Workflow execution method and device, nonvolatile storage medium and electronic equipment

Abstract

The invention discloses a workflow execution method and device, a nonvolatile storage medium and electronic equipment. The method comprises the following steps: determining scene description information corresponding to a task text; determining a working scene corresponding to the task text according to the scene description information, and determining a verification rule set according to the working scene; generating an initial workflow according to the working scene and the task text, and verifying the initial workflow according to the verification rule set; under the condition that the verification is passed, determining that the initial workflow is a workflow; under the condition that the verification is not passed, generating modification suggestion information according to a verification result, and displaying the modification suggestion information to the target object; and modifying the initial workflow according to a modification instruction sent by the target object. According to the method and the device, the technical problems of relatively high workflow execution cost and relatively long time consumption caused by manual configuration of related parameters of the workflow in related technologies are solved.

Description

Technical Field

[0001] This application relates to the field of industrial automation, and more specifically, to a workflow execution method, apparatus, non-volatile storage medium, and electronic device. Background Technology

[0002] Workflow systems in related technologies require significant manual configuration and maintenance when facing complex industrial scenarios. Furthermore, workflow parameters must be configured manually by professionals based on experience, making it impossible to achieve automated workflow generation and execution in these technologies.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a workflow execution method, apparatus, non-volatile storage medium, and electronic device to at least solve the technical problem that the workflow execution cost is high and the execution time is long due to the need for manual configuration of relevant parameters in the related art.

[0005] According to one aspect of the embodiments of this application, a workflow execution method is provided, including:

[0006] Optionally, generating a workflow based on the task text includes: determining the scenario description information corresponding to the task text; determining the work scenario corresponding to the task text based on the scenario description information, and determining a set of verification rules based on the work scenario; generating an initial workflow based on the work scenario and the task text, and verifying the initial workflow based on the set of verification rules; determining the initial workflow as a workflow if the verification passes; generating modification suggestion information based on the verification result if the verification fails, and displaying the modification suggestion information to the target object; and modifying the initial workflow based on the modification instructions sent by the target object.

[0007] Optionally, determining the scenario description information corresponding to the task text includes: determining multiple alternative scenario description information based on the semantic information of the task text; determining the work information of the target object, wherein the work information includes at least one of the following: job information, department information, project information; determining the degree of matching between the alternative scenario description information and the work information, and determining the alternative scenario description information with the highest degree of matching with the work information among the multiple alternative scenario description information as the scenario description information.

[0008] Optionally, generating modification suggestion information based on the verification results includes: identifying the verification items that failed in the verification results; identifying the target verification rules corresponding to the verification items; and generating modification suggestion information based on the target verification rules.

[0009] Optionally, determining the standardized capability unit corresponding to the work task node includes: determining the execution target corresponding to the work task node; and determining the standardized capability unit corresponding to the work task node based on the execution target.

[0010] Optionally, the second description information includes the device status; determining the parameter set corresponding to the standardized capability unit based on the first description information of the work task node and the second description information of the standardized capability unit includes: determining target historical task data in the historical database based on the first description information of the work task node, wherein the matching degree between the target historical task data and the first description information is higher than a preset threshold; constructing a scene context vector based on the target historical task data, the first description information and the second description information; and determining the parameter set based on the scene context vector.

[0011] Optionally, the method further includes: determining the confidence level of the workflow; if the confidence level is lower than a preset confidence threshold, instructing the target object to modify the workflow, or providing the target object with a preset workflow template.

[0012] According to another aspect of the embodiments of this application, a workflow execution apparatus is also provided, comprising: a workflow generation module, configured to acquire task text sent by a target object and generate a workflow based on the task text, wherein the workflow includes multiple task nodes; a matching module, configured to determine a standardized capability unit corresponding to a task node, wherein the standardized capability unit is used to execute a preset data processing task; a parameter generation module, configured to determine a parameter set corresponding to a standardized capability unit based on first description information of the task node and second description information of the standardized capability unit; and an execution module, configured to call the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set.

[0013] According to another aspect of the embodiments of this application, a non-volatile storage medium is also provided, wherein a program is stored in the non-volatile storage medium, and the program controls the device where the non-volatile storage medium is located to execute a workflow execution method when it runs.

[0014] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program executes a workflow execution method when it runs.

[0015] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps of a workflow execution method.

[0016] In this embodiment, the method involves obtaining the task text sent by the target object and generating a workflow based on the task text. The workflow includes multiple task nodes. A standardized capability unit corresponding to each task node is determined, whereby the standardized capability unit executes a preset data processing task. A parameter set corresponding to the standardized capability unit is determined based on the first description information of the task node and the second description information of the standardized capability unit. The standardized capability unit is then invoked to execute the task node corresponding to the standardized capability unit based on the parameter set. By determining the standardized capability unit corresponding to the task node and the parameter set corresponding to the standardized capability unit based on the first and second description information of the task node, the method achieves the goal of automatically determining the relevant parameters of the workflow. This realizes the technical effect of automatically generating and executing the workflow, thereby solving the technical problem of high workflow execution cost and long execution time caused by the need for manual configuration of workflow parameters in related technologies. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 This is a schematic diagram of the structure of a computer terminal (mobile device) according to an embodiment of this application;

[0019] Figure 2 This is a schematic diagram of the structure of a workflow execution system according to an embodiment of this application;

[0020] Figure 3 This is a flowchart illustrating a workflow execution method according to an embodiment of this application;

[0021] Figure 4 This is a flowchart illustrating a workflow generation process according to an embodiment of this application;

[0022] Figure 5 This is a schematic diagram of a parameter generation and verification process provided according to an embodiment of this application;

[0023] Figure 6 This is a schematic diagram of the execution flow of a work task node according to an embodiment of this application;

[0024] Figure 7 This is a schematic diagram of a display interface provided according to an embodiment of this application;

[0025] Figure 8 This is a schematic diagram of an interactive interface provided according to an embodiment of this application;

[0026] Figure 9 This is a schematic diagram of a workflow execution summary process provided according to an embodiment of this application;

[0027] Figure 10 This is a schematic diagram of a token threshold control process provided according to an embodiment of this application;

[0028] Figure 11 This is a schematic diagram of a workflow execution device provided according to an embodiment of this application. Detailed Implementation

[0029] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0031] To better understand the embodiments of this application, the technical terms involved in the embodiments of this application are explained below:

[0032] Workflow: A directed graph consisting of multiple nodes and edges, representing the execution flow of business logic.

