Full-scene elastic computing power service method and system based on edge reasoning

Abstract

The invention provides a full-scene elastic computing power service method and system based on edge reasoning, and the method comprises the steps: obtaining field operation data corresponding to each scene site, constructing a resource demand list corresponding to the scene site, deducing a resource execution process corresponding to the resource demand list according to a data source corresponding to the field operation data, and obtaining a resource execution process corresponding to the resource demand list; constructing a resource operation process corresponding to each scene site, correcting the resource demand list according to operation abnormal information in each resource operation process, configuring corresponding effective resources for each scene site according to a correction result, and according to the resource operation information corresponding to each scene site, configuring the corresponding effective resources for each scene site. According to the embodiment of the invention, parameter optimization is carried out on the resources of the corresponding scene site, and an optimized resource list corresponding to each scene site is generated, so that an optimized resource list can be constructed for each site of an enterprise in a short time, and related personnel are assisted to complete site layout and parameter adjustment work.

Description

Technical Field

[0001] This invention relates to the field of computing power allocation technology, and in particular to a method and system for providing elastic computing power services across all scenarios based on edge inference. Background Technology

[0002] In recent years, due to the rapid development of technology, many enterprises have expanded their production scale or diversified their product range to keep pace with the times. This inevitably requires the introduction of a large number of new equipment. Since each enterprise has its own production plan, the new equipment introduced is mostly different, and the functions and production tasks of the equipment also vary. This requires relevant personnel to spend a lot of time analyzing how to lay out and debug the equipment. Although many enterprises send relevant personnel to learn from experienced companies or institutions and then optimize the existing layout according to their own situation, this can reduce the equipment debugging time to a certain extent. However, a lot of time is still needed for subsequent debugging. Improper debugging may lead to factory shutdowns and affect the enterprise's production efficiency. Therefore, how to lay out resources for an enterprise in a short period of time has become an urgent problem to be solved.

[0003] Therefore, this invention provides a method and system for providing elastic computing power services across all scenarios based on edge inference. Summary of the Invention

[0004] This invention provides a full-scenario elastic computing power service method based on edge inference, which can build an optimized resource list for each site of an enterprise in a short time, assisting relevant personnel in completing site layout and parameter adjustment.

[0005] This invention provides a method for providing elastic computing power services across all scenarios based on edge inference, comprising: Step 1: Obtain the on-site operation data corresponding to each scenario, and construct a resource requirement list for each scenario based on the data type; Step 2: Based on the data source corresponding to the on-site operation data, deduce the resource execution process corresponding to the resource requirement list, and construct the resource operation process corresponding to each scenario on-site. Step 3: Correct the resource requirement list based on the operational anomaly information during the operation of each resource, and configure the corresponding effective resources for each scenario based on the correction results; Step 4: Based on the resource operation information corresponding to each scenario, optimize the parameters of the resources corresponding to the scenario, generate an optimized resource list for each scenario, and display it.

[0006] In one feasible approach Step 1 includes: Step 11: Obtain the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. Step 12: Identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. Step 13: Periodically monitor each of the field operation data using a preset sliding time window to obtain the data change characteristics corresponding to each scene, and obtain the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. Step 14: Mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

[0007] In one feasible approach Step 14 includes: Step 141: Determine the changed data bit contained in the corresponding field operation data according to the data change characteristics, determine the change feedback source in the data producer information according to the change feedback characteristics, match the change feedback source with the changed data bit, and determine the mark position of each changed resource in the field operation data. Step 142: Based on the marking results, filter several unmarked locations and corresponding passive sub-operation data in the field operation data; construct a first matching condition based on the resource matching type corresponding to the unmarked location; and construct a second matching condition based on the data period corresponding to the passive sub-operation data. Step 143: Using the first matching condition and the second matching condition, find several matching resources for the unmarked position in the data producer information, arrange and filter the matching resources, and determine the fixed resource that matches each passive sub-run data according to the repulsion between different matching resources; Step 144: Identify the first resource execution information corresponding to each of the variable resources and the second resource execution information corresponding to each of the fixed resources in the data producer information, and construct a resource demand list corresponding to the scene.

