AI-based multi-dimensional resource intelligent scheduling method and system

Abstract

The application relates to the technical field of resource scheduling, and discloses a multi-dimensional resource intelligent scheduling method and system based on AI, which comprises the following steps: when there are multiple demand objects with data processing requirements, based on the actual data processing requirements of each demand object and the corresponding demand purpose, the target data resources and the first computing power resources required by each demand object are determined; for each demand object, based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements, the first computing power resources of the demand object are corrected to obtain the second computing power resources; and the use permissions of each demand object, the corresponding target data resources and the second computing power resources are opened to trigger the demand object to execute the data processing operation corresponding to the actual data processing requirement corresponding to the demand object. It can be seen that the application can improve the allocation flexibility, timeliness and accuracy of computing power resources.

Description

Technical Field

[0001] This invention relates to the field of resource scheduling technology, and in particular to an AI-based multi-dimensional intelligent resource scheduling method and system. Background Technology

[0002] In today's digital age, the demand for data processing is exploding, encompassing various fields such as finance, healthcare, transportation, and scientific research. With the rapid increase in data volume and the growing complexity of data processing tasks, higher demands are being placed on the performance and efficiency of data processing systems. In existing data processing systems, resource scheduling and management face numerous challenges when dealing with multiple entities with data processing needs (hereinafter referred to as demand entities).

[0003] In terms of computing resource allocation, current common methods often allocate computing resources to each user based on rough estimates or fixed rules. For example, in some cloud computing platforms, computing resources are allocated according to a certain proportion based on the scale of the tasks submitted by users in advance and the expected running time. However, this conventional resource allocation method is rather rigid and cannot adaptively allocate resources according to the actual situation of the users. It suffers from the problem of allocating too much computing power to simple data statistics tasks, while allocating insufficient computing power to complex artificial intelligence model training tasks. In other words, it suffers from low flexibility and accuracy in the allocation of computing resources.

[0004] Meanwhile, the existing data resources, the data resources to be allocated, and the computing power resources of the demanding entities are not static. In practical applications, these factors dynamically adjust with time, business development, or changes in the external environment. For example, in financial transaction data processing, real-time changes in market conditions can lead to changes in the types and scale of data to be processed, thus affecting the demand for computing power. However, most existing resource scheduling systems lack effective dynamic adjustment mechanisms and cannot promptly detect these changes and adjust the allocation of computing power resources accordingly. This results in low timeliness and low accuracy in computing power resource allocation.

[0005] Therefore, it is particularly important to provide a corresponding solution to the aforementioned technical problems existing in the current resource scheduling system. Summary of the Invention

[0006] This invention provides an AI-based multi-dimensional intelligent resource scheduling method and system, which can improve the flexibility and accuracy of computing resource allocation, as well as the timeliness and precision of computing resource allocation.

[0007] The first aspect of this invention discloses an AI-based multi-dimensional intelligent resource scheduling method, the method comprising:

[0008] When there are multiple objects with data processing needs, the target data resources and the first computing power resources required by each object are determined based on the actual data processing needs of each object and the purpose corresponding to the actual data processing needs of each object.

[0009] For each of the aforementioned demand objects, based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object.

[0010] For each of the aforementioned demand objects, grant access to the demand object and its corresponding target data resource and the second computing power resource, thereby triggering the demand object to perform data processing operations corresponding to its actual data processing needs based on its corresponding target data resource and the second computing power resource.

[0011] As an optional implementation, in the first aspect of the present invention, determining the target data resources and first computing power resources required by each of the demand objects based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object includes:

[0012] For each of the actual data processing requirements of the aforementioned objects, a feature extraction operation is performed on the actual data processing requirement to obtain a feature extraction result for that actual data processing requirement; the feature extraction result includes at least one feature among data type, data volume, data processing frequency, and data processing real-time requirements;

[0013] For each actual data processing requirement of the required object, an information analysis operation is performed on the required purpose to obtain at least one sub-purpose for the required purpose, and a priority sorting operation is performed on all the sub-purposes corresponding to the required purpose to obtain the priority sorting result corresponding to all the sub-purposes corresponding to the required purpose.

[0014] For each of the aforementioned demand objects, candidate data resources matching the feature extraction results and priority ranking results corresponding to the demand object are determined from the data resource library. A filtering operation and a data evaluation operation are then performed on the candidate data resources to obtain the target data resources corresponding to the demand object. The filtering operation is used to remove redundant data; the data evaluation operation is used to assess data quality and usability.

[0015] For each of the aforementioned demand objects, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to the preset computing power demand model to obtain the first computing power resources allocated to the demand object.

[0016] As an optional implementation, in the first aspect of the present invention, for each of the demand objects, performing a computing power analysis operation on the feature extraction results corresponding to the demand object and the target data resources according to a preset computing power demand model to obtain the first computing power resources allocated to the demand object includes:

[0017] Determine the computing power analysis factors for each of the aforementioned demand objects, wherein the computing power analysis factors include at least one of the following: task complexity, data volume, and data processing frequency;

[0018] For each of the aforementioned demand objects, based on the preset computing power demand model and combined with the computing power analysis factors corresponding to the demand object, the first analysis operation is performed on the feature extraction results and the target data resources corresponding to the demand object to obtain the initial computing power resources allocated to the demand object.

[0019] Obtain historical data processing information and computing power usage records for each of the aforementioned demand objects;

[0020] For each of the aforementioned demand objects, based on the historical data processing status and computing power usage records corresponding to the demand object, a second analysis operation is performed on the initial computing power resources corresponding to the demand object to obtain the first computing power resources allocated to the demand object; the second analysis operation includes a computing power resource evaluation operation and its corresponding computing power resource adjustment operation, and the second analysis operation is used to optimize the initial computing power resources corresponding to the demand object.

[0021] As an optional implementation, in the first aspect of the present invention, the step of performing a correction operation on the first computing power resources required by the demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, to obtain the second computing power resources required by the demand object, includes:

[0022] Obtain the existing data resources of the demand object, which include at least the cached data resources;

[0023] The resource information that is missing or overlapped with the target data resource corresponding to the demand object is determined respectively to obtain the first resource information and the second resource information, which are used as resource difference information;

[0024] Based on the resource difference information corresponding to the demand object, the correction type of the first computing power resource required for the demand object is determined, and the correction type includes a first correction type or a second correction type; the correction type is used to indicate whether to increase or decrease the first computing power resource required for the demand object.

[0025] Based on the resource difference information and correction type corresponding to the demand object, and in conjunction with the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resource required by the demand object to obtain the second computing power resource required by the demand object.

[0026] As an optional implementation, in the first aspect of the present invention, when the first resource information corresponding to the demand object indicates that the existing data resources of the demand object do not overlap with the target data resources corresponding to the demand object, the correction type is the first correction type.

[0027] When the first resource information indicates that the existing data resources of the demand object are missing data resources relative to the target data resources corresponding to the demand object, and the second resource information indicates that the existing data resources of the demand object have or do not have overlapping data resources relative to the target data resources corresponding to the demand object, the correction type is the second correction type.

[0028] As an optional implementation, in the first aspect of the present invention, the step of performing a correction operation on the first computing power resource required by the demand object based on the resource difference information corresponding to the demand object and the correction type, combined with the actual data processing demand corresponding to the demand object, to obtain the second computing power resource required by the demand object, includes:

[0029] When the correction type is the first correction type, based on the resource difference information corresponding to the demand object and the actual data processing demand corresponding to the demand object, the first computing power resource required by the demand object is subjected to a first computing power evaluation operation to obtain the first computing power resource required by the demand object.

