Multi-source computing power collaborative optimization method and system based on artificial intelligence

Abstract

According to the multi-source computing power collaborative optimization method and system based on artificial intelligence, multi-source sensing and joint modeling are carried out on the energy supply state, the computing power equipment operation state and the computing power task access state, intelligent scheduling and dynamic optimization control of computing power resources are achieved on the premise that multiple operation constraint conditions are met, and the multi-source computing power collaborative optimization method and system based on artificial intelligence are achieved. Therefore, the comprehensive operation efficiency and the business income level of the computing power system are improved.

Description

Technical Field

[0001] This invention relates to the field of computing power optimization technology, and in particular to a multi-source computing power collaborative optimization method and system based on artificial intelligence, specifically applied to a scenario where computing power systems and computing power task access sides operate collaboratively under energy constraints. Background Technology

[0002] As computing power systems continue to expand in scale, the energy consumption and operating costs they generate during operation are becoming increasingly prominent. Existing computing power management solutions typically handle energy management, computing power scheduling, and revenue management separately, lacking cross-layer collaborative optimization mechanisms, resulting in low overall system energy efficiency and insufficient resource utilization.

[0003] Especially in computing centers that utilize gas-fired power generation or distributed energy supply, fluctuations in energy supply, changes in computing load, and uncertainties in the revenue of computing task access are interdependent, making it difficult for existing methods to achieve globally optimal control. Therefore, it is necessary to propose a computing power optimization technology that can perform unified modeling and intelligent scheduling of the operation of the computing system and the computing task access side under energy constraints. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-source computing power collaborative optimization method and system based on artificial intelligence, so as to solve the problems of low energy efficiency, extensive scheduling and unstable returns in existing computing power systems.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: 1. A multi-source computing power collaborative optimization method based on artificial intelligence, characterized by the following steps: S1. Collecting multi-source operating data of the computing power system according to the energy layer monitoring module, computing power layer monitoring module, computing power access layer monitoring module, and network status monitoring module to obtain the original operating dataset; S2. Performing time alignment, anomaly detection, and preprocessing on the multi-source operating data to generate standardized operating data; S3. Extracting feature parameters for describing the system operating state based on the standardized operating data to construct a system state feature set; S4. Based on the system state feature set, performing coarse-grained optimization of the system operating state in future time periods. S5. Based on the rough estimation results, perform a fine estimation of the computing power energy consumption status, computing power load status, and computing power task access side revenue status to obtain a fine estimation result; S6. Perform a feasibility verification of the fine estimation results according to energy supply constraints, computing power equipment security constraints, and system operation reliability constraints; S7. If the constraints are met, generate corresponding computing power scheduling strategies and operating parameter adjustment instructions; S8. Execute the control of computing power equipment and its computing power resources according to the computing power scheduling strategy, and feed the execution results back to the artificial intelligence decision-making module to achieve dynamic optimization and adjustment.

[0006] 2. The method according to claim 1, wherein the energy layer operation data in the multi-source operation data includes energy supply power, gas or electricity parameters, power generation efficiency, and energy cost information.

[0007] 3. The method according to claim 1, wherein the artificial intelligence decision-making process is implemented using at least one machine learning model, reinforcement learning model, or intelligent optimization model based on a combination of rules and models for multi-source operation state prediction and scheduling decision-making.

[0008] 4. The method according to claim 1, wherein the computing power scheduling strategy is generated through a hierarchical decision-making process, including computing power access layer decision, computing power layer decision and energy layer decision.

[0009] 5. A multi-source computing power collaborative optimization system based on artificial intelligence, characterized in that it comprises: an energy layer monitoring module for collecting operational status data from the energy supply side; a computing power layer monitoring module for collecting operational status data from computing power devices; a computing power access layer monitoring module for collecting operational status data from the computing power task access side; an artificial intelligence decision-making module for performing state fusion, predictive analysis, and scheduling decisions on the multi-source operational data; and an execution control module for controlling the access status of computing power devices and computing power tasks according to the scheduling decision results.

[0010] 6. The system according to claim 5, wherein the artificial intelligence decision-making module includes functional units for performing state prediction, scheduling decision and strategy verification, wherein the functional units are respectively used for trend prediction of system operating state, generation of computing power scheduling strategy and feasibility verification of the scheduling strategy under constraints.

[0011] 7. The system according to any one of claims 5 or 6, characterized in that the system adopts hierarchical optimized control logic, including computing power access layer control logic, computing power layer control logic and energy layer control logic.

[0012] 8. A computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing the method of any one of claims 1 to 4.