[0008] In one feasible approach Step 2 includes: Step 21: Obtain several data sources corresponding to each of the field operation data, and construct the source logic relationship between the data sources included in the scene based on the flow path of the field operation data between each of the data sources; Step 22: Set the corresponding resource logic relationship for the resource requirement list according to the source logic relationship, construct the resource execution process corresponding to the resource requirement list in combination with the resource achievable function corresponding to each resource, simulate the resource execution process, and obtain the resource achievable function corresponding to each resource; Step 23: Based on the functions already implemented by the resources, extract the resource operation data corresponding to each resource from the field operation data, and perform dimensionality reduction processing on each resource operation data to obtain several principal components corresponding to each resource operation data; Step 24: During the resource execution process, filter the master information corresponding to each principal component, determine several functional operation modes corresponding to each resource, arrange the functional operation modes according to the resource logical relationship, and obtain the resource operation process of the scene.

[0009] In one feasible approach Step 3 includes: Step 31: Collect the operation information corresponding to each of the resource operation processes in a collaborative manner, perform Markov jump training on each of the operation information, obtain several key features of resource operation corresponding to the scene, and construct a Markov chain corresponding to the scene. Step 32: Identify several resource function transfer processes corresponding to the scene in the Markov chain, evaluate the coherence of each resource function transfer process, and determine several stuttering anomaly information corresponding to the scene. Step 33: Construct a scatter plot corresponding to the scene based on the feature value corresponding to each key feature of resource operation, obtain the discrete resource corresponding to each discrete point in the scatter plot, and filter the corresponding discrete anomaly information in the key features of the operation of the discrete resource. Step 34: Determine the missing computing speed and computing power of the corresponding scene based on the stuttering anomaly information, determine the missing computing offloading computing power of the scene based on the discrete anomaly information, and generate a list of effective resource requirements corresponding to the scene.

[0010] In one feasible approach Step 3 further includes: Step 35: Identify the updated resources included in the list of valid resource requirements, and use artificial intelligence technology to analyze the resource response data of the corresponding updated resources based on the on-site operation data; Step 36: Set parameters for the corresponding updated resource based on the resource response data to obtain valid resources and configure them in the corresponding scene.

[0011] In one feasible approach Step 4 includes: Step 41: Obtain the resource operation information corresponding to each of the above scenarios, deduce the on-site operation parameters corresponding to each on-site resource, and use artificial intelligence to evaluate the parameter utilization rate corresponding to each of the above scenarios. Step 42: Determine the parameter adjustment direction and parameter adjustment amount for the corresponding field resources based on the parameter utilization rate, optimize the corresponding field operation parameters, obtain and display the optimized resource list for each scene.

[0012] In one feasible approach Also includes: The calculation allocation for the corresponding scene is derived from the optimized resource list and then displayed.

[0013] This invention provides a full-scenario elastic computing power service system based on edge inference, comprising: The resource identification module is used to acquire the on-site operation data corresponding to each scenario and construct a resource requirement list for each scenario based on the data type. The process derivation module is used to deduce the resource execution process corresponding to the resource requirement list based on the data source corresponding to the on-site operation data, and to construct the resource operation process corresponding to each on-site scenario. The resource requirement list correction module is used to correct the resource requirement list based on the operation anomaly information during the operation of each resource, and to configure the corresponding effective resources for each scenario based on the correction results. The resource optimization module is used to optimize the parameters of the resources corresponding to each scenario based on the resource operation information of each scenario, generate an optimized resource list for each scenario and display it.

