An intelligent center energy efficiency optimization system and method

Abstract

The application relates to the technical field of data processing, in particular to a wisdom calculation center energy efficiency optimization system and method. The system comprises a data acquisition module which acquires a plurality of historical calculation resource usage data and historical temperature change graphs; a prediction module which analyzes the change trend of any historical calculation resource usage data to predict calculation resource allocation data; a calibration module which calibrates the plurality of predicted calculation resource allocation data based on the comparison result of the predicted calculation resource allocation data and the total actual calculation resource; a data collection module which collects any actual resource utilization rate; a temperature collection module which collects any actual temperature change graph and compares the actual temperature change graph with the historical temperature change graph to determine whether the actual temperature change graph is abnormal; and a resource adjustment module which adjusts the calibrated calculation resource allocation data or adjusts the historical period based on the comparison result of the actual resource utilization rate of any load and the preset resource utilization rate and the abnormal analysis result of the actual temperature change graph. The application improves the effect of wisdom calculation center energy efficiency optimization.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an intelligent computing center energy efficiency optimization system and method. Background Art

[0002] Intelligent computing centers are the primary computing power generators of the intelligent era. Leveraging converged computing systems and data resources, they leverage powerful computing power to drive AI models for deep data processing, continuously generating a variety of intelligent computing services and providing them to organizations and individuals via the internet as cloud services. With the rapid development of big data, cloud computing, and artificial intelligence technologies, energy efficiency management has become a significant challenge for intelligent computing centers, core data processing and computing facilities.

[0003] The patent document of Chinese patent application publication number CN117539726A discloses a method and system for optimizing energy efficiency of green intelligent computing center. The method includes: based on the historical energy consumption records of the intelligent computing center, using association rule mining and cluster analysis algorithms to analyze energy consumption patterns and generate energy consumption analysis reports; based on the energy consumption analysis reports, using time series analysis and neural network algorithms to predict the future load demand of the intelligent computing center and generate a load prediction model; based on the load prediction model, using constraint-based optimization algorithms and genetic algorithms to dynamically allocate computing resources and generate resource scheduling strategies; based on the resource scheduling strategies, using decision tree algorithms and heuristic algorithms to select data. According to the processing path, a data processing path plan is generated; based on the data processing path plan, polling and minimum connection load balancing algorithms are used to perform task allocation and generate energy-efficiency optimized load distribution; based on the energy-efficiency optimized load distribution, deep learning and network traffic analysis algorithms are used to generate optimized data traffic management; based on the optimized data traffic management, energy management and renewable energy integration strategies are adopted to implement energy-saving measures and form a comprehensive energy-saving execution plan; it can be seen that in the existing technology, due to the possible differences between the predicted and actual operating conditions, if there is no timely calibration and adjustment mechanism, it will lead to over-allocation or under-allocation of resources, thereby causing energy waste. Summary of the Invention

[0004] To this end, the present invention provides an intelligent computing center energy efficiency optimization system and method, which can solve the problem of energy waste.

[0005] To achieve the above objectives, the present invention provides an intelligent computing center energy efficiency optimization system, which includes:

[0006] A data acquisition module is used to obtain a number of historical computing resource usage data corresponding to a number of loads in a historical period and a number of historical temperature change graphs corresponding to a number of loads;

[0007] a prediction module, connected to the data acquisition module, for analyzing a change trend of a plurality of the historical computing power resource usage data corresponding to any load, so as to predict computing power resource allocation data within a preset period based on the change trend analysis results and obtain predicted computing power resource allocation data;

[0008] a calibration module, connected to the prediction module, configured to calibrate the plurality of predicted computing power resource allocation data based on a comparison result of the plurality of predicted computing power resource allocation data with the total actual computing power resources, to obtain a plurality of calibrated computing power resource allocation data;

[0009] a data collection module, connected to the prediction module, for collecting actual resource utilization corresponding to any of the loads;

[0010] a temperature acquisition module connected to the data acquisition module, configured to acquire a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, compare the actual temperature change graph with the historical temperature change graph corresponding to the load, determine an abnormal area based on the comparison result, determine whether the actual temperature change graph is abnormal based on a comparison result of the proportion of the abnormal temperature area with the proportion of the preset area, or analyze the change trend of the abnormal temperature area to determine whether the actual temperature change graph is abnormal;

[0011] A resource adjustment module is connected to the data acquisition module and the temperature acquisition module, and is used to determine the adjustment direction of the calibration computing power resource allocation data according to the abnormal situation of the actual temperature change diagram, and to determine the adjustment parameters of the calibration computing power resource allocation data according to the comparison result of any actual resource utilization rate and the preset resource utilization rate, or to adjust the historical period.

[0012] Furthermore, the prediction module includes:

[0013] a graph obtaining unit, configured to obtain a plurality of historical CPU usage rates and a plurality of historical memory usage rates corresponding to any of the loads, and plot the plurality of historical CPU usage rates and the plurality of historical memory usage rates in the same coordinate system based on a time series to obtain a historical resource change graph;

[0014] a trend analysis unit connected to the graph acquisition unit, configured to analyze the historical resource change graph, identify an ascending segment, a descending segment, and a stable segment based on the analysis results, construct piecewise function models based on the ascending segment, the descending segment, and the stable segment, and determine an actual prediction model based on the plurality of piecewise function models;

[0015] A prediction unit is connected to the trend analysis unit and is used to predict the predicted CPU usage and predicted memory occupancy of the load within a preset period according to the actual prediction model.

