sequence based on the prompt word template of the large model and the production plan to be processed, and output it to the tool layer 840. The data layer 830 is communicatively connected to the tool layer 840 and is used to output the stored business data to the tool layer 840. The tool layer 840 is communicatively connected to the data layer 830 and the orchestration layer 820, respectively, and is used to call the processing tool to generate a production plan modification scheme based on the business data and the tool call sequence, and return it to the orchestration layer 820.

[0138] In one implementation, the interaction layer 810 includes: Input interface 811 is used to receive input adjustment instructions and parse them to obtain the identifier of the business segment to be adjusted and the adjustment period in the adjustment instructions; wherein, the adjustment instructions include at least one of natural language instructions and structured instructions; Analysis interface 812 is used to receive the identifier of the business segment and the adjustment period from the input interface, generate the priority result of the business segment and send it to the confirmation interface. The confirmation interface 813 is used to receive a confirmation instruction for the priority result from the analysis interface, and generate a production plan to be processed based on the confirmation instruction and send it to the orchestration layer.

[0139] In one implementation, the arrangement layer 820 includes: The prompt word module 821 is used to generate prompt word templates based on the large model; wherein, the prompt word templates include process guidance logic templates, tool call rule templates, data parsing rule templates, user interaction templates, and confirmation templates; The tool orchestration module 822 is used to generate a tool call sequence based on the prompt word template and the production plan to be processed through the large model; The confirmation module 823 is used to confirm the invocation of the tool based on the tool invocation sequence and the prompt word template.

[0140] In one implementation, the tool layer 840 includes: Extraction module 841 is used to extract the data table corresponding to the business data based on the tool's call sequence; wherein, the data table includes at least one of the following: basic information table, production plan table, revenue data table, resource dependency table, version management table, and constraint rule table; The calculation module 842 is used to calculate the average utilization rate and maximum utilization rate of each business segment based on the business data, and generate the ranking result of the business segment based on the average utilization rate and the maximum utilization rate. The generation module 843 is used to generate the production plan modification scheme by calling the processing tool based on the sorting result.

[0141] In one implementation, the system further includes: The tool layer 840 is also used to extract revenue data based on the production plan modification scheme using the processing tool; The orchestration layer 820 is also used to verify the revenue target achievement rate of the production plan modification scheme based on the revenue data through the large model, and obtain the verification result; The tool layer 840 also uses the verification result to update the version identifier of the production plan modification scheme in the data layer 830.

[0142] In one implementation, the orchestration layer 820 is further configured to: calculate, based on the revenue data, the average and maximum utilization rates of the business sub-segments within the business segment of the production plan modification scheme using the large model; extract the resource dependencies of the business sub-segments using the large model based on the revenue data; calculate the resource gap of the business sub-segments using the large model based on the revenue data; wherein the resource gap includes direct resource gaps and indirect resource gaps; calculate the revenue target achievement rate based on the average utilization rate, the maximum utilization rate, the resource dependencies, and the resource gap; and verify the revenue target achievement rate of the production plan modification scheme using the large model to obtain a verification result.

[0143] The preferred embodiments of this disclosure have been described in detail above with reference to the accompanying drawings. However, this disclosure is not limited to the specific details of the above embodiments. Within the scope of the technical concept of this disclosure, various simple modifications can be made to the technical solutions of this disclosure, and these simple modifications all fall within the protection scope of this disclosure. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of this disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures have been omitted in the above description. Furthermore, it should be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, this disclosure will not describe the various possible combinations separately.

[0144] Furthermore, various different embodiments of this disclosure can be combined in any way, as long as they do not violate the spirit of this disclosure, they should also be regarded as the content disclosed in this disclosure.

Claims

1. A production planning processing method, comprising: The system receives adjustment instructions and parses them to obtain the business segments to be adjusted, and generates a production plan to be processed based on the business segments; wherein, the adjustment instructions are used to indicate the business segments to be adjusted. Based on the prompt word template of the large model and the production plan to be processed, the tool call sequence is generated using the large model; Based on business data and the tool call sequence, the processing tool is invoked to generate a production plan modification scheme.

