Model calculation scheduling method and device, equipment, medium and product

Abstract

The invention discloses a model calculation scheduling method and device, equipment, a medium and a product. The method comprises the following steps: determining position information of a basic section of a calculation unit for each calculation unit in a plurality of calculation units; determining the type of a model processing unit of the basic section, an input task of the model processing unit, an output task of the model processing unit and calculation process information corresponding to the model processing unit based on the position information of the calculation unit in the upstream and downstream topological relation; in response to calculation indication information for the drainage basin, generating a calculation task scheduling instruction based on the type of each model processing unit; in response to the calculation task scheduling instruction, obtaining identification information of a scheduling scheme of the drainage basin; starting a target calculation process corresponding to each target model processing unit on each target calculation node based on the identification information of the scheduling scheme; and each target model processing unit is controlled to operate to generate a calculation result.

Description

Technical Field

[0001] The present application relates to the technical field of water conservancy engineering, and in particular to a model calculation and scheduling method, device, equipment, medium and product. Background Art

[0002] With the development of big data and artificial intelligence technology, large-scale distributed computing has become an important means of processing massive data and complex models, especially in the water conservancy industry, where informatization is relatively late, and traditional single-host single-solution model calculations can no longer meet the current large-area, real-time computing requirements.

[0003] Existing parallel computing scheduling is mostly based on fixed regions, single models, and multi-threaded scheduling, which makes it inconvenient to debug the model independently and parameters, and does not support the rapid addition of different models. This leads to the following disadvantages: (1) poor compatibility with multiple models, mostly for a single model; (2) existing multi-threaded scheduling of models is mostly based on the code level, which easily makes the debugging of the model itself inconvenient; (3) the existing framework lacks compatibility with dynamic expansion of large regions, dynamic expansion of models, and different precisions of the same model. Summary of the invention

[0004] In view of this, the embodiments of the present application provide a model computing scheduling method, device, equipment, medium and product, which aim to realize the computing scheduling of multi-model, multi-process and multi-host computing nodes in a large watershed.

[0005] The technical solution of the embodiment of the present application is implemented as follows:

[0006] In a first aspect, an embodiment of the present application provides a model calculation scheduling method, including:

[0007] Determine a plurality of computing units of the watershed based on an upstream and downstream topological relationship between each sub-watershed partition of the watershed and area data of each of the sub-watershed partitions; the upstream and downstream topological relationship is generated based on encoding each sub-watershed partition of the watershed;

[0008] For each of the plurality of computing units, determining the location information of the basic section of the computing unit; and based on the location information of the computing unit in the upstream and downstream topological relationship, determining the type of the model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit, and the computing process information corresponding to the model processing unit; the number of the model processing units includes a plurality;

[0009] In response to the calculation instruction information for the watershed, generating a calculation task scheduling instruction based on the type of each of the model processing units;

[0010] In response to the computing task scheduling instruction, obtaining identification information of a scheduling scheme of the watershed; the scheduling scheme includes a plurality of target model processing units, computing process information corresponding to each of the target model processing units, and a target computing node corresponding to each of the target model processing units, wherein the target model processing unit is one of the plurality of model processing units;

[0011] Based on the identification information of the scheduling scheme, on each of the target computing nodes, the target computing process corresponding to each of the target model processing units is started; and each of the target model processing units is controlled to run to generate the computing result of the watershed;

[0012] The model processing unit is a standardized and packaged model processing unit.

[0013] In one embodiment, the method further comprises:

[0014] For each computing unit among the plurality of computing units, obtaining location information of an egress node corresponding to the computing unit;

[0015] Based on the position information of the exit node, the position information of the basic section of the calculation unit is determined.

[0016] In some embodiments, controlling each of the target model processing units to run and generating the calculation results of the watershed includes:

[0017] For each target model processing unit in each target model processing unit in the scheduling scheme, based on the type information of the target model processing unit, generating a start instruction corresponding to the type information;

[0018] In response to the startup instruction, the computing package of the target model processing unit is started to run and the model computing result is generated; wherein the computing package includes input file information, output file information, output directory information, log directory and configuration file; wherein the configuration file is used to store the input path information and output path information of the target model processing unit based on a set text format; the set text format includes JSON format;

[0019] Based on the calculation results of each of the models, the calculation results of the watershed are generated.

[0020] In some embodiments, the method further comprises:

[0021] Based on the spatial geometric information of the multiple computing units, the computing results are classified to generate classification result data; and in response to a result display request, the classification result data is displayed.

[0022] In some embodiments, the method further comprises:

[0023] Get monitoring request;

[0024] In response to the monitoring request, at least one of the following is displayed:

[0025] The identification information of the scheduling scheme, the calculation process information of each of the model processing units, the calculation status information of each of the model processing units and the calculation progress information of each of the model processing units, wherein the calculation process information includes the number of processes and the process status.

[0026] In some embodiments, the method further comprises:

[0027] Obtaining a parameter change request for the model processing unit;

[0028] In response to the parameter change request, the initial parameters of the model processing unit are updated to target parameters.

[0029] In some embodiments, the method further comprises:

[0030] Obtain instruction information for adding a model processing unit; the instruction information includes the type of the model processing unit to be added;

[0031] In response to the indication information, based on the type of the model processing unit to be added, determining a target computing unit corresponding to the model processing unit to be added, and adding the model to be added to the target computing unit;

[0032] Based on the position information of the basic section of the target computing unit, the input task corresponding to the to-be-added model processing unit, the output task corresponding to the to-be-added model processing unit and the process information of the to-be-added model processing unit are determined.

[0033] In a second aspect, an embodiment of the present application provides a model calculation scheduling device, the device comprising:

[0034] A first determination module is used to determine a plurality of computing units of the watershed based on an upstream and downstream topological relationship between each sub-watershed partition of the watershed and area data of each of the sub-watershed partitions; the upstream and downstream topological relationship is generated based on encoding each sub-watershed partition of the watershed;

[0035] A second determination module is used to determine, for each of the plurality of calculation units, the location information of the basic section of the calculation unit; and based on the location information of the calculation unit in the upstream and downstream topological relationship, determine the type of the model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit, and the calculation process information corresponding to the model processing unit; the number of the model processing units includes a plurality;

[0036] A generating module, configured to generate a computing task scheduling instruction in response to the computing instruction information for the watershed and based on the type of each of the model processing units;

[0037] an acquisition module, configured to acquire identification information of a scheduling scheme of the watershed in response to the computing task scheduling instruction; the scheduling scheme includes a plurality of target model processing units, computing process information corresponding to each of the target model processing units, and a target computing node corresponding to each of the target model processing units, wherein the target model processing unit is one of the plurality of model processing units;

[0038] A scheduling module, configured to start the target computing process corresponding to each target model processing unit on each target computing node based on the identification information of the scheduling scheme; and control the operation of each target model processing unit to generate a computing result;

[0039] The model processing unit is a standardized and packaged model processing unit.