[0033] Tool: The smallest unit of execution in the system, with clearly defined input and output and functional description.

[0034] Node: The execution unit in a workflow, which includes module name, tool name, configuration information, etc.

[0035] Edge: A directed connection that connects workflow nodes, defining the execution order and conditions.

[0036] JSON Schema: A specification for defining and validating JSON data structures.

[0037] LLM (Large Language Model): A large language model used for natural language understanding and generation.

[0038] MCP (Model Context Protocol): A protocol used to standardize the interaction between AI models and external systems.

[0039] OPC UA: Industrial communication standard used for data exchange between industrial devices.

[0040] Vector Database: A database system for storing and retrieving high-dimensional vector data.

[0041] x-source field: An extended field in JSON Schema used to identify the source and context of data.

[0042] Industrial workflow systems in related technologies (such as BPEL and BPMN) mainly rely on predefined static processes and lack intelligence and adaptability. When faced with complex industrial scenarios, they require a large amount of manual configuration and maintenance.

[0043] Furthermore, industrial sites contain numerous heterogeneous industrial control systems (DCS, APC, AAS, PLC, SCADA, MES, ERP, etc.), lacking unified interface standards, resulting in high integration costs and long development cycles. Moreover, industrial equipment parameter configuration typically requires manual settings by specialized engineers based on experience, lacking intelligent parameter generation and optimization mechanisms. In addition, systems in these technologies struggle to understand complex industrial scenarios and cannot dynamically adjust execution strategies based on historical data and real-time status. Furthermore, industrial systems in these technologies suffer from complex human-machine interfaces, high operational barriers, and a lack of intelligent interaction mechanisms.

[0044] In summary, the methods in the relevant technologies have at least the following technical problems:

[0045] Lack of intelligence: Existing workflow systems cannot automatically generate workflows based on user needs, requiring a large amount of manual design.

[0046] Parameter generation is difficult: the system cannot intelligently generate device parameters and relies on manual experience for configuration.

[0047] Weak context processing capabilities: It cannot effectively handle complex industrial scenario context and state information.

[0048] Limited integration capabilities: It is difficult to unify and integrate various heterogeneous industrial control systems and equipment.

[0049] Poor user experience: It lacks intelligent user interaction mechanisms and is complicated to operate.

[0050] To address the aforementioned issues, this application provides relevant solutions, which are detailed below.

[0051] According to an embodiment of this application, a method embodiment of a workflow execution method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0052] The methods and embodiments provided in this application can be executed on mobile terminals, computer terminals, or similar computing devices. Figure 1 A hardware block diagram of a computer terminal (or mobile device) for implementing a workflow execution method is shown. Figure 1 As shown, the computer terminal 10 (or mobile device 10) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0053] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0054] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the workflow execution method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned workflow execution method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0055] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0056] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0057] According to an embodiment of this application, a workflow execution system is also provided, which can be used to execute... Figure 3 The workflow execution method is shown below. Figure 2As can be seen, the system includes a Human-Computer Interaction Layer 20, a Workflow Generation & Scenario Recognition Layer 22, a Tool Abstraction Layer 24, an Intelligent Parameter Generation Layer 26, a Parameter Validation & Error Correction Layer 28, a Data Integration Layer 210, and a Result Processing & Summarization Layer 212.

[0058] In some embodiments of this application, the human-computer interaction layer 22 includes an interaction management module, an interface generation module, a dynamic interaction mechanism module, and a dynamic page and form module. The interaction management module supports various interaction types, including form input, confirmation operations, custom pages, and data display. This module specifically provides manual editing functionality for low-confidence workflows; when the confidence level of the system-generated workflow is low, users can manually adjust and optimize it through an intuitive interface. The module also implements a secondary confirmation mechanism for sensitive operations to ensure the security of important operations. For complex interaction scenarios, such as the visualization of the model identification process, the module can provide a rich interactive experience to help users better understand the system's workflow.

[0059] The interface generation module automatically generates user interfaces based on workflow and context information, significantly reducing the workload of interface development. This module supports dynamic form generation, automatically adjusting form structure and fields according to different workflow requirements. It also provides chart visualization and real-time data monitoring functions to help users intuitively understand the system's running status and execution results. The interface adopts a responsive design, adapting to different device types, including desktop computers, tablets, and mobile phones. The module also supports theme customization and personalized configuration, allowing users to adjust the interface appearance and functional layout according to their preferences and needs.

[0060] The dynamic interaction mechanism module supports dynamic adjustment of displayed content via push notifications during runtime. The system provides three main view operations: `add_view` for adding new view components, `update_view` for updating existing view content, and `append_view` for appending content to an existing view. These operations support various display formats, including card (card layout), markdown (rich text format), echarts (interactive charts), and plotly (advanced visualization charts), as well as flexible combinations of these formats to meet the display needs of different industrial scenarios.

[0061] The dynamic page and form modules support real-time interface adjustments during capability execution. When the workflow reaches a specific node, the system can dynamically generate corresponding pages and forms based on the execution status and context information. These dynamic interfaces not only display the current execution status but also receive user operation results and pass these results to subsequent capability nodes, achieving closed-loop execution of human-machine collaboration. This mechanism is particularly suitable for industrial scenarios requiring manual confirmation, parameter adjustment, or anomaly handling, ensuring stable workflow execution in complex environments.

[0062] In some embodiments of this application, the workflow generation and scene recognition layer 22 is responsible for scene recognition, workflow generation and verification, and processing of low-confidence workflows. This layer integrates scene recognition and workflow generation into a unified processing flow, improving the system's intelligence and execution efficiency.

