A cloud-computing-based computer room energy-saving control method and system

Abstract

This invention relates to the technical field of data center energy management, and discloses a cloud computing-based energy-saving control method and system for data centers. It utilizes a multi-dimensional sensor network to collect real-time equipment operation data, establishes a power consumption prediction model and an energy allocation optimization model, extracts key features using principal component analysis, constructs a multivariate regression model to determine the energy efficiency coefficient, and employs a long short-term memory neural network to predict future power demand. A genetic algorithm is used to optimize the energy allocation scheme, achieving precise power control and dynamic balance. This invention also establishes an adaptive energy management decision-making mechanism, analyzing energy consumption trends through a sliding window, initiating emergency adjustment procedures, and using a Kalman filter algorithm to correct state estimates and dynamically adjust prediction model parameters. This method enables intelligent management of data center energy utilization, effectively reducing overall energy costs, improving energy efficiency, and providing technical support for energy conservation and emission reduction in large data centers.

Description

Technical Field

[0001] This invention relates to the technical field of data center energy consumption management, and in particular to a cloud computing-based data center energy-saving control method and system. Background Technology

[0002] With the rapid development of technologies such as cloud computing and artificial intelligence, the power consumption of data center equipment is increasing exponentially. Energy costs account for a significant proportion of data center operating costs, making precise and efficient energy consumption management technologies urgently needed. Current mainstream data center energy management solutions mainly rely on traditional monitoring systems and static energy allocation strategies, which are inadequate when faced with complex and ever-changing equipment operating conditions.

[0003] Meanwhile, the diverse types of equipment in a data center environment and their complex and variable operating states make it difficult for traditional monitoring methods to construct a complete digital representation of equipment energy consumption status. This results in an information gap between energy consumption data and the actual equipment status. This lack of digital representation further restricts the ability to deeply analyze equipment power variation patterns and energy conversion efficiency, making it difficult for managers to accurately grasp the true energy consumption characteristics and trends of different equipment under various operating conditions. This leads to the adoption of a crude, static allocation model, failing to dynamically optimize energy configuration based on real-time operating status, resulting in significant energy waste.

[0004] Therefore, existing technologies generally face problems such as insufficient data collection, lack of depth in energy consumption analysis, and delayed decision response, which makes it difficult to achieve refined management of data center energy consumption. Summary of the Invention

[0005] This invention provides a cloud computing-based energy-saving control method and system for data centers, enabling refined management of data center energy consumption.

[0006] Firstly, in order to solve the above-mentioned technical problems, the present invention provides a cloud computing-based data center energy-saving control method, comprising:

[0007] Acquire real-time data and historical power consumption patterns of data center equipment, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain a matrix of equipment operating parameters.

[0008] The equipment load is obtained according to the equipment operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the equipment is obtained according to the comparison result, and the power demand of each equipment is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained and the optimal power allocation scheme of each equipment is determined according to the power consumption prediction matrix.

[0009] A real-time energy scheduling system is established based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and continuously monitor it. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision mechanism is obtained.

[0010] Preferably, the step of acquiring real-time data and historical power consumption patterns of the data center equipment, establishing a basic data set including power peak, average, and fluctuation amplitude, and obtaining a device operating parameter matrix includes:

[0011] The power consumption, temperature distribution and load status of the equipment in the computer room are collected in real time through a multi-dimensional sensor network to obtain raw data streams covering multiple operating conditions. The raw data streams are then stored in a pre-established database in a time series format to obtain structured equipment operation records.

[0012] Based on the structured equipment operation records, the historical change trajectory is segmented, and the power peak, average power and fluctuation amplitude in each time period are calculated to form the first basic data.

[0013] If the fluctuation amplitude exceeds the preset fluctuation threshold, the first basic data is marked as abnormal, and the operating parameters under abnormal conditions are obtained by combining the load status and temperature distribution data.

[0014] The basic data set marked with anomalies is compared with the operating condition coverage requirements, and the equipment operating parameter matrix is ​​organized for subsequent monitoring.

[0015] Preferably, the step of obtaining the device load based on the device operating parameter matrix and comparing it with a preset load threshold, obtaining the energy conversion efficiency coefficient of the device based on the comparison result, predicting and calculating the power demand of each device, and if the predicted power exceeds the current power supply capacity, obtaining a power consumption prediction matrix and determining the optimal power allocation scheme for each device based on the power consumption prediction matrix includes:

[0016] Based on the equipment operating parameter matrix, key feature vectors are extracted, and a multivariate regression model between equipment status and energy consumption level is constructed. If the equipment load rate exceeds the preset load threshold, it is marked as a high energy consumption state. The energy conversion efficiency coefficient of various types of equipment under different load conditions is determined.