[0040] In a third aspect, an embodiment of the present application provides an electronic device, comprising: a processor and a memory for storing a computer program that can be run on the processor, wherein the processor, when used to run the computer program, executes the steps of the method of the first aspect of the embodiment of the present application.

[0041] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in the first aspect are implemented.

[0042] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the method in the first aspect.

[0043] The technical solution provided by the embodiment of the present application is a model calculation scheduling method, which includes: determining multiple calculation units of the basin based on the upstream and downstream topological relationship between each sub-basin partition of the basin and the area data of each sub-basin; the upstream and downstream topological relationship is generated based on the encoding of each sub-basin partition of the basin; for each calculation unit in each calculation unit, determining the location information of the basic section of the calculation unit; and based on the location information of the calculation unit in the upstream and downstream topological relationship, determining the type of model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit and the calculation process information corresponding to the model processing unit; the number of model processing units includes multiple; in response to the target For the calculation indication information of the watershed, a calculation task scheduling instruction is generated based on the type of each model processing unit; in response to the calculation task scheduling instruction, the identification information of the scheduling scheme of the watershed is obtained; the scheduling scheme includes multiple target model processing units, the calculation process information corresponding to each target model processing unit, and the target computing node corresponding to each target model processing unit, and the target model processing unit is one of the multiple model processing units; based on the identification information of the scheduling scheme, the target calculation process corresponding to each target model processing unit is started on each target computing node; and the operation of each target model processing unit is controlled to generate a calculation result; wherein the model processing unit is a model processing unit after standardized encapsulation.

[0044] Thus, the embodiment of the present application, from a global perspective, significantly improves the efficiency, resource utilization and system stability of large basin computing tasks by accurately dividing the computing units of the basin, constructing upstream and downstream topological relationships, realizing multi-process parallel computing, and standardizing the encapsulation of model processing units. Specifically, (1) through the topological coding and computing unit division of the basin sub-basin, it is ensured that the tasks of each computing unit can be executed collaboratively, thereby avoiding resource waste and improving computing efficiency. (2) Multi-process parallel computing can make full use of the resources of multiple computing nodes and shorten the overall computing time. (3) Standardized encapsulated model processing units ensure that different models can run on independent nodes without interfering with each other, making the system more stable and easy to expand and maintain. This standardized processing also ensures that multiple models can run in parallel on different computing nodes, and the stability of the overall system will not be affected by the update or debugging of a single model; thereby achieving efficient computing scheduling of multi-model, multi-process, and multi-host computing nodes in a large basin. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 A schematic diagram of a flow chart of a model-based scheduling method provided in an embodiment of the present application;

[0046] Figure 2 A schematic diagram of the structure of the model assembly on the basic section provided in the embodiment of the present application;

[0047] Figure 3A schematic diagram of a structured computing package provided in an embodiment of the present application;

[0048] Figure 4 A schematic diagram of the interface for calculating progress provided in an embodiment of the present application;

[0049] Figure 5 The system structure and flow chart of a large-scale distributed model hybrid computing scheduling solution based on basic sections provided for an application example of this application;

[0050] Figure 6 A schematic diagram of the structure of a model calculation scheduling device provided in an embodiment of the present application;

[0051] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0054] The present invention provides a scheduling method based on model calculation, which includes the following steps:

[0055] Step 110: Based on the upstream and downstream topological relationship between each sub-basin partition of the basin and the area data of each sub-basin, determine multiple calculation units of the basin; the upstream and downstream topological relationship is generated based on encoding each sub-basin partition of the basin.

[0056] It can be understood that a watershed refers to the area where water flows in a region, which can be divided into multiple sub-basins (such as rivers, tributaries, etc.). Each sub-basin can be regarded as a hydrological unit with different characteristics such as size, shape, location, etc. The area data of each sub-basin reflects its geographical characteristics.

[0057] Here, the upstream and downstream topological relationship refers to the water flow transmission relationship between different sub-basins, for example, the water flow in one sub-basin will affect the downstream basin. This relationship is determined based on the flow direction and convergence mode of the water flow in the basin.

[0058] It can be understood that by encoding each sub-basin partition, these sub-basins and their upstream and downstream relationships can be identified in a structured manner, so that the topological relationship of each sub-basin (i.e., its relative position in the basin) can be clearly identified and used for the division of computing tasks.

[0059] Here, according to the upstream and downstream topological relationship and area data, the watershed can be divided into multiple computing units. Each computing unit represents a computing area of ​​the watershed, and their division basis may involve factors such as topography, climate, and hydrological characteristics. The computing units can be assigned to different computing resources for processing in subsequent steps.

[0060] For example, in order to support multiple models in the same area, we adopt the method of reducing the large to the small. First, for the entire target area, according to the collected regional small watershed data and upstream and downstream topology, according to different watersheds and different divisions, from upstream to downstream, every 200-500 square kilometers is a calculation unit.

[0061] Taking the national region as an example, basic data of small watersheds were collected, and seven major watershed divisions were selected and determined; within each watershed division, 200 calculation units were divided according to the area and number of small watersheds.

[0062] Step 120: For each of the multiple computing units, determine the location information of the basic section of the computing unit; and based on the location information of the computing unit in the upstream and downstream topological relationship, determine the type of model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit, and the calculation process information corresponding to the model processing unit; the number of model processing units includes multiple.

[0063] Here, the basic section refers to a specific geographical section (such as a river section or a hydrological section) used to describe the calculation unit. These sections are usually used to describe the distribution, velocity, water level and other hydrological data of the water flow in the basin. The basic section can be used to further understand the water flow dynamics in the calculation unit. The basic section is the starting and ending position for basin calculation and analysis.