[0063] In some embodiments of this application, the workflow generation and scene recognition layer 22 further includes an LLM fine-tuning module, a workflow verification module, and a low-confidence processing module. The LLM fine-tuning module is used to construct a training dataset specifically for industrial scenarios by collecting a large amount of historical workflow data, business scenario descriptions, and corresponding execution results from the industrial field. This dataset contains key information such as user questions, scenario descriptions, generated workflows, and execution effects. The system uses a large language model based on the Transformer architecture for fine-tuning, training the model to deeply understand the semantic features of industrial scenarios and generate workflow structures that conform to industrial standards. After fine-tuning, this module serves as a fundamental component of the system, providing services for all workflow generation requests. This module also supports confidence calculation, enabling the evaluation of the reliability of generated workflows and providing important basis for subsequent processing decisions. The model's input includes user questions, scenario descriptions, and historical context; the output is a structured JSON-formatted workflow definition and the corresponding confidence score.

[0064] The workflow verification module is responsible for performing a comprehensive quality check on the workflows generated by the LLM fine-tuning module. This module first performs syntax and semantic validation to ensure the generated workflow is structurally correct. It then checks the validity of node connections and the completeness of parameter configurations to verify that the workflow complies with industry safety standards and best practices. The module also performs constraint rule analysis to ensure the workflow meets various industry specifications. When problems are found, this module can provide workflow optimization suggestions and error correction solutions to ensure the final output workflow has high quality and reliability.

[0065] The low-confidence handling module addresses low-confidence situations that arise during workflow generation. This module continuously monitors the confidence score of the generated workflow, and when the confidence score falls below a preset threshold, a special processing procedure is initiated. In this situation, the system provides users with multiple options, including directly executing the currently generated workflow, editing an existing workflow template, selecting a historical workflow case, or choosing a preset workflow template. This module also provides an interactive workflow editing interface, allowing users to manually adjust and optimize the workflow. Simultaneously, the system can intelligently retrieve relevant cases from the historical workflow library and preset workflow templates, providing users with references and options.

[0066] In some embodiments of this application, the capability abstraction layer 24 is used to abstract various heterogeneous industrial control system functions and third-party services into a standardized capability interface, providing a unified service for the execution of upper-layer workflows.

[0067] Optionally, the capability abstraction layer 24 includes a capability standardization module, a capability execution engine, and a third-party capability access module. The capability standardization module is responsible for abstracting the functions of various industrial control systems (including DCS, APC, AAS, PLC, SCADA, MES, ERP, etc.) into standardized capability units. Each standardized capability unit contains complete metadata information (i.e., secondary description information), including name, description, input schema, output schema, and execution interface. This module supports multiple access methods, including Python scripts, RESTful APIs, MCP protocol, OPC UA, etc., ensuring effective integration with various industrial control systems. The module also provides a comprehensive capability registration, discovery, and version management mechanism, supporting dynamic capability registration and version control to ensure system scalability and maintainability.

[0068] The capability execution engine is built upon a concurrent execution framework based on the Actor model, ensuring high performance and reliability for capability execution. This engine supports asynchronous capability execution, handling concurrent calls to multiple capabilities simultaneously, significantly improving system efficiency. It also implements robust timeout control and error handling mechanisms to ensure timely response and processing in case of exceptions during capability execution. The engine provides data transfer and state synchronization mechanisms between capabilities, supporting data flow and state sharing between different capabilities in complex workflows. Furthermore, the engine supports dynamic loading and hot updating of capabilities, enabling the system to update and expand capabilities without downtime, enhancing system flexibility and availability.

[0069] The third-party capability access module is specifically responsible for integrating and managing third-party service capabilities, including cloud service APIs, external data sources, and third-party AI services. This module abstracts these external services into standardized capabilities, enabling seamless integration with internal industrial control system capabilities. The module supports various access methods for third-party services, including RESTful APIs, GraphQL, and WebSocket, and provides unified authentication, authorization, and error handling mechanisms. Through this module, the system can expand its capabilities, integrate more external services, and provide users with richer functional support.

[0070] In some embodiments of this application, the intelligent parameter generation layer 26 is used to intelligently generate the optimal parameter configuration based on the execution context and historical data, which can reduce the workload of manual configuration.

[0071] Optionally, the intelligent parameter generation layer 26 includes a context understanding module, a parameter generation module, a script assistance module, and a vector retrieval module. The context understanding module is responsible for in-depth analysis of multi-dimensional information such as the current execution state, historical data, and device status, extracting key features and patterns to construct a scene context vector. This module uses an attention mechanism to understand the importance of different contextual information, automatically identifying the information most relevant to parameter generation. The module ultimately generates a structured context representation, providing accurate input for subsequent parameter generation.

[0072] The parameter generation module intelligently generates input parameters for standardized capability units based on contextual scenario vectors and workflows. When the amount of contextual information is too large, the module generates parameter transformation functions instead of direct parameters. These transformation functions are then executed to process historical data, yielding the final parameters. The module supports step-by-step parameter generation and incremental updates, dynamically adjusting parameters based on feedback during execution. Furthermore, the module provides confidence assessment of parameter generation, helping the system determine the quality and reliability of the generated parameters.

[0073] The script assistance module provides dynamic data processing capabilities, such as converting relative time descriptions into specific absolute times. It supports intelligent parsing of various time expressions, such as converting "today" to the current date, "yesterday" to the previous day, and "the last 7 days" to a specific date range. The module also supports dynamic calculation of relative values, such as intelligent parsing and calculation of expressions like "120% of current output" and "1.5 times the average." Furthermore, the module provides a rich script template library, including time conversion scripts, numerical calculation scripts, and data aggregation scripts, enabling the rapid generation of corresponding data processing scripts based on different industrial scenario requirements, greatly improving the flexibility and accuracy of parameter generation.

[0074] The vector retrieval module uses the `x-source` field for vectorized representation, establishing an efficient vector database to store historical parameter configurations and their corresponding execution effects. It employs similarity retrieval technology to find the most relevant historical configurations, supporting fuzzy parameter matching and intelligent recommendation functions. For older systems, especially those lacking fuzzy matching capabilities, vector retrieval enables precise matching, resolving the issue of systems being unable to handle fuzzy parameter matching in related technologies. When generating new parameters, the vector retrieval module can quickly search for similar historical cases, providing valuable reference information for parameter generation.

