Green computing power intelligent scheduling optimization system for computing network integration

Abstract

The invention relates to the technical field of computing network fusion, in particular to a green computing power intelligent scheduling optimization system for computing network fusion, which comprises a theft source sensing unit, a computing power distribution optimization unit, an energy consumption monitoring regulation and control unit and a task dynamic adaptation unit. A dynamic resource distribution graph is generated through multi-dimensional data collection, task allocation is optimized based on an improved fast transmission algorithm, and efficient and flexible task scheduling and energy consumption control are achieved in combination with hierarchical energy consumption management and a dynamic task adaptation strategy. According to the method, the computing power resource utilization efficiency can be remarkably improved, the energy consumption is greatly reduced while the service quality is ensured, and a brand new solution is provided for green computing in a computing network integration environment.

Description

Technical Field

[0001] The present invention belongs to the field of information technology and communication technology, and specifically relates to a green computing power intelligent scheduling optimization system for computing-network integration. Background Art

[0002] With the rapid development of information technology, computing power scheduling requirements in computing-network convergence scenarios are becoming increasingly complex and variable, posing new challenges to green computing and efficient resource utilization. Traditional computing power scheduling methods primarily rely on static allocation and local optimization strategies, but these methods have certain limitations in terms of flexibility and adaptability. Due to the dynamic changes and diverse demands of application scenarios, traditional methods have difficulty responding to complex real-time scheduling requirements in a timely manner. In addition, although some existing intelligent scheduling systems have introduced big data analysis and machine learning technologies, they still face problems such as high energy consumption and insufficient scheduling efficiency when processing multi-dimensional, large-scale computing power resources. Therefore, there is an urgent need for an intelligent optimization system that can balance green energy conservation and efficient scheduling to better meet the diverse needs of computing-network convergence environments. Summary of the Invention

[0003] The present invention relates to the field of computing-network integration technology, and specifically to a green computing power intelligent scheduling and optimization system for computing-network integration. The system of the present invention includes a resource perception unit, a computing power allocation optimization unit, an energy consumption monitoring and control unit, and a task dynamic adaptation unit. The resource perception unit collects multi-dimensional computing power resource data in real time, and generates a dynamic resource distribution map in combination with historical operation records; the computing power allocation optimization unit performs hierarchical classification processing on tasks based on the distribution map, and introduces a lightweight machine learning model to predict the optimal solution for the task execution path; the energy consumption monitoring and control unit monitors the energy consumption status of each node in real time through a distributed sensor network, and adjusts the working mode of the computing node according to a preset threshold; the task dynamic adaptation unit is responsible for re-planning the task priority and execution order according to the current task requirements and resource usage.

[0004] In this invention, the resource perception unit achieves comprehensive awareness of computing resources through micro-sensor modules deployed on computing nodes. These sensor modules are connected to the computing nodes using a high-speed serial communication interface to ensure low latency and high reliability of data transmission. The micro-sensor modules integrate temperature sensors, power consumption monitors, and performance counters, which are used to collect the node's thermal status, energy consumption level, and computing performance indicators, respectively. All collected data is integrated through a centralized data aggregation module to form a multidimensional resource distribution map. This map is stored and updated in the form of a time series, providing basic support for subsequent computing power allocation.

[0005] The computing power allocation optimization unit adopts a task scheduling strategy based on an improved genetic algorithm. The core of this strategy is to gradually optimize the task allocation scheme by simulating the natural selection process. Specifically, the initial population consists of multiple randomly generated task allocation schemes, and each scheme corresponds to a mapping relationship between a set of tasks and computing power nodes. During the evolution process, the fitness function comprehensively considers the three key indicators of task completion time, node energy consumption, and resource utilization. By performing crossover and mutation operations on individuals in the population, the optimal task allocation scheme is gradually screened out. In order to improve the convergence speed of the algorithm, the present invention also introduces an adaptive parameter adjustment mechanism to dynamically adjust the crossover probability and mutation probability according to the diversity of the current population, thereby avoiding falling into the local optimal solution.

