A computing power scheduling adaptive adjustment method and system based on policy optimization

Abstract

This invention belongs to the field of computing power scheduling technology, and relates to a method and system for adaptive adjustment of computing power scheduling based on strategy optimization. The method includes: acquiring the current operating status parameters of each computing server, extracting backup parameters based on task processing status, generating verification parameters to construct multiple candidate operating intervals, and predicting the future operating status of the computing server; calculating the difference between the calculated temperature and the reference temperature to determine the verification execution time, and evaluating the carrying capacity and potential risks of the computing server under different task loads; calculating power standby parameters as delay parameters, and determining scheduling capabilities. This invention adjusts the scheduling strategy according to the real-time operating status of the computing server and task processing requirements to ensure efficient utilization of computing resources; it provides sufficient risk warning and avoidance before scheduling tasks are issued, reducing the probability of task execution interruption and ensuring the completion of scheduled tasks.

Description

Technical Field

[0001] This invention belongs to the field of computing power scheduling technology, specifically relating to a computing power scheduling adaptive adjustment method and system based on strategy optimization. Background Technology

[0002] With the development of technologies such as cloud computing, big data, and artificial intelligence, the demand for server computing power has surged. As a key means to optimize server resource allocation and improve computing efficiency, computing power scheduling technology manages and dynamically allocates massive and heterogeneous computing resources in a unified manner. By ensuring the operation of various complex business scenarios, computing power scheduling technology reduces the operating pressure on servers, thereby ensuring business continuity and achieving intensive resource management.

[0003] Existing computing power scheduling technologies mostly rely solely on task priorities and the instantaneous load of servers. Due to the continuous and buffered nature of server states, scheduling based on instantaneous indicators often struggles to accurately predict equipment health trends. When formulating scheduling schemes, it is difficult to fully consider server maintenance cycles and historical operating trajectories, and thus impossible to adjust according to fluctuations in computing load. In actual operation, extreme situations can easily occur where multiple normally operating servers simultaneously enter a shutdown maintenance state, disrupting the stability of computing power supply. In severe cases, this can easily lead to increased response latency and decreased service quality, making it difficult to meet the requirements of high-reliability services for a continuous and stable supply of computing power.

[0004] In view of this, the present invention provides a method and system for adaptive adjustment of computing power scheduling based on strategy optimization. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for adaptive adjustment of computing power scheduling based on strategy optimization, so as to solve the problem in the prior art that the entire computing power scheduling adjustment scheme cannot be optimized based on the historical operating status of the computing power server.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] An adaptive adjustment method for computing power scheduling based on policy optimization performs the following steps when a scheduling task to be processed is detected:

[0008] Determine the task processing status of each computing server based on its current operating status parameters;

[0009] Extract backup parameters corresponding to the task processing status; the backup parameters include the backup power, backup load, and backup temperature of the computing server;

[0010] The acquired backup power, backup load, and backup temperature are set as reference parameters, and the updated backup parameters are used to replace the reference parameters to generate multiple verification parameters based on the updated backup parameters; wherein, the multiple verification parameters are simulated temperature values;

[0011] Based on the backup parameters, multiple verification parameters are generated to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals;

[0012] Obtain the power required for each scheduled task, and calculate the power standby parameters by combining the historical task standby time and the equivalent execution time; then determine the power standby parameters as delay parameters.

[0013] Based on the matching results of the delay parameters of the scheduling task and multiple candidate running intervals, it is determined whether the computing server has scheduling capabilities.

[0014] When scheduling capability is determined, a scheduling task is issued to the computing server; when scheduling capability is determined not to be determined, a task redirection operation is performed.

[0015] Preferably, constructing multiple candidate running intervals includes:

[0016] Obtain the real-time operating power and reference temperature of the computing server, and set the reference temperature as the first temperature; adjust the real-time operating power to the backup power in the backup parameters, calculate the second temperature under simulated operation, and calculate the deviation value between the first temperature and the second temperature; generate multiple simulated temperature values ​​based on the first temperature and the deviation value, and determine the multiple simulated temperature values ​​as multiple verification parameters.

[0017] Preferably, the task processing state includes a running state, an idle state, and a delayed state;

[0018] When the task processing status is running, extract the backup temperature from the backup parameters, and generate multiple simulated temperature values ​​based on the time-series fluctuation range of the backup temperature, which serve as multiple verification parameters.