[0075] In some embodiments of this application, the parameter verification and error correction layer 28 is used to ensure that the generated parameters meet the specification requirements and to automatically correct and optimize when problems occur.

[0076] Optionally, the parameter validation and error correction layer 28 includes a JSON Schema validation module and an intelligent error correction module. The JSON Schema validation module performs rigorous validity validation on the generated parameters based on the input Schema. This module not only checks the basic format and type of the parameters but also performs in-depth semantic validation to ensure the parameters are correct in business logic. When validation fails, the module provides detailed error information and specific correction suggestions to help the system quickly locate and resolve problems. The module also supports custom validation rules and business logic checks, enabling flexible validation configuration according to different industrial scenario requirements.

[0077] The intelligent error correction module activates when parameter validation fails. By deeply analyzing the cause of the error and combining error information with contextual information, it intelligently regenerates the parameters or transformation function. The intelligent error correction module supports multiple rounds of error correction and parameter optimization, enabling iterative improvement of parameter quality. It also records the complete error correction process, including error type, correction strategy, and final result. This information is used for continuous model improvement and optimization, enhancing the overall performance and reliability of the system.

[0078] In some embodiments of this application, the data access layer 210 is responsible for acquiring and processing data from various data sources to provide unified data services to the upper-layer system.

[0079] Optionally, the data access layer 210 includes a field data access module and a file upload module. The field data access module supports multiple industrial protocols, including OPC UA, Modbus, and Ethernet / IP, enabling communication with various industrial equipment and control systems. This module provides a standardized data acquisition interface, supporting the acquisition of real-time and historical data to ensure the system obtains a complete data view. The module also supports multiple database connections, including MySQL, PostgreSQL, and InfluxDB, enabling data acquisition from different data storage systems. To ensure data quality, the module also provides data preprocessing and cleaning functions, automatically identifying and repairing outliers, missing values, and other issues in the data.

[0080] The file upload module supports uploading and processing various file formats, including common formats such as Excel, CSV, JSON, and XML. This module provides intelligent file parsing and data extraction capabilities, automatically identifying file structure and extracting valid data. It supports batch file processing and asynchronous uploads, enabling efficient handling of large numbers of files. To ensure data security and traceability, the module also provides file version management and access control functions, supporting file version tracking and access permission management.

[0081] In some embodiments of this application, the result processing and summarizing layer 212 is used to intelligently process, analyze and summarize the workflow execution results, providing users with clear and useful execution reports.

[0082] Optionally, the result processing and summarization layer 212 includes a step-by-step summary module, a final summary generation module, a token threshold control module, and an auxiliary summary module. The step-by-step summary module performs real-time summarization and status recording of the execution results at each node in the workflow, ensuring the integrity and traceability of the execution process. This module maintains intermediate states and key information during execution, providing a detailed data foundation for subsequent analysis and summarization. The module also provides node-level result analysis and anomaly detection functions, enabling timely identification of problems and anomalies during execution.

[0083] The final summary generation module summarizes and analyzes the overall results after the workflow execution is complete. This module integrates the execution results of each node to generate a comprehensive execution report, providing users with a complete overview of the execution status. The module also provides execution effectiveness evaluation and optimization suggestions to help users understand the workflow's execution performance and identify improvement opportunities.

[0084] The auxiliary summary module is used for complex data analysis and result formatting using program scripts (such as Python scripts), providing the system with powerful data processing capabilities. This module supports custom data processing logic and summary templates, enabling the generation of personalized summary reports based on different needs. The module also provides high-performance data processing and visualization functions, supporting multiple output formats, including text, charts, and reports, to meet the needs of different users.

[0085] Under the above operating environment, embodiments of this application provide a workflow execution method, such as... Figure 3 As shown, the method includes the following steps:

[0086] Step S302: Obtain the task text sent by the target object, and generate a workflow based on the task text, wherein the workflow includes multiple task nodes;

[0087] In the technical solution provided in step S302, the step of generating a workflow based on the task text includes: determining the scenario description information corresponding to the task text; determining the work scenario corresponding to the task text based on the scenario description information, and determining the set of verification rules based on the work scenario; generating an initial workflow based on the work scenario and the task text, and verifying the initial workflow based on the set of verification rules; determining the initial workflow as a workflow if the verification passes; generating modification suggestion information based on the verification result if the verification fails, and displaying the modification suggestion information to the target object; and modifying the initial workflow based on the modification instructions sent by the target object.

[0088] As an optional implementation, the step of determining the scene description information corresponding to the task text includes: determining multiple candidate scene description information based on the semantic information of the task text; determining the work information of the target object, wherein the work information includes at least one of the following: job information, department information, project information; determining the degree of matching between the candidate scene description information and the work information, and determining the candidate scene description information with the highest degree of matching with the work information among the multiple candidate scene description information as the scene description information.

[0089] The target object's work information can help determine the scene description information corresponding to the task text input by the target object. For example, if the input task text contains abbreviations that can be interpreted in multiple ways, the possible meanings of the abbreviations can be determined based on the target object's work information and the context of the abbreviations in the task text, or the probability of selecting the various meanings corresponding to the abbreviations can be determined.

[0090] In some embodiments of this application, the step of generating modification suggestion information based on the verification results includes: determining the verification items that failed in the verification results; determining the target verification rule corresponding to the verification item; and generating modification suggestion information based on the target verification rule.

[0091] In some embodiments of this application, the method further includes: determining the confidence level of the workflow; and if the confidence level is lower than a preset confidence threshold, instructing the target object to modify the workflow, or providing the target object with a preset workflow template.

[0092] In some embodiments of this application, it is assumed that the user question contained in the task text is Q, the scene description information is S, and the historical context determined based on the task text is C, such as... Figure 4 As shown, the workflow can be determined using the following process:

[0093] The first step is model fine-tuning and preparation.

[0094] Model fine-tuning and preparation are completed before system deployment. In this step, the system specifically fine-tunes the large language model based on existing industrial data, enabling it to support the generation of original workflows that conform to industry standards. During this process, the system constructs a comprehensive training dataset containing historical workflows, scenario descriptions, and execution results. Supervised learning and reinforcement learning are used to train the model, allowing it to deeply understand the semantic features and workflow structure of industrial scenarios. After fine-tuning, the model serves as a fundamental component of the system, providing services for all workflow generation requests.