[0017] The power consumption prediction model of the device is established by the energy conversion efficiency coefficient and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the load balancing mechanism is triggered to obtain a power consumption prediction matrix that includes time and device dimensions.

[0018] An energy allocation optimization model is constructed based on the power consumption prediction matrix. The objective function is set as minimizing the overall energy consumption cost. The constraints include the minimum operating power requirement of the equipment and the upper limit of power supply. The optimal power allocation scheme for each device is determined by calculation.

[0019] Preferably, the step of extracting key feature vectors based on the equipment operating parameter matrix, constructing a multiple regression model between equipment status and energy consumption level, and marking the equipment load rate as a high-energy-consuming state if it exceeds a preset load threshold, and determining the energy conversion efficiency coefficients of various types of equipment under different load conditions, includes:

[0020] The power changes and operating parameters in the equipment operating parameter matrix are standardized and organized to obtain the data content corresponding to the load. The second basic dataset is then merged and compared with the load threshold.

[0021] If the energy consumption exceeds the limit, the corresponding device status will be marked as the high energy consumption state, and the marked status data group will be determined.

[0022] Data under different load conditions are processed in layers to obtain the energy consumption distribution corresponding to the device status, and a set of classified efficiency data is obtained.

[0023] The efficiency data set is correlated with the power change to obtain the energy conversion efficiency coefficient.

[0024] Preferably, the step of establishing a device power consumption prediction model based on the energy conversion efficiency coefficient and predicting the power demand of each device, triggering a load balancing mechanism if the predicted power exceeds the current power supply capacity, yields a power consumption prediction matrix that includes time and device dimensions, including:

[0025] Based on the energy conversion efficiency coefficient, and combining the time and equipment dimensions, the historical operation records are classified and organized to obtain the feature content related to power demand, thus obtaining the third basic data set.

[0026] By combining the third basic dataset with the temporal characteristics of the data, the power demand for future periods is analyzed and processed to obtain the predicted power demand distribution results.

[0027] Based on the power demand distribution results and combined with the power supply capacity limit, data verification is performed. If the limit is exceeded, the load balancing mechanism is triggered to obtain the adjusted power allocation strategy.

[0028] Based on the power allocation strategy, and combining the time dimension and the device dimension, a prediction matrix containing multi-dimensional information is constructed to obtain the power consumption prediction matrix.

[0029] Preferably, the step of constructing an energy allocation optimization model based on the power consumption prediction matrix, setting the objective function as minimizing the overall energy consumption cost, and the constraints including the minimum operating power requirements of the equipment and the upper limit of power supply, and determining the optimal power allocation scheme for each device through calculation, includes:

[0030] Based on the power consumption prediction matrix, the predicted power data of each device is classified and processed to obtain the power demand distribution information related to the operation of the device, and a classified power demand set is obtained.

[0031] Based on the power demand set combined with the preset supply limit and power requirements, the matching of the power demand of each device with the constraints is verified. If the predicted power of one device exceeds the power supply limit, the power allocation ratio of that device is adjusted, and the adjusted power demand list is determined.

[0032] Using the power demand list, combined with the objective function and the energy consumption cost data, the power allocation scheme for each device is processed to obtain a preliminary allocation result that minimizes cost.

[0033] Based on the preliminary allocation results and the actual operating requirements of the equipment, an optimized power allocation scheme for each device is generated, resulting in the optimal power allocation scheme.

[0034] Preferably, the step of establishing a real-time energy dispatching system based on the optimal power allocation scheme, and adjusting the operating parameters of the equipment in the computer room to obtain the energy configuration status if the actual power consumption of one of the devices exceeds a preset range, includes:

[0035] The system continuously collects the actual power consumption of each device, determines whether the actual power consumption exceeds the preset range, and if so, automatically adjusts the operating parameters. In combination with the requirements of intelligent management, it generates an adjusted parameter configuration scheme and determines a temporary operating mode suitable for each device.

[0036] Based on the adjusted parameter configuration scheme, each device is precisely allocated, the energy configuration data feedback is updated in real time, and a power allocation list that matches the current device status is obtained.