[0016] Furthermore, the calibration module includes:

[0017] a computing power comparison unit, configured to calculate a total predicted computing power resource based on the plurality of predicted computing power resource allocation data, and compare the total predicted computing power resource with the total actual computing power resource to obtain a comparison result;

[0018] A calibration unit is connected to the computing power comparison unit and is used to determine whether to calibrate the predicted computing power resource allocation data according to the processing priority of several of the loads based on the comparison result of the total predicted computing power resources and the total actual computing power resources, or to evenly distribute the remaining computing power resources to several of the loads.

[0019] Furthermore, the calibration unit includes:

[0020] a priority determination subunit, configured to determine the processing priority based on the order of the plurality of predicted CPU usage rates and the plurality of predicted memory occupancy rates;

[0021] A calibration subunit is connected to the priority determination subunit, and is used to determine a calibration strategy based on the processing priority and the difference between the total predicted computing power resources and the total actual computing power resources, so as to calibrate the predicted CPU usage and the predicted memory occupancy to obtain a calibrated CPU usage and a calibrated memory occupancy.

[0022] Furthermore, the data acquisition module includes:

[0023] A collection unit, configured to collect actual CPU usage and actual memory usage corresponding to any of the loads;

[0024] A calculation unit is connected to the collection unit and is used to calculate the actual resource utilization based on the actual CPU usage and the calibrated CPU usage, the actual memory occupancy and the calibrated memory occupancy.

[0025] Furthermore, the temperature acquisition module includes:

[0026] a trend comparison unit, configured to compare the actual temperature change curve in the actual temperature change graph with the historical temperature change curve in the historical temperature change graph, and identify different areas between the actual temperature change curve and the historical temperature change curve based on the comparison result to obtain an abnormal area;

[0027] An abnormality judgment unit is connected to the trend comparison unit and is used to calculate the proportion of the abnormal area in the actual temperature change curve as the abnormal area proportion, and compare it with the preset area proportion to judge whether the actual temperature change graph is abnormal based on the comparison result, or analyze the change trend of the temperature abnormal area to judge whether the actual temperature change graph is abnormal.

[0028] Furthermore, the resource adjustment module includes:

[0029] a temperature anomaly adjustment unit, configured to determine whether the actual temperature change graph is a temperature increase anomaly or a temperature decrease anomaly, and to determine whether to adjust the calibration computing power resource allocation data to increase or decrease the adjustment according to the temperature increase anomaly or the temperature decrease anomaly;

[0030] The utilization rate adjustment unit is configured to compare the actual resource utilization rate with a preset resource utilization rate, determine an adjustment parameter based on the utilization rate comparison result, or adjust the historical period.

[0031] Furthermore, the abnormality judgment unit includes:

[0032] The length calculation subunit is used to calculate the length of the curve corresponding to the temperature anomaly area;

[0033] The trend judgment subunit is used to draw a length change graph based on the lengths of several curves, judge the change trend corresponding to the length change graph, and judge whether the actual temperature change graph is abnormal.

[0034] Furthermore, the utilization rate adjustment unit includes:

[0035] a period adjustment subunit, configured to increase the historical period when the actual resource utilization is less than or equal to the preset resource utilization;

[0036] a difference calculation subunit, configured to calculate a difference between the actual resource utilization and the preset resource utilization when the actual resource utilization is greater than the preset resource utilization, to obtain a utilization difference;

[0037] The period adjustment subunit is connected to the difference calculation subunit and is used to calculate an adjustment coefficient based on the utilization difference and the preset resource utilization.

[0038] On the other hand, the present invention also provides a method for optimizing the energy efficiency of an intelligent computing center, the method comprising:

[0039] Obtaining historical computing resource usage data corresponding to several loads within a historical period and historical temperature change graphs corresponding to several loads;

[0040] Analyze the change trend of the historical computing resource usage data corresponding to any load, and predict the computing resource allocation data within a preset period based on the change trend analysis results to obtain predicted computing resource allocation data;

[0041] Calibrate the plurality of predicted computing resource allocation data based on a comparison result of the plurality of predicted computing resource allocation data with the total actual computing resource to obtain a plurality of calibrated computing resource allocation data;

[0042] Collect the actual resource utilization corresponding to any load;

[0043] collecting a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, comparing the actual temperature change graph with the historical temperature change graph corresponding to the load, determining an abnormal area based on the comparison result, judging whether the actual temperature change graph is abnormal based on a comparison result of the proportion of the abnormal temperature area with the proportion of the preset area, or analyzing a change trend of the abnormal temperature area to judge whether the actual temperature change graph is abnormal;

[0044] Determine the adjustment direction of the calibration computing power resource allocation data based on the abnormal situation of the actual temperature change graph, and determine the adjustment parameters of the calibration computing power resource allocation data based on the comparison result of any of the actual resource utilization and the preset resource utilization, or adjust the historical period.