[0019] When the task processing status is idle, the computing server is marked as the scheduling server, the current time point is recorded as the base time, the second temperature reached under the backup power in the backup parameters is obtained, and multiple simulated temperature values ​​are generated in combination with the time-series fluctuation range of the second temperature as the multiple verification parameters.

[0020] When the task processing status is delayed, the allowable fluctuation range of the reference temperature is obtained, the reference temperature is extracted, and multiple simulated temperature values ​​are generated by combining the time-series fluctuation range of the reference temperature as multiple verification parameters. The delayed period is set as the default candidate running interval, and risk warning information is generated at the same time.

[0021] Preferably, the multiple verification parameters generated based on the backup parameters, used to calibrate the verification execution time of the computing power server and construct multiple candidate operating intervals, include:

[0022] Set the reference temperature as the reference value and set the allowable fluctuation range corresponding to the reference temperature; sort multiple verification parameters in chronological order and set them as calculation temperatures, calculate the difference between the calculation temperature and the reference temperature in turn until the difference exceeds the allowable fluctuation range; determine the time point before the difference exceeds the allowable fluctuation range as the verification execution time; subtract the verification execution time from the reference time to generate multiple candidate operating ranges.

[0023] Preferably, determining whether the computing server has scheduling capabilities based on the matching results of the scheduling task's delay parameters and multiple candidate running intervals includes:

[0024] Based on the power required by the scheduling task, a test point is selected from multiple candidate running intervals; the verification time point is determined in conjunction with the current scheduling time, and the verification time point is compared with the termination time points of multiple candidate running intervals; if the verification time point is within multiple candidate running intervals, the computing power server is determined to have scheduling capability.

[0025] Preferably, the step of performing task redirection when it is determined that scheduling capability is not available includes:

[0026] Collect backup servers for historical tasks with similar scheduling parameters that were interrupted when the task was interrupted; set the backup servers as the target servers for the current scheduled task to be redirected.

[0027] An adaptive adjustment system for computing power scheduling based on policy optimization includes the following modules:

[0028] The status awareness module is used to collect the current operating status parameters of each computing server; and to determine the task processing status of the computing server based on the current operating status parameters of each computing server.

[0029] The parameter extraction module is used to extract backup parameters corresponding to the task processing status.

[0030] The interval construction module is used to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals based on multiple verification parameters generated from the backup parameters.

[0031] The scheduling decision module is used to obtain the power required for each scheduling task, and calculate the power standby parameters by combining the historical task standby time and the equivalent execution time; the power standby parameters are determined as delay parameters; and the computing server is determined to have scheduling capability based on the matching results of the delay parameters of the scheduling task and multiple candidate running intervals.

[0032] The task execution module is used to issue scheduling tasks to the computing server when it is determined that scheduling capability is available; and to perform task redirection operations when it is determined that scheduling capability is not available.

[0033] Preferably, the multiple verification parameters generated based on the backup parameters, used to calibrate the verification execution time of the computing power server and construct multiple candidate operating intervals, include:

[0034] Set the reference temperature as the reference value and set the allowable fluctuation range corresponding to the reference temperature; sort multiple verification parameters in chronological order and set them as calculation temperatures, calculate the difference between the calculation temperature and the reference temperature in turn until the difference exceeds the allowable fluctuation range; determine the time point before the difference exceeds the allowable fluctuation range as the verification execution time; subtract the verification execution time from the reference time to generate multiple candidate operating ranges.

[0035] Preferably, determining whether the computing server has scheduling capabilities based on the matching results of the scheduling task's delay parameters and multiple candidate running intervals includes:

[0036] Based on the power required by the scheduling task, a test point is selected from multiple candidate running intervals; the verification time point is determined in conjunction with the current scheduling time, and the verification time point is compared with the termination time points of multiple candidate running intervals; if the verification time point is within multiple candidate running intervals, the computing power server is determined to have scheduling capability.

[0037] Preferably, the step of performing task redirection when it is determined that scheduling capability is not available includes:

[0038] Collect backup servers for historical tasks with similar scheduling parameters that were interrupted when the task was interrupted; set the backup servers as the target servers for the current scheduled task to be redirected.