[0095] The second step is intent recognition and scene analysis.

[0096] In this step, by analyzing user input, the system intelligently identifies the user's true intent and the type and complexity of the business scenario. The system directly processes user questions, scenario descriptions, and historical context, extracting key business elements and constraints, and analyzing the similarity between the current scenario and historical cases to provide accurate contextual information for subsequent workflow generation. This stage simplifies the processing flow, eliminating intermediate natural language understanding and context encoding steps, thus improving processing efficiency. The historical context includes execution cases of relevant historical workflows retrieved based on user text.

[0097] The third step is workflow generation and confidence calculation.

[0098] In this step, an initial workflow is generated from a finely tuned large language model, and the confidence score of the generated workflow is calculated. The system performs constraint rule analysis to verify whether the generated workflow conforms to industry standards, and extracts relevant cases from the historical workflow library and pre-built workflow templates for reference.

[0099] Step 4: Low Confidence Handling and User Selection

[0100] In this step, when the system determines that the generated workflow has insufficient confidence, it provides the user with several processing options for selection, including directly executing the currently generated workflow, editing an existing workflow template, selecting a historical workflow case, or selecting a preset workflow template. The system also provides an interactive workflow editing interface, supporting manual adjustments and optimizations. This stage integrates low-confidence handling and user selection decisions into a unified processing flow, improving system response efficiency and user experience.

[0101] Step 5: Workflow Optimization

[0102] This step involves the final verification and optimization of the generated workflow. The system verifies the correctness and security of the workflow, optimizes execution efficiency and resource utilization, and ensures that the final output workflow has high quality and reliability.

[0103] Step S304: Determine the standardized capability unit corresponding to the work task node, wherein the standardized capability unit is used to execute the preset data processing task;

[0104] In the technical solution provided in step S304, determining the standardized capability unit corresponding to the work task node includes: determining the execution target corresponding to the work task node; and determining the standardized capability unit corresponding to the work task node based on the execution target.

[0105] In some embodiments of this application, the execution target of a task node includes the data that the task node expects to obtain after execution, or the action that the device to be invoked is expected to perform.

[0106] Step S306: Based on the first description information of the work task node and the second description information of the standardized capability unit, determine the parameter set corresponding to the standardized capability unit.

[0107] In the technical solution provided in step S306, the second description information includes the device status; determining the parameter set corresponding to the standardized capability unit based on the first description information of the work task node and the second description information of the standardized capability unit includes: determining target historical task data in the historical database based on the first description information of the work task node, wherein the matching degree between the target historical task data and the first description information is higher than a preset threshold; constructing a scene context vector based on the target historical task data, the first description information and the second description information; and determining the parameter set based on the scene context vector.

[0108] Optionally, the aforementioned device status can be the status of the physical device corresponding to the standardized capability unit, or the status of a virtual device corresponding to the standardized capability unit for performing a specific data processing task or computing task.

[0109] In some embodiments of this application, the parameter generation and verification process is as follows: Figure 5 As shown, it includes the following stages:

[0110] Data collection phase: Collect various data during the workflow operation, including historical execution data, current status information, equipment parameters, etc., to provide a complete data foundation for parameter generation.

[0111] Information extraction stage: Extract relevant information based on the x-source field, including key information such as data source, context information, and historical configuration, to provide accurate input data for parameter generation.

[0112] Schema assembly phase: Assemble the JSON schema of the transformation function, define the input and output formats and constraints of the transformation function, and ensure that the generated transformation function conforms to the standard specification.

[0113] Output Schema Extraction Stage: Extract the JSON Schema of the transformation function output (target capability input) to clarify the parameter format and validation rules required for the target capability.

[0114] Context assembly phase: The collected data, extracted information, and schema definitions are assembled into a complete context, providing comprehensive input information for parameter generation.

[0115] Parameter generation phase: Based on the assembled context, the system generates a transformation function or directly generates parameters. The system selects the most suitable generation strategy according to data complexity and processing requirements.

[0116] Verification and Retry Phase: The generated parameters are verified to ensure they meet the target capability requirements. If verification fails, the system will regenerate the parameters until compliant parameters are generated.

[0117] Step S308: Based on the parameter set, call the standardized capability unit to execute the work task node corresponding to the standardized capability unit.

[0118] In some embodiments of this application, the execution flow of the work task node is as follows: Figure 6As shown, the process includes steps such as capability invocation request, parameter generation, context analysis, vector retrieval, parameter verification, and error correction. During the context analysis phase, the context size is analyzed. If the context size is not less than a preset threshold, a parameter set is not directly generated; instead, a transformation function is generated and executed to process historical data to obtain the parameter set. If the context size is less than the preset threshold, parameters are directly generated using an LLM model. Optionally, parameters can be dynamically adjusted based on feedback during execution. Furthermore, confidence assessment is performed on the parameters during generation to help the system determine the quality and reliability of the generated parameters.

[0119] Optionally, the context analysis step also requires in-depth analysis of multi-dimensional information such as the current execution state, historical data, and device status to extract key features and patterns and construct a scene context vector. Furthermore, an attention mechanism can be used to understand the importance of different contextual information, automatically identifying the information most relevant to parameter generation. The module ultimately generates a structured context representation, providing accurate input for subsequent parameter generation.

[0120] In some embodiments of this application, when performing vector retrieval, the most relevant historical configurations can be found using similarity retrieval techniques to provide a reference for the parameter generation process.

[0121] During the parameter validation and error correction step, the generated parameters can be rigorously validated based on the input schema of the capability. This includes not only checking the basic format and type of the parameters but also performing in-depth semantic validation to ensure the parameters are logically correct. When validation fails, detailed error information and specific correction suggestions can be provided to help the system quickly locate and resolve problems. Furthermore, validation rules and business logic checks can be customized, allowing for flexible validation configurations based on the needs of different industrial scenarios.