[0045] Compared with the prior art, the beneficial effects of the present invention are that, by setting the data acquisition module to provide comprehensive historical computing resource usage data and temperature change diagrams, a reliable basis is provided for prediction and calibration; by setting the prediction module to predict computing resource allocation data within a preset period, a basis is provided for resource scheduling and optimization; computing resources are effectively allocated through prediction, resource idleness or overuse is avoided, thereby improving resource utilization efficiency; by setting the calibration module to calibrate the comparison results of the predicted computing resource allocation data with the total actual computing resource, the accuracy of the predicted data is improved; the calibration process takes into account the limitations of the actual computing resources, so that the prediction results are more in line with the actual situation, the adaptability of the system is enhanced, and it is helpful to achieve further optimal utilization of resources; by setting the data acquisition module to collect the actual data of the load in real time Source utilization provides timely information support for resource adjustment. Setting up a temperature acquisition module helps to detect temperature anomalies in a timely manner and prevent equipment failure or performance degradation due to overheating. By analyzing the changing trend of the temperature anomaly area, the abnormal temperature change can be judged more accurately, providing an important basis for subsequent adjustment and optimization. By setting up a resource adjustment module, according to the abnormal conditions of the actual temperature change graph and the comparison results of the actual resource utilization rate with the preset resource utilization rate, the adjustment direction and adjustment parameters of the calibrated computing power resource allocation data are determined. According to real-time data and abnormal conditions, the allocation of computing power resources is dynamically adjusted to ensure the energy efficiency and stability of the system. At the same time, by adjusting the historical cycle, it can adapt to the changes in load more flexibly, further improve energy efficiency, and improve resource utilization efficiency, providing strong support for the energy efficiency optimization of the intelligent computing center.

[0046] In particular, by setting the graphic acquisition unit to obtain the historical CPU usage and memory occupancy of the load, and plotting them in the same coordinate system, the changing trend of resource usage over time can be intuitively observed. By setting the trend analysis unit to analyze these historical data, the rising segment, the falling segment and the stable segment can be identified, which helps to capture the dynamic characteristics of resource usage more accurately. The piecewise function model constructed based on the segment can be closer to the actual changes in resource usage, thereby improving the accuracy of the prediction. By setting the prediction unit to predict the CPU usage and memory occupancy of the load within a preset period according to the actual prediction model, resources can be effectively allocated to avoid waste caused by excessive resource allocation or performance degradation caused by insufficient resources.

[0047] In particular, setting up the computing power comparison unit helps to identify the difference between the predicted computing power resource allocation data and the actual resource demand, providing basic data for subsequent calibration. By setting up the calibration unit, an allocation strategy is formulated to ensure that the load can fully utilize the remaining resources, improve resource utilization efficiency, ensure that the predicted computing power resource allocation data matches the actual resource demand, avoid resource waste, and at the same time improve the energy efficiency and operation efficiency of the system.

[0048] In particular, by setting up a trend comparison unit to compare the actual temperature change curve with the historical temperature change curve, the different areas between the two, that is, potential temperature anomaly areas, are accurately identified, which helps to timely discover subtle differences in temperature changes and provide an accurate basis for subsequent judgment and adjustment. By setting the abnormal judgment unit to calculate the proportion of abnormal areas in the actual temperature change curve, that is, the abnormal area proportion, the severity of the temperature anomaly can be quantitatively evaluated. According to the comparison result of the abnormal area proportion and the preset area proportion, it is judged whether the actual temperature change graph is abnormal. The abnormal situation is further confirmed by analyzing the change trend of the temperature anomaly area, which helps to improve the accuracy and reliability of the judgment and help to ensure the stable operation and energy efficiency optimization of the intelligent computing center. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A structural block diagram of the energy efficiency optimization system for the intelligent computing center provided by an embodiment of the present invention;

[0050] Figure 2 A structural block diagram of a prediction module in the intelligent computing center energy efficiency optimization system provided by an embodiment of the present invention;

[0051] Figure 3 This is a structural block diagram of the data acquisition module in the intelligent computing center energy efficiency optimization system provided by an embodiment of the present invention;

[0052] Figure 4 A flow chart of the method for optimizing energy efficiency of an intelligent computing center provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.

[0054] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0055] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be understood as a limitation on the present invention.

[0056] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0057] See also Figure 1 As shown, an embodiment of the present invention provides an intelligent computing center energy efficiency optimization system, which includes:

[0058] A data acquisition module 10 is used to acquire a plurality of historical computing resource usage data corresponding to a plurality of loads in a historical period and a plurality of historical temperature change graphs corresponding to the plurality of loads;

[0059] The prediction module 20 is connected to the data acquisition module 10 and is used to analyze the change trend of the historical computing resource usage data corresponding to any load, so as to predict the computing resource allocation data within a preset period based on the change trend analysis results and obtain the predicted computing resource allocation data;

[0060] a calibration module 30 connected to the prediction module 20, configured to calibrate the plurality of predicted computing power resource allocation data based on a comparison result of the plurality of predicted computing power resource allocation data with the total actual computing power resources, thereby obtaining a plurality of calibrated computing power resource allocation data;

[0061] a data collection module 40 connected to the prediction module 20 for collecting actual resource utilization corresponding to any of the loads;

[0062] a temperature acquisition module 50 connected to the data acquisition module 40 and configured to acquire a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, compare the actual temperature change graph with the historical temperature change graph corresponding to the load, determine an abnormal area based on the comparison result, determine whether the actual temperature change graph is abnormal based on a comparison result of the proportion of the abnormal temperature area with the proportion of the preset area, or analyze the change trend of the abnormal temperature area to determine whether the actual temperature change graph is abnormal;

[0063] The resource adjustment module 60 is connected to the data acquisition module 40 and the temperature acquisition module 50, and is used to determine the adjustment direction of the calibration computing power resource allocation data according to the abnormal situation of the actual temperature change diagram, and to determine the adjustment parameters of the calibration computing power resource allocation data according to the comparison result of any actual resource utilization rate and the preset resource utilization rate, or to adjust the historical period.