[0064] For each calculation unit in each calculation unit, there is a corresponding basic section, so that the location information of the basic section of the calculation unit can be determined. Here, the basic section refers to the key position in the model calculation, usually located at the exit or key node (such as the water confluence point) of the calculation unit.

[0065] Here, the location information of each computing unit is its positional relationship in the upstream and downstream chain of the watershed, usually referring to its position in the upstream and downstream chain of the watershed. For example, the output of the computing unit located upstream may be used as the input of the downstream computing unit. The downstream computing unit needs to receive the necessary data from the upstream to ensure the continuity of the computing logic of the entire watershed.

[0066] Here, the location data of the calculation unit refers to the specific location of each calculation unit in the watershed, such as whether it is located upstream, midstream or downstream. The location data is usually the result of geographic information data (GIS data), which can help the system accurately locate each calculation unit.

[0067] Here, the type of the model processing unit, the input task, and the output task corresponding to each of the computing units are selected according to the position information of the computing unit in the upstream and downstream topological relationship.

[0068] It is understandable that the type of model processing unit can be selected according to the characteristics of the basic section and the calculation requirements. The types of model processing units include: API model and SKBY model. For example, hydrological model, climate model or geological model.

[0069] Here, the input task of each model processing unit refers to the data it needs, which usually comes from the upstream computing unit or external input source; and the output task is the result generated after the model runs, which may be used as the input of the downstream computing unit or as the final computing output.

[0070] Here, each model processing unit has independent computing process information (such as process ID, execution status, etc.) so as to (1) run independently to avoid mutual interference between tasks and (2) support parallel processing to improve computing efficiency.

[0071] Here, based on the upstream and downstream topological relationship of the computing unit, the system matches the input task, output task and computing process information with the specific model processing unit. This matching ensures that each model processing unit can smoothly obtain the required data. The output results can be seamlessly connected to downstream tasks or output as independent results.

[0072] For example, Figure 2 As shown, Figure 2 The structural diagram of the model assembly on the basic section. For the calculation area, a basic section with upstream and downstream topology is established. Basic section 1 flows to basic section 2, and basic section 2 flows to basic section 3. The same or different calculation models can be selected on different basic sections. Basic section 1 uses the API model, basic section 2 uses the SKBY model, basic section 3 uses the SKBY model, and basic section 4 uses the API model, which facilitates the mixed assembly of different models.

[0073] Here, the traditional method is to select the same calculation model for the entire area. For example, the API model can be used for the entire area, or the SKBY model can be used for the entire area. If they cannot be mixed, the advantages of each model cannot be brought into play. The soil type, soil texture, terrain, etc. in different regions are different, and the optimal model for application may be different. If one model is used uniformly, the optimal result cannot be obtained.

[0074] The present application solves this problem well by assembling the model based on the basic section of the partition. The optimal model that matches the terrain, geology and other characteristics can be selected in different regions to achieve the optimal solution.

[0075] Step 130: In response to the calculation instruction information for the watershed, a calculation task scheduling instruction is generated based on the type of each model processing unit.

[0076] Here, computing instruction information refers to specific requirements or instructions for computing tasks, which come from users, scheduling systems or other control centers. The instruction information includes the goals of watershed computing, task priorities, computing resource requirements, etc.

[0077] Here, in response to the calculation instruction information, the embodiment of the present application needs to generate corresponding calculation task scheduling instructions. This process arranges the calculation order and allocates computing resources according to the type of each model processing unit (such as hydrological simulation, climate model, etc.), and ensures that the calculation tasks are executed in the order of topological relationships.

[0078] Here, by properly scheduling computing tasks, we ensure that each model processing unit runs on a suitable computing node and can obtain input data and generate output data in a timely manner. The computing task scheduling instruction is the core control logic of the system, which is used to manage the order of task execution and resource allocation.

[0079] Step 140: In response to the computing task scheduling instruction, obtain a scheduling plan for the watershed; the scheduling plan includes multiple target model processing units, computing process information corresponding to each target model processing unit, and target computing nodes corresponding to each target model processing unit, and the target model processing unit is one of the multiple model processing units.

[0080] Here, the scheduling scheme is a set of detailed computing task scheduling schemes generated by the system after receiving the computing task scheduling instructions. This scheme will list the specific execution plan of each target model processing unit, including which computing node to execute on, when to execute, and how to execute. After determining the scheduling scheme, the identification information of the scheduling scheme can be obtained.

[0081] Here, the target model processing unit refers to the model processing unit to be executed for the model processing task, and the target model processing unit is one of the multiple model processing units. The scheduling scheme needs to list all target model processing units, as well as their corresponding target computing nodes and computing process information.

[0082] Here, each target model processing unit will be assigned to a specific target computing node, which is a hardware resource with computing capabilities. The process information indicates how the target model processing unit runs on the computing node, including tasks such as startup, monitoring, and control.

[0083] Step 150: Based on the identification information of the scheduling scheme, on each target computing node, start the target computing process corresponding to each target model processing unit; and control each target model processing unit to run and generate computing results;

[0084] The model processing unit is a standardized and encapsulated model processing unit.

[0085] It is understandable that once the scheduling scheme is determined, the corresponding process can be started on each target computing node based on the identification information of the scheduling scheme. Each process corresponds to a model processing unit, which will start executing according to the predetermined computing task.

[0086] It is understandable that after starting the process on the target computing node, the system will monitor the running status of each model processing unit to ensure that it executes tasks according to the scheduling plan. If a problem occurs during the calculation process of a model processing unit, the system will take corresponding error correction or adjustment measures.

[0087] It can be understood that the standardized encapsulated model processing unit means that each model is encapsulated into an independent unit with a unified interface and data format. This enables different model processing units to run on different computing nodes and are not affected by other model updates or debugging during the scheduling process. Standardized encapsulation can improve the compatibility, scalability and maintainability of the system.

[0088] By developing a web service interface, each computing node can be monitored. This interface mainly implements the startup, shutdown, and status monitoring of the program (model processing module) in the specified model package directory. The model processing module is a secondary encapsulation package for each model developed in C# language. The purpose is to continuously expand the model and provide support for the scheduling module in a consistent manner. Through this model processing module middle layer, the scheduling management only needs to manage this unique model processing module program.

[0089] In addition, this standardized packaging supports multiple platforms (such as Windows and Linux systems) and uses a consistent interface to implement the scheduling and management of model processing programs on servers with different architectures.

