Data Interface and Service Development Methods for Heterogeneous Mechanism Models in the Nuclear Power Industry

Abstract

This invention discloses a data interface and service development method for heterogeneous mechanism models in the nuclear power industry, comprising the following steps: Step S1, creating a dataset for model training, testing, and verification, and creating multiple service components as the development basis for different types of models; Step S2, creating a project for project management, creating models under the project, calling the corresponding service components according to the model classification for model development and debugging, and uploading the debugged model to the mechanism model library; Step S3, creating a model service, selecting a model from the mechanism model library and binding it to the model service, calling the corresponding service components according to the selected model classification for model service operation and deployment. This invention provides an integrated development model for mechanism models in the nuclear power industry.

Description

Technical Field

[0001] This invention relates to the field of nuclear power industrial internet platform optimization technology, and in particular to a data interface and service development method for heterogeneous mechanism models in the nuclear power industry. Background Technology

[0002] Currently, there are numerous nuclear power plants in China, each with a large number of operating units and a wide range of models. In addition, many power plants have been operating for many years, resulting in a large number and variety of mechanistic models involved in nuclear power operations. How to effectively manage and utilize these nuclear power mechanistic models, and how to meet the programming and debugging needs of various mechanistic models and improve user experience have become urgent problems to be solved. Summary of the Invention

[0003] In view of this, it is necessary to provide data interfaces and service development methods for heterogeneous mechanism models in the nuclear power industry, so as to effectively solve the technical problem of the lack of integrated development and management of mechanism models in the nuclear power industry.

[0004] This invention provides a data interface and service development method for heterogeneous mechanism models in the nuclear power industry, including the following steps:

[0005] Step S1: Create a dataset for model training, testing, and validation, and create multiple service components as the foundation for developing different types of models;

[0006] Step S2: Create a project for project management, create a model under the project, call the corresponding service components according to the model classification to develop and debug the model, and upload the debugged model to the mechanism model library;

[0007] Step S3: Create a model service. Select a model from the mechanism model library and bind it to the model service. Call the corresponding service components to run and publish the model service according to the classification of the selected model.

[0008] Preferably, in step S1, creating a dataset for model training, testing, and validation specifically involves:

[0009] The dataset includes public datasets and user-created custom datasets; users use the data in the public datasets by collecting them; users create the custom datasets by inputting basic dataset information and uploading data, and the custom datasets can be set to public or private.

[0010] Preferably, the dataset in step S1 further includes an external dataset;

[0011] When accessing the external dataset, if the accessed data is asset data, an asset package is first applied for in the data asset platform, and then the data is queried through the HTTP interface; if the accessed data is sample data shared in the sample tool, the data is queried and downloaded through the sample tool's HTTP interface.

[0012] Preferably, the service components in step S1 specifically include a real-time computing service component, an AI service component, a cloud-based simulation service component, and an industrial software service component.

[0013] Preferably, in step S2, creating a project for project management specifically involves:

[0014] The basic information of the project is input to create the project. The data between projects is isolated and called through open interfaces. The project includes a general project and an application project. The general project provides a general model service. The application calls the general model service through the industrial service open platform. The application project is used to provide corresponding model services for specific applications. The application directly calls the corresponding model service.

[0015] Preferably, in step S2, a model is created under the project, and the corresponding service components are called according to the model's classification for model development and debugging, specifically as follows:

[0016] Input the basic information of the model. The models are divided into real-time computing models, AI models, cloud simulation models and industrial software models. According to the model classification, call the interface of the corresponding service component to complete the model creation.

[0017] After the model is created, the corresponding service component provides development functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to develop the model. If the service component does not provide algorithm configuration functions, the user will upload the model file to develop the model.

[0018] After the model development is completed, the corresponding service component provides debugging functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to debug the model. If the service component does not provide algorithm configuration functions, the user will configure the model parameters through the platform, select the dataset to be debugged, and then call the model running interface of the corresponding service component to submit the debugging-related data to the corresponding service component to run and debug the model.

[0019] Preferably, in step S3, the corresponding service component is invoked to run the model service according to the classification of the selected model, specifically as follows:

[0020] If the service component corresponding to the model category provides a configuration page, you will be redirected to the configuration page to set the running parameters. If the service component corresponding to the model category does not provide a configuration page, you will need to enter the running parameters or upload a running parameter file. Select the running environment according to the model category. Configure the model running mode according to the model category.