[0014] In one feasible approach The resource identification module includes: The data decomposition unit is used to acquire the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. The data interaction unit is used to identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. The change analysis unit is used to periodically monitor each of the field operation data using a preset sliding time window, obtain the data change characteristics corresponding to each scene, acquire the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. The fixed analysis unit is used to mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

[0015] The beneficial effects of the above technical solution are as follows: In order to quickly configure effective resources and adjust parameters for each scene, the required resource list is first determined based on the on-site operation data of each scene. Then, by deriving the operation process of resources in the scene, the abnormalities generated during the operation are identified, thereby correcting the resource requirement list and configuring effective resources for the scene. In order to ensure that each scene can operate stably and continuously, the parameters of the resources are improved and optimized based on the resource operation information of the scene during its operation, thereby coordinating the adjustment of the corresponding resources and ensuring the normal operation of the scene. In this way, the layout of the scene can be realized, and the parameter adjustment can be carried out in a short time, ensuring the normal operation of the scene and improving the production efficiency of the enterprise.

[0016] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram illustrating the workflow of a full-scenario elastic computing power service method based on edge inference in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the composition of a full-scenario elastic computing power service system based on edge reasoning in an embodiment of the present invention. Detailed Implementation

[0019] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0020] Example 1: This example provides a method for providing elastic computing power services across all scenarios based on edge inference, such as... Figure 1 As shown, it includes: Step 1: Obtain the on-site operation data corresponding to each scenario, and construct a resource requirement list for each scenario based on the data type; Step 2: Based on the data source corresponding to the on-site operation data, deduce the resource execution process corresponding to the resource requirement list, and construct the resource operation process corresponding to each scenario on-site. Step 3: Correct the resource requirement list based on the operational anomaly information during the operation of each resource, and configure the corresponding effective resources for each scenario based on the correction results; Step 4: Based on the resource operation information corresponding to each scenario, optimize the parameters of the resources corresponding to the scenario, generate an optimized resource list for each scenario, and display it.

[0021] In this example, a scene site represents a company's production scene, and a company can have one or more scene sites; In this example, the resource requirements list represents a summary of all resources needed when executing the production plan on-site in this scenario; In this example, the resource execution process represents the operation of each resource during production planning; In this example, resources include: devices, sensors, terminals, etc. In this example, parameter optimization refers to the process of adjusting the operating parameters of resources.

[0022] The working principle and beneficial effects of the above technical solution are as follows: In order to quickly configure effective resources and adjust parameters for each scene, the required resource list is first determined based on the on-site operation data of each scene. Then, by deriving the operation process of resources in the scene, anomalies generated during the operation are identified, thereby correcting the resource requirement list and configuring effective resources for the scene. To ensure the stable and continuous operation of each scene, the parameters of the resources are improved and optimized based on the resource operation information of the scene during operation, thereby coordinating the adjustment of the corresponding resources and ensuring the normal operation of the scene. In this way, the layout of the scene can be realized, and parameter adjustment can be carried out in a short time, ensuring the normal operation of the scene and improving the production efficiency of the enterprise.

[0023] Example 2: Based on Example 1, the method for providing full-scenario elastic computing power services based on edge inference, step 1 includes: Step 11: Obtain the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. Step 12: Identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. Step 13: Periodically monitor each of the field operation data using a preset sliding time window to obtain the data change characteristics corresponding to each scene, and obtain the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. Step 14: Mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

[0024] In this example, the real-time data stream represents the data stream generated during the operation of the scene; In this example, the data producer information includes the source name that generated the data, the data format, and basic information about the data source; In this example, the data classification rule represents the rule for dividing the corresponding real-time data stream according to the data production patterns of the data producer; In this example, the distributed data queue represents the result of temporarily storing sub-data segments flowing in different directions; In this example, the data interaction features represent the characteristics that emerge when different sub-data segments interact with each other. In this example, the preset sliding time window represents a time window with a predetermined window size, used to acquire periodic data from the field operation data; In this example, variable resources refer to resources whose data changes within a period, while fixed resources refer to resources whose data does not change within a period. In this example, passive sub-run data refers to sub-run data whose source is temporarily uncertain.