[0064] Specifically, the embodiment of the present invention provides comprehensive historical computing power resource usage data and temperature change diagrams by setting the data acquisition module, providing a reliable basis for prediction and calibration, and provides a basis for resource scheduling and optimization by setting the prediction module to predict computing power resource allocation data within a preset period. By predicting, computing power resources are effectively allocated to avoid idle or excessive use of resources, thereby improving resource utilization efficiency. By setting the calibration module to calibrate the comparison results of the predicted computing power resource allocation data with the total actual computing power resources, the accuracy of the predicted data is improved. The calibration process takes into account the limitations of the actual computing power resources, so that the prediction results are more in line with the actual situation, enhances the adaptability of the system, and helps to achieve further optimal utilization of resources. By setting the data acquisition module to collect the actual resource utilization of the load in real time , providing timely information support for resource adjustment. Setting up a temperature acquisition module helps to detect temperature anomalies in a timely manner and prevent equipment failure or performance degradation due to overheating. By analyzing the changing trend of the temperature anomaly area, the abnormal temperature change can be judged more accurately, providing an important basis for subsequent adjustment and optimization. By setting up a resource adjustment module, according to the abnormal conditions of the actual temperature change graph and the comparison results of the actual resource utilization rate with the preset resource utilization rate, the adjustment direction and adjustment parameters of the calibrated computing power resource allocation data are determined. According to real-time data and abnormal conditions, the allocation of computing power resources is dynamically adjusted to ensure the energy efficiency and stability of the system. At the same time, by adjusting the historical cycle, it can adapt to the changes in load more flexibly, further improve energy efficiency, improve resource utilization efficiency, and provide strong support for the energy efficiency optimization of the intelligent computing center.

[0065] It is understandable that in the embodiment of the present invention, the preset period and the historical period have the same time length. For example, if the historical period is the entire month of November last year, then the preset period is the entire month of November this year.

[0066] See also Figure 2 As shown, the prediction module 20 includes:

[0067] A graph obtaining unit 21 is configured to obtain a plurality of historical CPU usage rates and a plurality of historical memory usage rates corresponding to any of the loads, and plot the plurality of historical CPU usage rates and the plurality of historical memory usage rates in the same coordinate system based on a time series to obtain a historical resource change graph;

[0068] a trend analysis unit 22 connected to the graph acquisition unit 21, configured to analyze the historical resource change graph, identify an ascending segment, a descending segment, and a stable segment based on the analysis results, construct piecewise function models based on the ascending segment, the descending segment, and the stable segment, and determine an actual prediction model based on the plurality of piecewise function models;

[0069] The prediction unit 23 is connected to the trend analysis unit 22 and is used to predict the predicted CPU usage and the predicted memory occupancy of the load within a preset period according to the actual prediction model.

[0070] Specifically, the embodiment of the present invention obtains the historical CPU usage and memory occupancy of the load by setting the graphic acquisition unit, and plots them in the same coordinate system, so as to intuitively observe the changing trend of resource usage over time. By setting the trend analysis unit to analyze these historical data, the rising segment, the falling segment and the stable segment are identified, which helps to more accurately capture the dynamic characteristics of resource usage. The piecewise function model constructed based on the segment can be closer to the actual changes in resource usage, thereby improving the accuracy of the prediction. By setting the prediction unit to predict the CPU usage and memory occupancy of the load within a preset period according to the actual prediction model, resources can be effectively allocated to avoid waste caused by excessive resource allocation or performance degradation caused by insufficient resources.

[0071] It can be understood that the trend analysis unit described in the embodiment of the present invention determines the rising segment, the falling segment and the stable segment by calculating the change rate of the CPU usage and the memory occupancy between adjacent time points to identify the changing trend of the resource usage. The change rate can be calculated by dividing the difference between adjacent data points by the time interval. According to the statistical characteristics of the historical data, the threshold of the change rate is set. When the change rate exceeds a certain positive threshold, the resource usage is considered to be in the rising segment; when the change rate is lower than a certain negative threshold, the resource usage is considered to be in the falling segment; when the change rate is between the positive and negative thresholds, the resource usage is considered to be in the stable segment. The entire time series data is traversed to continuously detect whether the change rate exceeds or is lower than the set threshold, thereby dividing the rising segment, the falling segment and the stable segment. This is a prior art and will not be repeated here.

[0072] It can be understood that, in the embodiment of the present invention, the positive threshold value can be set to the mean value of the plurality of change threshold values ​​+ 2× the standard deviation of the plurality of change threshold values;

[0073] The negative threshold may be set to the mean of the plurality of change thresholds - 2× the standard deviation of the plurality of change thresholds.

[0074] It can be understood that the trend analysis unit described in the embodiment of the present invention selects a suitable function type to describe the changing trend of resource utilization rate according to the characteristics of the rising segment, the falling segment and the stable segment. For example, linear functions, exponential functions or logarithmic functions can be used in the rising segment and the falling segment; constant functions or random fluctuation functions can be used in the stable segment. Among them, the parameters of the piecewise function model are estimated by least squares method, nonlinear regression and other methods, so that it can accurately describe the trend characteristics in the historical resource change graph. This is the existing technology and will not be repeated here.