[0039] Beneficial effects

[0040] This invention obtains the current operating status parameters of each computing server, extracts backup parameters based on the task processing status, generates verification parameters based on the backup parameters to construct multiple candidate operating intervals, predicts the future operating status of the computing server, identifies and marks computing servers with scheduling capabilities before the scheduling task is issued, improves the accuracy and predictability of computing resource scheduling, and avoids scheduling errors caused by information lag.

[0041] When the task processing status is running, this invention extracts the backup temperature as a verification parameter; when the task processing status is idle, the computing server is marked as a scheduling server, and the first temperature reached under backup power is obtained as a verification parameter. The scheduling strategy is adjusted according to the real-time running status of the computing server and the task processing requirements to ensure the efficient utilization of computing resources.

[0042] When generating candidate operating intervals, this invention obtains the real-time operating power and reference temperature of the computing server and adjusts it to a standby power level to calculate the second temperature and deviation value under simulated operating conditions, thereby generating verification parameters. When determining the verification execution time, the default temperature is set as the reference temperature and an allowable fluctuation range is set. The verification execution time is determined by calculating the difference between the calculated temperature and the reference temperature. This invention assesses the carrying capacity and potential risks of the computing server under different task loads, provides sufficient risk warnings and avoidance before the scheduling task is issued, reduces the probability of task execution interruption, and ensures the completion of the scheduling task.

[0043] This invention obtains the power required for each scheduled task, calculates the power standby parameter as a delay parameter by combining the standby time of historical tasks with the equivalent execution time, and matches it with candidate running intervals to determine scheduling capability. If scheduling capability is not available, a task redirection operation is performed, and the backup server corresponding to the historical task with similar scheduling parameters during the interruption is collected as the target server. This enables the management and intelligent redirection of scheduled tasks, optimizes the allocation efficiency of computing resources, improves the robustness and fault tolerance of the entire computing power scheduling, and ensures business continuity. Attached Figure Description

[0044] Figure 1 This is a flowchart of the method of the present invention;

[0045] Figure 2 This is a system structure diagram of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0047] Example 1

[0048] Please refer to Figure 1 This embodiment provides a computing power scheduling adaptive adjustment method based on strategy optimization, including the following steps:

[0049] S1. Operational Status Acquisition: Acquire the current operational status parameters and current task status of each computing server, and input these parameters and task status into a preset operational status identification logic for classification processing. The operational status parameters include CPU utilization, memory usage, network bandwidth utilization, disk I / O rate, real-time power consumption, and server chassis temperature. In practical use, a combination of at least one or more operational status parameters can be selected to characterize the real-time workload and physical health of the computing server.

[0050] The operation status identification logic is based on multi-dimensional hardware monitoring indicators. By comparing the operation status parameters with the preset load stage threshold range, the current load stage of the server can be identified.

[0051] The preset load phase threshold range is determined by empirical values ​​or historical data, including low load range, medium load range and high load range; and when the operating status parameters are in the low load range and medium load range, the server will not suffer hardware damage or drastic performance degradation.

[0052] Specifically, the low load range includes: CPU utilization ≤ 50%, memory utilization ≤ 50%, network bandwidth utilization ≤ 50%, SSD disk input / output I / O rate ≤ 300MB / s, real-time power consumption ≤ 50% of rated power, and server chassis temperature ≤ 55℃. The server's current load stage when the operating status parameters are in the low load range is defined as the low load stage.

[0053] High load range includes: CPU utilization ≥80%, memory utilization ≥80%, network bandwidth utilization ≥80%, SSD disk I / O rate ≥500MB / s, real-time power consumption ≥80% of rated power, server chassis temperature ≥75℃. The server's current load stage when the operating status parameters are in the high load range is defined as the high load stage.

[0054] The range between low load and high load is the medium load range. When the server's operating status parameters are in the medium load range, the current load stage is defined as the medium load stage.

[0055] S2. Backup parameter extraction: Based on the output of the running status identification logic, determine the task processing status of the computing power server, and extract backup parameters for quantifying the remaining available resources of the computing power server based on the task processing status.

[0056] The backup power, backup load, and backup temperature of the computing server are collected, and these three parameters are set as baseline parameters and used for dynamic update operations to obtain the updated backup parameters.