[0021] The model structure file, the running parameters, and the running mode are encapsulated and downloaded to the running environment. Based on the model computing engine of the running environment, a running task is created and the model bound to the model service is run.

[0022] Preferably, in step S3, the corresponding service component is invoked to publish the model service according to the classification of the selected model, specifically as follows:

[0023] After the model service is completed, the corresponding service components are called according to the model category to create a model service API interface for publication. After publication, the model service provides an authorized model service interface to the outside world to provide the model's computing capabilities.

[0024] If the model service belongs to a general project, the model service interface is published to the industrial service open platform, and the model service is called through the industrial service open platform; if the model service belongs to a non-general project, access permissions are set for the model service interface.

[0025] Preferably, step S3 further includes,

[0026] After the model service is running, the model's running status is monitored, and online interactive instructions are sent to the running model task. The results of the model service are saved as a new dataset for subsequent training, testing, and validation of the model.

[0027] Compared with the prior art, the present invention has the following advantages: The present invention creates datasets and service components as the basis for model integration and development. Different service components are called according to different model classifications to develop and debug models, meeting the orchestration and debugging needs of various mechanistic models. Different service components are called to manage the operation and release of model services. After the model service is released, users can easily call the model service they need, improving the user experience. At the same time, projects are created to manage different models and divide models from a functional perspective. Users can also select model services according to the project. Attached Figure Description

[0028] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0029] Figure 1 is a flowchart of an embodiment of the data interface and service development method for heterogeneous mechanism models in the nuclear power industry provided by the present invention.

[0030] Figure 2 yes Figure 1 The diagram shown illustrates the integrated development and operation of an embodiment of a data interface and service development method for a heterogeneous mechanism model in the nuclear power industry.

[0031] Figure 3a yes Figure 1 The illustrated embodiment is a schematic diagram of model addition, deletion, and modification in an embodiment of the data interface and service development method for heterogeneous mechanism models in the nuclear power industry.

[0032] Figure 3b yes Figure 1 The illustrated embodiment is a schematic diagram of a model development method for a data interface and service development method for a heterogeneous mechanism model in the nuclear power industry.

[0033] Figure 3c yes Figure 1 The illustrated embodiment shows a schematic diagram of model debugging for a data interface and service development method for a heterogeneous mechanism model in the nuclear power industry.

[0034] Figure 4a yes Figure 1 The illustrated embodiment is a schematic diagram of the model service operation of an embodiment of the data interface and service development method for heterogeneous mechanism models in the nuclear power industry.

[0035] Figure 4b yes Figure 1 The illustrated embodiment is a schematic diagram of model service publishing for an embodiment of the data interface and service development method for heterogeneous mechanism models in the nuclear power industry. Detailed Implementation

[0036] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0037] Example 1

[0038] Please see Figure 1 The data interface and service development method for the heterogeneous mechanism model of the nuclear power industry in this embodiment includes the following steps:

[0039] Step S1: Create a dataset for model training, testing, and validation, and create multiple service components as the foundation for developing different types of models;

[0040] Step S2: Create a project for project management, create a model under the project, call the corresponding service components according to the model classification to develop and debug the model, and upload the debugged model to the mechanism model library;

[0041] Step S3: Create a model service. Select a model from the mechanism model library and bind it to the model service. Call the corresponding service components to run and publish the model service according to the classification of the selected model.

[0042] like Figure 2 As shown, this embodiment creates a dataset and service components as the foundation for model integration and development. The dataset provides a collection of data for model training, testing, and validation, and can be of various types such as text, images, audio, or multimodal data. Users can call different service components according to different model classifications to develop and debug models, meeting the orchestration and debugging needs of various mechanistic models. Different service components are called to manage the operation and release of model services. After the model service is released, users can easily call the required model service, improving the user experience. At the same time, projects are created to manage different models, dividing models from a functional perspective. Users can also select model services according to the project.

[0043] Specifically, step S1 involves creating a dataset for model training, testing, and validation, specifically as follows:

[0044] The dataset includes public datasets and user-created custom datasets; users use the data in the public datasets by collecting them; users create the custom datasets by inputting basic dataset information and uploading data, and the custom datasets can be set to public or private.