[0030] When the first computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first upward adjustment operation is used to increase the allocation of the first computing power resources required by the demand object.

[0031] When the first computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first reduction operation is used to reduce the allocation of the first computing power resources required by the demand object.

[0032] As an optional implementation, in the first aspect of the present invention, the step of performing a correction operation on the first computing power resource required by the demand object based on the resource difference information corresponding to the demand object and the correction type, combined with the actual data processing demand corresponding to the demand object, to obtain the second computing power resource required by the demand object, includes:

[0033] When the correction type is the second correction type, according to the second resource information corresponding to the demand object, a data optimization operation is performed on the target data resource corresponding to the demand object, and according to the optimized target data resource corresponding to the demand object, resource optimization information for the first computing power resource required by the demand object is generated.

[0034] Based on the first resource information corresponding to the demand object, and combined with the actual data processing demand corresponding to the demand object, a second computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain a second computing power evaluation result for the first computing power resources required by the demand object.

[0035] When the second computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a second upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second upward adjustment operation is used to increase the allocation of the first computing power resources.

[0036] When the second computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, according to the second computing power assessment result and the resource optimization information, a second reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second reduction operation is used to reduce the allocation of the first computing power resources.

[0037] A second aspect of this invention discloses an AI-based multi-dimensional intelligent resource scheduling system, the system comprising:

[0038] The determination module is used to determine the target data resources and the first computing power resources required by each of the multiple demand objects with data processing needs, based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object.

[0039] The correction module is used to perform a correction operation on the first computing power resources required by each of the demand objects based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, so as to obtain the second computing power resources required by the demand object.

[0040] The access control module is used to grant each of the aforementioned request objects access rights to the target data resource and the second computing power resource, thereby triggering the request object to perform data processing operations corresponding to its actual data processing needs based on the target data resource and the second computing power resource.

[0041] As an optional implementation, in the second aspect of the present invention, the method by which the determining module determines the target data resources and the first computing power resources required by each of the demand objects based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object specifically includes:

[0042] For each of the actual data processing requirements of the aforementioned objects, a feature extraction operation is performed on the actual data processing requirement to obtain a feature extraction result for that actual data processing requirement; the feature extraction result includes at least one feature among data type, data volume, data processing frequency, and data processing real-time requirements;

[0043] For each actual data processing requirement of the required object, an information analysis operation is performed on the required purpose to obtain at least one sub-purpose for the required purpose, and a priority sorting operation is performed on all the sub-purposes corresponding to the required purpose to obtain the priority sorting result corresponding to all the sub-purposes corresponding to the required purpose.

[0044] For each of the aforementioned demand objects, candidate data resources matching the feature extraction results and priority ranking results corresponding to the demand object are determined from the data resource library. A filtering operation and a data evaluation operation are then performed on the candidate data resources to obtain the target data resources corresponding to the demand object. The filtering operation is used to remove redundant data; the data evaluation operation is used to assess data quality and usability.

[0045] For each of the aforementioned demand objects, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to the preset computing power demand model to obtain the first computing power resources allocated to the demand object.

[0046] As an optional implementation, in the second aspect of the present invention, the determining module performs a computing power analysis operation on each demand object according to a preset computing power demand model, based on the feature extraction results and the target data resources corresponding to the demand object, to obtain the first computing power resource allocated to the demand object. Specifically, this includes:

[0047] Determine the computing power analysis factors for each of the aforementioned demand objects, wherein the computing power analysis factors include at least one of the following: task complexity, data volume, and data processing frequency;

[0048] For each of the aforementioned demand objects, based on the preset computing power demand model and combined with the computing power analysis factors corresponding to the demand object, the first analysis operation is performed on the feature extraction results and the target data resources corresponding to the demand object to obtain the initial computing power resources allocated to the demand object.

[0049] Obtain historical data processing information and computing power usage records for each of the aforementioned demand objects;

[0050] For each of the aforementioned demand objects, based on the historical data processing status and computing power usage records corresponding to the demand object, a second analysis operation is performed on the initial computing power resources corresponding to the demand object to obtain the first computing power resources allocated to the demand object; the second analysis operation includes a computing power resource evaluation operation and its corresponding computing power resource adjustment operation, and the second analysis operation is used to optimize the initial computing power resources corresponding to the demand object.

[0051] As an optional implementation, in a second aspect of the present invention, the correction module performs a correction operation on the first computing power resources required by the demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, to obtain the second computing power resources required by the demand object. Specifically, this includes:

[0052] Obtain the existing data resources of the demand object, which include at least the cached data resources;

[0053] The resource information that is missing or overlapped with the target data resource corresponding to the demand object is determined respectively to obtain the first resource information and the second resource information, which are used as resource difference information;

[0054] Based on the resource difference information corresponding to the demand object, the correction type of the first computing power resource required for the demand object is determined, and the correction type includes a first correction type or a second correction type; the correction type is used to indicate whether to increase or decrease the first computing power resource required for the demand object.

[0055] Based on the resource difference information and correction type corresponding to the demand object, and in conjunction with the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resource required by the demand object to obtain the second computing power resource required by the demand object.

[0056] As an optional implementation, in a second aspect of the present invention, when the first resource information corresponding to the demand object indicates that the existing data resources of the demand object do not overlap with the target data resources corresponding to the demand object, the correction type is the first correction type.

[0057] When the first resource information indicates that the existing data resources of the demand object are missing data resources relative to the target data resources corresponding to the demand object, and the second resource information indicates that the existing data resources of the demand object have or do not have overlapping data resources relative to the target data resources corresponding to the demand object, the correction type is the second correction type.

[0058] As an optional implementation, in a second aspect of the present invention, the correction module performs a correction operation on the first computing power resource required by the demand object based on the resource difference information corresponding to the demand object and the correction type, combined with the actual data processing demand corresponding to the demand object, to obtain the second computing power resource required by the demand object. Specifically, this includes:

[0059] When the correction type is the first correction type, based on the resource difference information corresponding to the demand object and the actual data processing demand corresponding to the demand object, the first computing power resource required by the demand object is subjected to a first computing power evaluation operation to obtain the first computing power resource required by the demand object.

[0060] When the first computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first upward adjustment operation is used to increase the allocation of the first computing power resources required by the demand object.

[0061] When the first computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first reduction operation is used to reduce the allocation of the first computing power resources required by the demand object.

[0062] As an optional implementation, in a second aspect of the present invention, the correction module performs a correction operation on the first computing power resource required by the demand object based on the resource difference information corresponding to the demand object and the correction type, combined with the actual data processing demand corresponding to the demand object, to obtain the second computing power resource required by the demand object. Specifically, this includes:

[0063] When the correction type is the second correction type, according to the second resource information corresponding to the demand object, a data optimization operation is performed on the target data resource corresponding to the demand object, and according to the optimized target data resource corresponding to the demand object, resource optimization information for the first computing power resource required by the demand object is generated.

[0064] Based on the first resource information corresponding to the demand object, and combined with the actual data processing demand corresponding to the demand object, a second computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain a second computing power evaluation result for the first computing power resources required by the demand object.

[0065] When the second computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a second upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second upward adjustment operation is used to increase the allocation of the first computing power resources.

[0066] When the second computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, according to the second computing power assessment result and the resource optimization information, a second reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second reduction operation is used to reduce the allocation of the first computing power resources.

[0067] A third aspect of this invention discloses an AI-based multi-dimensional intelligent resource scheduling device, the device comprising:

[0068] Memory containing executable program code;

[0069] A processor coupled to the memory;

[0070] The processor calls the executable program code stored in the memory to execute some or all of the steps in the AI-based multi-dimensional intelligent resource scheduling method according to any of the first aspects of the present invention.