[0013] Compared with the prior art, the present invention has at least the following beneficial effects: 1. Achieve cross-layer collaborative optimization of energy supply, computing power execution, and revenue on the computing power task access side; 2. Improve the overall energy efficiency of the computing system and reduce energy consumption per unit of computing power; 3. Improve the stability of computing power revenue and reduce losses caused by energy and network fluctuations; 4. Applicable to centralized or distributed computing centers, facilitating deployment and expansion in computing systems of different scales. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the system structure of the multi-source computing power collaborative optimization method and system based on artificial intelligence of the present invention, used to illustrate the relationship between multi-source monitoring, state fusion, staged estimation, decision generation and execution control in the system; Figure 2 This is a flowchart illustrating the multi-source computing power collaborative optimization method and system based on artificial intelligence of the present invention. It is used to explain the processing flow of preprocessing, state feature construction, coarse estimation, fine estimation, constraint verification, scheduling strategy generation and execution feedback based on multi-source operating data. Figure 3 This is a schematic diagram of the hierarchical optimization control logic of the multi-source computing power collaborative optimization method and system based on artificial intelligence of the present invention, used to illustrate the collaborative control relationship between the computing power access layer, the computing power layer and the energy layer. Detailed Implementation

[0015] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. These embodiments are only for illustrating the technical solutions of the present invention and do not constitute a limitation on the scope of protection of the present invention. Those skilled in the art can make various modifications or substitutions without departing from the technical concept of the present invention, and all such modifications or substitutions should fall within the scope of protection of the present invention.

[0016] Reference Figure 1 The present invention provides a multi-source computing power collaborative optimization method and system based on artificial intelligence, including an energy layer monitoring module, a computing power layer monitoring module, a computing power access layer monitoring module, a network status monitoring module, an artificial intelligence decision-making module, and an execution control module.

[0017] The energy layer monitoring module is used to collect operational data from the energy supply side, including but not limited to power generation, gas or electricity parameters, energy supply stability indicators, and energy cost information. In some embodiments, the energy layer monitoring module includes an energy status acquisition unit and an energy cost assessment unit, used to acquire energy supply status information and assess the corresponding energy usage costs, respectively.

[0018] The computing power layer monitoring module is used to collect operating status data of computing devices, including device power consumption, operating frequency, chip temperature, cooling system status, and computing power output data. In some embodiments, the computing power layer monitoring module includes an operating frequency acquisition unit and a temperature status acquisition unit, used to acquire the operating frequency information and temperature status information of the computing devices.

[0019] The computing power access layer monitoring module is used to acquire operational information from the computing power task access side. This operational information includes the computing power task access side's profitability, network difficulty, network latency, and computing power access status information. In some embodiments, the computing power access layer monitoring module includes a profitability status acquisition unit and a network latency acquisition unit, used to acquire profitability status and network communication status information after the computing power task accesses the system.

[0020] The network status monitoring module is used to obtain network operation status information of the computing power system. In some implementations, the network status monitoring module includes a communication bandwidth monitoring unit and a network latency monitoring unit, which are used to monitor the communication bandwidth and network latency of the computing power system.

[0021] The artificial intelligence decision-making module is communicatively connected to the energy layer monitoring module, computing power layer monitoring module, computing power access layer monitoring module, and network status monitoring module, respectively, for comprehensive analysis, prediction, and optimization decision-making based on collected multi-source operational data. In some embodiments, the artificial intelligence decision-making module includes a state fusion module, a coarse estimation module, a fine estimation module, and functional units for performing state prediction, scheduling decisions, and policy verification. Specifically, the functional units for performing state prediction, scheduling decisions, and policy verification are used to predict system operation trends, generate computing power scheduling strategies, and verify the feasibility of scheduling strategies.

[0022] In practical implementation, the AI ​​decision-making module can be deployed on a cloud server to utilize centralized computing resources for joint analysis and optimization decisions based on multi-source data. In other implementations, the AI ​​decision-making module can be deployed on a local server or edge computing device to reduce data transmission latency and improve the system's real-time response capability. Those skilled in the art can choose an appropriate deployment method based on the scale of the computing system and network conditions.

[0023] The execution control module is communicatively connected to the artificial intelligence decision-making module and is used to control and adjust the operating parameters of the computing power equipment and the access status of the computing power task access side according to the computing power scheduling strategy output by the artificial intelligence decision-making module. In specific implementation, the execution control module adjusts the operating frequency, voltage parameters, start / stop status, and computing power allocation ratio of the computing power equipment according to the received control commands, and completes the corresponding control operations through the computing power equipment interface or the computing power task access side interface. The execution control module can also feed back the actual operating status of the computing power equipment and the control execution results to the artificial intelligence decision-making module to achieve closed-loop optimization control.