[0045] The platform provides public datasets, which users can bookmark and use later; users can also create their own datasets and upload data, which can then be published on the platform's public datasets.

[0046] The datasets publicly available on the platform originate from user-created and publicly shared data. Display fields include the dataset's Chinese name, English name, description, tags, creation time, and number of favorites. Datasets can be searched using keywords such as dataset name and tag name. Clicking the "favorite" button allows users to favorite a dataset; clicking it again cancels the favorite. Users can favorite datasets and also make their created datasets public. Clicking on a dataset opens its details page. The dataset details include basic information, dataset description, dataset files, and a data preview. Basic information includes the dataset's Chinese name, English name, tags, creation time, and number of favorites. The dataset description tab displays a brief introduction to the dataset, explaining its background, contents, source, uses, and usage instructions. The dataset files tab displays a list of dataset files, including filenames, file sizes, and update times. Folder display is supported. The data preview tab is used to preview dataset files. Users can select a previewable dataset file; only CSV format data files are supported for previewing.

[0047] Dataset maintenance allows users to search for created and favorited datasets using keywords such as dataset name and tag name. The "Create" tab displays user-created datasets in paginated format, with fields including Chinese and English dataset name, dataset description, tags, creation date, public / private status, and edit and delete buttons. The "Favorites" tab displays user-favorited datasets in paginated format, with fields including Chinese and English dataset name, dataset description, tags, creation date, and number of favorites buttons.

[0048] The "Create Dataset" button opens the Dataset Creation page. Dataset fields include the Chinese name, English name, labels, a brief description, and whether it's public. Users can upload one or more dataset files, which can be text, images, audio, or multimodal data. Multiple files can be organized as folders or compressed archives. Basic information about user-created datasets is stored in a database, while uploaded dataset files are stored in an object storage service such as MinIO. Users can modify their created datasets, including changing basic information and adding or deleting dataset files. Deleting a user-created dataset permanently removes all files from the object storage service. Dataset sharing is also possible; user-created datasets can be shared to public datasets by setting whether they are public.

[0049] Specifically, such as Figure 2 As shown, the dataset in step S1 also includes an external dataset;

[0050] When accessing the external dataset, if the accessed data is asset data, an asset package is first applied for in the data asset platform, and then the data is queried through the HTTP interface; if the accessed data is sample data shared in the sample tool, the data is queried and downloaded through the sample tool's HTTP interface.

[0051] In addition to the created datasets, external datasets can also be accessed. Data from data assets requires applying for an asset package on the data asset platform; after approval, the data can be queried via an HTTP interface. Samples shared in the sample tool can be queried and downloaded via the sample tool's HTTP interface. Sample files are in CSV format, with multiple files stored in ZIP format. First, call the sample tool's token acquisition interface, then call the sample tool's query interface to retrieve the samples shared by the sample tool. Select the sample data you need, obtain the sample data ID, and then call the sample tool's download sample interface to download the sample file.

[0052] Specifically, such as Figure 2 As shown, the service components in step S1 specifically include a real-time computing service component, an AI service component, a cloud-based simulation service component, and an industrial software service component.

[0053] Specifically, such as Figure 2 As shown, step S2 involves creating a project for project management, specifically as follows:

[0054] The basic information of the project is input to create the project. The data between projects is isolated and called through open interfaces. The project includes a general project and an application project. The general project provides a general model service. The application calls the general model service through the industrial service open platform. The application project is used to provide corresponding model services for specific applications. The application directly calls the corresponding model service.

[0055] The platform allows users to create and manage projects, with data isolation between projects and access via open APIs. Basic project settings include adding, modifying, deleting, and querying. Basic project authorization management includes adding and revoking authorizations, as well as authorizing data interfaces based on permissions. Model management is where all platform models under a project are managed; models are categorized into real-time computing models, cloud simulation models, industrial software models, and AI models.

[0056] Users can search for information on models they have created or downloaded by model category, model name, and tag name, and view detailed model information, including basic model information and model descriptions. Model information sources include models created by the user and models downloaded from the mechanistic model library IML.

[0057] Self-created models can be modified, deleted, developed, debugged, and uploaded. If a model category does not support configuration development, the develop button will not be displayed. Models downloaded from the mechanism model library can be modified, deleted, and debugged; users can create, modify, and delete models. Based on the model category, the system calls the CRUD (Create, Read, Update, Delete) model interfaces of the four component services. After a successful interface call, the platform model's CRUD operations are then executed.