[0071] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps in the AI-based multi-dimensional intelligent resource scheduling method described in any of the first aspects of the present invention.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] This invention provides an AI-based multi-dimensional intelligent resource scheduling method, comprising: when there are multiple demand objects with data processing needs, determining the target data resources and first computing power resources required by each demand object based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object; for each demand object, performing a correction operation on the first computing power resources required by the demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing needs corresponding to the demand object, to obtain the second computing power resources required by the demand object; for each demand object, granting access permissions to the demand object and its corresponding target data resources and second computing power resources, so as to trigger the demand object to perform data processing operations corresponding to its corresponding actual data processing needs according to its corresponding target data resources and second computing power resources. As can be seen, by comprehensively considering the actual needs and purposes of the target users and accurately matching resources, the implementation of this invention improves the targeting and accuracy of matching target data resources and primary computing power resources for different target users, avoids resource waste, and improves the accuracy of resource allocation. Furthermore, by setting a dynamic correction mechanism, computing power resources can be flexibly adjusted according to the actual situation, avoiding task failure due to insufficient computing power resources or resource idleness due to excess computing power resources. In addition, by accurately opening permissions to ensure the scientific and accurate management of resource allocation, the smooth completion of data processing tasks is ensured, and the overall performance and response speed of the system are improved. Attached Figure Description

[0074] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0075] Figure 1 This is a flowchart illustrating an AI-based multi-dimensional intelligent resource scheduling method disclosed in an embodiment of the present invention.

[0076] Figure 2 This is a flowchart illustrating another AI-based multi-dimensional intelligent resource scheduling method disclosed in an embodiment of the present invention;

[0077] Figure 3 This is a schematic diagram of the structure of an AI-based multi-dimensional intelligent resource scheduling system disclosed in an embodiment of the present invention;

[0078] Figure 4 This is a schematic diagram of the structure of an AI-based multi-dimensional intelligent resource scheduling device disclosed in an embodiment of the present invention. Detailed Implementation

[0079] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0080] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0081] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0082] This invention discloses an AI-based multi-dimensional intelligent resource scheduling method and system. By comprehensively considering the actual needs and objectives of the target users, it accurately matches resources, improving the targeting and accuracy of matching target data resources and primary computing power resources for different target users, avoiding resource waste, and improving resource allocation accuracy. Furthermore, by setting a dynamic correction mechanism, it can flexibly adjust computing power resources according to the actual situation, avoiding task failure due to insufficient computing power resources or resource idleness due to excess computing power resources. In addition, by accurately opening permissions, it ensures the scientific and accurate management of resource allocation, ensuring the smooth completion of data processing tasks and improving the overall performance and response speed of the system.

[0083] To better understand the AI-based multi-dimensional intelligent resource scheduling method and system described in this invention, the applicable scenarios for the AI-based multi-dimensional intelligent resource scheduling method are first described. Specifically, these scenarios can be intelligent traffic resource management scenarios in the transportation industry for urban traffic flow management and real-time traffic condition monitoring; medical resource scheduling scenarios in the medical industry for intelligent analysis of patient data and intelligent scheduling of medical devices; and transaction data intelligent scheduling scenarios in the financial industry for risk assessment and credit decision resource scheduling. Taking the intelligent traffic resource management scenario as an example: In a city's traffic management center, real-time analysis and processing of traffic flow data from multiple areas are required. Specifically, the core management objectives are to achieve intelligent control of traffic lights, traffic congestion early warning, and traffic planning optimization. Based on this, the target of this intelligent traffic management scenario can be different areas within the city, such as commercial areas, residential areas, and industrial areas.

[0084] Taking a commercial area as an example, the actual data processing requirement of this commercial area could be the real-time acquisition and analysis of traffic flow and pedestrian flow data at major intersections in the area. The corresponding objective of this data processing requirement could be to determine real-time traffic conditions based on traffic flow and pedestrian flow data, and then adjust traffic light durations accordingly to alleviate congestion. Based on this actual data processing requirement and objective, the required target data resources could include real-time video stream data from cameras at various intersections in the area, traffic flow sensor data, etc. Simultaneously, the initial computing power resource can be determined after a preliminary estimate based on the data volume and analysis complexity of the target data resources. For example, the initial computing power resource could correspond to the computing power capable of processing 10 high-definition video streams in real time and performing simple analysis.

[0085] Based on the pre-determined target data resources and primary computing power resources corresponding to the commercial area, existing data resources of its data nodes can be further acquired for subsequent correction processing. These existing data resources for the commercial area can be recent records (e.g., partial historical traffic data from the last 3 days or 7 days). After acquiring these existing data resources, they are used as historical computing power allocation and usage requirements. Combined with the target data resources (such as the aforementioned real-time video stream and traffic flow sensor data) and actual data processing needs (real-time analysis and adjustment of traffic lights), a computing power matching analysis is performed. It is found that relying solely on the initially estimated primary computing power resources may not be sufficient to meet the real-time analysis requirements for the target data resources. Because real-time video stream processing may require high computational performance, the first computing power resource is adjusted by increasing the number of hardware devices such as graphics processing units (GPUs) that are activated, thus obtaining a second computing power resource (this second computing power resource includes at least Y GPUs activated in area X of the commercial area). Alternatively, edge computing can be introduced to migrate some traffic data processing tasks of the commercial area edge node to the cloud. For example, for some tasks that require a lot of computing resources but do not require real-time performance, such as long-term trend analysis of traffic data and model training, they can be submitted to cloud computing platforms such as Alibaba Cloud and Tencent Cloud for execution. This eliminates the need to purchase / configure a large number of hardware devices, freeing up the data processing computing power of the commercial area edge node for data processing tasks that do not require real-time performance, allowing the commercial area edge node to allocate more data processing computing power to other traffic data.

[0086] After obtaining the second set of computing resources and target data resources for the commercial area through calibration, access to the commercial area and its corresponding target data resources (real-time video streams and traffic flow sensor data) as well as the second set of computing resources (calibrated graphics processing units) is granted. The commercial area management module then performs real-time traffic data analysis based on these resources and computing power, and adjusts traffic light durations according to the analysis results.

[0087] It should be noted that the application scenarios mentioned above are only to illustrate the applicable scenarios for the AI-based multi-dimensional intelligent resource scheduling method. The specific data types / contents of the target data, the actual data processing requirements of the target data, the purpose of the actual data processing requirements, the actual target data resources, the first computing power resources, the corrected second computing power resources, the authorized use of the target data resources, and the control form of the second computing power resources can be adaptively adjusted according to the actual scenario. The scenario descriptions mentioned above do not limit this.

[0088] The above describes the application scenarios applicable to the AI-based multi-dimensional intelligent resource scheduling method. The following section provides a detailed description of the AI-based multi-dimensional intelligent resource scheduling method and system.

[0089] Example 1

[0090] Please see Figure 1 , Figure 1 This is a flowchart illustrating a multi-dimensional intelligent resource scheduling method based on AI disclosed in an embodiment of the present invention. Figure 1 The described AI-based multi-dimensional intelligent resource scheduling method can be applied to AI-based multi-dimensional intelligent resource scheduling systems or devices, and the embodiments of this invention are not limited thereto. Figure 1 As shown, this AI-based multi-dimensional intelligent resource scheduling method can include the following operations:

[0091] 101. When there are multiple objects with data processing needs, the target data resources and the first computing power resources required by each object are determined based on the actual data processing needs of each object and the purpose of the actual data processing needs of each object.