[0122] During the error correction phase, parameters or transformation functions can be intelligently regenerated by deeply analyzing the causes of errors and combining error information with contextual information. Furthermore, multiple rounds of error correction and parameter optimization can be performed during error correction, thereby gradually improving parameter quality through iteration. Additionally, the entire error correction process can be recorded, including error type, error correction strategy, and final result. This information is used for continuous model improvement and optimization, enhancing the overall performance and reliability of the system.

[0123] In some embodiments of this application, a method such as... is also provided. Figure 7 The display interface shown, and as Figure 8 The interactive interface shown is used to demonstrate how target users (including staff) can easily interact with the system to determine the execution results of workflows and modify them.

[0124] In some embodiments of this application, such as Figure 9 As shown, each step of the workflow can be summarized during its execution, and the final summary result of the workflow can be generated based on the summary results of each step.

[0125] Additionally, during the summary process, to avoid consuming too many tokens, you can use methods such as... Figure 10 The method shown controls the tokens. Optionally, if the number of tokens required for the context to be summarized is less than a preset threshold, the summary can be performed directly. Otherwise, the context is compressed and summarized using programming-assisted analysis (Python, etc.).

[0126] In some embodiments of this application, the following are also provided: Figure 10 The workflow shown is as follows. Figure 10 The workflow shown can be divided into the following stages:

[0127] User Input Processing Phase: This is the starting point for system execution. The system receives the user's question or scenario description, uses natural language understanding technology to identify intent, and accurately understands the user's true needs. During this process, the system extracts key information and constraints, providing an accurate input basis for subsequent processing.

[0128] Scene Identification and Decomposition Phase: This phase requires in-depth analysis of user needs. The system analyzes the complexity of the scene, identifies the capabilities and resources required to complete the task, and generates a detailed execution plan. This phase ensures that the system accurately understands the task requirements and provides guidance for subsequent workflow generation.

[0129] Workflow Generation Phase: This phase involves calling a fine-tuned large language model to generate a workflow that meets the requirements. The system verifies the correctness of the generated workflow, ensuring its logical rationality, and optimizes the workflow structure to improve execution efficiency.

[0130] Capability Execution Phase: This phase employs a cyclical execution mechanism, executing each node sequentially according to the workflow definition. The system first initializes the workflow execution environment, then iteratively retrieves the next node to be executed. For each node, the system generates corresponding input parameters and calls the appropriate capability interface to complete the specific task. After execution, the system directly determines whether the node was successfully executed. If successful, it checks if the branching conditions are met and decides whether to continue executing subsequent nodes based on those conditions. If execution fails or the branching conditions are not met, the system directly enters the final summary phase. This cyclical process continues until all nodes have been executed or a termination condition is encountered.

[0131] Final summary stage: In this stage, the execution results need to be fully processed and analyzed to generate complete execution results.

[0132] In some embodiments of this application, assuming the application scenario requires optimization of PID parameters, and the received task text is "The pH fluctuation in the neutralization tank is a bit large, I want to tune the PID parameters," the system will use natural language understanding technology to identify that this is a PID optimization scenario and automatically generate the following workflow:

[0133] PID / Loop Configuration Query → PID / Data Verification → PID / Pre-tuning → PID / Control Loop Performance Evaluation and Diagnosis → PID / Intelligent Wizard Tuning

[0134] The edges in the workflow define conditional constraints: after a pre-tuning node executes successfully, the `valid` field in the output JSON Schema must be true for subsequent performance evaluation and smart wizard tuning steps to continue. The system calculates the workflow confidence level at 0.92, which falls within the high confidence range.

[0135] The following steps will involve the gradual implementation and interactive demonstration of capabilities:

[0136] Step 1: PID / Loop Configuration Query

[0137] The system generates input parameters for PID / loop configuration query capabilities. The JSON schema includes fields such as device name, loop name, and description. The device name, loop name, and description fields contain x-source identifiers, pointing to different vector libraries. The system performs similarity matching in the vector library of the description field through vector retrieval, successfully matching the configuration related to "pH value of the neutralization tank" and finding the corresponding loop name "PH-001".

[0138] After execution, the system uses the add_view operation to add a card format view, displaying key information such as the loop name "PH-001", the device name "neutralization tank", the current PID parameters (Kp=2.5, Ki=0.8, Kd=1.2), and the control target value (PH=7.0±0.2).

[0139] Step 2: PID / Data Verification

[0140] The system invokes the PID / data verification capability to verify the historical data quality and integrity of loop PH-001, ensuring that the data meets the pre-tuning requirements. The verification results show that the data integrity is 98.5%, the data quality is good, and the pre-tuning requirements are met.

[0141] The system uses ECharts format to display the pH value change curve over the past 24 hours, including three curves: setpoint, actual value, and control output. This clearly shows the current control effect and fluctuations, providing a visual basis for subsequent analysis.

[0142] Step 3: PID / Pre-tuning

[0143] The system performs PID / pre-tuning, automatically calculating preliminary PID parameters based on historical data. The pre-tuning algorithm analyzes historical data, identifies system response characteristics, and calculates preliminary parameters (Kp=2.6, Ki=1.0, Kd=1.0). After pre-tuning, the system checks the valid field in the output, confirming a value of true, thus meeting the conditions for continuing execution.

[0144] The system uses Markdown format to display the technical details of the pre-tuning process, including the algorithms used, the amount of data analyzed, and the key parameters calculated.

[0145] Step 4: PID / Control Loop Performance Evaluation and Diagnosis

[0146] The system utilizes PID / control loop performance evaluation and diagnostic capabilities to comprehensively analyze the current control loop performance, identifying existing problems and areas for improvement. The evaluation results show the following current performance indicators: overshoot 15.2%, settling time 8.5 minutes, steady-state error ±0.15, and oscillation count 3. The main problem is oscillation caused by excessively short integral time.

[0147] The system uses the plotly format to display detailed performance analysis results, including time-domain response curves, frequency-domain characteristic analysis, and control performance index comparison charts, intuitively showing the current control effect and existing problems.

[0148] Step 5: PID / Smart Wizard Tuning

[0149] Based on the pre-tuning results and performance evaluation report, the system executes PID / intelligent wizard tuning capabilities to generate final optimized parameter suggestions. The intelligent algorithm comprehensively considers system characteristics, performance requirements, and industrial experience to generate optimized parameters (Kp=2.8, Ki=1.2, Kd=0.9), predicting performance indicators of overshoot 8.5%, settling time 6.2 minutes, and steady-state error ±0.08, significantly improving control performance.