[0090] Thus, the embodiment of the present application, from a global perspective, significantly improves the efficiency, resource utilization and system stability of large basin computing tasks by accurately dividing the computing units of the basin, constructing upstream and downstream topological relationships, realizing multi-process parallel computing, and standardizing the encapsulation of model processing units. Specifically, (1) through the topological coding and computing unit division of the basin sub-basin, it is ensured that the tasks of each computing unit can be executed collaboratively, thereby avoiding resource waste and improving computing efficiency. (2) Multi-process parallel computing can make full use of the resources of multiple computing nodes and shorten the overall computing time. (3) Standardized encapsulated model processing units ensure that different models can run on independent nodes without interfering with each other, making the system more stable and easy to expand and maintain. This standardized processing also ensures that multiple models can run in parallel on different computing nodes, and the stability of the overall system will not be affected by the update or debugging of a single model; thereby achieving efficient computing scheduling of multi-model, multi-process, and multi-host computing nodes in a large basin.

[0091] In some embodiments, the method further comprises:

[0092] For each computing unit among the plurality of computing units, obtaining location information of an exit node corresponding to the computing unit;

[0093] Based on the position information of the exit node, the position information of the basic section of the calculation unit is determined.

[0094] It is understood that the exit node refers to the boundary position of each computational unit, usually located at the exit of the computational unit, where the data flows out of the unit. In hydrological or watershed calculations, the exit node may involve a watershed, an outfall, or other key hydrological location.

[0095] Here, based on the location information of the exit node, the specific location of the basic section can be calculated. The exit node is usually the boundary of the calculation unit. According to the geographical coordinates of this node and the water flow characteristics of the basin, the location of the basic section to be used in the hydrological model within the calculation unit can be determined.

[0096] Here, there are many types of basic sections, such as cross sections, longitudinal sections, etc. Different basic section types are suitable for different hydrological calculation tasks. According to the type information of the basic section, the system can further select the corresponding model processing unit (such as water flow model or water quality model) to ensure the accuracy of the calculation task.

[0097] In some embodiments, controlling each target model processing unit to run and generate calculation results of the watershed includes:

[0098] For each target model processing unit in each target model processing unit in the scheduling scheme, based on the type information of the target model processing unit, a start instruction corresponding to the type information is generated;

[0099] In response to the start instruction, the calculation package of the target model processing unit is started to run and the model calculation result is generated; wherein the calculation package includes input file information, output file information, output directory information, log directory and configuration file; wherein the configuration file is used to store the input path information and output path information of the target model processing unit based on a set text format; the set text format includes JSON format;

[0100] Based on the calculation results of each model, the calculation results of the watershed are generated.

[0101] Here, each target model processing unit has different types and computing requirements, so the system generates corresponding startup instructions based on its type information. The startup instructions specifically specify how to start the computing task of the model processing unit, such as the input files and model configurations that need to be loaded.

[0102] Here, the computing task of each model processing unit is managed by a computing package, which includes: Input file information: the input data required by the model (such as hydrological data, meteorological data, etc.). Output file information: the result file generated during the calculation process. Output directory information: the directory where the calculation results are stored. Log directory: used to store log files generated during the execution of the model, which is convenient for subsequent debugging and viewing the execution status. The configuration file contains the configuration parameters required for the model to run, such as input path, output path and other information. These configuration files are usually stored in text format to ensure that the model can read and write data correctly.

[0103] Here, the set text format includes the JSON format, which has strong compatibility, readability, and is easy to interact with other systems.

[0104] Here, after the model runs, relevant calculation results will be generated, which will be used for subsequent data analysis and decision support.

[0105] Exemplarily, to realize the mixed assembly of different computing models based on the basic section, it is first necessary to ensure that the computing package of each model follows the unified specifications and interfaces, so that the computing engine can process different computing packages with consistent operations. The structure of the computing package is designed as 4 core directories: input directory (storing input data), output directory (storing output results), log directory (storing log files), sim configuration file directory (storing configuration files in JSON format). Each JSON file includes the input and output paths required for model calculation, the computing resources used (such as GPU index), the calculation time and other parameters. In this way, when executing, whether using the API model or the SKBY model, the corresponding input file can be loaded and calculated through a consistent command (such as API.exe appsetting.json) to ensure the smooth operation of the model.

[0106] In detail, Figure 3 As shown, Figure 3 A schematic diagram of a structured computing package. Figure 3 In the , sim directory contains the appsetting.json file. The input directory contains input files. The output directory contains output files. The log directory contains log files. API.exe and SKBY.exe are executable files. API.jar is the Java version of the executable file.

[0107] All packages are designed to be text-based, including four directories: input directory, output directory, log directory, and sim configuration file directory. The sim configuration file directory contains json-formatted configuration files, which store the input and output file paths involved in model calculations, the gpu index number used, and the calculation time. Taking the API and SKBY calculation programs as examples, the API.exe appsetting.json command can be used to read the input files in the input from the json file and output them to the output directory. The SKBY.exe appsetting.json command can be used to read the input files in the input from the json file and output them to the output file. If it is a java version of the calculation package, the java-jarAPI.jar appsetting.json method can also be used to parse the input files in the input and generate output files in the output. This file-based standardized structure requirement can support calculation programs in different languages, and it is also convenient to view and debug inputs and troubleshoot outputs through text files.

[0108] In this way, this computing management method based on standardized packaging and configuration files enables seamless docking of different models and greatly improves the compatibility, scalability and maintainability of the models. Whether adding new models or updating existing models, it can quickly achieve and maintain the stability of the system.

[0109] In some embodiments, the method further comprises:

[0110] Based on the spatial geometric information of the plurality of computing units, the computing results are classified to generate classified result data; and in response to a request for achievement display, the classified result data is displayed.

[0111] Here, spatial geometric information usually includes three basic geometric elements: points, lines and surfaces. Each spatial element represents a different type of geographic entity. A point represents a specific location on the earth's surface, without width or depth, and is usually used to represent a single geographical location (such as a measurement point or a river confluence). A line represents a geometric object with length but no width, and is used to represent roads, rivers, watershed boundaries, etc. A surface represents a geometric object with area, and is usually used to describe an area, such as a geographical area of ​​a calculation unit or watershed, a soil type area, etc.