[0092] In this embodiment of the invention, when determining the target data resources and primary computing power resources required by each demand object, the actual data processing needs of each demand object and the corresponding purpose of these needs can be comprehensively considered. Unlike the one-sidedness of allocating resources based on only a single factor, this comprehensive method of determining target data resources and primary computing power resources can deeply clarify the specific needs of the demand object. For example, different demand objects may have the same data cleaning needs, but one may be for market trend analysis while the other is for customer profile construction. Based on these different purposes, the required data range, data quality, and data processing accuracy will differ. By comprehensively considering the actual data processing needs and purposes, the targeting and accuracy of matching target data resources and primary computing power resources to different demand objects can be improved, avoiding resource waste and simultaneously improving resource allocation accuracy and resource utilization efficiency.

[0093] 102. For each demand object, based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, perform a correction operation on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object.

[0094] In this embodiment of the invention, based on the determination of the first computing power resource in step 101, a further correction process for the first computing power resource is set up. In practical applications, the existing data resources of the requesting object will affect its data processing. For example, if the requesting object already has some relevant data, the workload of data acquisition and integration required to process the same request will be reduced, and correspondingly, the required computing power resources will also change. Through the dynamic correction mechanism set in step 102, the computing power resources can be flexibly adjusted according to the actual situation, ensuring that each requesting object has sufficient and suitable computing power resources to complete its data processing tasks, avoiding processing task failure due to insufficient computing power resources or resource idleness due to excess computing power resources.

[0095] 103. For each demand object, grant access to the demand object and its corresponding target data resources and second computing power resources, so as to trigger the demand object to perform data processing operations corresponding to its actual data processing needs based on its corresponding target data resources and second computing power resources.

[0096] In this embodiment of the invention, by setting up an access control mechanism, the user can obtain the required resources and perform data processing operations in a timely manner according to their own needs. This achieves reasonable control of the access control mechanism, improves the overall resource management system, and helps to enhance the practicality and control stability of the resource management system.

[0097] It is evident that implementation Figure 1 The described AI-based multi-dimensional intelligent resource scheduling method accurately matches resources by comprehensively considering the actual needs and objectives of the target users, improving the targeting and accuracy of matching target data resources and primary computing power resources for different target users, avoiding resource waste, and improving resource allocation precision. Furthermore, by setting a dynamic correction mechanism, it can flexibly adjust computing power resources according to actual conditions, avoiding task failures due to insufficient computing power resources or resource idleness due to excess computing power resources. In addition, by accurately opening permissions, it ensures the scientific and accurate management of resource allocation, ensuring the smooth completion of data processing tasks and improving the overall performance and response speed of the system.

[0098] In an optional embodiment, the method by which step 101 above determines the target data resources and the first computing power resources required by each demand object based on the actual data processing needs of each demand object and the corresponding demand purpose of the actual data processing needs of each demand object specifically includes:

[0099] For each actual data processing requirement of a demand object, a feature extraction operation is performed on the actual data processing requirement to obtain the feature extraction result for that actual data processing requirement; the feature extraction result includes at least one feature among data type, data volume, data processing frequency, and data processing real-time requirements;

[0100] For each actual data processing requirement of a requirement object, perform information analysis on the requirement purpose to obtain at least one sub-purpose for the requirement purpose, and perform priority sorting on all sub-purposes corresponding to the requirement purpose to obtain the priority sorting result corresponding to all sub-purposes corresponding to the requirement purpose.

[0101] For each demand object, select candidate data resources that match the feature extraction results and priority ranking results corresponding to the demand object from the data resource library, and perform filtering and data evaluation operations on the candidate data resources respectively to obtain the target data resources corresponding to the demand object; the filtering operation is used to remove redundant data; the data evaluation operation is used to evaluate data quality and usability.

[0102] For each demand object, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to the preset computing power demand model to obtain the first computing power resource allocated to the demand object.

[0103] In this optional embodiment, multi-dimensional feature extraction enables a comprehensive and detailed analysis of the data processing needs of the target audience. For example, in financial transaction data processing scenarios, common data types may involve transaction records, account information, etc.; the data volume may vary depending on the transaction scale; the data processing frequency may require real-time updates of transaction data; and the real-time requirements of data processing are high to ensure the timeliness and accuracy of transactions. By accurately extracting these features, a deep understanding of the data processing needs of each target audience can be gained, providing precise data basis for subsequent matching of primary computing power resources.

[0104] In this optional embodiment, by subdividing and prioritizing the purpose of the demand, the key user needs at different stages or under different circumstances can be clarified, thereby helping to improve the targeting of resource allocation.

[0105] In this optional embodiment, based on the feature extraction results obtained in advance and the priority ranking results, the combination of the two can quickly and accurately filter candidate resource data related to the needs from massive data resources, that is, it is beneficial to improve the efficiency and accuracy of data filtering.

[0106] In this optional embodiment, the filtering operation set above can filter out redundant data, reduce unnecessary data processing burden, and improve data processing efficiency; the data evaluation operation set above performs quality and usability evaluation on the remaining data to ensure that the data used is accurate, complete, and reliable, and can meet the needs of data processing.

[0107] In this optional embodiment, calculating and allocating computing resources based on the configured computing power demand model can improve the accuracy of determining the first computing power resource. Specifically, for data processing tasks with large data volumes, high processing frequencies, and strong real-time requirements, the trained computing power demand model can adaptively allocate more computing power resources; while for tasks with small data volumes and low processing frequencies, fewer computing power resources will be allocated. Through this precise computing power allocation method based on the computing power demand model, it can be ensured that each demand object has sufficient and suitable computing power resources to complete its data processing task, avoiding task failure due to insufficient computing power or resource waste due to excessive computing power, further improving the accuracy of determining and allocating computing power.

[0108] As can be seen, in this optional embodiment, the analytical accuracy of data processing requirements is improved by comprehensive and detailed feature extraction and scientific segmentation and prioritization of needs; the accuracy of determining target data resources is improved by efficiently screening target data resources and ensuring their quality and availability; and the accuracy of determining and allocating the first computing power resource is greatly improved by scientifically and accurately allocating the first computing power resource based on the computing power demand model.

[0109] In another optional embodiment, the method of performing computing power analysis operations on the feature extraction results and target data resources corresponding to each demand object according to the preset computing power demand model to obtain the first computing power resource allocated to the demand object specifically includes:

[0110] Identify the computing power analysis factors for each demand object. The computing power analysis factors include at least one of the following factors: task complexity, data volume, and data processing frequency.

[0111] For each demand object, based on the preset computing power demand model and combined with the computing power analysis factors corresponding to the demand object, the first analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object to obtain the initial computing power resources allocated to the demand object.

[0112] Retrieve historical data processing records and computing power usage records for each demand object;

[0113] For each demand object, based on the historical data processing and computing power usage records corresponding to the demand object, a second analysis operation is performed on the initial computing power resources corresponding to the demand object to obtain the first computing power resources allocated to the demand object; the second analysis operation includes a computing power resource evaluation operation and its corresponding computing power resource adjustment operation, and the second analysis operation is used to optimize the initial computing power resources corresponding to the demand object.

[0114] In this optional embodiment, by clarifying multiple factors for computing power analysis, the dimensions and granularity of the analysis of computing power requirements are broadened. For example, in image recognition tasks, if the image resolution is high and the features are complex, the task complexity is high, requiring more computing power for feature extraction and model training. Regarding data volume, processing large-scale gene sequencing data obviously requires more powerful computing power than processing small amounts of user behavior data. Regarding data processing frequency, real-time video stream processing obviously requires continuous and high-speed computing power to ensure smooth video playback and analysis, while the real-time requirements for processing periodic report generation data are relatively lower.