[0006] The energy consumption monitoring and control unit has designed a hierarchical energy consumption management mechanism. Its core concept is to flexibly adjust the working mode of the computing node according to the priority and resource requirements of different tasks. Specifically, when the real-time energy consumption of a node exceeds the preset safety threshold, the control unit will trigger a frequency reduction operation, reducing energy consumption by lowering the processor's main frequency; if the energy consumption requirements cannot be met after the frequency reduction, it will further switch to sleep mode and migrate non-critical tasks to other idle nodes. In addition, to ensure the stability and reliability of the system, the control unit is also equipped with a redundant backup module, which can quickly take over tasks when the main node fails, avoiding service interruptions caused by node switching.

[0007] The task dynamic adaptation unit achieves dynamic adjustment of the task queue through a task priority evaluation method based on a sliding window. The method first divides the task queue into several time windows, each of which contains a certain number of task instances. For each task instance, the adaptation unit calculates a comprehensive priority score based on its resource requirements, execution time and task type, and uses the score result as the sorting basis to rearrange the task queue. In order to further improve the adaptation efficiency, the present invention also proposes a task merging strategy based on task similarity, that is, merging tasks with similar resource requirements and execution paths into a composite task, thereby reducing the complexity of task scheduling.

[0008] Through the above-mentioned technical means, the present invention effectively solves the deficiencies of traditional computing power scheduling methods in terms of flexibility and adaptability. The multi-dimensional data acquisition capability of the resource perception unit enables the system to fully grasp the real-time status of computing power resources, providing reliable data support for subsequent optimization; the improved genetic algorithm of the computing power allocation optimization unit significantly improves the efficiency and accuracy of task allocation while reducing overall energy consumption; the hierarchical management mechanism of the energy consumption monitoring and control unit minimizes energy waste while ensuring system stability; the priority evaluation and task merging strategy of the task dynamic adaptation unit further improve the flexibility and response speed of task scheduling.

[0009] In summary, this paper constructs a green computing power intelligent scheduling optimization system for computing-network convergence by introducing key technologies such as multi-dimensional resource perception, an improved genetic algorithm, hierarchical energy consumption management, and dynamic task adaptation. This system not only significantly improves the utilization efficiency of computing resources but also significantly reduces energy consumption while ensuring service quality, providing a new solution for green computing and efficient scheduling in a computing-network convergence environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention, showing the logical relationship and interaction process between the resource perception unit, computing power allocation optimization unit, energy consumption monitoring and control unit and task dynamic adaptation unit.

[0011] Figure 2 This is a structural diagram of the resource sensing unit, which details the connection between the micro-sensing module and the computing power node, as well as the integrated layout of the temperature sensor, power consumption monitor and performance counter inside the module.

[0012] Figure 3 This is a flowchart of the task scheduling strategy for the computing power allocation optimization unit, focusing on the task allocation scheme optimization process based on the improved genetic algorithm, including initial population generation, fitness evaluation and adaptive parameter adjustment mechanism.

[0013] Figure 4 This is a schematic diagram of the hierarchical energy consumption management mechanism of the energy consumption monitoring and control unit, showing the logical control path of the energy consumption threshold triggering frequency reduction operation and sleep mode switching, and also marking the workflow of the redundant backup module.

[0014] Figure 5 This is a schematic diagram of the task priority evaluation method for the task dynamic adaptation unit, presenting the specific implementation steps of sliding window division, comprehensive priority score calculation, and task queue reordering.

[0015] Figure 6 This is the execution flow chart of the task merging strategy, which describes the task merging process based on task similarity and its optimization effect on task scheduling complexity.

[0016] The accompanying drawings are numbered as follows:

[0017] 1. Resource perception unit; 2. Computing power allocation optimization unit; 3. Energy consumption monitoring and control unit; 4. Task dynamic adaptation unit; 5. Micro sensor module; 6. Temperature sensor; 7. Power consumption monitor; 8. Performance counter; 9. Centralized data aggregation module; 10. Improved genetic algorithm module; 11. Hierarchical energy consumption management module; 12. Redundant backup module; 13. Sliding window module; 14. Task merging module. DETAILED DESCRIPTION

[0018] The present invention discloses a green computing power intelligent scheduling optimization system for computing-network integration, and its specific implementation method is described in detail with reference to the accompanying drawings. Figure 1 This is a schematic diagram of the overall system architecture, showing the logical relationship and interaction between the resource perception unit 1, computing power allocation optimization unit 2, energy consumption monitoring and control unit 3, and task dynamic adaptation unit 4. These units are connected via a high-speed communication network to ensure real-time and reliable data transmission.