[0058] like Figure 3a As shown, to create a model: Click the "Create Model" button to bring up the model creation page. Users can create models themselves. Basic model information includes the model's Chinese name, English name, description, version number, tags, and category. If the model category does not support configuration development, you can directly upload the model file for debugging and running. After filling in the basic model information, call the "Add Model" interface of the four component services according to the model category. After a successful interface call, the component-created model ID is returned. Then, add platform model data and bind the model ID returned by the interface.

[0059] Modify Model: Select the platform model data to be modified and obtain the corresponding model IDs of the four component services. Then, based on the model category, call the modify model interface of the four component services, submitting the model IDs of the four component services and the basic information of the model to be modified to the modify model interface of the four component services. After the interface call is successful, the basic information of the platform model is modified.

[0060] Deleting a model: Select the platform model data to be deleted and obtain the corresponding model IDs from the four component services. Then, based on the model category, call the delete model interface of the four component services, submitting the model IDs of the four component services to their respective delete model interfaces. After the interface call is successful, the platform model data is deleted.

[0061] Once the basic model information is created, different component services can be called to develop the model according to different model categories.

[0062] Specifically, in step S2, a model is created under the project, and the corresponding service components are called according to the model's classification for model development and debugging, specifically as follows:

[0063] Input the basic information of the model. The models are divided into real-time computing models, AI models, cloud simulation models and industrial software models. According to the model classification, call the interface of the corresponding service component to complete the model creation.

[0064] After the model is created, the corresponding service component provides development functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to develop the model. If the service component does not provide algorithm configuration functions, the user will upload the model file to develop the model.

[0065] After the model development is completed, the corresponding service component provides debugging functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to debug the model. If the service component does not provide algorithm configuration functions, the user will configure the model parameters through the platform, select the dataset to be debugged, and then call the model running interface of the corresponding service component to submit the debugging-related data to the corresponding service component to run and debug the model.

[0066] like Figure 3b As shown, different model categories are developed using corresponding component services. If a component service provides algorithm configuration functionality, the user will be redirected to the corresponding algorithm configuration page based on the model category to develop the model. After development, the model file and other information are saved. If a component service does not provide algorithm configuration functionality, the user can upload the model file. The platform will call the component service's upload model file interface based on the model category to upload the model file.

[0067] like Figure 3c As shown, model debugging is provided by the corresponding component service for different model categories. If the component service provides algorithm configuration functionality, the user will be redirected to the corresponding algorithm configuration page for model debugging based on the model category. Parameters for each node of the model can be configured, and the debugging dataset can be selected through the platform's dataset interface. Finally, the model, parameters, etc., are submitted to the component service to run the model. If the component service does not provide algorithm configuration functionality, the model parameters can be configured through the platform, the debugging dataset can be selected, and then the model run interface of the component service can be called to submit the model ID, parameters, etc., to the component service to run the model. The component service provides an interface for querying model run results, with result formats including forms, JSON files, visualization interfaces, logs, etc.

[0068] Model upload allows users to upload their created models to the Mechanism Model Library (IML). Uploaded information includes basic model information and model files; for example, real-time algorithm model files are XML and RTT files, while AI model files are JSON files. Users select the model to upload, and the platform, based on the model category, calls the upload IML interfaces of the four component services, sending the model ID to the component service. The component service's upload IML interface, based on the passed model ID, finds the corresponding basic model information and model file, calls the upload interface of the Mechanism Model Library, and uploads the basic model information and model file to the Mechanism Model Library. Users can then query or download the model by calling the Mechanism Model Library interface.

[0069] To download a model, click the "Download Model" button on the model management page to open the mechanism model list page. Users can query the model information that has been approved in the mechanism model library and download it. The platform calls the interface provided by the mechanism model library IML to query the model information that the user has approved in the mechanism model library in a paginated manner. The list fields include the model's Chinese name, English name, model category, version, description, and status. Downloaded models are displayed as "Downloaded," while models that have not been downloaded are displayed as "Not Downloaded but Available for Download." When the user clicks the download button after the model information, the four component services call the download IML and create model interface according to the model category, sending the mechanism model library model ID to be downloaded to the component service. The component service's download IML and create model interface first calls the mechanism model library's download model interface to download the model, then creates the model information and returns the model ID. After the platform successfully calls the component service's download IML and create model interface, it synchronously creates the platform model information. After the entire download operation is completed, the user can debug and run the model.