[0075] It can be understood that the determination of the actual prediction model in the trend analysis unit of the embodiment of the present invention can be based on the use of cross-validation, leave-one-out method and other methods to verify the prediction performance of several piecewise function models to ensure that they can accurately predict the resource utilization rate in the future period of time, and select the optimal piecewise function model as the actual prediction model based on the verification results.

[0076] In a specific example, assume that the intelligent computing center serves an e-commerce company. This company experiences peak business during the annual "Double 11" shopping festival, while business volume is relatively stable during normal times. The graph acquisition unit collects hourly historical CPU usage and memory utilization data for this workload over the past 11 months. This data covers data points from daily operations, promotional events, and special circumstances (such as brief periods of fluctuation after system upgrades). Then, based on a time series, this historical CPU usage and memory utilization data are plotted on the same coordinate system, forming a historical resource change graph. This graph shows distinct patterns across time periods. For example, during "Double 11," CPU usage and memory utilization experience significant peaks, while remaining at lower levels during low-traffic times such as the early morning hours. Rising period: In the hours before the promotional event, as users begin browsing products and adding items to their carts, and tasks such as order processing and inventory queries increase, CPU usage and memory utilization begin to gradually rise. In particular, between midnight and 2 a.m. on "Double 11," the data exhibits a rapid upward trend, as a large number of orders simultaneously enter the system. Stable Segment: During normal, non-promotional periods, such as weekday afternoons and evenings, order volume is relatively stable, and CPU and memory usage remain relatively stable with minimal fluctuations. Declining Segment: During the period after a promotion ends, as order processing gradually completes, the system load decreases, and CPU and memory usage also decrease. Based on these identified segments, a piecewise function model is constructed. For example, an exponential function model might be used to describe the rapidly growing portion of the rising segment; a constant function or a simple linear function (accommodating small fluctuations) might be used for the stable segment; and an appropriate decreasing function model is selected for the declining segment based on the speed and pattern of the decline. Combining these piecewise function models, an actual forecasting model is determined. The forecasting unit uses this actual forecasting model to predict the CPU and memory usage for this workload during the following year's "Double 11" shopping festival. According to the model, from midnight to 2:00 a.m. on "Double 11," CPU usage is expected to exceed 90%, and memory usage will approach system limits. This is due to the simultaneous processing of a large number of orders and complex data analysis tasks (such as real-time sales data statistics and inventory adjustment analysis). In the hours before a promotion, CPU usage and memory occupancy will gradually rise from daily levels to an intermediate value, such as CPU usage from 30% to 60%, and memory usage from 40% to 70%. The allocation of computing resources will be adjusted in real time based on the predicted results.

[0077] Specifically, the calibration module 30 includes:

[0078] a computing power comparison unit, configured to calculate a total predicted computing power resource based on the plurality of predicted computing power resource allocation data, and compare the total predicted computing power resource with the total actual computing power resource to obtain a comparison result;

[0079] A calibration unit is connected to the computing power comparison unit and is used to calibrate the predicted computing power resource allocation data based on the processing priorities of the several loads when the total predicted computing power resources are greater than the total actual computing power resources, and to evenly distribute the remaining computing power resources to the several loads when the total predicted computing power resources are less than the total actual computing power resources. The calibration of the predicted computing power resource allocation data is determined based on the processing priorities of the several loads according to the comparison result of the total predicted computing power resources and the total actual computing power resources, or to evenly distribute the remaining computing power resources to the several loads.

[0080] It can be understood that the calibration unit of the embodiment of the present invention calibrates the predicted computing power resource allocation data based on the processing priorities of the several loads when the total predicted computing power resources are greater than the total actual computing power resources, and evenly distributes the remaining computing power resources to the several loads when the total predicted computing power resources are less than the total actual computing power resources.

[0081] It is understandable that the total predicted computing power resources described in the embodiment of the present invention are the total amount of computing power currently actually available to the system.

[0082] Specifically, the embodiment of the present invention helps to identify the difference between the predicted computing power resource allocation data and the actual resource demand by setting the computing power comparison unit, providing basic data for subsequent calibration, and formulating an allocation strategy by setting the calibration unit to ensure that the load can fully utilize the remaining resources, improve resource utilization efficiency, ensure that the predicted computing power resource allocation data matches the actual resource demand, avoid resource waste, and at the same time improve the energy efficiency and operation efficiency of the system.

[0083] Specifically, the calibration unit includes:

[0084] a priority determination subunit, configured to determine the processing priority based on the descending order of the plurality of predicted CPU usage rates and the plurality of predicted memory occupancy rates;

[0085] A calibration subunit is connected to the priority determination subunit, and is used to determine a calibration strategy based on the processing priority and the difference between the total predicted computing power resources and the total actual computing power resources, so as to calibrate the predicted CPU usage and the predicted memory occupancy to obtain a calibrated CPU usage and a calibrated memory occupancy.

[0086] It can be understood that the priority determination subunit described in the embodiment of the present invention sorts the several predicted CPU usage rates and the several predicted memory occupancy rates from large to small, and determines their sorting numbers based on the sorting results. For example, the sorting number of the largest predicted CPU usage rate among the several predicted CPU usage rates is 1. Similarly, the sum of the sorting numbers corresponding to the sorting results of the predicted CPU usage rates and the predicted memory occupancy rates is calculated, and the priorities of the several loads are determined from high to low based on the sum of the sorting numbers from small to large.