[0019] Resource sensing unit 1 is one of the core modules of the system, and its structure is as follows: Figure 2 As shown, it includes a micro-sensor module 5, a temperature sensor 6, a power consumption monitor 7, a performance counter 8, and a centralized data aggregation module 9. The micro-sensor module 5 is deployed on the computing power node and connected to the computing power node through a high-speed serial communication interface to achieve low-latency and high-reliability data collection. The temperature sensor 6 is used to collect the thermal status information of the node, the power consumption monitor 7 is used to measure the energy consumption level of the node, and the performance counter 8 is responsible for recording the computing performance indicators of the node. All collected data is integrated through the centralized data aggregation module 9 to form a multi-dimensional resource distribution map. The map is stored and updated in the form of a time series, providing basic support for subsequent computing power allocation. The various components inside the micro-sensor module 5 are integrated on a circuit board, where the temperature sensor 6 is located close to the processor heat source, the power consumption monitor 7 is directly connected to the power management module, and the performance counter 8 communicates with the processor through a dedicated interface. This layout ensures the accuracy and comprehensiveness of data collection.

[0020] The task scheduling strategy of computing power allocation optimization unit 2 is as follows: Figure 3 As shown, its core is the task allocation scheme optimization process based on the improved genetic algorithm. The initial population consists of multiple randomly generated task allocation schemes, each of which corresponds to a mapping relationship between a set of tasks and computing power nodes. The fitness function comprehensively considers three key indicators: task completion time, node energy consumption, and resource utilization, and gradually screens out the optimal solution by performing crossover and mutation operations on individuals in the population. The adaptive parameter adjustment mechanism dynamically adjusts the crossover probability and mutation probability according to the diversity of the current population to avoid falling into the local optimal solution. The improved genetic algorithm module 10 is implemented by a software program, runs on a high-performance computing server, and interacts with the resource perception unit 1 and the task dynamic adaptation unit 4 through a network interface. The module 10 receives a multi-dimensional resource distribution map from the resource perception unit 1, and sends the optimized task allocation scheme to the task dynamic adaptation unit 4, thereby achieving efficient scheduling of tasks.

[0021] The hierarchical energy consumption management mechanism of the energy consumption monitoring and control unit 3 is as follows: Figure 4As shown, its core idea is to flexibly adjust the working mode of the computing node according to the priority and resource requirements of different tasks. The hierarchical energy consumption management module 11 monitors the energy consumption status of each node in real time through a distributed sensor network, and triggers frequency reduction operation or sleep mode switching according to the preset threshold. When the real-time energy consumption of a node exceeds the safety threshold, the control unit first reduces the processor main frequency to reduce energy consumption; if the requirements cannot be met after the frequency reduction, it will further switch to sleep mode and migrate non-critical tasks to other idle nodes. The redundant backup module 12 is equipped to quickly take over tasks when the main node fails, avoiding service interruptions caused by node switching. The hierarchical energy consumption management module 11 and the redundant backup module 12 work together through hardware circuits and software programs. The hardware circuit is responsible for real-time monitoring of the energy consumption status and performing frequency reduction or sleep operations, and the software program is responsible for the logical control of task migration and redundant backup.

[0022] The task priority evaluation method of the task dynamic adaptation unit 4 is as follows: Figure 5 As shown, its core is a dynamic adjustment mechanism of task queues based on sliding windows. The sliding window module 13 divides the task queue into several time windows, each of which contains a certain number of task instances. For each task instance, the adaptation unit calculates a comprehensive priority score based on its resource requirements, execution time and task type, and rearranges the task queue based on the score result. The task merging module 14 merges tasks with similar resource requirements and execution paths into a composite task by analyzing task similarity, thereby reducing the complexity of task scheduling. The sliding window module 13 and the task merging module 14 are implemented by software programs, run on the task management server, and interact with the computing power allocation optimization unit 2 and the energy consumption monitoring and control unit 3 through a network interface. Module 13 receives the task allocation plan from the computing power allocation optimization unit 2, and sends the reordered task queue to the energy consumption monitoring and control unit 3, thereby realizing dynamic adaptation of tasks.