[0024] Reference Figure 2The multi-source computing power collaborative optimization method and system based on artificial intelligence of the present invention includes the following steps: S1, Data acquisition step. Multi-source operational data of the computing power system are collected through the energy layer monitoring module, computing power layer monitoring module, computing power access layer monitoring module, and network status monitoring module to obtain the original operational dataset. S2, Data preprocessing step. The artificial intelligence decision-making module performs time alignment, anomaly detection, and preprocessing on the multi-source operational data to generate standardized operational data, thereby eliminating the impact of outliers and sampling differences on subsequent analysis. S3, State feature construction step. Based on the standardized operational data, feature parameters describing the system's operational state are extracted to construct a system state feature set. S4, Coarse estimation step. Based on the system state feature set, a coarse estimate of the system's operational state for future periods is performed to obtain a coarse estimation result. S5, Fine estimation step. Based on the coarse estimation result, a fine estimation of the computing power energy consumption state, computing power load state, and computing power task access side revenue state is performed to obtain a fine estimation result. S6, Constraint verification step. Based on energy supply constraints, computing power equipment security constraints, and system operational reliability constraints, the feasibility of the fine estimation result is verified. S7, Scheduling Generation Step. Under the condition that the constraints are met, generate the corresponding computing power scheduling strategy and operating parameter adjustment instructions. S8, Execution Feedback Step. Execute the control of computing power equipment and its computing power resources according to the computing power scheduling strategy, and feed the execution results back to the artificial intelligence decision-making module to achieve dynamic optimization and adjustment.

[0025] Reference Figure 3 This invention employs a hierarchical optimization control approach to optimize and manage a computing power system, including computing power access layer control logic, computing power layer control logic, and energy layer control logic. In the computing power access layer control logic, the system determines the computing power access strategy and computing power allocation ratio based on the predicted revenue results from computing power task access and network status information to optimize the overall revenue level. In the computing power layer control logic, the system schedules and controls the operating frequency, power parameters, and start / stop status of computing power devices based on the predicted computing power load to improve the utilization efficiency of computing power resources. In the energy layer control logic, the system combines energy supply capacity and power constraints to adjust the computing power load scale to ensure the overall security and stability of the system operation.

[0026] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any equivalent substitutions or modifications made by those skilled in the art to the above embodiments without departing from the technical concept of the present invention should fall within the scope of protection of the present invention.

Claims

1. A multi-source computing power collaborative optimization method based on artificial intelligence, characterized in that, Includes the following steps: S1. Collect multi-source operational data of the computing power system based on the energy layer monitoring module, computing power layer monitoring module, computing power access layer monitoring module, and network status monitoring module to obtain the raw operational dataset; S2. Perform time alignment, anomaly detection, and preprocessing on the multi-source operational data to generate standardized operational data; S3. Extract feature parameters to describe the system's operational status based on the standardized operational data to construct a system status feature set; S4. Based on the system state feature set, make a rough estimate of the system operating state for future time periods and obtain the rough estimate result; S5. Based on the rough estimation results, perform a fine estimation of the computing power energy consumption status, computing power load status, and computing power task access side revenue status to obtain a fine estimation result; S6. Perform a feasibility verification of the fine estimation results based on energy supply constraints, computing power equipment security constraints, and system operation reliability constraints. S7. Under the condition of satisfying the constraints, generate the corresponding computing power scheduling strategy and operation parameter adjustment instructions; S8. Execute the control of computing devices and their computing resources according to the computing power scheduling strategy, and feed the execution results back to the artificial intelligence decision-making module to achieve dynamic optimization and adjustment.

2. The method according to claim 1, characterized in that, The energy layer operation data in the multi-source operation data includes energy supply power, gas or electricity parameters, power generation efficiency, and energy cost information.

3. The method according to claim 1, characterized in that, The artificial intelligence decision-making process is implemented using at least one machine learning model, reinforcement learning model, or intelligent optimization model based on a combination of rules and models for multi-source operation state prediction and scheduling decision-making.

4. The method according to claim 1, characterized in that, The computing power scheduling strategy is generated through a hierarchical decision-making process, including computing power access layer decision, computing power layer decision, and energy layer decision.

5. A multi-source computing power collaborative optimization system based on artificial intelligence, characterized in that, include: The energy layer monitoring module is used to collect operational status data from the energy supply side. The computing power layer monitoring module is used to collect operating status data of computing power equipment; The computing power access layer monitoring module is used to collect the operating status data of the computing power task access side; The artificial intelligence decision-making module is used to perform state fusion, predictive analysis, and scheduling decisions on multi-source operational data; the execution control module is used to control the access status of computing power devices and computing power tasks based on the scheduling decision results.

6. The system according to claim 5, characterized in that, The artificial intelligence decision-making module includes functional units for performing state prediction, scheduling decisions, and strategy verification. These functional units are respectively used to predict the trend of system operation status, generate computing power scheduling strategies, and verify the feasibility of the scheduling strategies under constraints.

7. The system according to any one of claims 5 or 6, characterized in that, The system adopts a hierarchical optimized control logic, including computing power access layer control logic, computing power layer control logic, and energy layer control logic.

8. A computer-readable storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing the method of any one of claims 1 to 4.

Images