[0070] Specifically, in step S3, the corresponding service component is invoked to run the model service according to the classification of the selected model, which is as follows:

[0071] If the service component corresponding to the model category provides a configuration page, you will be redirected to the configuration page to set the running parameters. If the service component corresponding to the model category does not provide a configuration page, you will need to enter the running parameters or upload a running parameter file. Select the running environment according to the model category. Configure the model running mode according to the model category.

[0072] The model structure file, the running parameters, and the running mode are encapsulated and downloaded to the running environment. Based on the model computing engine of the running environment, a running task is created and the model bound to the model service is run.

[0073] Model service management is where all model services under the project are managed, supporting the creation, deletion, modification, querying, running, publishing, and monitoring of model services.

[0074] Model service maintenance and query: Users can query model services under a project, supporting the creation, modification, deletion, operation, deployment, and monitoring of model services. Filtering is supported by model service name, model category, status, etc. Display fields include model service name, model service description, model name, model category, model version, creator, creation time, and status. Model service creation: Clicking the "Create Model Service" button will bring up the model service creation page. Fill in the basic information of the model service. Basic information includes the Chinese name, English name, and service description. Then, select a model from the project's platform model list. The model and model service are bound together to support subsequent model service operation and deployment. Model service modification: Supports modification of basic model service information. Model service deletion: If the model service is already running or deployed, the interfaces of the four component services are called to take the model running task offline, and then the model service information on the platform is deleted synchronously.

[0075] To run the model service, click the "Run" button next to the model service to open the Run Model Service page. Running the model service means running the model bound to it. You need to fill in or upload information such as model runtime parameters, runtime environment, and runtime method, and then download and run it in the runtime environment. For example... Figure 4a As shown, if the component service corresponding to the model category provides a configuration page for setting parameters, you will be redirected to the configuration page to set the parameters. After setting, you will obtain the model running parameters. If the component service corresponding to the model category does not provide a configuration page for setting parameters, you can fill in the model running parameters through a form on the platform or directly upload the model running parameter file. After setting the model running parameters, select the running environment according to the model category. Different model categories have different running environments, including different model computing engines, runtime schedulers, and dependency libraries. After selecting the running environment, configure the model's running mode according to the model category. The running mode is divided into periodic running and aperiodic running. Periodic running requires configuring a scheduling strategy, and the model computing engine will schedule the run according to the configured scheduling strategy. The platform will encapsulate the model structure file, model running parameters, running mode, etc., of the model to be run and download them to the selected running environment. The model computing engine of the running environment creates a running task and runs the model, while recording the running log and running results. The platform can view the model's running status, running log, and running results.

[0076] Specifically, in step S3, the corresponding service component is invoked to publish the model service according to the classification of the selected model, which is as follows:

[0077] After the model service is completed, the corresponding service components are called according to the model category to create a model service API interface for publication. After publication, the model service provides an authorized model service interface to the outside world to provide the model's computing capabilities.

[0078] If the model service belongs to a general project, the model service interface is published to the industrial service open platform, and the model service is called through the industrial service open platform; if the model service belongs to a non-general project, access permissions are set for the model service interface.

[0079] Model services can be published. These services support authorized service interfaces, allowing them to provide computational capabilities, receive input parameters, and return results. Existing model services can be published to provide external model service interfaces.

[0080] like Figure 4b As shown, after the model service runs, clicking the "Publish" button calls the creation model service API interfaces of the four component services according to the model category, obtaining information such as the interface address, request method, and request parameters. Then, the "Publish Model Service" page pops up and is displayed. If the model service belongs to a general project, the model service interface is published to the industrial service open platform, and applications call the model service through the industrial service open platform. If the model service belongs to a non-general project, access permissions can be set for the API, including setting access tokens (access_token), request rate limits, request count limits, robot detection, and IP geofencing. After setting access permissions, it can be opened to external applications, allowing them to directly call the model service.

[0081] Specifically, such as Figure 2 As shown, step S3 further includes,

[0082] After the model service is running, the model's running status is monitored, and online interactive instructions are sent to the running model task. The results of the model service are saved as a new dataset for subsequent training, testing, and validation of the model.