[0087] It will be appreciated that the calibration subunit described in this embodiment of the present invention sorts all loads according to their determined priorities, with the lowest-priority load first. Starting with the lowest-priority load, its predicted computing power resource allocation data is reduced by a certain percentage. The reduction percentage can be determined based on the extent to which the total predicted computing power resources exceed the total actual computing power resources. For example, if the excess is small, a smaller percentage (e.g., 5% to 10%) can be reduced; if the excess is large, a larger percentage (e.g., 20% to 30%) can be reduced. During the reduction process, the relationship between the total predicted computing power resources and the total actual computing power resources is continuously checked until the total predicted computing power resources do not exceed the total actual computing power resources. During the resource reduction process, if the predicted resources of a low-priority load are reduced to a certain level that seriously affects its basic functions (e.g., causing frequent errors or inability to complete tasks), the reduction of that load is suspended and consideration is given to continuing with the next lowest priority load. Among them, the larger or smaller amount of excess can be determined by comparing the difference between the total predicted computing power resources and the total actual computing power resources with the preset difference. When the difference is less than the preset difference, the excess is smaller. When the difference is greater than or equal to the preset difference, the excess is larger. The preset difference can be 1 / 20 of the total actual computing power resources.

[0088] See also Figure 3 As shown, the data acquisition module 40 includes:

[0089] A collection unit 41 is used to collect the actual CPU usage and actual memory occupancy corresponding to any of the loads;

[0090] The calculation unit 42 is connected to the collection unit 41 and is used to calculate the actual resource utilization based on the actual CPU usage and the calibrated CPU usage, the actual memory occupancy and the calibrated memory occupancy.

[0091] Specifically, the embodiment of the present invention sets a calculation unit to calculate the actual resource utilization based on the collected data, and provides the resource adjustment unit with accurate load resource usage information to facilitate effective resource optimization. By calculating the actual resource utilization and comparing it with the preset resource utilization, resource allocation is adjusted according to actual needs, thereby improving the energy efficiency and operating efficiency of the system, enabling the system to dynamically adjust resource allocation according to real-time changes in the load, thereby ensuring the rationality of resource utilization and the stability of the system.

[0092] It can be understood that the actual resource utilization described in the embodiment of the present invention=actual CPU usage / calibrated CPU usage+actual memory occupancy / calibrated memory occupancy.

[0093] Specifically, the temperature acquisition module 50 includes:

[0094] a trend comparison unit, configured to compare the actual temperature change curve in the actual temperature change graph with the historical temperature change curve in the historical temperature change graph, identify different areas between the actual temperature change curve and the historical temperature change curve based on the comparison result, and mark them on the actual temperature change curve to obtain abnormal areas;

[0095] An abnormality judgment unit is connected to the trend comparison unit and is used to calculate the proportion of the abnormal area in the actual temperature change curve as the abnormal area proportion, and compare it with the preset area proportion to judge whether the actual temperature change graph is abnormal based on the comparison result, or analyze the change trend of the temperature abnormal area to judge whether the actual temperature change graph is abnormal.

[0096] It can be understood that the abnormality judgment unit described in the embodiment of the present invention judges that the actual temperature change graph is abnormal when the abnormal area proportion is greater than the preset area proportion, and analyzes the change trend of the temperature abnormal area when the abnormal area proportion is greater than 0 and less than the preset area proportion.

[0097] Specifically, the embodiment of the present invention sets a trend comparison unit to compare the actual temperature change curve with the historical temperature change curve, accurately identifies the different areas between the two, that is, potential temperature anomaly areas, which helps to timely discover subtle differences in temperature changes and provide an accurate basis for subsequent judgment and adjustment. By setting the abnormal judgment unit to calculate the proportion of abnormal areas in the actual temperature change curve, that is, the abnormal area proportion, the severity of the temperature anomaly can be quantitatively evaluated. According to the comparison result of the abnormal area proportion and the preset area proportion, it is judged whether the actual temperature change graph is abnormal. The abnormal situation is further confirmed by analyzing the change trend of the temperature anomaly area, which helps to improve the accuracy and reliability of the judgment and help to ensure the stable operation and energy efficiency optimization of the intelligent computing center.

[0098] It can be understood that, in the present invention, the preferred proportion of the preset area is 1 / 5.

[0099] Specifically, the resource adjustment module 60 includes:

[0100] a temperature anomaly adjustment unit, configured to determine whether the actual temperature change graph is a temperature increase anomaly or a temperature decrease anomaly, and to determine whether to adjust the calibration computing power resource allocation data to increase or decrease the adjustment according to the temperature increase anomaly or the temperature decrease anomaly;

[0101] The utilization rate adjustment unit is configured to compare the actual resource utilization rate with a preset resource utilization rate, determine an adjustment parameter based on the utilization rate comparison result, or adjust the historical period.

[0102] Specifically, the embodiment of the present invention can adjust the calibration CPU usage and the calibration memory occupancy according to the comparison results of the actual resource utilization and the temperature change trend by setting a resource adjustment unit, thereby realizing dynamic optimization allocation of resources and improving the energy efficiency and operation efficiency of the system. By setting a period adjustment unit, when the actual temperature change graph is normal and the actual resource utilization is greater than or equal to the preset resource utilization, the historical period is adjusted to collect historical data more comprehensively.

[0103] It can be understood that the preset resource utilization rate in the embodiment of the present invention is 75%.

[0104] Specifically, the abnormality judgment unit includes:

[0105] The length calculation subunit is used to calculate the length of the curve corresponding to the temperature anomaly area;

[0106] The trend judgment subunit is used to draw a length change graph based on the lengths of several curves, judge the change trend corresponding to the length change graph, and judge whether the actual temperature change graph is abnormal.