[0023] In actual application, this system is deployed in a data center environment with integrated computing and network. The computing nodes in the data center are interconnected through a high-speed network, and each node is equipped with a micro-sensor module 5 to achieve multi-dimensional resource perception. The centralized data aggregation module 9 is deployed next to the core switch of the data center, responsible for collecting and integrating the resource data of all nodes. The improved genetic algorithm module 10 runs on the central scheduling server of the data center, and generates the optimal task allocation plan based on the multi-dimensional resource distribution map. The hierarchical energy consumption management module 11 and the redundant backup module 12 are deployed on each computing node, and energy consumption monitoring and task migration are realized through the local controller. The sliding window module 13 and the task merging module 14 run on the task management server, responsible for dynamically adjusting the task queue and optimizing the complexity of task scheduling. The entire system is configured and monitored through a unified management platform to ensure the coordinated operation of each module.

[0024] The resource sensing unit 1 and the computing power allocation optimization unit 2 are connected via a high-speed communication link. The resource sensing unit 1 transmits the collected multi-dimensional resource distribution map to the computing power allocation optimization unit 2 in real time. The computing power allocation optimization unit 2 generates a task allocation plan based on the improved genetic algorithm and sends it to the task dynamic adaptation unit 4 via a network interface. The task dynamic adaptation unit 4 replans the task priority and execution order based on the current task requirements and resource usage. The energy consumption monitoring and control unit 3 monitors the energy consumption status of each node in real time through a distributed sensor network and feeds the energy consumption data back to the task dynamic adaptation unit 4. The task dynamic adaptation unit 4 adjusts the task execution strategy based on the energy consumption status. This collaborative working mechanism between modules ensures the efficient operation of the system and the optimal utilization of resources.

[0025] During the operation of the system, the resource perception unit 1 continuously collects multi-dimensional resource data of the computing power nodes and generates a multi-dimensional resource distribution map through the centralized data aggregation module 9. The computing power allocation optimization unit 2 sends the task allocation plan generated based on the map to the task dynamic adaptation unit 4 through the network interface. The task dynamic adaptation unit 4 re-plans the task priority and execution order according to the current task requirements and resource usage. At the same time, the energy consumption monitoring and control unit 3 monitors the energy consumption status of each node in real time through a hierarchical energy consumption management mechanism, and adjusts the node working mode according to the energy consumption data. The task dynamic adaptation unit 4 dynamically adjusts the task queue based on the energy consumption data and the task priority evaluation results to ensure the efficient execution of tasks and the rational use of resources. This complete operation process realizes the intelligent scheduling of computing power resources and the refined management of energy consumption.

[0026] In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is further supplemented below with reference to a specific application scenario.

[0027] In a data center environment where computing and networking converge, after the system is started, the resource perception unit 1 first collects multi-dimensional resource data of the computing power node in real time through the temperature sensor 6, power consumption monitor 7, and performance counter 8 in the micro-sensor module 5. This data includes the node's thermal status information, energy consumption level, and computing performance indicators. All collected data is integrated through the centralized data aggregation module 9 to form a multi-dimensional resource distribution map. This map is stored and updated in the form of a time series, providing basic support for subsequent task allocation. The various components within the micro-sensor module 5 are integrated on a circuit board. The temperature sensor 6 is located close to the processor's heat source to ensure accurate collection of thermal status information; the power consumption monitor 7 is directly connected to the power management module to provide real-time feedback on energy consumption changes; and the performance counter 8 communicates with the processor through a dedicated interface to record computing performance indicators. This layout design ensures the accuracy and comprehensiveness of data collection, thus laying a solid foundation for subsequent optimized scheduling.