[0025] The working principle and beneficial effects of the above technical solution are as follows: To clearly identify the resources required for each scene and facilitate advance preparation by relevant personnel, data classification rules are first constructed based on the real-time data stream and its corresponding data producer information. The real-time data stream is divided into several sub-data segments, thereby generating a distributed data queue. The distributed storage characteristics of this queue are used to identify the data interaction features between different sub-data segments. Combined with the real-time data stream, the on-site operation data of each scene is constructed. Then, through periodic monitoring, the data change characteristics of the scene are extracted to deduce the changing resources of the scene. Further filtering is performed on sub-operation data whose source is temporarily uncertain, thereby finding the corresponding fixed resources in the producer information. Finally, the changing resources and fixed resources are summarized to obtain the resource requirement list of the scene. In this way, not only can the resources of the scene be identified, but their status can also be determined. Moreover, this method can identify all resources in the scene, assisting relevant personnel in completing the layout preparation work in a short time.

[0026] Example 3: Based on Example 2, the method for providing elastic computing power services across all scenarios based on edge inference, step 14 includes: Step 141: Determine the changed data bit contained in the corresponding field operation data according to the data change characteristics, determine the change feedback source in the data producer information according to the change feedback characteristics, match the change feedback source with the changed data bit, and determine the mark position of each changed resource in the field operation data. Step 142: Based on the marking results, filter several unmarked locations and corresponding passive sub-operation data in the field operation data; construct a first matching condition based on the resource matching type corresponding to the unmarked location; and construct a second matching condition based on the data period corresponding to the passive sub-operation data. Step 143: Using the first matching condition and the second matching condition, find several matching resources for the unmarked position in the data producer information, arrange and filter the matching resources, and determine the fixed resource that matches each passive sub-run data according to the repulsion between different matching resources; Step 144: Identify the first resource execution information corresponding to each of the variable resources and the second resource execution information corresponding to each of the fixed resources in the data producer information, and construct a resource demand list corresponding to the scene.

[0027] In this example, the change feedback source refers to the resource source that generates the data change characteristics; In this example, the first matching condition means: perform resource matching for unmarked locations based on resource matching type, and the second matching condition means: perform resource matching for unmarked locations based on data period. In this example, the matching resource represents the resource that can fulfill the first matching condition and the second condition; In this example, sorting and filtering refers to the process of sorting matching resources; In this example, the first resource execution information represents the information generated when the variable resource performs a task in the scene, and the second resource execution information represents the information generated when the fixed resource performs a task in the scene.

[0028] The working principle and beneficial effects of the above technical solution are as follows: By utilizing the characteristics of data change, the variable data position is located in the field operation data. Based on the change feedback source in the data production information, the variable data is matched with it to determine the location of the variable resource and mark it. This indirectly determines the unmarked position and the corresponding passive sub-operation data. Then, corresponding matching conditions are established for the unmarked position. The matching conditions are used to find the corresponding matching resources in the producer information. By arranging and filtering and combining the repulsion between each arrangement method, the fixed resources of each passive sub-operation data are determined. Finally, the resource requirement list of the corresponding scene is determined based on the execution information of each resource. In this way, both variable resources and fixed resources in the scene can be processed to construct a list with reference value.

[0029] Example 4: Based on Example 1, the method for providing full-scenario elastic computing power services based on edge inference, step 2 includes: Step 21: Obtain several data sources corresponding to each of the field operation data, and construct the source logic relationship between the data sources included in the scene based on the flow path of the field operation data between each of the data sources; Step 22: Set the corresponding resource logic relationship for the resource requirement list according to the source logic relationship, construct the resource execution process corresponding to the resource requirement list in combination with the resource achievable function corresponding to each resource, simulate the resource execution process, and obtain the resource achievable function corresponding to each resource; Step 23: Based on the functions already implemented by the resources, extract the resource operation data corresponding to each resource from the field operation data, and perform dimensionality reduction processing on each resource operation data to obtain several principal components corresponding to each resource operation data; Step 24: During the resource execution process, filter the master information corresponding to each principal component, determine several functional operation modes corresponding to each resource, arrange the functional operation modes according to the resource logical relationship, and obtain the resource operation process of the scene.