[0057] Among them, the reserve power is the difference between the rated power and real-time power consumption of the computing server under the current task load, representing the power increment corresponding to the additional computing capacity that can still be provided; the reserve load is the amount of additional tasks that the computing server can bear based on the percentage of remaining hardware resources without affecting the response latency of existing tasks; the reserve temperature is the temperature margin of the core components of the computing server, such as the processor core and memory, from the safe trigger threshold under the current ambient temperature and cooling fan speed.

[0058] The dynamic update operation includes replacing the baseline parameters with the backup parameters that are collected and updated in real time, and generating verification parameters based on the updated backup parameters. This dynamic update operation ensures the real-time nature of the backup parameters and accurately reflects the actual available resources of the computing server.

[0059] Task processing status includes running status, idle status, and delayed status; the task processing status is determined according to the current load stage of the server, with low load range, medium load range, and high load range corresponding to idle status, delayed status, and running status, respectively;

[0060] When the task processing status is running, the backup temperature is extracted from the backup parameters, and multiple simulated temperature values ​​are generated by combining the time-series fluctuation range of the backup temperature as multiple verification parameters to assess the risk of temperature rise when the computing server continues to carry tasks under the existing load.

[0061] When the task processing status is idle, it means that the computing server is currently not executing any tasks. The computing server is marked as a scheduling server, and the current time point is recorded as the base time. The second temperature reached under the backup power in the backup parameters is obtained. Multiple simulated temperature values ​​are generated by combining the time-series fluctuation range of the second temperature as multiple verification parameters to evaluate its thermodynamic performance after starting a new task. The thermodynamic performance includes the trend of temperature change over time, heat dissipation efficiency, and thermal equilibrium state.

[0062] When the task processing status is delayed, the allowable fluctuation range is obtained, the baseline temperature is extracted, and multiple simulated temperature values ​​are generated by combining the time-series fluctuation range of the baseline temperature as multiple verification parameters. The delayed period in the delayed state is set as the default candidate running range. At the same time, risk warning information is generated to remind the scheduler that the server may have response lag or hardware fatigue.

[0063] The allowable fluctuation range is the normal deviation range of the server's operating temperature from the set target value. The set target value can be selected as 40℃ based on empirical data, and the allowable fluctuation range can be set to 37-55℃. Other temperature ranges can also be set based on empirical data. The reference temperature is the measured steady-state temperature of the server under the current environment. The risk warning information includes alarm information that the server has potential performance bottlenecks or hardware vulnerabilities.

[0064] S3. Operational range generation: Based on the spare parameters of each computing server, verification parameters are generated through the operation trend evaluation logic. These verification parameters are used to simulate the operation trend of the computing server in the future.

[0065] The verification parameters are simulated temperature values. The process of generating verification parameters through the operation trend evaluation logic includes: obtaining the real-time operating power and reference temperature of the computing server, and setting the reference temperature as the first temperature; adjusting the real-time operating power to the backup power in the backup parameters, obtaining the second temperature under simulated high-load operation through the thermal resistance calculation process, and calculating the deviation value between the first temperature and the second temperature; generating multiple simulated temperature values ​​based on the first temperature and the deviation value, and determining these multiple simulated temperature values ​​as multiple verification parameters.

[0066] The thermal resistance calculation process involves using the heat conduction equation, combined with chip power consumption, thermal resistance of the heat dissipation medium, and ambient temperature, to calculate the junction temperature of electronic components or the chassis temperature. Specifically, the second temperature = the first temperature + (standby power - actual operating power) × thermal resistance coefficient, where the thermal resistance coefficient is the overall thermal resistance of the chassis, typically 0.05℃ / W for servers.

[0067] By simulating the temperature value generation process, the heat accumulation trend of the computing server after undertaking new tasks can be predicted. Based on these verification parameters, the time period during which the computing server can continue to run without triggering frequency reduction protection can be calculated in a time conversion method, which serves as the verification execution time.

[0068] Specifically, determining the verification execution time includes: setting the default safe temperature as the reference temperature, and setting the reference temperature to define the allowable fluctuation range of the upper and lower limits of the computing server under safe operation; arranging the verification parameters in chronological order, and setting this time series data as the calculation temperature; calculating the difference between the calculation temperature and the reference temperature at each time point in turn, until the difference exceeds the allowable fluctuation range.

[0069] The time point before the difference exceeds the allowable fluctuation range is determined as the verification execution time. The verification execution time represents the longest time that the computing server can maintain stable operation under the current level of spare resources.