[0115] In this optional embodiment, the aforementioned computing power resource assessment operation can comprehensively evaluate the effectiveness of the initial computing power resources, determining whether they meet actual needs and their utilization efficiency. The aforementioned computing power resource adjustment operation can then optimize and adjust the initial computing power resources based on the assessment results. For example, if historical data shows that a certain user has excess initial computing power resources when handling similar tasks, the computing power allocated to that user can be appropriately reduced; conversely, if insufficient initial computing power leads to low task processing efficiency, the computing power allocation can be increased. In this way, computing power allocation can be more closely aligned with actual needs, improving the rationality and efficiency of resource utilization.

[0116] As can be seen, in this optional embodiment, by comprehensively considering multi-dimensional computing power analysis factors such as task complexity, data volume, and data processing frequency, and combining them with a preset computing power demand model, the initial computing power resources can be determined quickly and accurately. On this basis, by using historical data processing and computing power usage records, the initial computing power resources are evaluated and adjusted and optimized, realizing secondary, refined allocation and dynamic optimization of computing power resources, further improving the scientificity and accuracy of the final determination and allocation of the first computing power resources.

[0117] Example 2

[0118] Please see Figure 2 , Figure 2 This is a flowchart illustrating another AI-based multi-dimensional intelligent resource scheduling method disclosed in an embodiment of the present invention. Figure 2The described AI-based multi-dimensional intelligent resource scheduling method can be applied to AI-based multi-dimensional intelligent resource scheduling systems or devices, and the embodiments of this invention are not limited thereto. Figure 2 As shown, this AI-based multi-dimensional intelligent resource scheduling method can include the following operations:

[0119] 201. When there are multiple objects with data processing needs, the target data resources and the first computing power resources required by each object are determined based on the actual data processing needs of each object and the purpose of the actual data processing needs of each object.

[0120] 202. For each demand object, obtain the existing data resources of the demand object, which shall include at least the cached data resources.

[0121] In this embodiment of the invention, in a video stream processing scenario, the cached data resource may be a portion of video segments already cached on the user's device.

[0122] 203. Determine the resource information that is missing or overlapped with the target data resource corresponding to the demand object based on the existing data resource of the demand object, and obtain the first resource information and the second resource information as resource difference information.

[0123] In this embodiment of the invention, taking scientific research data analysis as an example, the target data resource may be a comprehensive experimental dataset in a certain field, while the existing data resource may be some relevant data previously collected by researchers. Through this comparison, it is clear which data is missing and which is duplicated. Furthermore, this precise resource difference information can accurately reflect the changes in computing power requirements of the target audience during data processing. If there is a large amount of missing key data, it may mean that more computing power is needed for data acquisition, integration, and preliminary processing; while if there is a large amount of overlapping data, it may reduce the computing power consumption of some redundant calculations.

[0124] 204. Based on the resource difference information corresponding to the demand object, determine the correction type of the first computing power resource required for the demand object.

[0125] In this embodiment of the invention, the correction type includes a first correction type or a second correction type; the correction type is used to indicate whether the first computing power resource required for the demand object is increased or decreased.

[0126] In this embodiment of the invention, when the first resource information corresponding to the demand object indicates that the existing data resources of the demand object do not overlap with the target data resources corresponding to the demand object, the correction type is the first correction type.

[0127] In this embodiment of the invention, when the first resource information indicates that the existing data resources of the demand object are missing data resources relative to the target data resources corresponding to the demand object, and the second resource information indicates that the existing data resources of the demand object have or do not have overlapping data resources relative to the target data resources corresponding to the demand object, the correction type is the second correction type.

[0128] 205. Based on the resource difference information and correction type corresponding to the demand object, and combined with the actual data processing requirements corresponding to the demand object, perform a correction operation on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object.

[0129] In this embodiment of the invention, by comprehensively considering resource difference information, correction type and actual data processing needs corresponding to the demand object, computing resources can be precisely adjusted so that the second computing resources can not only meet the actual needs of data processing, but also ensure the efficient completion of processing tasks.

[0130] 206. For each demand object, grant access to the demand object and its corresponding target data resources and second computing power resources, so as to trigger the demand object to perform data processing operations corresponding to its actual data processing needs based on its corresponding target data resources and second computing power resources.

[0131] For further descriptions of steps 201 and 206 in this embodiment of the invention, please refer to the other specific descriptions of steps 101 and 103 in Embodiment 1. These descriptions will not be repeated in this embodiment of the invention.

[0132] It is evident that implementation Figure 2 The described AI-based multi-dimensional intelligent resource scheduling method provides an accurate basis for computing power correction by comprehensively acquiring existing data resources of the demand object and accurately determining resource difference information. On this basis, the correction type is scientifically determined according to resource differences, clarifying the direction of computing power adjustment. Finally, by integrating resource difference information, correction type and actual data processing needs, the method achieves accurate correction of the first computing power resource, improving the scientificity and accuracy of the correction of the first computing power resource, and at the same time improving the accuracy and scientificity of the determination of the second computing power resource.

[0133] In an optional embodiment, step 205 above, based on the resource difference information and correction type corresponding to the demand object, and in conjunction with the actual data processing requirements corresponding to the demand object, performs a correction operation on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object. Specifically, this includes:

[0134] When the correction type is the first correction type, based on the resource difference information corresponding to the demand object and the actual data processing demand corresponding to the demand object, the first computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain the first computing power evaluation result of the first computing power resources required by the demand object.

[0135] When the first computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing needs corresponding to the demand object, a first upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first upward adjustment operation is used to increase the allocation of the first computing power resources required by the demand object.

[0136] When the first computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first reduction operation is used to reduce the allocation of the first computing power resources required by the demand object.

[0137] In this optional embodiment, the method of performing a first increase operation on the computing power resources required by the first computing power resource needed by the demand object to obtain the second computing power resource required by the demand object specifically includes:

[0138] Determine the current resource allocation status of the demand object, which at least indicates the remaining unused hardware computing power resources and their quantity in the demand object;

[0139] Calculate the difference between the first computing power resource and the first computing power requirement for computing power processing;

[0140] Determine whether the hardware computing power resource is higher than the first computing power difference. If it is determined that the hardware computing power resource is higher than or equal to the first computing power difference, determine the target hardware computing power resource to be started and the corresponding number of target hardware from the hardware computing power resource according to the first computing power difference, and use it as the second computing power resource required by the demand object.

[0141] Optionally, when it is determined that the hardware computing power resources are lower than the first computing power difference, a software resource scheduling strategy is generated for the demand object based on the current resource allocation, hardware computing power resources, and the first computing power difference. The existing data resources, target data resources, and first computing power resources of the demand object are dynamically adjusted according to the software resource scheduling strategy to update the existing data resources, target data resources, and first computing power resources of the demand object. At the same time, the updated first computing power resources are determined as the second computing power resources required by the demand object.

[0142] As can be seen, in this optional embodiment, a processing flow is set up to perform a first increase operation on the first computing power resource. By intelligently analyzing the current resource allocation of the demand object and comparing it with the calculated first computing power difference, and determining the corresponding target hardware computing power resources and the corresponding number of target hardware according to different judgment results, or by generating and executing software resource scheduling strategies, the execution flexibility and accuracy of the first increase operation are improved.

[0143] In this optional embodiment, by comprehensively considering resource differences and actual needs, the first computing power assessment operation can comprehensively and accurately determine whether the current first computing power resources meet the actual processing requirements, providing a data basis for subsequent computing power adjustments.