[0037] By comparing the power allocation list with the real-time monitoring data, the power control strategy for each device is determined, and the energy configuration status is obtained.

[0038] Preferably, the mechanism for updating energy allocation parameters and obtaining an adaptive energy consumption management decision-making mechanism upon detecting abnormal fluctuations includes:

[0039] If abnormal fluctuations are detected, the trend of energy consumption changes is analyzed, and in combination with the preset threshold range, it is determined whether the conditions for initiating emergency regulation are met, and the specific state of the triggering conditions is determined.

[0040] Based on the specific state of the triggering condition, the allocation strategy information that matches the current fluctuation characteristics is extracted from the historical experience database, the adjusted parameter scheme is generated, the updated configuration data is obtained and fed back to the monitoring process, it is determined whether the preset stability standard has been reached, and the final operating state is obtained.

[0041] Preferably, the method further includes:

[0042] A closed-loop feedback control system is established based on the adaptive energy consumption management decision-making mechanism. The deviation between actual energy consumption and expected targets is compared, the weight parameters of the prediction model are dynamically adjusted, and a continuously optimized energy management and control scheme is determined.

[0043] Specifically:

[0044] According to the energy consumption management decision mechanism, actual energy consumption data is obtained, compared with the expected target, and it is determined whether it exceeds the preset threshold range to obtain the initial deviation result. If it exceeds the threshold, real-time information is collected and combined with historical operating data to obtain the adjusted status record.

[0045] Based on the adjusted status records, a matching scheme is extracted from the pre-established database to determine the updated weight value;

[0046] The energy management scheme is dynamically adjusted using the updated weight values. To achieve continuous optimization, the adjusted scheme data is fed back to the closed-loop process to determine whether it meets the preset conditions and obtain the final configuration result.

[0047] Secondly, a cloud computing-based data center energy-saving control system includes:

[0048] The detection end is used to acquire real-time data and historical power consumption changes of the equipment in the computer room, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain the equipment operating parameter matrix.

[0049] On the processing end, the device load is obtained according to the device operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the device is obtained according to the comparison result, and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained, and the optimal power allocation scheme of each device is determined according to the power consumption prediction matrix.

[0050] The decision-making end is used to establish a real-time energy scheduling system based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and perform continuous monitoring. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision-making mechanism is obtained.

[0051] Compared with existing technologies, this invention provides a cloud computing-based data center energy-saving control method and system, which has the following beneficial effects:

[0052] 1. This invention utilizes a multi-dimensional sensor network to collect real-time data on power consumption, temperature distribution, and load status of data center equipment, covering raw data streams under various operating conditions. This multi-dimensional, multi-condition data acquisition method, compared to traditional single-data acquisition methods, can more comprehensively construct a digital representation of the equipment's energy consumption status. For example, traditional methods may only collect power consumption data, while this invention also incorporates information such as temperature and load status, making data acquisition more comprehensive. This solves the information gap problem caused by insufficient data acquisition in existing technologies, achieving a complete digital representation of the energy consumption status of data center equipment and providing a comprehensive data foundation for subsequent accurate energy consumption analysis and management.

[0053] 2. This invention establishes a basic dataset including peak power, average power, and fluctuation amplitude to obtain a device operating parameter matrix. Then, it extracts key feature vectors from this matrix and constructs a multiple regression model between device status and energy consumption level. This approach allows for in-depth analysis of the energy conversion efficiency coefficient of the device under different load conditions. For example, traditional methods may simply calculate the average power consumption of the device, while this invention, through a multiple regression model, can uncover the complex relationship between device load rate and energy consumption, thereby determining the energy conversion efficiency of the device under various load conditions. This solves the problem of existing technologies' difficulty in deeply analyzing the power variation patterns and energy conversion efficiency of devices, providing strong support for precise energy consumption management.

[0054] 3. This invention establishes a real-time energy dispatching system. If the actual power consumption of one device exceeds a preset range, the system adjusts the operating parameters of the equipment in the data center and continuously monitors the system. Furthermore, by constructing an adaptive energy management decision-making mechanism, energy allocation parameters can be updated promptly when abnormal fluctuations are detected. For example, when the actual power consumption of a device exceeds the preset range, the system can react quickly and adjust its operating parameters, unlike traditional methods which require a longer time to detect and handle abnormal energy consumption. This effectively solves the problem of delayed decision-making response in existing technologies, improves the speed of decision-making response, and enables timely adjustment of energy allocation based on the actual operating status of the equipment, reducing energy waste. Attached Figure Description

[0055] Figure 1 This is a flowchart of a cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0056] Figure 2 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0057] Figure 3 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0058] Figure 4 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0059] Figure 5 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0060] Figure 6 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0061] Figure 7 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0062] Figure 8 This is a flowchart of another cloud computing-based data center energy-saving control method provided in an embodiment of the present invention;

[0063] Figure 9 This is a flowchart of another cloud computing-based data center energy-saving control method provided in this embodiment of the invention.