[0107] It can be understood that, in the embodiment of the present invention, when the change trend corresponding to the length change graph increases, the actual temperature change graph is abnormal; in other cases, the actual temperature change graph is normal.

[0108] Specifically, the utilization rate adjustment unit includes:

[0109] a period adjustment subunit, configured to increase the historical period when the actual resource utilization is less than or equal to the preset resource utilization;

[0110] a difference calculation subunit, configured to calculate a difference between the actual resource utilization and the preset resource utilization when the actual resource utilization is greater than the preset resource utilization, to obtain a utilization difference;

[0111] The period adjustment subunit is connected to the difference calculation subunit and is used to calculate an adjustment coefficient based on the utilization difference and the preset resource utilization.

[0112] It can be understood that in an embodiment of the present invention, the historical period can be increased by calculating an increase coefficient based on the actual resource utilization and the preset resource utilization to increase the historical period, where the increase coefficient is (1+actual resource utilization divided by the preset resource utilization).

[0113] It can be understood that the adjustment coefficient in the embodiment of the present invention = utilization difference / preset resource utilization.

[0114] See also Figure 4 As shown, an embodiment of the present invention further provides a method for optimizing an energy efficiency system of an intelligent computing center, the method comprising:

[0115] Step S100, obtaining a number of historical computing resource usage data corresponding to a number of loads within a historical period and a number of historical temperature change graphs corresponding to the number of loads;

[0116] Step S200: analyzing a change trend of a plurality of the historical computing resource usage data corresponding to any load, and predicting computing resource allocation data within a preset period based on the change trend analysis results to obtain predicted computing resource allocation data;

[0117] Step S300: calibrating the plurality of predicted computing resource allocation data based on a comparison result between the plurality of predicted computing resource allocation data and the total actual computing resource to obtain a plurality of calibrated computing resource allocation data;

[0118] Step S400, collecting the actual resource utilization corresponding to any load;

[0119] Step S500: collecting a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, comparing the actual temperature change graph with the historical temperature change graph corresponding to the load, determining an abnormal area based on the comparison result, determining whether the actual temperature change graph is abnormal based on a comparison result of the ratio of the abnormal temperature area with the ratio of the preset area, or analyzing a change trend of the abnormal temperature area to determine whether the actual temperature change graph is abnormal;

[0120] Step S600, determining the adjustment direction of the calibration computing power resource allocation data according to the abnormal situation of the actual temperature change diagram, and determining the adjustment parameters of the calibration computing power resource allocation data according to the comparison result of any of the actual resource utilization and the preset resource utilization, or adjusting the historical period.

[0121] Specifically, the method of the intelligent computing center energy efficiency optimization system provided in the embodiment of the present invention can be applied to the above-mentioned intelligent computing center energy efficiency optimization system to achieve the same technical effect, which will not be repeated here.

[0122] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

[0123] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. An intelligent computing center energy efficiency optimization system, characterized in that: include: A data acquisition module is used to obtain a number of historical computing resource usage data corresponding to a number of loads in a historical period and a number of historical temperature change graphs corresponding to a number of loads, wherein the historical computing resource usage data includes historical CPU usage and a number of historical memory usage rates; a prediction module, connected to the data acquisition module, for analyzing a change trend of a plurality of the historical computing power resource usage data corresponding to any load, so as to predict computing power resource allocation data within a preset period based on the change trend analysis results, and obtain predicted computing power resource allocation data, wherein the predicted computing power resource allocation data includes a predicted CPU usage rate and a predicted memory occupancy rate; a calibration module, connected to the prediction module, configured to calibrate the plurality of predicted computing power resource allocation data based on a comparison result of the plurality of predicted computing power resource allocation data with the total actual computing power resources, to obtain a plurality of calibrated computing power resource allocation data, wherein the calibrated computing power resource allocation data includes a calibrated CPU usage rate and a calibrated memory occupancy rate; a data collection module, connected to the calibration module, for collecting actual resource usage data corresponding to any of the loads, and calculating actual resource utilization based on the actual resource usage data and the calibration computing power resource allocation data, wherein the actual resource usage data includes actual CPU usage and actual memory occupancy; a temperature acquisition module connected to the data acquisition module, configured to acquire a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, compare the actual temperature change graph with the historical temperature change graph corresponding to the load, determine an abnormal area based on the comparison result, determine whether the actual temperature change graph is abnormal based on a comparison result of the proportion of the abnormal temperature area with the proportion of the preset area, or analyze the change trend of the abnormal temperature area to determine whether the actual temperature change graph is abnormal; a resource adjustment module, connected to the data acquisition module and the temperature acquisition module, for determining an adjustment direction for adjusting the calibrated computing power resource allocation data based on an abnormality in the actual temperature change graph, and determining an adjustment parameter for adjusting the calibrated computing power resource allocation data based on a comparison result between any of the actual resource utilization rates and a preset resource utilization rate, or adjusting the historical period; The prediction module includes: a graph obtaining unit, configured to obtain a plurality of historical CPU usage rates and a plurality of historical memory usage rates corresponding to any of the loads, and plot the plurality of historical CPU usage rates and the plurality of historical memory usage rates in the same coordinate system based on a time series to obtain a historical resource change graph; a trend analysis unit connected to the graph acquisition unit, configured to analyze the historical resource change graph to identify an ascending segment, a descending segment, and a stable segment, construct piecewise function models based on the ascending segment, the descending segment, and the stable segment, and determine an actual prediction model based on the plurality of piecewise function models; a prediction unit, connected to the trend analysis unit, for predicting a predicted CPU usage rate and a predicted memory occupancy rate of the load within a preset period according to the actual prediction model; The calibration module includes: a computing power comparison unit, configured to calculate a total predicted computing power resource based on the plurality of predicted computing power resource allocation data, and compare the total predicted computing power resource with the total actual computing power resource to obtain a comparison result; a calibration unit connected to the computing power comparison unit, configured to determine, based on a comparison result of the total predicted computing power resources and the total actual computing power resources, whether to calibrate the predicted computing power resource allocation data according to the processing priorities of the plurality of loads, or to evenly distribute the remaining computing power resources to the plurality of loads; The calibration unit comprises: a priority determination subunit, configured to determine the processing priority based on the order of the plurality of predicted CPU usage rates and the plurality of predicted memory occupancy rates; A calibration subunit is connected to the priority determination subunit, and is used to determine a calibration strategy based on the processing priority and the difference between the total predicted computing power resources and the total actual computing power resources, so as to calibrate the predicted CPU usage and the predicted memory occupancy to obtain a calibrated CPU usage and a calibrated memory occupancy.