[0144] In this optional embodiment, a corresponding processing mechanism is set for different first computing power evaluation results, corresponding to the first upward or downward adjustment operation of the first computing power resources, thereby improving the adjustment flexibility of the first computing power resources.

[0145] As can be seen, in this optional embodiment, by performing a first computing power assessment operation based on a comprehensive consideration of resource differences and actual data processing needs, an accurate judgment of the degree of matching between the first computing power resources and actual needs is achieved. On this basis, a first upward or downward adjustment operation is flexibly performed according to different first computing power assessment results. This ensures the smooth completion of data processing tasks while avoiding the waste of computing power resources. In other words, it helps to improve the accuracy and flexibility of adjusting the first computing power resources, while also improving the reliability and accuracy of the finally determined second computing power resources.

[0146] In another optional embodiment, step 205 above, based on the resource difference information and correction type corresponding to the demand object, and in conjunction with the actual data processing requirements corresponding to the demand object, performs a correction operation on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object. Specifically, this includes:

[0147] When the correction type is the second correction type, according to the second resource information corresponding to the demand object, data optimization operation is performed on the target data resource corresponding to the demand object, and according to the optimized target data resource corresponding to the demand object, resource optimization information for the first computing power resource required for the demand object is generated.

[0148] Based on the first resource information corresponding to the demand object, and combined with the actual data processing needs corresponding to the demand object, a second computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain the second computing power evaluation result of the first computing power resources required by the demand object.

[0149] When the second computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing needs corresponding to the demand object, a second upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second upward adjustment operation is used to increase the allocation of the first computing power resources.

[0150] When the second computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a second reduction operation of computing power resources is performed on the first computing power resources required by the demand object based on the second computing power assessment result and resource optimization information, so as to obtain the second computing power resources required by the demand object; the second reduction operation is used to reduce the allocation of the first computing power resources.

[0151] In this optional embodiment, in a big data analytics scenario, existing data resources may contain some pre-processed or filtered data that overlaps with the target data resources. By optimizing the overlapping data, such as removing duplicate data, standardizing the data, and extracting key features, redundant computations during data processing can be reduced. Taking image data processing as an example, if existing cached image data has a large amount of overlap with the target image data, optimizing these duplicate images by removing unnecessary pixel information or compressing them can reduce the computational demands on subsequent image recognition and analysis tasks.

[0152] In this optional embodiment, the resource optimization information for the first computing power resource may include key indicators such as the degree of simplification of data processing steps and the percentage reduction in data volume. Based on this resource optimization information, a precise reference can be provided for subsequent computing power adjustments. For example, if the data optimization operation reduces the data volume of the target data resource by 30%, then when assessing computing power requirements later, the allocation of computing power resources can be appropriately reduced based on this information, thereby improving the utilization efficiency of computing power resources.

[0153] In this optional embodiment, similar to the above-described scheme for processing the first correction type, an evaluation process for the matching degree between the first computing power resources and the actual data processing needs is also set up, and the second computing power evaluation results are clearly divided into two different cases, thereby achieving accurate correction for the second correction type.

[0154] In this optional embodiment, unlike the first correction type, when the second computing power assessment result indicates that the first computing power resource is higher than or equal to the actual data processing demand, in addition to considering the second assessment result, a second reduction operation can also be performed in conjunction with resource optimization information. For example, if the data optimization operation reduces the computing power demand of the data processing task by 20%, and the second computing power assessment result shows that the first computing power resource is available, then the second reduction operation directly performs the reduction allocation of computing power according to the resource optimization information, omitting the step of calculating the reduction amount, which to some extent helps to improve the execution efficiency and accuracy of the second reduction operation.

[0155] As can be seen, in this optional embodiment, a correction mechanism is set for the target data resources and the first computing power resources under the second correction type. Specifically: resource optimization information is generated by data optimization operations based on the second resource information, thereby realizing the correction and optimization of the target data resources and improving the accuracy of the determination of the target data resources; it can also accurately determine the matching degree between the first computing power resources and the actual needs by performing the second computing power evaluation operation; and flexibly execute the second upward adjustment or the second downward adjustment operation combined with the resource optimization information according to the evaluation results, thereby realizing the dynamic optimization configuration of the first computing power resources under the second correction type, improving the flexibility and targeting of the adjustment of the first computing power resources under different correction types, and at the same time, improving the reliability and accuracy of the finally determined second computing power resources.

[0156] Example 3

[0157] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an AI-based multi-dimensional intelligent resource scheduling system disclosed in an embodiment of the present invention. This AI-based multi-dimensional intelligent resource scheduling system can be a medical resource intelligent scheduling system applied to medical resource scheduling scenarios, a transaction data intelligent scheduling system applied to financial scenarios for risk assessment and credit decision scheduling, or a traffic resource scheduling system applied to urban traffic management scenarios for real-time traffic flow analysis and traffic control. The specific application scenario of this AI-based multi-dimensional intelligent resource scheduling system can be adjusted according to actual needs, and the embodiments of the present invention do not limit it. Figure 3 As shown, the AI-based multi-dimensional intelligent resource scheduling device may include a determination module 301, a correction module 302, and an access control module 303, wherein:

[0158] The determination module 301 is used to determine the target data resources and the first computing power resources required by each demand object when there are multiple demand objects with data processing needs, based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object.

[0159] The correction module 302 is used to perform a correction operation on the first computing power resources required by each demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, so as to obtain the second computing power resources required by the demand object.

[0160] The access control module 303 is used to grant each request object access permissions to its corresponding target data resources and secondary computing power resources, so as to trigger the request object to perform data processing operations corresponding to its actual data processing needs based on its corresponding target data resources and secondary computing power resources.

[0161] It is evident that implementation Figure 3 The described AI-based multi-dimensional intelligent resource scheduling system accurately matches resources by comprehensively considering the actual needs and objectives of the target users, improving the targeting and accuracy of matching target data resources and primary computing power resources for different target users, avoiding resource waste, and improving resource allocation precision. Furthermore, by setting a dynamic correction mechanism, it can flexibly adjust computing power resources according to actual conditions, avoiding task failures due to insufficient computing power resources or resource idleness due to excess computing power resources. In addition, by accurately opening permissions, it ensures the scientific and accurate management of resource allocation, ensuring the smooth completion of data processing tasks and improving the overall performance and response speed of the system.

[0162] In an optional embodiment, the method by which the determining module 301 determines the target data resources and the first computing power resources required by each demand object based on the actual data processing needs of each demand object and the corresponding demand purpose of the actual data processing needs of each demand object specifically includes:

[0163] For each actual data processing requirement of a demand object, a feature extraction operation is performed on the actual data processing requirement to obtain the feature extraction result for that actual data processing requirement; the feature extraction result includes at least one feature among data type, data volume, data processing frequency, and data processing real-time requirements;

[0164] For each actual data processing requirement of a requirement object, perform information analysis on the requirement purpose to obtain at least one sub-purpose for the requirement purpose, and perform priority sorting on all sub-purposes corresponding to the requirement purpose to obtain the priority sorting result corresponding to all sub-purposes corresponding to the requirement purpose.

[0165] For each demand object, select candidate data resources that match the feature extraction results and priority ranking results corresponding to the demand object from the data resource library, and perform filtering and data evaluation operations on the candidate data resources respectively to obtain the target data resources corresponding to the demand object; the filtering operation is used to remove redundant data; the data evaluation operation is used to evaluate data quality and usability.

[0166] For each demand object, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to the preset computing power demand model to obtain the first computing power resource allocated to the demand object.