[0070] Using the verification execution time as the baseline time point and combining it with the preset time step node, candidate running intervals are generated as a reference for subsequent task scheduling. The candidate running interval is obtained by subtracting the verification execution time from the baseline time. Then, according to the segmentation strategy, the total duration is divided into continuous time segments. Each time segment represents the window in which the computing power server can be used to execute tasks within a specific time period. The segmentation strategy divides the total duration into segments according to the preset time step node, which can be 1 minute.

[0071] S4. Filter the candidate running intervals and extract the corresponding delay parameters according to the scheduling requirements of the scheduling task to be scheduled;

[0072] The delay parameters are obtained by: acquiring the power required for each scheduled task, and calculating the power standby parameters by combining the historical task standby time and the equivalent execution time; the historical task standby time is the time required for the computing server to switch from idle standby state to task execution state, or it can be selected as the average time for similar historical scheduled tasks to switch from idle standby state to task execution stage; the equivalent execution time is the additional equivalent time corresponding to the unit load power of the computing server in the formal execution state of the task; the power standby parameter = (power required for each scheduled task × equivalent execution time) + historical task standby time;

[0073] The power standby parameter is defined as a delay parameter. The power standby parameter is used to reflect the power demand of the computing server on the scheduling task and the response delay of the server switching from a low power state to a high load state.

[0074] The delay parameters are matched with the candidate running intervals, and the matching results are used to determine whether the computing server has scheduling capabilities; ensure that the scheduling of scheduling tasks not only considers the availability of the server, but also the physical latency of switching scheduling tasks;

[0075] The matching method between the delay parameter and the candidate running interval includes: selecting a timestamp representing the expected start time of the scheduled task as the test point within the candidate running interval based on the power required by the scheduled task; determining the verification time point in combination with the current scheduling time, and comparing the verification time point with the end time point of the candidate running interval; if the verification time point is within the start and end range of the candidate running interval, it is determined that the computing power server has scheduling capability; otherwise, it is determined that the computing power server does not have scheduling capability.

[0076] S5. Scheduling and execution: If the computing power server has scheduling capabilities, a scheduling task is sent to the computing power server and the scheduling task is executed.

[0077] If the computing server does not have scheduling capabilities, a task redirection operation is performed to retrieve other idle servers or servers with redundant resources as target nodes and reallocate the execution path of the scheduling task.

[0078] The task redirection operation includes: collecting backup servers corresponding to historical scheduled tasks with similar hardware configuration parameters from a pre-stored list of associated devices when the task was interrupted; wherein the backup server is a device that is highly compatible with the original server in terms of network topology or computing architecture; setting the backup server as the target server for the current scheduled task to be redirected, thereby realizing the rapid migration of the scheduled task.

[0079] Example 2

[0080] Please refer to Figure 2 This embodiment provides a strategy-optimized adaptive adjustment system for computing power scheduling. This system can be deployed in the control core of a cloud computing center, edge computing node, or distributed computing power management platform. It communicates in real-time with each computing power server within the cluster via a network interface. The system includes the following modules:

[0081] The status awareness module is used to collect the current operating status parameters of each computing server and determine the task processing status of the computing server.

[0082] Continuously monitor the hardware indicators of each computing server to identify whether it is in a running state, an idle state, or a delayed state. When the task processing status is identified as an idle state, mark the corresponding computing server as a scheduling server and record the current time point as the base time to provide a starting anchor point for time-scale calculations.

[0083] If the task processing status is delayed, obtain the allowable fluctuation range of the baseline temperature, extract the baseline temperature, and generate multiple simulated temperature values ​​based on the time-series fluctuation range of the baseline temperature as multiple verification parameters. Set the delayed period as the default candidate running range, and generate risk warning information to warn of potential scheduling risks.

[0084] The parameter extraction module is used to extract the corresponding backup parameters in response to the task processing status determined by the status perception module. The backup parameters include the backup power, backup load and backup temperature of the computing server.

[0085] During the extraction process, the initial standby power, standby load, and standby temperature are set as baseline parameters, and the updated standby parameters are used to replace the baseline parameters based on the real-time feedback from the server to ensure that the subsequently generated verification parameters are timely.

[0086] When the task processing status is running, the backup temperature in the backup parameters is directly extracted as a verification parameter to assess the risk of temperature rise and thermodynamic performance of continuing to carry out the task under the existing load.