[0112] For example, points may include sites, hydrological stations, reservoir stations, dams, etc. Lines may include rivers, levees, and pipe networks. Surfaces may include rainfall surfaces, watershed surfaces, and flooded surfaces.

[0113] Here, the calculation results will be classified into different areas according to the spatial geometric information (points, lines, and surfaces). This allows the results to match the geographical features within the basin more accurately. For example, the calculation of the upstream area may focus on the analysis of precipitation, while the downstream area focuses on the change of water level. Such classification enhances the comprehensibility and practicality of the data.

[0114] The classified calculation results can be displayed in different ways to help users view the required calculation data according to geographical location and task requirements. For example, point data can display the calculation results of weather stations and monitoring points; line data can display the changes in water flow on rivers, dams and other paths; surface data can display relevant hydrological information in different areas of the basin, such as water level, precipitation, etc.

[0115] In some embodiments, the method further comprises:

[0116] Get monitoring request;

[0117] In response to the monitoring request, at least one of the following is displayed:

[0118] The identification information of the scheduling scheme, the calculation process information of each model processing unit, the calculation status information of each model processing unit and the calculation progress information of each model processing unit, the calculation process information includes the number of processes and the process status.

[0119] Here, monitoring requests are usually initiated by users or system administrators to check the operating status of the model processing unit and ensure the stability of the system and the smooth progress of computing tasks.

[0120] Here, after the system responds to the monitoring request, it displays at least one of the following:

[0121] Process information provides information about the calculation process, including the current number of processes (the number of processes running simultaneously) and the status of each process (for example, whether it is running, waiting, completed, crashed, etc.). The calculation status reflects the running progress of each model processing unit, whether there are errors or exceptions, or whether the calculation task is successfully completed. Calculation progress information (for example, the overall progress percentage, the specific progress of the partition or task).

[0122] Exemplarily, the computing status and process data displayed by the system help users understand the execution status of computing tasks in real time. The specific display content includes: the currently used scheduling scheme information (such as the scheduling scheme ID). Computation status information, which clearly indicates the current status of the task (such as computing, normal exit, waiting for computing, crash exit, etc.). Computation progress information (for example, the overall progress percentage, the specific progress of the partition or task). Computation log records, which provide detailed logs of the current computing task to help users understand possible problems during execution.

[0123] In practical applications, when responding to monitoring requests, process data and computing status can help users or system administrators quickly assess the health of the system and the progress of computing tasks. Computation progress and logs can provide further details to help track down problems, optimize computing processes, or adjust strategies.

[0124] like Figure 4 As shown, Figure 4 This is the calculation progress monitoring interface. The specific contents are as follows:

[0125] Scheduling scheme (sk_001): The scheduling scheme currently in use can help users understand the current task configuration and execution plan.

[0126] Current computing status: lists different computing statuses (such as computing, normal exit, waiting for computing, etc.), providing users with clear task status information.

[0127] Calculation progress: Through the progress percentage and process details, users can understand the specific progress of each area and process. The calculation progress of multiple areas (such as Zhangwei Xinhe and Si Nusi) is displayed here, and the completion status of each process is indicated (such as process 1 has been completed, and processes 2 to 6 have not yet started).

[0128] Calculation log: records the various states of the calculation process in real time, making it easier to track the execution of tasks and solve potential problems.

[0129] In some embodiments, the method further comprises:

[0130] Obtain parameter change requests for model processing units;

[0131] In response to the parameter change request, the initial parameters of the model processing unit are updated to the target parameters.

[0132] It is understandable that users or administrators may need to adjust the parameters of the model (such as initial water level, precipitation, temperature, etc.). The system automatically updates the initial parameters of the model processing unit based on the parameter change request. This process ensures that the model can be calculated based on the latest data and requirements, maintaining high accuracy and adaptability.

[0133] In some embodiments, the method further comprises:

[0134] Obtain instruction information for adding a model processing unit; the instruction information includes the type of the model processing unit to be added;

[0135] In response to the indication information, based on the type of the model processing unit to be added, determine the target computing unit corresponding to the model processing unit to be added, and add the model to be added to the target computing unit;

[0136] Based on the position information of the basic section of the target computing unit, the input task corresponding to the model processing unit to be added, the output task corresponding to the model processing unit to be added, and the process information of the model processing unit to be added are determined.

[0137] Below, the embodiment of the present application is described in detail with reference to an application example.

[0138] With the development of big data and artificial intelligence technology, large-scale distributed computing has become an important means of processing massive data and complex models. Especially in the water conservancy industry, where informatization is relatively late, traditional single-host single-solution model computing can no longer meet the current large-area, real-time computing requirements. Existing parallel computing scheduling is mostly scheduled in a fixed area, single model, and multi-threaded manner, which makes independent debugging and parameter debugging of models inconvenient, and does not support the rapid addition of different models.

[0139] This application example is based on the water conservancy industry. It uses a unified regional topology encoding and repackages multiple models to extract input, output, and upstream and downstream relationships. It also debugs the models according to independent processes and finally assembles them into multi-process topology-based parallel computing. It implements computing scheduling for multiple models, large regions, multiple processes, and multiple host computing nodes.

[0140] It can be seen that the disadvantages of the related technology include:

[0141] (1) Poor compatibility with multiple models, mostly for a single model

[0142] (2) Most existing models are based on multi-threaded scheduling at the code level, which easily makes the debugging of the model itself inconvenient

[0143] (3) The existing framework lacks compatibility with dynamic expansion of large areas, dynamic expansion of models, and different precisions of the same model.

[0144] Based on this, this application example provides a large-scale distributed model hybrid computing scheduling method based on basic sections. It is based on the overall consideration from local to basin to national region, adopts a unified basin upstream and downstream topological relationship, forms computing partitions, and each partition can bind multiple models, different precisions of the same model, etc. By encapsulating the three elements of input, output, and upstream and downstream topological processing of the model in the computing partition, an independent multi-process scheduling management program is written. It realizes multiple models in different computing partitions in the region, and synchronously calculates in the form of multiple processes on multiple computing nodes, which is convenient for independent debugging and testing of models, greatly improves computing time, can continuously add new models, update model precision scales, and continuously expand the computing area, and finally realizes hybrid computing of multiple models and output of model results in the national region.