[0064] Figure 10 This is a schematic diagram of a cloud computing-based energy-saving control system for a data center, provided in an embodiment of the present invention. Detailed Implementation

[0065] 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.

[0066] Reference Figure 1 The first embodiment of the present invention provides a flowchart of a cloud computing-based data center energy-saving control method, including the following steps:

[0067] S1, acquire real-time data and historical power consumption changes of the equipment in the computer room, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain the equipment operating parameter matrix;

[0068] S2, obtain the equipment load according to the equipment operating parameter matrix and compare it with the preset load threshold. Obtain the energy conversion efficiency coefficient of the equipment according to the comparison result and predict the power demand of each equipment. If the predicted power exceeds the current power supply capacity, obtain the power consumption prediction matrix and determine the optimal power allocation scheme for each equipment according to the power consumption prediction matrix.

[0069] S3. Establish a real-time energy scheduling system based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, adjust the operating parameters of the equipment in the computer room, obtain the energy configuration status and continuously monitor it. If abnormal fluctuations are found, update the energy allocation parameters and obtain an adaptive energy consumption management decision mechanism.

[0070] In step S1, real-time data and historical power consumption change trajectories of the equipment in the computer room are obtained, a basic data set including power peak, average, and fluctuation amplitude is established, and the equipment operation parameter matrix is ​​obtained.

[0071] refer to Figure 2 Step S1 includes:

[0072] S11, Real-time acquisition of power consumption, temperature distribution and load status of the equipment in the computer room is carried out through a multi-dimensional sensor network to obtain raw data streams covering multiple operating conditions, and the raw data streams are stored in a pre-established database in a time series format to obtain structured equipment operation records;

[0073] S12, based on the structured equipment operation records, the historical change trajectory is segmented, and the power peak, average power and fluctuation amplitude in each time period are calculated to form the first basic data;

[0074] S13, If the fluctuation amplitude exceeds the preset fluctuation threshold, the first basic data is marked as abnormal, and the operating parameters under abnormal conditions are obtained by combining the load status and temperature distribution data.

[0075] S14, compare the basic data set after anomaly marking with the operating condition coverage requirements, and organize the equipment operating parameter matrix for subsequent monitoring.

[0076] For example, in data center equipment monitoring scenarios, real-time data acquisition of equipment power, temperature, and load status via a multi-dimensional sensor network is a core component. Sensors can be deployed next to each critical piece of equipment within the data center, recording data streams in real time, such as collecting power, temperature, and load percentage data every minute, forming time-series data stored in a database. This approach ensures data continuity and integrity, providing a reliable foundation for subsequent analysis.

[0077] Specifically, for segmented processing of historical data, a day's operational data can be divided into 24 time periods by hour, calculating the peak power, average power, and fluctuation range within each time period. For example, if a device's peak power is 5000 watts and the average power is 4500 watts between 9:00 AM and 10:00 AM, with a fluctuation range of 10%, and a preset fluctuation range threshold of 8%, the data for that time period will be marked as abnormal. This segmented analysis helps to quickly locate problematic time periods and improves troubleshooting efficiency.

[0078] For example, after anomaly identification, further analysis of potential risks can be conducted by combining load status and temperature distribution data. Suppose that power fluctuations are abnormal during a certain period, while the load rate reaches 90%, and the temperature distribution shows that the core area of ​​the equipment reaches 50 degrees Celsius, far exceeding the normal range of 40 degrees Celsius, then it can be determined that there is a risk of overload or poor heat dissipation. Through this multi-dimensional data correlation analysis, the root cause of the problem can be identified more accurately, preventing equipment damage.

[0079] Specifically, when constructing a comprehensive matrix, multiple parameters such as power, temperature, and load can be organized into a data matrix. For example, each row of the matrix represents a point in time, and each column corresponds to a parameter, such as the first column being the power value and the second column being the temperature value. Through this structured organization, the correlation between data becomes immediately apparent, facilitating the training of subsequent monitoring models and anomaly prediction, and significantly improving the intelligence level of the monitoring system.