[0150] The system uses the plotly format to display parameter comparisons and performance predictions before and after optimization, including parameter change comparison charts, performance index improvement prediction charts, and control effect simulation comparisons, providing users with an intuitive display of optimization results.

[0151] The system can then generate a dynamic form that users can interact with to select "Apply new parameters," "Retune," or "Keep original parameters." When the user selects "Apply new parameters," the system passes the user's selection to the subsequent PID parameter configuration update capability, completing the entire optimization process.

[0152] In some embodiments of this application, it is assumed that the application scenario is a predictive analytics scenario, and the received task text is "predict the output of H2S in the next 24 hours". Based on the received task text, the system will identify the application scenario as a predictive analytics scenario and automatically generate the following workflow:

[0153] ML / Modeling Problems and Standardization → ML / Inference Computation

[0154] The system calculates the workflow confidence level as 0.89, which is in the high confidence range, and proceeds directly to the execution phase.

[0155] The following steps will involve the gradual implementation and interactive demonstration of capabilities:

[0156] Step 1: ML / Modeling Issues and Standardization

[0157] The system utilizes ML / modeling and standardization capabilities, first matching relevant data source information in the vector library, including historical H2S production data, related process parameters, and environmental conditions. The matching results show that 30 days of historical data were found, including eight key parameters such as temperature, pressure, flow rate, and catalyst activity.

[0158] When the system detects missing key data, it pushes an update_view operation to the user through a dynamic interaction mechanism, prompting the user to manually upload the relevant data files in card format. After the user uploads the supplementary data, the system confirms that the data integrity reaches 95%, meeting the modeling requirements.

[0159] The system preprocesses and performs feature engineering on the data, automatically identifies outliers and missing values, cleans and standardizes the data, and constructs a suitable machine learning model. After feature selection and model selection, an LSTM neural network is chosen for time series forecasting.

[0160] The system uses Markdown format to display the technical details of data preprocessing, including data cleaning rules, feature engineering methods, model selection criteria, training data volume (30 days of historical data, a total of 720 data points), and other information.

[0161] The system performed model training, using the first 25 days of data as the training set and the last 5 days of data as the validation set. After training, the model performed well on the validation set with R²=0.89, MAE=45.2kg / h, and RMSE=62.8kg / h.

[0162] The system uses the echarts format to display the model training process, including training loss curves, validation loss curves, model convergence status, etc., allowing users to understand the detailed process of model training.

[0163] Step 2: ML / Inference Computation

[0164] The system performs ML / inference computations and predicts H2S production for the next 24 hours based on a trained model. The prediction results show that the current production is 1250 kg / h, and the predicted production after 24 hours is 1180 kg / h, showing a downward trend, with an average predicted production of 1215 kg / h.

[0165] The system uses the add_view operation to display the H2S production forecast curve for the next 24 hours in echarts format, including historical 7-day data comparison, forecast value, upper and lower limits of the 95% confidence interval, and intuitively displays the forecast trend and uncertainty range.

[0166] The system appends specific prediction results to the existing interface using the append_view operation. Key indicators of the prediction results are displayed in card format, including average predicted yield (1215 kg / h), maximum predicted yield (1280 kg / h), minimum predicted yield (1150 kg / h), prediction accuracy (R²=0.89), and model confidence (92.3%).

[0167] By acquiring the task text sent by the target object and generating a workflow based on the task text, wherein the workflow includes multiple task nodes; determining the standardized capability unit corresponding to the task node, wherein the standardized capability unit is used to execute a preset data processing task; determining the parameter set corresponding to the standardized capability unit based on the first description information of the task node and the second description information of the standardized capability unit; and calling the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set, the purpose of automatically determining the relevant parameters of the workflow is achieved. This realizes the technical effect of automatically generating and executing the workflow, and solves the technical problem of high workflow execution cost and long execution time caused by the need for manual configuration of relevant parameters in related technologies.

[0168] This application provides a workflow execution device. Figure 11 This is a schematic diagram of the device. From Figure 11 As can be seen from the diagram, the device includes: a workflow generation module 110, used to acquire the task text sent by the target object and generate a workflow based on the task text, wherein the workflow includes multiple task nodes; a matching module 112, used to determine the standardized capability unit corresponding to the task node, wherein the standardized capability unit is used to execute a preset data processing task; a parameter generation module 114, used to determine the parameter set corresponding to the standardized capability unit based on the first description information of the task node and the second description information of the standardized capability unit; and an execution module 116, used to call the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set.

[0169] In some embodiments of this application, the workflow generation module 110 generates a workflow based on a task text, including: determining the scenario description information corresponding to the task text; determining the work scenario corresponding to the task text based on the scenario description information, and determining a set of verification rules based on the work scenario; generating an initial workflow based on the work scenario and the task text, and verifying the initial workflow based on the set of verification rules; determining the initial workflow as a workflow if the verification passes; generating modification suggestion information based on the verification result if the verification fails, and displaying the modification suggestion information to the target object; and modifying the initial workflow based on the modification instruction sent by the target object.

[0170] In some embodiments of this application, the step of the workflow generation module 110 in determining the scene description information corresponding to the task text includes: determining multiple alternative scene description information based on the semantic information of the task text; determining the work information of the target object, wherein the work information includes at least one of the following: job information, department information, project information; determining the degree of matching between the alternative scene description information and the work information, and determining the alternative scene description information with the highest degree of matching with the work information among the multiple alternative scene description information as the scene description information.

[0171] In some embodiments of this application, the step of the workflow generation module 110 generating modification suggestion information based on the verification results includes: determining the verification items that failed in the verification results; determining the target verification rule corresponding to the verification item; and generating modification suggestion information based on the target verification rule.

[0172] In some embodiments of this application, the workflow generation module 110 is further configured to: determine the confidence level of the workflow; and, if the confidence level is lower than a preset confidence threshold, instruct the target object to modify the workflow or provide the target object with a preset workflow template.