2. The intelligent computing center energy efficiency optimization system according to claim 1, characterized in that: The data acquisition module includes: A collection unit, configured to collect actual CPU usage and actual memory usage corresponding to any of the loads; A calculation unit is connected to the collection unit and is used to calculate the actual resource utilization based on the actual CPU usage and the calibrated CPU usage, the actual memory occupancy and the calibrated memory occupancy.

3. The intelligent computing center energy efficiency optimization system according to claim 2, characterized in that: The temperature acquisition module includes: a trend comparison unit, configured to compare the actual temperature change curve in the actual temperature change graph with the historical temperature change curve in the historical temperature change graph, and identify different areas between the actual temperature change curve and the historical temperature change curve based on the comparison result to obtain an abnormal area; An abnormality judgment unit is connected to the trend comparison unit and is used to calculate the proportion of the abnormal area in the actual temperature change curve as the abnormal area proportion, and compare it with the preset area proportion to judge whether the actual temperature change graph is abnormal based on the comparison result, or analyze the change trend of the temperature abnormal area to judge whether the actual temperature change graph is abnormal.

4. The intelligent computing center energy efficiency optimization system according to claim 3, characterized in that: The resource adjustment module includes: a temperature anomaly adjustment unit, configured to determine whether the actual temperature change graph is a temperature increase anomaly or a temperature decrease anomaly, and to determine whether to adjust the calibration computing power resource allocation data to increase or decrease the adjustment according to the temperature increase anomaly or the temperature decrease anomaly; The utilization rate adjustment unit is configured to compare the actual resource utilization rate with a preset resource utilization rate, determine an adjustment parameter based on the utilization rate comparison result, or adjust the historical period.

5. The intelligent computing center energy efficiency optimization system according to claim 4, characterized in that: The abnormality judgment unit includes: The length calculation subunit is used to calculate the length of the curve corresponding to the temperature anomaly area; The trend judgment subunit is used to draw a length change graph based on the lengths of several curves, judge the change trend corresponding to the length change graph, and judge whether the actual temperature change graph is abnormal.

6. The intelligent computing center energy efficiency optimization system according to claim 5, characterized in that: The utilization rate adjustment unit includes: a period adjustment subunit, configured to increase the historical period when the actual resource utilization is less than or equal to the preset resource utilization; a difference calculation subunit, configured to calculate a difference between the actual resource utilization and the preset resource utilization when the actual resource utilization is greater than the preset resource utilization, to obtain a utilization difference; The period adjustment subunit is connected to the difference calculation subunit and is used to calculate an adjustment coefficient based on the utilization difference and the preset resource utilization.

7. An energy efficiency optimization method applied to the energy efficiency optimization system of the intelligent computing center according to any one of claims 1 to 6, characterized in that: include: Obtaining historical computing resource usage data corresponding to several loads within a historical period and historical temperature change graphs corresponding to several loads; Analyze the change trend of the historical computing resource usage data corresponding to any load, and predict the computing resource allocation data within a preset period based on the change trend analysis results to obtain predicted computing resource allocation data; Calibrate the plurality of predicted computing resource allocation data based on a comparison result of the plurality of predicted computing resource allocation data with the total actual computing resource to obtain a plurality of calibrated computing resource allocation data; Collecting actual resource usage data corresponding to any of the loads, and calculating actual resource utilization based on the actual resource usage data and the calibrated computing power resource allocation data; collecting a plurality of real-time temperature values ​​corresponding to any of the loads in real time within a preset period to draw an actual temperature change graph, comparing the actual temperature change graph with the historical temperature change graph corresponding to the load, determining an abnormal area based on the comparison result, judging whether the actual temperature change graph is abnormal based on a comparison result of the proportion of the abnormal temperature area with the proportion of the preset area, or analyzing a change trend of the abnormal temperature area to judge whether the actual temperature change graph is abnormal; Determine the adjustment direction of the calibration computing power resource allocation data based on the abnormal situation of the actual temperature change graph, and determine the adjustment parameters of the calibration computing power resource allocation data based on the comparison result of any of the actual resource utilization and the preset resource utilization, or adjust the historical period.

Images