[0087] The interval construction module is used to generate verification parameters based on the backup parameters and mark the verification execution time in order to construct candidate running intervals;

[0088] When generating verification parameters, the real-time operating power and reference temperature of the computing server are obtained, and the reference temperature is set as the first temperature. The real-time operating power is adjusted to the backup power in the backup parameters. The second temperature under simulated high load operation is obtained through thermal resistance calculation, and the deviation between the first temperature and the second temperature is calculated. Multiple simulated temperature values ​​are generated based on the first temperature and the deviation value, and these multiple simulated temperature values ​​are determined as multiple verification parameters.

[0089] When calibrating the execution time, the reference temperature is set as the reference value and allowable fluctuation range. The calibration parameters are arranged in chronological order and set as the calculation temperature. The difference between the calculation temperature and the reference temperature at each time point is calculated in turn. When the difference exceeds the allowable fluctuation range, the time point before the difference is exceeded is determined as the calibration execution time.

[0090] The total duration is obtained by subtracting the verification execution time from the baseline time, and multiple candidate running intervals reflecting the server's sustainable service capabilities are generated according to the segmentation strategy.

[0091] The scheduling decision module is used to respond to pending scheduling tasks, match the delay parameters of the scheduling tasks with the candidate running intervals, and determine whether the computing server has scheduling capabilities.

[0092] The power required for the scheduled task is obtained, and the power standby parameter is calculated by combining the historical task standby time and the equivalent execution time, and then determined as the delay parameter.

[0093] During the matching process, test points are selected within the candidate operating range based on the required power, and the verification time point is determined in conjunction with the current scheduling time. The verification time point is then compared with the end time point of the candidate operating range. If the verification time point is within the start and end range of the candidate operating range, the computing power server is determined to have scheduling capability.

[0094] The task execution module is used to issue scheduling tasks to computing power servers with scheduling capabilities based on the judgment result of the scheduling decision module, or to perform task redirection operations when it is determined that no scheduling capability is available. When performing task redirection operations, the module collects the backup servers corresponding to historical scheduling tasks with similar hardware configuration parameters when interrupted from the list of associated devices, and sets the backup server as the target server for the current scheduling task to be redirected, so as to realize the smooth migration and reliable execution of tasks.

[0095] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand and implement the present invention. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A computational power scheduling adaptive adjustment method based on strategy optimization, characterized in that, When a pending scheduled task is detected, the following steps are performed: Determine the task processing status of each computing server based on its current operating status parameters; Extract backup parameters corresponding to the task processing status; the backup parameters include the backup power, backup load, and backup temperature of the computing server; The acquired backup power, backup load, and backup temperature are set as reference parameters, and the updated backup parameters are used to replace the reference parameters to generate multiple verification parameters based on the updated backup parameters; wherein, the multiple verification parameters are simulated temperature values; Based on the backup parameters, multiple verification parameters are generated to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals; Obtain the power required for each scheduled task, and calculate the power standby parameters by combining the historical task standby time and the equivalent execution time; then determine the power standby parameters as delay parameters. Based on the matching results of the delay parameters of the scheduling task and multiple candidate running intervals, it is determined whether the computing server has scheduling capabilities. When scheduling capability is determined, a scheduling task is issued to the computing server; when scheduling capability is determined not to be determined, a task redirection operation is performed.

2. The adaptive adjustment method for computing power scheduling based on strategy optimization according to claim 1, characterized in that, The construction of multiple candidate running intervals includes: Obtain the real-time operating power and reference temperature of the computing server, and set the reference temperature as the first temperature; adjust the real-time operating power to the backup power in the backup parameters, calculate the second temperature under simulated operation, and calculate the deviation value between the first temperature and the second temperature; generate multiple simulated temperature values ​​based on the first temperature and the deviation value, and determine the multiple simulated temperature values ​​as multiple verification parameters.

3. The adaptive adjustment method for computing power scheduling based on strategy optimization according to claim 1, characterized in that, The task processing status includes running status, idle status, and delayed status; When the task processing status is running, extract the backup temperature from the backup parameters, and generate multiple simulated temperature values ​​based on the time-series fluctuation range of the backup temperature, which serve as multiple verification parameters. When the task processing status is idle, the computing server is marked as the scheduling server, the current time point is recorded as the base time, the second temperature reached under the backup power in the backup parameters is obtained, and multiple simulated temperature values ​​are generated in combination with the time-series fluctuation range of the second temperature as the multiple verification parameters. When the task processing status is delayed, the allowable fluctuation range of the reference temperature is obtained, the reference temperature is extracted, and multiple simulated temperature values ​​are generated by combining the time-series fluctuation range of the reference temperature as multiple verification parameters. The delayed period is set as the default candidate running interval, and risk warning information is generated at the same time.