[0173] In some embodiments of this application, the step of matching module 112 in determining the standardized capability unit corresponding to the work task node includes: determining the execution target corresponding to the work task node; and determining the standardized capability unit corresponding to the work task node based on the execution target.

[0174] In some embodiments of this application, the second description information includes device status; the step of parameter generation module 114 determining the parameter set corresponding to the standardized capability unit based on the first description information of the work task node and the second description information of the standardized capability unit includes: determining target historical task data in a historical database based on the first description information of the work task node, wherein the matching degree between the target historical task data and the first description information is higher than a preset threshold; constructing a scene context vector based on the target historical task data, the first description information and the second description information; and determining the parameter set based on the scene context vector.

[0175] It should be noted that each module in the above-mentioned workflow execution device can be a program module (for example, a set of program instructions that implement a certain function) or a hardware module. For the latter, it can be manifested in the following forms, but is not limited to them: each of the above modules is manifested as a processor, or the functions of each of the above modules are implemented by a processor.

[0176] This application provides a non-volatile storage medium storing a program. During program execution, the device containing the non-volatile storage medium performs the following workflow execution method: acquiring task text sent by a target object and generating a workflow based on the task text, wherein the workflow includes multiple task nodes; determining the standardized capability unit corresponding to each task node, wherein the standardized capability unit is used to execute a preset data processing task; determining a parameter set corresponding to the standardized capability unit based on a first description of the task node and a second description of the standardized capability unit; and calling the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set.

[0177] This application provides an electronic device, including a memory and a processor. The processor is used to run a program stored in the memory. When the program runs, it executes the following workflow execution method: obtaining task text sent by a target object and generating a workflow based on the task text, wherein the workflow includes multiple task nodes; determining a standardized capability unit corresponding to each task node, wherein the standardized capability unit is used to execute a preset data processing task; determining a parameter set corresponding to the standardized capability unit based on first description information of the task node and second description information of the standardized capability unit; and calling the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set.

[0178] This application provides a computer program product, including a computer program, characterized in that, when executed by a processor, the computer program implements the following workflow execution method steps: obtaining task text sent by a target object and generating a workflow based on the task text, wherein the workflow includes multiple task nodes; determining the standardized capability unit corresponding to the task node, wherein the standardized capability unit is used to execute a preset data processing task; determining the parameter set corresponding to the standardized capability unit based on the first description information of the task node and the second description information of the standardized capability unit; and calling the standardized capability unit to execute the task node corresponding to the standardized capability unit based on the parameter set.

[0179] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0180] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0181] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0182] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0183] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0184] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A workflow execution method, characterized in that, include: Obtain the task text sent by the target object, and generate a workflow based on the task text, wherein the workflow includes multiple task nodes; Determine the standardized capability unit corresponding to the work task node, wherein the standardized capability unit is used to execute a preset data processing task; Based on the first description information of the work task node and the second description information of the standardized capability unit, determine the parameter set corresponding to the standardized capability unit; Based on the parameter set, the standardized capability unit is invoked to execute the work task node corresponding to the standardized capability unit.

2. The workflow execution method according to claim 1, characterized in that, The workflow generation based on the task text includes: Determine the scene description information corresponding to the task text; Based on the scenario description information, determine the work scenario corresponding to the task text, and determine the set of verification rules based on the work scenario; An initial workflow is generated based on the work scenario and the task text, and the initial workflow is validated according to the set of validation rules. If the verification passes, the initial workflow is determined to be the workflow. If the verification fails, modification suggestions are generated based on the verification results and displayed to the target object. The initial workflow is modified according to the modification instructions sent by the target object.

3. The workflow execution method according to claim 2, characterized in that, Determining the scene description information corresponding to the task text includes: Based on the semantic information of the task text, multiple alternative scenario descriptions are determined; Determine the work information of the target object, wherein the work information includes at least one of the following: job information, department information, and project information; Determine the degree of matching between the candidate scenario description information and the work information, and determine the candidate scenario description information with the highest degree of matching with the work information among the multiple candidate scenario description information as the scenario description information.

4. The workflow execution method according to claim 2, characterized in that, Based on the verification results, the generated modification suggestions include: Identify the verification items that failed in the verification results; Determine the target verification rule corresponding to the verification item; The modification suggestion information is generated based on the target verification rules.

5. The workflow execution method according to claim 1, characterized in that, Determining the standardized capability unit corresponding to the work task node includes: Determine the execution target corresponding to the work task node; The standardized capability unit corresponding to the work task node is determined based on the execution objective.

6. The workflow execution method according to claim 1, characterized in that, The second description information includes the device status; based on the first description information of the work task node and the second description information of the standardized capability unit, the parameter set corresponding to the standardized capability unit is determined to include: Based on the first description information of the work task node, target historical task data is determined in the historical database, wherein the degree of matching between the target historical task data and the first description information is higher than a preset threshold. Based on the target historical task data, the first description information, and the second description information, a scene context vector is constructed; The parameter set is determined based on the scene context vector.

7. The workflow execution method according to claim 1, characterized in that, The method further includes: Determine the confidence level of the workflow; If the confidence level is lower than a preset confidence threshold, instruct the target object to modify the workflow, or provide the target object with a preset workflow template.

8. A workflow execution device, characterized in that, include: The workflow generation module is used to obtain the task text sent by the target object and generate a workflow based on the task text, wherein the workflow includes multiple task nodes; A matching module is used to determine the standardized capability unit corresponding to the work task node, wherein the standardized capability unit is used to execute a preset data processing task; The parameter generation module is used to determine the parameter set corresponding to the standardized capability unit based on the first description information of the work task node and the second description information of the standardized capability unit. The execution module is used to call the standardized capability unit to execute the work task node corresponding to the standardized capability unit based on the parameter set.

9. A non-volatile storage medium, characterized in that, The non-volatile storage medium stores a program, wherein when the program is executed, it controls the device where the non-volatile storage medium is located to execute the workflow execution method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A memory and a processor, the processor being configured to run a program stored in the memory, wherein the program, when running, executes the workflow execution method according to any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the workflow execution method according to any one of claims 1 to 7.

Images