[0145] Specifically, taking the national region as an example, basic data of small watersheds are collected, and seven major watershed divisions are selected; each watershed division is divided into 200 computing units according to the area and number of small watersheds; the exit nodes of the 200 computing units are extracted as basic sections, and 200 basic sections are determined to control the calculation; the scheduling engine controls a certain watershed on multiple specified computer nodes, and starts 200 processes at the same time to calculate the computing units controlled by the 200 control sections; the calculation engine dynamically updates the model input, updates the model parameters, searches for the upstream and downstream topology of the waiting model, and parses the results into the database after the calculation is completed; at the same time, the scheduling engine monitors the startup, shutdown, and status of the 200 processes; the results of the model are stored as time series data according to the feature types of points, lines, and surfaces for application display.

[0146] Next, combine Figure 5 , and give a detailed introduction to this plan. Figure 5The architecture and workflow of the solution are presented in a modular form. From watershed division to model calculation and then to result presentation, each step has a clear division of labor and management, ensuring the efficient operation of the system and accurate processing of data.

[0147] 1. Based on the watershed units in the region, divide the calculation units to form the basic sections.

[0148] like Figure 5 As shown in the figure, in order to support multiple models in the same area, we adopt the method of reducing the large to the small. First, for the entire target area, according to the collected regional small watershed data and upstream and downstream topology, according to different watersheds and different divisions, from upstream to downstream, every 200-500 square kilometers is a calculation unit, and the key position of the calculation unit (such as the outlet) is selected as a basic section. The purpose is to provide a basis for binding n models to each basic section in the next step and subsequent calculation scheduling based on this unified basic section.

[0149] Specifically, it includes: 1. Determine the number of basins: First determine the number of basins in the study area. 2. Form basin partitions: Divide the entire basin into multiple partitions, each partition corresponds to a basic section. 3. Divide calculation units: In each basin partition, further divide the calculation units, which are the basic units of model calculation. 4. Form basic sections: Determine the basic sections of each calculation unit, which are the starting and end points of the model calculation.

[0150] 2. Bind n model processing units to each basic section.

[0151] After the basic section is determined, 0-n model processing modules (processes) are mapped and bound to each basic section. Simply put, each basic section corresponds to a calculation unit, and a calculation unit contains multiple small watersheds. Each basic section is mapped to n model calculation modules, and each model calculation module corresponds to a calculation process. As long as the basic section is managed well, the basic object of the entire calculation scheduling is managed well.

[0152] Based on the model configuration of the basic section (station), the mixed assembly of multiple models is realized: Figure 2As shown in the figure, for the calculation area, a basic section with upstream and downstream topology is established. The rapid section 1 flows to the basic section 2, and the basic section 2 flows to the basic section 3. The same or different calculation models can be selected on different basic sections. The basic section 1 uses the API model, the basic section 2 uses the SKBY model, the basic section 3 uses the SKBY model, and the basic section 4 uses the API model, which conveniently realizes the mixed assembly of different models. The traditional method is to select the same calculation model for the entire area. For example, the entire area can be calculated using the API model, or the entire area can be calculated using the SKBY model. If it cannot be mixed, the advantages of each model cannot be brought into play. The soil type, soil texture, terrain, etc. in different regions are different, and the optimal model applicable may be different. If a unified model is used, the optimal result cannot be obtained. The model assembly based on the partitioned basic section solves this problem very well. The optimal model matching the terrain, geology and other characteristics can be selected in different regions to achieve the optimal solution.

[0153] Specifically, the following steps are included: 1. Update the model package input: configure model processing units for each basic section, which are responsible for specific calculation tasks. 2. Update the model package parameters: update the parameters of the model package according to the characteristics of the basic section to adapt to different calculation requirements. 3. Update the upstream and downstream topological data: update the upstream and downstream topological data related to the basic section, including water source, working conditions, flood diversion and other information. 4. Parse the model results and store them in the database: parse the results of the model calculation and store them in the database.

[0154] 3. Process scheduling management module.

[0155] like Figure 5 As shown, based on the process scheduling management module, the following can be achieved: 1. Start the calculation process and start the calculation task of the model. 2. Process shutdown: After the calculation task is completed, the corresponding process is closed. 3. Process status monitoring: Real-time monitoring of the status of each calculation process to ensure the smooth progress of the calculation task.

[0156] The process scheduling management module develops a web service interface to monitor each computing node. It mainly implements the startup, shutdown, and status monitoring of the program (model processing module) in the specified model package directory. The model processing module is a secondary encapsulation package for each model developed in C# language. The purpose is to continuously expand the model and provide support for the scheduling module in a consistent manner. Through this model processing module middle layer, the scheduling management only needs to manage this unique model processing module program. It supports Windows and Linux systems, and uses a consistent interface to implement the scheduling management of model processing programs on servers with different architectures.

[0157] The most critical thing in the process scheduling management module is: unified standard model calculation package: to achieve mixed assembly of different calculation models based on basic sections, it is necessary to have unified standard requirements for the algorithm package of the calculation model, so that the calculation engine can process different calculation packages with consistent operations. Figure 3 As shown in the figure, all packages are designed to be based on text interaction, including 4 directories: input directory, output directory, log directory, and sim configuration file directory. The sim configuration file directory contains json-formatted configuration files, which store the input and output file paths involved in the model calculation, the gpu index number used, the calculation time and other information. Taking the API and SKBY calculation programs as examples, the API.exe appsetting.json command can be used to read the input file in the input from the json file and output it to the output directory. The SKBY.exe appsetting.json command can be used to read the input file in the input from the json file and output it to the output file. If it is a java version of the calculation program package, the java-jarAPI.jar appsetting.json method can also be used to parse the input file in the input and generate the output file in the output. Through this file-based standardized structure requirement, calculation programs in different languages ​​can be supported, and it is also convenient to view and debug input and troubleshoot output through text files.

[0158] 4. Results query and display module.

[0159] Here, the result display module can display the calculation results of points, lines and surfaces;

[0160] Specifically include: Points (sites, hydrological stations, reservoir stations, dams, etc.): display the calculation results related to the site. Lines (rivers, levees, pipe networks): display the calculation results related to the river, levees, and pipe networks. Surfaces (rainfall surfaces, watershed surfaces, flooded surfaces): display the calculation results related to the rainfall surfaces, watershed surfaces, and flooded surfaces.