[0030] In this example, the flow path represents the process of on-site operational data flowing between various data sources; In this example, the source logical relationship represents the working logical relationship between different data sources; In this example, the function operation mode refers to the method used when a resource implements a function; In this example, the resource realizeable function refers to the function that a resource can realize during operation, while the resource realized function refers to the function that a resource has already realized during operation.

[0031] The working principle and beneficial effects of the above technical solution are as follows: First, the flow path of on-site operational data between various data sources is analyzed to determine the source logic relationship between data sources in a scenario. From this, the resource logic relationship can be inferred. Then, the resource execution process is constructed by combining the achievable functions of each resource, and the realized functions of the resources are determined. Furthermore, the principal components are obtained by performing dimensionality reduction processing on the resource operation data of each resource, thereby determining the functional operation mode of the resources. Combined with the corresponding resource logic relationship, the resource operation process of the scenario is constructed. In this way, the resources of the scenario can be comprehensively analyzed, so that the obtained resource operation process not only fits the actual operation process, but also optimizes the resources from the side.

[0032] Example 5: Based on Example 1, the method for providing full-scenario elastic computing power services based on edge inference, step 3 includes: Step 31: Collect the operation information corresponding to each of the resource operation processes in a collaborative manner, perform Markov jump training on each of the operation information, obtain several key features of resource operation corresponding to the scene, and construct a Markov chain corresponding to the scene. Step 32: Identify several resource function transfer processes corresponding to the scene in the Markov chain, evaluate the coherence of each resource function transfer process, and determine several stuttering anomaly information corresponding to the scene. Step 33: Construct a scatter plot corresponding to the scene based on the feature value corresponding to each key feature of resource operation, obtain the discrete resource corresponding to each discrete point in the scatter plot, and filter the corresponding discrete anomaly information in the key features of the operation of the discrete resource. Step 34: Determine the missing computing speed and computing power of the corresponding scene based on the stuttering anomaly information, determine the missing computing offloading computing power of the scene based on the discrete anomaly information, and generate a list of effective resource requirements corresponding to the scene.

[0033] In this example, Markov jump training represents the process of extracting valuable information from runtime information; In this example, a Markov chain represents a mathematical model used to describe the state transitions between various resources in a scenario. In this example, missing computing speed means that the computing power related to computing speed is missing in the scene, and missing computing offloading means that the computing power related to data offloading is missing in the scene.

[0034] The working principle and beneficial effects of the above technical solution are as follows: In order to further improve the resource integration of the scene, the operation information generated during the resource operation is first trained to obtain the corresponding Markov chain. Several resource function transfer processes in the scene are identified in the chain. Through coherence evaluation, the stuttering anomaly information in the scene is determined. At the same time, the discrete anomaly information in the scene is identified by constructing a scatter plot of the scene. Finally, the corresponding computing power is allocated to the scene based on the anomaly information, and the effective resource demand list of the scene is obtained. In this way, different types of anomalies in the scene can be identified and the corresponding computing power can be matched to them, ensuring that the effective resource demand list can meet the needs of the scene.

[0035] Example 6: Based on Example 5, the method for providing full-scenario elastic computing power services based on edge inference, step 3 further includes: Step 35: Identify the updated resources included in the list of valid resource requirements, and use artificial intelligence technology to analyze the resource response data of the corresponding updated resources based on the on-site operation data; Step 36: Set parameters for the corresponding updated resource based on the resource response data to obtain valid resources and configure them in the corresponding scene.

[0036] In this example, updating resources refers to resources that are included in the list of valid resource requirements but not in the list of resource requirements.

[0037] The working principle and beneficial effects of the above technical solution are as follows: by analyzing the updated resources to determine the parameters that should be set, and then configuring the corresponding parameters, the entire process of resource configuration for the scene is completed.