[0028] Subsequently, the computing power allocation optimization unit 2 optimizes the task allocation scheme using the improved genetic algorithm module 10 based on the multi-dimensional resource distribution map generated by the resource sensing unit 1. The initial population consists of multiple randomly generated task allocation schemes, each of which corresponds to a mapping relationship between a set of tasks and computing power nodes. The fitness function comprehensively considers three key indicators: task completion time, node energy consumption, and resource utilization, and gradually screens out the optimal solution by performing crossover and mutation operations on individuals in the population. The adaptive parameter adjustment mechanism dynamically adjusts the crossover probability and mutation probability according to the diversity of the current population to avoid falling into a local optimal solution. The improved genetic algorithm module 10 runs on a high-performance computing server, receives the multi-dimensional resource distribution map from the resource sensing unit 1, and sends the optimized task allocation scheme to the task dynamic adaptation unit 4. This process is implemented through a software program to ensure the efficiency and accuracy of task allocation.

[0029] At the same time, the energy consumption monitoring and control unit 3 monitors the energy consumption status of each node in real time through the hierarchical energy consumption management module 11. When the real-time energy consumption of a node exceeds the preset safety threshold, the hierarchical energy consumption management module 11 first triggers the frequency reduction operation to reduce energy consumption by reducing the main frequency of the processor. If the requirements are still not met after the frequency reduction, it will further switch to the sleep mode and migrate non-critical tasks to other idle nodes. The redundant backup module 12 quickly takes over the task when the main node fails to avoid service interruption caused by node switching. The hierarchical energy consumption management module 11 and the redundant backup module 12 work together through hardware circuits and software programs. The hardware circuit is responsible for real-time monitoring of the energy consumption status and performing frequency reduction or sleep operations, while the software program is responsible for the logical control of task migration and redundant backup. This mechanism minimizes energy waste while ensuring system stability by flexibly adjusting the node working mode.

[0030] Next, the task dynamic adaptation unit 4 re-plans the task priority and execution order based on the current task requirements and resource usage. The sliding window module 13 divides the task queue into several time windows, each of which contains a certain number of task instances. For each task instance, the task dynamic adaptation unit 4 calculates a comprehensive priority score based on its resource requirements, execution time and task type, and uses the score result as the sorting basis to rearrange the task queue. The task merging module 14 analyzes the task similarity and merges tasks with similar resource requirements and execution paths into a composite task, thereby reducing the complexity of task scheduling. The sliding window module 13 and the task merging module 14 are implemented by software programs, running on the task management server, and interacting with the computing power allocation optimization unit 2 and the energy consumption monitoring and control unit 3 through a network interface. Module 13 receives the task allocation plan from the computing power allocation optimization unit 2, and sends the re-sorted task queue to the energy consumption monitoring and control unit 3, thereby realizing dynamic adaptation of tasks.

[0031] During actual operation, the resource perception unit 1 continuously collects multi-dimensional resource data of the computing power nodes and generates a multi-dimensional resource distribution map through the centralized data aggregation module 9. The computing power allocation optimization unit 2 sends the task allocation plan generated based on the map to the task dynamic adaptation unit 4 through the network interface. The task dynamic adaptation unit 4 re-plans the task priority and execution order according to the current task requirements and resource usage. At the same time, the energy consumption monitoring and control unit 3 monitors the energy consumption status of each node in real time through a hierarchical energy consumption management mechanism, and adjusts the node working mode according to the energy consumption data. The task dynamic adaptation unit 4 dynamically adjusts the task queue based on the energy consumption data and the task priority evaluation results to ensure the efficient execution of tasks and the rational use of resources.

[0032] Through the above steps, the system achieves intelligent scheduling of computing resources and refined management of energy consumption. The multidimensional data collection capabilities of Resource Perception Unit 1 enable the system to fully grasp the real-time status of computing resources, providing reliable data support for subsequent optimization. The improved genetic algorithm of Computing Power Allocation Optimization Unit 2 significantly improves the efficiency and accuracy of task allocation while reducing overall energy consumption. The hierarchical management mechanism of Energy Consumption Monitoring and Control Unit 3 minimizes energy waste while ensuring system stability. The priority assessment and task merging strategy of Task Dynamic Adaptation Unit 4 further enhance the flexibility and responsiveness of task scheduling. The collaborative working mechanism of the entire system ensures efficient operation and optimal resource utilization.