[0167] As can be seen, in this optional embodiment, the analytical accuracy of data processing requirements is improved by comprehensive and detailed feature extraction and scientific segmentation and prioritization of needs; the accuracy of determining target data resources is improved by efficiently screening target data resources and ensuring their quality and availability; and the accuracy of determining and allocating the first computing power resource is greatly improved by scientifically and accurately allocating the first computing power resource based on the computing power demand model.

[0168] In another optional embodiment, the determining module 301 performs a computing power analysis operation on each demand object based on the feature extraction results and target data resources corresponding to the demand object according to a preset computing power demand model, and obtains the first computing power resource allocated to the demand object in the following specific ways:

[0169] Identify the computing power analysis factors for each demand object. The computing power analysis factors include at least one of the following factors: task complexity, data volume, and data processing frequency.

[0170] For each demand object, based on the preset computing power demand model and combined with the computing power analysis factors corresponding to the demand object, the first analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object to obtain the initial computing power resources allocated to the demand object.

[0171] Retrieve historical data processing records and computing power usage records for each demand object;

[0172] For each demand object, based on the historical data processing and computing power usage records corresponding to the demand object, a second analysis operation is performed on the initial computing power resources corresponding to the demand object to obtain the first computing power resources allocated to the demand object; the second analysis operation includes a computing power resource evaluation operation and its corresponding computing power resource adjustment operation, and the second analysis operation is used to optimize the initial computing power resources corresponding to the demand object.

[0173] As can be seen, in this optional embodiment, by comprehensively considering multi-dimensional computing power analysis factors such as task complexity, data volume, and data processing frequency, and combining them with a preset computing power demand model, the initial computing power resources can be determined quickly and accurately. On this basis, by using historical data processing and computing power usage records, the initial computing power resources are evaluated and adjusted and optimized, realizing secondary, refined allocation and dynamic optimization of computing power resources, further improving the scientificity and accuracy of the final determination and allocation of the first computing power resources.

[0174] In another optional embodiment, the correction module 302 performs a correction operation on the first computing power resources required by the demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, to obtain the second computing power resources required by the demand object. Specifically, this includes:

[0175] Obtain the existing data resources of the demand object, which include at least the cached data resources;

[0176] The resource information that is missing or overlapped with the target data resource corresponding to the demand object is determined respectively to obtain the first resource information and the second resource information, which are used as resource difference information.

[0177] Based on the resource difference information corresponding to the demand object, determine the correction type of the first computing power resource required for the demand object. The correction type includes a first correction type or a second correction type. The correction type is used to indicate whether to increase or decrease the first computing power resource required for the demand object.

[0178] Based on the resource difference information and correction type corresponding to the demand object, and combined with the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resource required by the demand object to obtain the second computing power resource required by the demand object.

[0179] In this optional embodiment, when the first resource information corresponding to the demand object indicates that the existing data resources of the demand object do not overlap with the target data resources corresponding to the demand object, the correction type is the first correction type.

[0180] In this optional embodiment, when the first resource information indicates that the existing data resources of the demand object are missing data resources relative to the target data resources corresponding to the demand object, and the second resource information indicates that the existing data resources of the demand object have or do not have overlapping data resources relative to the target data resources corresponding to the demand object, the correction type is the second correction type.

[0181] As can be seen, in this optional embodiment, by comprehensively acquiring the existing data resources of the target object and accurately determining the resource difference information, an accurate basis is provided for computing power correction; on this basis, the correction type is scientifically determined according to the resource difference, and the direction of computing power adjustment is clarified; finally, by comprehensively considering the resource difference information, the correction type, and the actual data processing requirements, accurate correction of the first computing power resource is achieved, improving the scientificity and accuracy of the correction of the first computing power resource, while also improving the accuracy and scientificity of the determination of the second computing power resource.

[0182] In another optional embodiment, the correction module 302 performs a correction operation on the first computing power resource required by the demand object based on the resource difference information and correction type corresponding to the demand object, combined with the actual data processing requirements corresponding to the demand object, to obtain the second computing power resource required by the demand object. Specifically, this includes:

[0183] When the correction type is the first correction type, based on the resource difference information corresponding to the demand object and the actual data processing demand corresponding to the demand object, the first computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain the first computing power evaluation result of the first computing power resources required by the demand object.

[0184] When the first computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing needs corresponding to the demand object, a first upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first upward adjustment operation is used to increase the allocation of the first computing power resources required by the demand object.

[0185] When the first computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first reduction operation is used to reduce the allocation of the first computing power resources required by the demand object.

[0186] As can be seen, in this optional embodiment, by performing a first computing power assessment operation based on a comprehensive consideration of resource differences and actual data processing needs, an accurate judgment of the degree of matching between the first computing power resources and actual needs is achieved. On this basis, a first upward or downward adjustment operation is flexibly performed according to different first computing power assessment results. This ensures the smooth completion of data processing tasks while avoiding the waste of computing power resources. In other words, it helps to improve the accuracy and flexibility of adjusting the first computing power resources, while also improving the reliability and accuracy of the finally determined second computing power resources.

[0187] In another optional embodiment, the correction module 302 performs a correction operation on the first computing power resource required by the demand object based on the resource difference information and correction type corresponding to the demand object, combined with the actual data processing requirements corresponding to the demand object, to obtain the second computing power resource required by the demand object. Specifically, this includes:

[0188] When the correction type is the second correction type, according to the second resource information corresponding to the demand object, data optimization operation is performed on the target data resource corresponding to the demand object, and according to the optimized target data resource corresponding to the demand object, resource optimization information for the first computing power resource required for the demand object is generated.

[0189] Based on the first resource information corresponding to the demand object, and combined with the actual data processing needs corresponding to the demand object, a second computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain the second computing power evaluation result of the first computing power resources required by the demand object.

[0190] When the second computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing needs corresponding to the demand object, a second upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second upward adjustment operation is used to increase the allocation of the first computing power resources.

[0191] When the second computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a second reduction operation of computing power resources is performed on the first computing power resources required by the demand object based on the second computing power assessment result and resource optimization information, so as to obtain the second computing power resources required by the demand object; the second reduction operation is used to reduce the allocation of the first computing power resources.

[0192] As can be seen, in this optional embodiment, a correction mechanism is set for the target data resources and the first computing power resources under the second correction type. Specifically: resource optimization information is generated by data optimization operations based on the second resource information, thereby realizing the correction and optimization of the target data resources and improving the accuracy of the determination of the target data resources; it can also accurately determine the matching degree between the first computing power resources and the actual needs by performing the second computing power evaluation operation; and flexibly execute the second upward adjustment or the second downward adjustment operation combined with the resource optimization information according to the evaluation results, thereby realizing the dynamic optimization configuration of the first computing power resources under the second correction type, improving the flexibility and targeting of the adjustment of the first computing power resources under different correction types, and at the same time, improving the reliability and accuracy of the finally determined second computing power resources.

[0193] Example 4

[0194] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a multi-dimensional intelligent resource scheduling device based on AI, as disclosed in an embodiment of the present invention. Figure 4 As shown, this AI-based multi-dimensional intelligent resource scheduling device may include:

[0195] Memory 401 storing executable program code;

[0196] Processor 402 coupled to memory 401;

[0197] The processor 402 calls the executable program code stored in the memory 401 to execute some or all of the steps in any of the AI-based multi-dimensional intelligent resource scheduling described in Embodiment 1 or Embodiment 2 of the present invention.

[0198] Example 5

[0199] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in any of the AI-based multi-dimensional intelligent resource scheduling methods described in Embodiment 1 or Embodiment 2 of this invention.