[0038] Example 7: Based on Example 1, the method for providing full-scenario elastic computing power services based on edge inference, step 4 includes: Step 41: Obtain the resource operation information corresponding to each of the above scenarios, deduce the on-site operation parameters corresponding to each on-site resource, and use artificial intelligence to evaluate the parameter utilization rate corresponding to each of the above scenarios. Step 42: Determine the parameter adjustment direction and parameter adjustment amount for the corresponding field resources based on the parameter utilization rate, optimize the corresponding field operation parameters, obtain and display the optimized resource list for each scene.

[0039] The working principle and beneficial effects of the above technical solution are as follows: In order to ensure the long-term effective operation of resources, the parameters of the scene are adjusted according to the resource operation information of the scene, ensuring that resources are synchronized with the production progress, improving production efficiency and reducing caching time.

[0040] Example 8: Based on Example 7, the method for providing full-scenario elastic computing power services based on edge inference further includes: The calculation allocation for the corresponding scene is derived from the optimized resource list and then displayed.

[0041] The working principle and beneficial effects of the above technical solution are as follows: the resource optimization list is used to deduce the computing power allocation in the scene, providing relevant personnel with inspection reference.

[0042] Example 9: This example provides a full-scenario elastic computing power service system based on edge inference, such as... Figure 2 As shown, it includes: The resource identification module is used to acquire the on-site operation data corresponding to each scenario and construct a resource requirement list for each scenario based on the data type. The process derivation module is used to deduce the resource execution process corresponding to the resource requirement list based on the data source corresponding to the on-site operation data, and to construct the resource operation process corresponding to each on-site scenario. The resource requirement list correction module is used to correct the resource requirement list based on the operation anomaly information during the operation of each resource, and to configure the corresponding effective resources for each scenario based on the correction results. The resource optimization module is used to optimize the parameters of the resources corresponding to each scenario based on the resource operation information of each scenario, generate an optimized resource list for each scenario and display it.

[0043] In this example, a scene site represents a company's production scene, and a company can have one or more scene sites; In this example, the resource requirements list represents a summary of all resources needed when executing the production plan on-site in this scenario; In this example, the resource execution process represents the operation of each resource during production planning; In this example, resources include: devices, sensors, terminals, etc. In this example, parameter optimization refers to the process of adjusting the operating parameters of resources.

[0044] The working principle and beneficial effects of the above technical solution are as follows: In order to quickly configure effective resources and adjust parameters for each scene, the required resource list is first determined based on the on-site operation data of each scene. Then, by deriving the operation process of resources in the scene, anomalies generated during the operation are identified, thereby correcting the resource requirement list and configuring effective resources for the scene. To ensure the stable and continuous operation of each scene, the parameters of the resources are improved and optimized based on the resource operation information of the scene during operation, thereby coordinating the adjustment of the corresponding resources and ensuring the normal operation of the scene. In this way, the layout of the scene can be realized, and parameter adjustment can be carried out in a short time, ensuring the normal operation of the scene and improving the production efficiency of the enterprise.

[0045] Example 10: Based on Example 9, the resource identification module of the full-scenario elastic computing service system based on edge inference includes: The data decomposition unit is used to acquire the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. The data interaction unit is used to identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. The change analysis unit is used to periodically monitor each of the field operation data using a preset sliding time window, obtain the data change characteristics corresponding to each scene, acquire the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. The fixed analysis unit is used to mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

[0046] In this example, the real-time data stream represents the data stream generated during the operation of the scene; In this example, the data producer information includes the source name that generated the data, the data format, and basic information about the data source; In this example, the data classification rule represents the rule for dividing the corresponding real-time data stream according to the data production patterns of the data producer; In this example, the distributed data queue represents the result of temporarily storing sub-data segments flowing in different directions; In this example, the data interaction features represent the characteristics that emerge when different sub-data segments interact with each other. In this example, the preset sliding time window represents a time window with a predetermined window size, used to acquire periodic data from the field operation data; In this example, variable resources refer to resources whose data changes within a period, while fixed resources refer to resources whose data does not change within a period. In this example, passive sub-run data refers to sub-run data whose source is temporarily uncertain.