[0033] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A green computing power intelligent scheduling optimization system for computing-network integration, characterized by: It includes a resource sensing unit (1), a computing power allocation optimization unit (2), an energy consumption monitoring and control unit (3) and a task dynamic adaptation unit (4); The resource sensing unit (1) is used to collect multi-dimensional computing resource data in real time and generate a dynamic resource distribution map; The computing power allocation optimization unit (2) is used to perform hierarchical classification processing on tasks based on a dynamic resource distribution map and predict the optimal solution of the task execution path through an improved genetic algorithm; The energy consumption monitoring and control unit (3) is used to monitor the energy consumption status of each node through a distributed sensor network and adjust the working mode of the computing node according to a preset threshold; The task dynamic adaptation unit (4) is used to re-plan task priorities and execution sequences according to current task requirements and resource usage.

2. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 1 is characterized in that: The resource sensing unit (1) includes a micro sensor module (5), a temperature sensor (6), a power consumption monitor (7), a performance counter (8) and a centralized data aggregation module (9); The micro sensor module (5) is deployed on the computing power node and connected to the computing power node via a high-speed serial communication interface; The temperature sensor (6) is used to collect thermal status information of the node; The power consumption monitor (7) is used to measure the energy consumption level of the node; The performance counter (8) is used to record the computing performance index of the node; The centralized data aggregation module (9) is used to integrate the collected data and form a multi-dimensional resource distribution map.

3. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 1 is characterized in that: The computing power allocation optimization unit (2) includes an improved genetic algorithm module (10); the improved genetic algorithm module (10) is used to optimize the task allocation scheme through initial population generation, fitness evaluation and adaptive parameter adjustment mechanism; The fitness evaluation comprehensively considers three indicators: task completion time, node energy consumption and resource utilization.

4. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 3 is characterized in that: The adaptive parameter adjustment mechanism dynamically adjusts the crossover probability and the mutation probability according to the diversity of the current population.

5. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 1 is characterized in that: The energy consumption monitoring and control unit (3) includes a hierarchical energy consumption management module (11) and a redundant backup module (12); The hierarchical energy consumption management module (11) is used to trigger a frequency reduction operation or a sleep mode switch according to the real-time energy consumption status; The redundant backup module (12) is used to quickly take over tasks when the main node fails.

6. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 5 is characterized in that: The hierarchical energy consumption management module (11) first reduces the processor main frequency when the real-time energy consumption exceeds a preset safety threshold; if the energy consumption requirement cannot be met after the frequency reduction, it switches to a sleep mode and migrates non-critical tasks to other idle nodes.

7. The green computing power intelligent scheduling optimization system for computing-network integration according to claim 1 is characterized in that: The task dynamic adaptation unit (4) includes a sliding window module (13) and a task merging module (14); The sliding window module (13) is used to divide the task queue into a plurality of time windows and calculate a comprehensive priority score for each task instance; The task merging module (14) is used to merge tasks with similar resource requirements and execution paths into a composite task based on task similarity.

8. The green computing power intelligent scheduling and optimization system for computing-network integration according to claim 7 is characterized in that: The comprehensive priority score is calculated based on the task's resource requirements, execution time, and task type.

9. The green computing power intelligent scheduling optimization system for computing-network integration according to any one of claims 1 to 8, characterized in that: The resource sensing unit (1) is connected to the computing power allocation optimization unit (2) via a high-speed communication link; the computing power allocation optimization unit (2) is connected to the task dynamic adaptation unit (4) via a network interface; and the energy consumption monitoring and control unit (3) is connected to the task dynamic adaptation unit (4) via a distributed sensor network.

10. The green computing power intelligent scheduling optimization system for computing-network integration according to any one of claims 1 to 8, characterized in that: The system is deployed in a data center environment and configured and monitored through a unified management platform.

Images