[0083] Model service monitoring allows for the monitoring of model operation after the model service is running, including task information, running status, running logs, historical execution records, and running results. The platform can send online interactive commands to running model tasks, including common scheduling commands such as start, stop, and restart. It supports viewing and downloading running results in text and file formats, and allows for visualization. Model running results can be added to the user's dataset. Component services corresponding to model categories provide model running monitoring capabilities; the platform calls the component service's interface to monitor model operation.

[0084] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of the present invention.

Claims

1. A data interface and service development method for heterogeneous mechanism models in the nuclear power industry, characterized in that: Includes the following steps: Step S1: Create a dataset for model training, testing, and validation, and create multiple service components as the foundation for developing different types of models; Step S2: Create a project for project management, create a model under the project, call the corresponding service components according to the model classification to develop and debug the model, and upload the debugged model to the mechanism model library; Step S3: Create a model service. Select a model from the mechanism model library and bind it to the model service. Call the corresponding service components to run and publish the model service according to the classification of the selected model. In step S2, creating a project for project management specifically involves: The basic information of the project is input to create the project. The data between projects is isolated and called through open interfaces. The project includes a general project and an application project. The general project provides a general model service. The application calls the general model service through the industrial service open platform. The application project is used to provide corresponding model services for specific applications. The application directly calls the corresponding model service.

2. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, Step S1 involves creating a dataset for model training, testing, and validation, specifically as follows: The datasets include public datasets and user-created custom datasets; Users can use the data in the public dataset by collecting it; users can create the custom dataset by inputting basic dataset information and uploading data, and the custom dataset can be set to public or private.

3. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, The dataset in step S1 also includes an external dataset; When accessing the external dataset, if the accessed data is asset data, an asset package is first applied for in the data asset platform, and then the data is queried through the HTTP interface; if the accessed data is sample data shared in the sample tool, the data is queried and downloaded through the sample tool's HTTP interface.

4. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, The service components in step S1 specifically include a real-time computing service component, an AI service component, a cloud-based simulation service component, and an industrial software service component.

5. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, In step S2, a model is created under the project, and the corresponding service components are called according to the model's classification for model development and debugging. Specifically: Input the basic information of the model. The models are divided into real-time computing models, AI models, cloud simulation models and industrial software models. According to the model classification, call the interface of the corresponding service component to complete the model creation. After the model is created, the corresponding service component provides development functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to develop the model. If the service component does not provide algorithm configuration functions, the user will upload the model file to develop the model. After the model development is completed, the corresponding service component provides debugging functions according to the model type. If the corresponding service component provides algorithm configuration functions, the user will be redirected to the corresponding algorithm configuration page to debug the model. If the service component does not provide algorithm configuration functions, the user will configure the model parameters through the platform, select the dataset to be debugged, and then call the model running interface of the corresponding service component to submit the debugging-related data to the corresponding service component to run and debug the model.

6. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, In step S3, the corresponding service component is invoked to run the model service according to the classification of the selected model. Specifically: If the service component corresponding to the model classification provides a configuration page, you will be redirected to the configuration page to set the running parameters. If the service component corresponding to the model classification does not provide a configuration page, you will need to enter the running parameters or upload the running parameter file. Select the runtime environment based on the model category; configure the model runtime mode based on the model category. The model structure file, the running parameters, and the running mode are encapsulated and downloaded to the running environment. Based on the model computing engine of the running environment, a running task is created and the model bound to the model service is run.

7. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, In step S3, the corresponding service component is invoked to publish the model service according to the classification of the selected model. Specifically: After the model service is completed, the corresponding service components are called according to the model category to create a model service API interface for publication. After publication, the model service provides an authorized model service interface to the outside world to provide the model's computing capabilities. If the model service belongs to a general project, then the model service interface will be published to the industrial service open platform, and the model service will be called through the industrial service open platform. If the project to which the model service belongs is not a general project, then access permissions are set for the model service interface.

8. The data interface and service development method for heterogeneous mechanism models in the nuclear power industry according to claim 1, characterized in that, Step S3 further includes monitoring the model's operation after the model service runs, sending online interaction instructions to the running model task, and saving the model service's running results as a new dataset for subsequent model training, testing, and validation.

Images