2. The method according to claim 1, wherein, The process of receiving adjustment instructions, parsing them to obtain the business segments to be adjusted, and generating production plans to be processed based on the business segments includes: The system receives an input adjustment instruction and parses it to obtain the identifier of the business segment to be adjusted and the adjustment period in the adjustment instruction; wherein, the adjustment instruction includes at least one of natural language instructions and structured instructions; Based on the identifier of the business segment and the adjustment period, a priority result for the business segment is generated; Receive confirmation instructions for the priority results, and generate a production plan to be processed based on the confirmation instructions.

3. The method according to claim 1, wherein, The prompt word template based on the large model and the production plan to be processed, using the large model generation tool to call the sequence, include: Generate prompt word templates based on a large model; Based on the prompt word template and the production plan to be processed, the sequence is invoked through the large model generation tool; The tool call is confirmed based on the tool call sequence and the prompt word template.

4. The method according to claim 1, wherein, The step of generating a production plan modification scheme by invoking a processing tool based on business data and the tool call sequence includes: The data table corresponding to the business data is extracted based on the tool call sequence; wherein, the data table includes at least one of the following: basic information table, production plan table, revenue data table, resource dependency table, version management table, and constraint rule table; Based on the business data, calculate the average utilization rate and maximum utilization rate of each business segment, and generate a ranking result of the business segments based on the average utilization rate and the maximum utilization rate; Based on the sorting results, the processing tool is invoked to generate the production plan modification scheme.

5. The method according to claim 1, further comprising: Revenue data for the production plan modification scheme is extracted based on the processing tool; Based on the revenue data, the revenue target achievement rate of the production plan modification scheme is verified through the large model to obtain the verification result; The version identifier of the production plan modification scheme is updated based on the verification results.

6. The method according to claim 5, wherein, Based on the revenue data, the revenue target achievement rate of the production plan modification scheme is verified through the large model to obtain the verification result, including: Based on the revenue data, the average and maximum utilization rates of the business sub-segments of the business segment in the production plan modification scheme are calculated using the large model. Based on the revenue data, the resource dependencies of the business sub-segments are extracted using the large model; Based on the revenue data, the resource gap of the business sub-segment is calculated using the large model; wherein, the resource gap includes direct resource gap and indirect resource gap; The revenue target achievement rate is calculated based on the average utilization rate, the maximum utilization rate, the resource dependency relationship, and the resource gap; The revenue target achievement rate of the modified production plan is verified using the large model, and the verification results are obtained.

7. A production planning and processing system, the system comprising: The interaction layer is used to receive adjustment instructions and parse the business segments to be adjusted, generate production plans to be processed based on the business segments, and output the production plans to be processed to the orchestration layer. The adjustment instruction is used to specify the business segment to be adjusted; The orchestration layer is communicatively connected to the interaction layer and the tool layer, respectively, and is used to generate a tool call sequence based on the prompt word template of the large model and the production plan to be processed, and output it to the tool layer; The data layer, which is communicatively connected to the tool layer, is used to output stored business data to the tool layer. The tool layer is communicatively connected to the data layer and the orchestration layer, and is used to invoke processing tools to generate production plan modification schemes based on the business data and the tool call sequence, and return them to the orchestration layer.

8. The system according to claim 7, wherein, The interaction layer includes: An input interface is used to receive input adjustment instructions and parse them to obtain the identifier of the business segment to be adjusted and the adjustment period in the adjustment instructions; wherein, the adjustment instructions include at least one of natural language instructions and structured instructions; An analysis interface is used to receive the identifier of the business segment and the adjustment period from the input interface, generate the priority result of the business segment, and send it to the confirmation interface. A confirmation interface is used to receive confirmation instructions for the priority results from the analysis interface, and generate a production plan to be processed based on the confirmation instructions and send it to the orchestration layer.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, wherein, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images