4. The adaptive adjustment method for computing power scheduling based on strategy optimization according to claim 1, characterized in that, The multiple verification parameters generated based on the backup parameters are used to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals, including: Set the reference temperature as the reference value and set the allowable fluctuation range corresponding to the reference temperature; sort multiple verification parameters in chronological order and set them as calculation temperatures, calculate the difference between the calculation temperature and the reference temperature in turn until the difference exceeds the allowable fluctuation range; determine the time point before the difference exceeds the allowable fluctuation range as the verification execution time; subtract the verification execution time from the reference time to generate multiple candidate operating ranges.

5. The adaptive adjustment method for computing power scheduling based on strategy optimization according to claim 1, characterized in that, The step of determining whether the computing server has scheduling capabilities based on the matching results of the scheduling task's delay parameters and multiple candidate running intervals includes: Based on the power required by the scheduling task, a test point is selected from multiple candidate running intervals; the verification time point is determined in conjunction with the current scheduling time, and the verification time point is compared with the termination time points of multiple candidate running intervals; if the verification time point is within multiple candidate running intervals, the computing power server is determined to have scheduling capability.

6. The adaptive adjustment method for computing power scheduling based on strategy optimization according to claim 1, characterized in that, The step of performing task redirection when scheduling capability is determined to be unavailable includes: Collect backup servers for historical tasks with similar scheduling parameters that were interrupted when the task was interrupted; set the backup servers as the target servers for the current scheduled task to be redirected.

7. A computing power scheduling adaptive adjustment system based on strategy optimization, characterized in that, Includes the following modules: The status awareness module is used to collect the current operating status parameters of each computing server; and to determine the task processing status of the computing server based on the current operating status parameters of each computing server. The parameter extraction module is used to extract backup parameters corresponding to the task processing status. The interval construction module is used to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals based on multiple verification parameters generated from the backup parameters. The scheduling decision module is used to obtain the power required for each scheduling task, and calculate the power standby parameters by combining the historical task standby time and the equivalent execution time; the power standby parameters are then determined as delay parameters. Based on the matching results of the delay parameters of the scheduling task and multiple candidate running intervals, it is determined whether the computing server has scheduling capabilities. The task execution module is used to issue scheduling tasks to the computing server when it is determined that scheduling capability is available; and to perform task redirection operations when it is determined that scheduling capability is not available.

8. The adaptive adjustment system for computing power scheduling based on strategy optimization according to claim 7, characterized in that, The multiple verification parameters generated based on the backup parameters are used to calibrate the verification execution time of the computing power server and construct multiple candidate running intervals, including: Set the reference temperature as the reference value and set the allowable fluctuation range corresponding to the reference temperature; sort multiple verification parameters in chronological order and set them as calculation temperatures, calculate the difference between the calculation temperature and the reference temperature in turn until the difference exceeds the allowable fluctuation range; determine the time point before the difference exceeds the allowable fluctuation range as the verification execution time; subtract the verification execution time from the reference time to generate multiple candidate operating ranges.

9. The adaptive adjustment system for computing power scheduling based on strategy optimization according to claim 7, characterized in that, The step of determining whether the computing server has scheduling capabilities based on the matching results of the scheduling task's delay parameters and multiple candidate running intervals includes: Based on the power required by the scheduling task, a test point is selected from multiple candidate running intervals; the verification time point is determined in conjunction with the current scheduling time, and the verification time point is compared with the termination time points of multiple candidate running intervals; if the verification time point is within multiple candidate running intervals, the computing power server is determined to have scheduling capability.

10. The adaptive adjustment system for computing power scheduling based on strategy optimization according to claim 7, characterized in that, The step of performing task redirection when scheduling capability is determined to be unavailable includes: Collect backup servers for historical tasks with similar scheduling parameters that were interrupted when the task was interrupted; set the backup servers as the target servers for the current scheduled task to be redirected.

Images