[0047] The working principle and beneficial effects of the above technical solution are as follows: To clearly identify the resources required for each scene and facilitate advance preparation by relevant personnel, data classification rules are first constructed based on the real-time data stream and its corresponding data producer information. The real-time data stream is divided into several sub-data segments, thereby generating a distributed data queue. The distributed storage characteristics of this queue are used to identify the data interaction features between different sub-data segments. Combined with the real-time data stream, the on-site operation data of each scene is constructed. Then, through periodic monitoring, the data change characteristics of the scene are extracted to deduce the changing resources of the scene. Further filtering is performed on sub-operation data whose source is temporarily uncertain, thereby finding the corresponding fixed resources in the producer information. Finally, the changing resources and fixed resources are summarized to obtain the resource requirement list of the scene. In this way, not only can the resources of the scene be identified, but their status can also be determined. Moreover, this method can identify all resources in the scene, assisting relevant personnel in completing the layout preparation work in a short time.

[0048] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for providing elastic computing power services across all scenarios based on edge inference, characterized in that, include: Step 1: Obtain the on-site operation data corresponding to each scenario, and construct a resource requirement list for each scenario based on the data type; Step 2: Based on the data source corresponding to the on-site operation data, deduce the resource execution process corresponding to the resource requirement list, and construct the resource operation process corresponding to each scenario on-site. Step 3: Correct the resource requirement list based on the operational anomaly information during the operation of each resource, and configure the corresponding effective resources for each scenario based on the correction results; Step 4: Based on the resource operation information corresponding to each scenario, optimize the parameters of the resources corresponding to the scenario, generate an optimized resource list for each scenario, and display it.

2. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 1, characterized in that, Step 1 includes: Step 11: Obtain the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. Step 12: Identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. Step 13: Periodically monitor each of the field operation data using a preset sliding time window to obtain the data change characteristics corresponding to each scene, and obtain the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. Step 14: Mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

3. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 2, characterized in that, Step 14 includes: Step 141: Determine the changed data bit contained in the corresponding field operation data according to the data change characteristics, determine the change feedback source in the data producer information according to the change feedback characteristics, match the change feedback source with the changed data bit, and determine the mark position of each changed resource in the field operation data. Step 142: Based on the marking results, filter several unmarked locations and corresponding passive sub-operation data in the field operation data; construct a first matching condition based on the resource matching type corresponding to the unmarked location; and construct a second matching condition based on the data period corresponding to the passive sub-operation data. Step 143: Using the first matching condition and the second matching condition, find several matching resources for the unmarked position in the data producer information, arrange and filter the matching resources, and determine the fixed resource that matches each passive sub-run data according to the repulsion between different matching resources; Step 144: Identify the first resource execution information corresponding to each of the variable resources and the second resource execution information corresponding to each of the fixed resources in the data producer information, and construct a resource demand list corresponding to the scene.

4. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 1, characterized in that, Step 2 includes: Step 21: Obtain several data sources corresponding to each of the field operation data, and construct the source logic relationship between the data sources included in the scene based on the flow path of the field operation data between each of the data sources; Step 22: Set the corresponding resource logic relationship for the resource requirement list according to the source logic relationship, construct the resource execution process corresponding to the resource requirement list in combination with the resource achievable function corresponding to each resource, simulate the resource execution process, and obtain the resource achievable function corresponding to each resource; Step 23: Based on the functions already implemented by the resources, extract the resource operation data corresponding to each resource from the field operation data, and perform dimensionality reduction processing on each resource operation data to obtain several principal components corresponding to each resource operation data; Step 24: During the resource execution process, filter the master information corresponding to each principal component, determine several functional operation modes corresponding to each resource, arrange the functional operation modes according to the resource logical relationship, and obtain the resource operation process of the scene.

5. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 1, characterized in that, Step 3 includes: Step 31: Collect the operation information corresponding to each of the resource operation processes in a collaborative manner, perform Markov jump training on each of the operation information, obtain several key features of resource operation corresponding to the scene, and construct a Markov chain corresponding to the scene. Step 32: Identify several resource function transfer processes corresponding to the scene in the Markov chain, evaluate the coherence of each resource function transfer process, and determine several stuttering anomaly information corresponding to the scene. Step 33: Construct a scatter plot corresponding to the scene based on the feature value corresponding to each key feature of resource operation, obtain the discrete resource corresponding to each discrete point in the scatter plot, and filter the corresponding discrete anomaly information in the key features of the operation of the discrete resource. Step 34: Determine the missing computing speed and computing power of the corresponding scene based on the stuttering anomaly information, determine the missing computing offloading computing power of the scene based on the discrete anomaly information, and generate a list of effective resource requirements corresponding to the scene.

6. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 5, characterized in that, Step 3 further includes: Step 35: Identify the updated resources included in the list of valid resource requirements, and use artificial intelligence technology to analyze the resource response data of the corresponding updated resources based on the on-site operation data; Step 36: Set parameters for the corresponding updated resource based on the resource response data to obtain valid resources and configure them in the corresponding scene.

7. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 1, characterized in that, Step 4 includes: Step 41: Obtain the resource operation information corresponding to each of the above scenarios, deduce the on-site operation parameters corresponding to each on-site resource, and use artificial intelligence to evaluate the parameter utilization rate corresponding to each of the above scenarios. Step 42: Determine the parameter adjustment direction and parameter adjustment amount for the corresponding field resources based on the parameter utilization rate, optimize the corresponding field operation parameters, obtain and display the optimized resource list for each scene.

8. The method for providing full-scenario elastic computing power services based on edge inference as described in claim 7, characterized in that, Also includes: The calculation allocation for the corresponding scene is derived from the optimized resource list and then displayed.

9. A full-scenario elastic computing power service system based on edge inference, characterized in that, include: The resource identification module is used to acquire the on-site operation data corresponding to each scenario and construct a resource requirement list for each scenario based on the data type. The process derivation module is used to deduce the resource execution process corresponding to the resource requirement list based on the data source corresponding to the on-site operation data, and to construct the resource operation process corresponding to each on-site scenario. The resource requirement list correction module is used to correct the resource requirement list based on the operation anomaly information during the operation of each resource, and to configure the corresponding effective resources for each scenario based on the correction results. The resource optimization module is used to optimize the parameters of the resources corresponding to each scenario based on the resource operation information of each scenario, generate an optimized resource list for each scenario and display it.

10. The full-scenario elastic computing power service system based on edge inference as described in claim 9, characterized in that, The resource identification module includes: The data decomposition unit is used to acquire the real-time data stream corresponding to each of the aforementioned scenarios, identify the data producer information corresponding to each of the aforementioned real-time data streams, construct data classification rules corresponding to the aforementioned scenarios based on the data producer information, divide each of the aforementioned real-time data streams into several sub-data segments, and construct a distributed data queue corresponding to each of the aforementioned scenarios based on the data type corresponding to each of the aforementioned sub-data segments. The data interaction unit is used to identify the data interaction characteristics between different sub-data segments in the distributed data queue, construct the operation structure of each scene, transmit the real-time data stream to the corresponding operation structure for structure optimization, and obtain the scene operation data corresponding to each scene. The change analysis unit is used to periodically monitor each of the field operation data using a preset sliding time window, obtain the data change characteristics corresponding to each scene, acquire the change feedback characteristics corresponding to the data producer information, and use the change feedback characteristics to determine the change resources corresponding to the scene. The fixed analysis unit is used to mark each of the variable resources in the corresponding field operation data, filter the passive sub-operation data in the field operation data, find the fixed resources that match the passive sub-operation data in the corresponding data producer information, and generate a resource requirement list corresponding to the scene based on the corresponding variable resources.

Images