[0161] Here, different computational models focus on different objects and have different computational results. The spatial shape attributes are divided into three categories of results: points, lines, and surfaces. In the model processing module, the data in the result directory of each model is parsed and stored in the database according to type, and a result viewing interface is provided to users through the service interface, such as Figure 4 shown.

[0162] In addition, it also includes database modules, including: Real-time library: stores real-time data for real-time calculation and monitoring. Business library: stores business-related data, including model parameters, calculation results, etc. Role and permission library: manages user roles and permissions to ensure secure access to data.

[0163] In order to implement the method of the embodiment of the present application, the embodiment of the present application also provides a model calculation scheduling device, which corresponds to the above-mentioned model calculation scheduling method, and each step in the above-mentioned model calculation scheduling method embodiment is also fully applicable to the present model calculation scheduling device embodiment.

[0164] like Figure 6 As shown, the model calculation scheduling device 600 includes: a first determination module 601, a second determination module 602, a generation module 603, an acquisition module 604 and a scheduling module 605. The first determination module 601 is used to determine multiple calculation units of the basin based on the upstream and downstream topological relationship between each sub-basin partition of the basin and the area data of each sub-basin; the upstream and downstream topological relationship is generated based on encoding each sub-basin partition of the basin; the second determination module 602 is used to determine the location information of the basic section of the calculation unit for each calculation unit in the multiple calculation units; and based on the location information of the calculation unit in the upstream and downstream topological relationship, determine the type of model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit and the calculation process information corresponding to the model processing unit; the number of model processing units includes multiple; the generation module 603 is used to respond to the calculation for the basin Indication information, based on the type of each model processing unit, generates a computing task scheduling instruction; the acquisition module 604 is used to respond to the computing task scheduling instruction and obtain the identification information of the scheduling scheme of the watershed; the scheduling scheme includes multiple target model processing units, the computing process information corresponding to each target model processing unit, and the target computing node corresponding to each target model processing unit, and the target model processing unit is one of the multiple model processing units; the scheduling module 605 is used to start the target computing process corresponding to each target model processing unit on each target computing node based on the identification information of the scheduling scheme; and control the operation of each target model processing unit to generate a computing result; wherein the model processing unit is a model processing unit after standardized encapsulation.

[0165] In some embodiments, the acquisition module 604 is also used to acquire the location information of the exit node corresponding to each computing unit in the multiple computing units; the second determination module 602 is also used to determine the location information of the basic section of the computing unit based on the location information of the exit node.

[0166] In some embodiments, the scheduling module 605 is further used to generate a start instruction corresponding to the type information for each target model processing unit in each target model processing unit in the scheduling scheme based on the type information of the target model processing unit;

[0167] In response to a startup instruction, the computing package of the target model processing unit is started to run and a model calculation result is generated; wherein the computing package includes input file information, output file information, output directory information, a log directory and a configuration file; wherein the configuration file is used to store the input path information and output path information of the target model processing unit based on a set text format; the set text format includes a JSON format; based on the calculation results of each model, a calculation result of the watershed is generated.

[0168] In some embodiments, the generation module 603 is further used to classify the calculation results based on the spatial geometric information of multiple calculation units, generate classification result data; and display the classification result data in response to a result display request.

[0169] In some embodiments, the model calculation scheduling device also includes a display module 606, which is used to obtain a monitoring request; in response to the monitoring request, at least one of the following is displayed: identification information of the scheduling plan, calculation process information of each model processing unit, calculation status information of each model processing unit, and calculation progress information of each model processing unit, the calculation process information includes the number of processes and the process status.

[0170] In some embodiments, the model calculation scheduling device further includes an updating module 607 for obtaining a parameter change request for the model processing unit; and updating the initial parameters of the model processing unit to the target parameters in response to the parameter change request.

[0171] In some embodiments, the acquisition module 603 is also used to obtain indication information for adding a model processing unit; the indication information includes the type of the model processing unit to be added; the second determination module 602 is also used to respond to the indication information, based on the type of the model processing unit to be added, determine the target computing unit corresponding to the model processing unit to be added, and add the model to be added to the target computing unit; based on the position information of the basic section of the target computing unit, determine the input task corresponding to the model processing unit to be added, the output task of the model processing unit to be added, and the process information of the model processing unit to be added.

[0172] In practical applications, the first determination module 601, the second determination module 602, the generation module 603, the acquisition module 604, the scheduling module 605, the display module 606 and the update module 607 can be implemented by a processor in the model calculation scheduling device. Of course, the processor needs to run the computer program in the memory to implement its functions.

[0173] It should be noted that: the model calculation scheduling device provided in the above embodiment only uses the division of the above program modules as an example when performing model calculation scheduling. In actual applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device is divided into different program modules to complete all or part of the processing described above. In addition, the model calculation scheduling device provided in the above embodiment and the model calculation scheduling method embodiment belong to the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0174] Based on the hardware implementation of the above program modules and in order to implement the method of the embodiment of the present application, the embodiment of the present application also provides an electronic device. Figure 7 Only an exemplary structure of the electronic device is shown, not all structures, and it can be implemented as needed. Figure 7 Partial or complete structure shown. Figure 7 As shown, the electronic device 700 provided in the embodiment of the present application includes: at least one processor 701, a memory 702, a user interface 703 and at least one network interface 704. The various components in the electronic device 700 are coupled together through a bus system 705. It can be understood that the bus system 705 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 705 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, in Figure 7 Various buses are labeled as bus system 705.

[0175] The user interface 703 may include a display, a keyboard, a mouse, a trackball, a click wheel, keys, buttons, a touch pad or a touch screen.

[0176] The memory 702 in the embodiment of the present application is used to store various types of data to support the operation of the electronic device. Examples of such data include: any computer program used to operate on the electronic device.

[0177] The three-dimensional modeling method of the ore body of the electronic device disclosed in the embodiment of the present application can be applied to the processor 701, or implemented by the processor 701. The processor 701 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the three-dimensional modeling method of the ore body of the electronic device can be completed by the hardware integrated logic circuit or software instructions in the processor 701. The above-mentioned processor 701 can be a general-purpose processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 701 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present application. The general-purpose processor can be a microprocessor or any conventional processor, etc. In combination with the steps of the method disclosed in the embodiment of the present application, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in the memory 702, and the processor 701 reads the information in the memory 702, and completes the steps of the model calculation scheduling method of the electronic device provided in the embodiment of the present application in combination with its hardware.