[0200] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0201] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0202] Finally, it should be noted that the above embodiments are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-dimensional intelligent resource scheduling method based on AI, characterized in that, The method includes: When there are multiple objects with data processing needs, the target data resources and the first computing power resources required by each object are determined based on the actual data processing needs of each object and the purpose corresponding to the actual data processing needs of each object. For each of the aforementioned demand objects, based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object. For each of the aforementioned demand objects, grant access to the demand object and its corresponding target data resources and the second computing power resources, thereby triggering the demand object to perform data processing operations corresponding to its actual data processing needs based on its corresponding target data resources and the second computing power resources. The determination of the target data resources and first computing power resources required by each of the demand objects, based on the actual data processing needs of each demand object and the corresponding demand objectives, includes: For each of the actual data processing requirements of the aforementioned objects, a feature extraction operation is performed on the actual data processing requirement to obtain a feature extraction result for that actual data processing requirement; the feature extraction result includes at least one feature among data type, data volume, data processing frequency, and data processing real-time requirements; For each actual data processing requirement of the required object, an information analysis operation is performed on the required purpose to obtain at least one sub-purpose for the required purpose, and a priority sorting operation is performed on all the sub-purposes corresponding to the required purpose to obtain the priority sorting result corresponding to all the sub-purposes corresponding to the required purpose. For each of the aforementioned demand objects, candidate data resources matching the feature extraction results and priority ranking results corresponding to the demand object are determined from the data resource library. A filtering operation and a data evaluation operation are then performed on the candidate data resources to obtain the target data resources corresponding to the demand object. The filtering operation is used to remove redundant data; the data evaluation operation is used to assess data quality and usability. For each of the aforementioned demand objects, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to the preset computing power demand model to obtain the first computing power resources allocated to the demand object.

2. The AI-based multi-dimensional intelligent resource scheduling method according to claim 1, characterized in that, For each of the aforementioned demand objects, a computing power analysis operation is performed on the feature extraction results and target data resources corresponding to the demand object according to a preset computing power demand model to obtain the first computing power resource allocated to the demand object, including: Determine the computing power analysis factors for each of the aforementioned demand objects, wherein the computing power analysis factors include at least one of the following: task complexity, data volume, and data processing frequency; For each of the aforementioned demand objects, based on the preset computing power demand model and combined with the computing power analysis factors corresponding to the demand object, the first analysis operation is performed on the feature extraction results and the target data resources corresponding to the demand object to obtain the initial computing power resources allocated to the demand object. Obtain historical data processing information and computing power usage records for each of the aforementioned demand objects; For each of the aforementioned demand objects, based on the historical data processing status and computing power usage records corresponding to the demand object, a second analysis operation is performed on the initial computing power resources corresponding to the demand object to obtain the first computing power resources allocated to the demand object; the second analysis operation includes a computing power resource evaluation operation and its corresponding computing power resource adjustment operation, and the second analysis operation is used to optimize the initial computing power resources corresponding to the demand object.

3. The AI-based multi-dimensional intelligent resource scheduling method according to claim 1 or 2, characterized in that, The process of performing a correction operation on the first computing power resource required by the demand object based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, to obtain the second computing power resource required by the demand object, includes: Obtain the existing data resources of the demand object, which include at least the cached data resources; The resource information that is missing or overlapped with the target data resource corresponding to the demand object is determined respectively to obtain the first resource information and the second resource information, which are used as resource difference information; Based on the resource difference information corresponding to the demand object, the correction type of the first computing power resource required for the demand object is determined, and the correction type includes a first correction type or a second correction type; the correction type is used to indicate whether to increase or decrease the first computing power resource required for the demand object. Based on the resource difference information and correction type corresponding to the demand object, and in conjunction with the actual data processing requirements corresponding to the demand object, a correction operation is performed on the first computing power resource required by the demand object to obtain the second computing power resource required by the demand object.

4. The AI-based multi-dimensional intelligent resource scheduling method according to claim 3, characterized in that, When the first resource information corresponding to the demand object indicates that the existing data resources of the demand object do not overlap with the target data resources corresponding to the demand object, the correction type is the first correction type. When the first resource information indicates that the existing data resources of the demand object are missing data resources relative to the target data resources corresponding to the demand object, and the second resource information indicates that the existing data resources of the demand object have or do not have overlapping data resources relative to the target data resources corresponding to the demand object, the correction type is the second correction type.

5. The AI-based multi-dimensional intelligent resource scheduling method according to claim 3, characterized in that, The step of performing a correction operation on the first computing power resource required by the demand object based on the resource difference information and the correction type corresponding to the demand object, combined with the actual data processing requirements corresponding to the demand object, to obtain the second computing power resource required by the demand object includes: When the correction type is the first correction type, based on the resource difference information corresponding to the demand object and the actual data processing demand corresponding to the demand object, the first computing power resource required by the demand object is subjected to a first computing power evaluation operation to obtain the first computing power resource required by the demand object. When the first computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first upward adjustment operation is used to increase the allocation of the first computing power resources required by the demand object. When the first computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, a first reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the first reduction operation is used to reduce the allocation of the first computing power resources required by the demand object.

6. The AI-based multi-dimensional intelligent resource scheduling method according to claim 3, characterized in that, The step of performing a correction operation on the first computing power resource required by the demand object based on the resource difference information and the correction type corresponding to the demand object, combined with the actual data processing requirements corresponding to the demand object, to obtain the second computing power resource required by the demand object includes: When the correction type is the second correction type, according to the second resource information corresponding to the demand object, a data optimization operation is performed on the target data resource corresponding to the demand object, and according to the optimized target data resource corresponding to the demand object, resource optimization information for the first computing power resource required by the demand object is generated. Based on the first resource information corresponding to the demand object, and combined with the actual data processing demand corresponding to the demand object, a second computing power evaluation operation is performed on the first computing power resources required by the demand object to obtain a second computing power evaluation result for the first computing power resources required by the demand object. When the second computing power assessment result indicates that the first computing power resources required by the demand object are lower than the computing power processing requirements of the actual data processing demand corresponding to the demand object, a second upward adjustment operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second upward adjustment operation is used to increase the allocation of the first computing power resources. When the second computing power assessment result indicates that the first computing power resources required by the demand object are higher than or equal to the computing power processing requirements of the actual data processing demand corresponding to the demand object, according to the second computing power assessment result and the resource optimization information, a second reduction operation of computing power resources is performed on the first computing power resources required by the demand object to obtain the second computing power resources required by the demand object; the second reduction operation is used to reduce the allocation of the first computing power resources.

7. An AI-based multi-dimensional intelligent resource scheduling system, characterized in that, The system is used to execute the AI-based multi-dimensional intelligent resource scheduling method as described in any one of claims 1-6, and the system comprises: The determination module is used to determine the target data resources and the first computing power resources required by each of the multiple demand objects with data processing needs, based on the actual data processing needs of each demand object and the demand purpose corresponding to the actual data processing needs of each demand object. The correction module is used to perform a correction operation on the first computing power resources required by each of the demand objects based on the existing data resources of the demand object, the target data resources corresponding to the demand object, and the actual data processing requirements corresponding to the demand object, so as to obtain the second computing power resources required by the demand object. The access control module is used to grant each of the aforementioned request objects access rights to the target data resource and the second computing power resource, thereby triggering the request object to perform data processing operations corresponding to its actual data processing needs based on the target data resource and the second computing power resource.

8. An AI-based multi-dimensional intelligent resource scheduling device, characterized in that, The device includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the AI-based multi-dimensional intelligent resource scheduling method as described in any one of claims 1-6.

9. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the AI-based multi-dimensional intelligent resource scheduling method as described in any one of claims 1-6.

Images