[0178] In an exemplary embodiment, the electronic device may be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD), field programmable gate array (FPGA), general processor, controller, microcontroller (MCU), microprocessor, or other electronic components to execute the aforementioned method.

[0179] It can be understood that the memory 702 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or an electrically erasable programmable read-only memory (EEPROM).

[0180] Erasable Programmable Read-Only Memory), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disk, or compact disc read-only memory (CD-ROM); magnetic surface memory can be magnetic disk memory or. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), direct memory bus random access memory (DRRAM). The memory described in the embodiments of the present application is intended to include but is not limited to these and any other suitable types of memory.

[0181] In an exemplary embodiment, the present application also provides a computer storage medium, which can be a computer-readable storage medium, on which a computer program is stored, and the computer program can be executed by a processor to complete the steps of the method of the present application. The computer-readable storage medium can be a memory such as ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface storage, optical disk, or CD-ROM.

[0182] In an exemplary embodiment, the embodiment of the present application further provides a computer program product, including a computer program, and the above-mentioned computer program can be executed by the processor 701 of the electronic device to complete the steps of the method of the embodiment of the present application.

[0183] It should be noted that: "first", "second", etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0184] In addition, the technical solutions described in the embodiments of the present application can be combined arbitrarily without conflict.

[0185] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A model calculation scheduling method, characterized in that: include: Determine a plurality of computing units of the watershed based on the upstream and downstream topological relationship between each sub-watershed partition of the watershed and the area data of each sub-watershed partition; The upstream and downstream topological relationship is generated based on encoding each sub-basin partition of the basin; For each of the plurality of computing units, determining position information of a basic section of the computing unit; and based on the position information of the calculation unit in the upstream and downstream topological relationship, determining the type of the model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit, and the calculation process information corresponding to the model processing unit; the number of the model processing units includes multiple; In response to the calculation instruction information for the watershed, generating a calculation task scheduling instruction based on the type of each of the model processing units; In response to the computing task scheduling instruction, obtaining identification information of a scheduling scheme of the watershed; the scheduling scheme includes a plurality of target model processing units, computing process information corresponding to each of the target model processing units, and a target computing node corresponding to each of the target model processing units, wherein the target model processing unit is one of the plurality of model processing units; Based on the identification information of the scheduling scheme, on each of the target computing nodes, the target computing process corresponding to each of the target model processing units is started; and each of the target model processing units is controlled to run to generate the computing result of the watershed; The model processing unit is a standardized and packaged model processing unit.

2. The method according to claim 1, characterized in that The method further comprises: For each computing unit among the plurality of computing units, obtaining location information of an egress node corresponding to the computing unit; Based on the position information of the exit node, the position information of the basic section of the calculation unit is determined.

3. The method according to claim 1, characterized in that The controlling each of the target model processing units to run and generate the calculation result of the watershed includes: For each target model processing unit in each target model processing unit in the scheduling scheme, based on the type information of the target model processing unit, generating a start instruction corresponding to the type information; In response to the startup instruction, the computing package of the target model processing unit is started to run and the model computing result is generated; wherein the computing package includes input file information, output file information, output directory information, log directory and configuration file; wherein the configuration file is used to store the input path information and output path information of the target model processing unit based on a set text format; the set text format includes JSON format; Based on the calculation results of each of the models, the calculation results of the watershed are generated.

4. The method according to claim 1, characterized in that The method further comprises: Based on the spatial geometric information of the multiple computing units, the computing results are classified to generate classification result data; and in response to a result display request, the classification result data is displayed.

5. The method according to claim 1, characterized in that The method further comprises: Get monitoring request; In response to the monitoring request, at least one of the following is displayed: The identification information of the scheduling scheme, the calculation process information of each of the model processing units, the calculation status information of each of the model processing units and the calculation progress information of each of the model processing units, wherein the calculation process information includes the number of processes and the process status.

6. The method according to claim 1, characterized in that The method further comprises: Obtaining a parameter change request for the model processing unit; In response to the parameter change request, the initial parameters of the model processing unit are updated to target parameters.

7. The method according to claim 1, characterized in that The method further comprises: Obtaining instruction information for adding a model processing unit; the instruction information includes the type of the model processing unit to be added; In response to the indication information, based on the type of the model processing unit to be added, determining a target computing unit corresponding to the model processing unit to be added, and adding the model to be added to the target computing unit; Based on the position information of the basic section of the target computing unit, the input task corresponding to the to-be-added model processing unit, the output task corresponding to the to-be-added model processing unit and the process information of the to-be-added model processing unit are determined.

8. A model calculation scheduling device, characterized in that: The device comprises: A first determination module is used to determine a plurality of computing units of the watershed based on an upstream and downstream topological relationship between each sub-watershed partition of the watershed and area data of each of the sub-watershed partitions; the upstream and downstream topological relationship is generated based on encoding each sub-watershed partition of the watershed; A second determination module is used to determine, for each of the plurality of calculation units, the location information of the basic section of the calculation unit; and based on the location information of the calculation unit in the upstream and downstream topological relationship, determine the type of the model processing unit of the basic section, the input task of the model processing unit, the output task of the model processing unit, and the calculation process information corresponding to the model processing unit; the number of the model processing units includes a plurality; A generating module, configured to generate a computing task scheduling instruction in response to the computing instruction information for the watershed and based on the type of each of the model processing units; an acquisition module, configured to acquire identification information of a scheduling scheme of the watershed in response to the computing task scheduling instruction; the scheduling scheme includes a plurality of target model processing units, computing process information corresponding to each of the target model processing units, and a target computing node corresponding to each of the target model processing units, wherein the target model processing unit is one of the plurality of model processing units; A scheduling module, configured to start the target computing process corresponding to each target model processing unit on each target computing node based on the identification information of the scheduling scheme; and control the operation of each target model processing unit to generate a computing result; The model processing unit is a standardized and packaged model processing unit.

9. An electronic device, characterized in that: include: A processor and a memory for storing a computer program that can be executed on the processor, wherein: A processor, configured to execute the steps of any one of the methods of claims 1 to 7 when running a computer program.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of any one of the methods of claims 1 to 7 are implemented.

Images