[0080] For example, during the comparison of operating condition coverage requirements, if data for certain extreme operating conditions is found to be missing, such as operating parameters under high temperature and high load, targeted test data can be supplemented to ensure that the complete dataset covers all possible scenarios. This approach can effectively reduce monitoring blind spots, improve the system's adaptability to complex operating conditions, and provide a guarantee for the safe and stable operation of data center equipment.

[0081] Specifically, the beneficial effects of implementing the above methods include improved data analysis efficiency, reduced equipment failure rates, and optimized resource allocation. Through real-time data acquisition and structured storage, data processing time can be reduced by more than 30%; through anomaly labeling and risk assessment, the early warning rate of equipment failures can be increased to 90%; and through a comprehensive matrix and complete datasets, the monitoring system can achieve 95% coverage of complex operating conditions, providing strong technical support for data center management.

[0082] In step S2, the device load is obtained according to the device operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the device is obtained according to the comparison result, and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained, and the optimal power allocation scheme of each device is determined according to the power consumption prediction matrix.

[0083] In the field of data center equipment, power consumption prediction matrices are models or tools used to analyze and predict the power consumption of equipment or systems. Through multi-dimensional data correlation and algorithmic modeling, they help managers optimize energy use, plan capacity, and reduce operating costs.

[0084] refer to Figure 3 Step S2 includes:

[0085] S21. Based on the equipment operating parameter matrix, extract key feature vectors, construct a multivariate regression model between equipment status and energy consumption level, and mark the equipment load rate as high energy consumption state if it exceeds the preset load threshold, and determine the energy conversion efficiency coefficient of various types of equipment under different load conditions.

[0086] S22, A device power consumption prediction model is established through the energy conversion efficiency coefficient, and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the load balancing mechanism is triggered to obtain a power consumption prediction matrix that includes time and device dimensions.

[0087] S23. Construct an energy allocation optimization model based on the power consumption prediction matrix, set the objective function as minimizing the overall energy consumption cost, and set the constraints as the minimum operating power requirement of the equipment and the upper limit of power supply. Determine the optimal power allocation scheme for each equipment through calculation.

[0088] refer to Figure 4 S21 includes:

[0089] S211, standardize and organize the power changes and operating parameters in the equipment operating parameter matrix, obtain the data content corresponding to the load, and merge the second basic dataset with the load threshold for data comparison.

[0090] S212, if the condition is exceeded, the corresponding device status is marked as the high energy consumption state, and the marked status data group is determined;

[0091] S213, perform hierarchical processing on the data under different load conditions, obtain the energy consumption distribution corresponding to the device status, and obtain the classified efficiency data set;

[0092] S214, The efficiency data set is correlated with the power change to obtain the energy conversion efficiency coefficient.

[0093] For example, in the scenario of monitoring equipment in a data center, the processing and analysis of the equipment operating parameter matrix can be carried out from multiple dimensions. By combining key data such as power changes and load conditions, the comprehensiveness and accuracy of data processing can be ensured.

[0094] For example, standardizing multidimensional data involves formatting collected data such as power, temperature, and load rate. Assuming a server in a data center has power data in watts, load rate as a percentage, and temperature in degrees Celsius, standardization converts these data into a unified numerical range, facilitating subsequent comparison and analysis. This approach eliminates dimensional differences and ensures the comparability of different data types.

[0095] For example, in the comparison of load rate values ​​with threshold settings, the load rate threshold can be set to 85%. If a device's load rate reaches 88% within a certain time period, the system will automatically mark its status as high-energy-consuming. This marking mechanism helps to quickly identify potentially high-risk devices, providing a basis for subsequent management.

[0096] For example, hierarchical processing of status data groups can be used to further subdivide the energy consumption distribution under different load conditions. Suppose that under high energy consumption conditions (load rate above 80%), the energy consumption distribution of a certain device shows that its main energy consumption is concentrated in the processor module, while under load rates below 50%, the energy consumption distribution is more even. This classification analysis can clearly reveal the energy consumption characteristics of the device under different operating conditions, providing a reference for optimizing operating strategies.

[0097] For example, in processing efficiency datasets, when correlating power changes with feature vectors, the relationship between power fluctuations and parameters such as load rate and temperature can be analyzed. Suppose a device experiences frequent power fluctuations at high load rates, and its temperature also rises simultaneously; then it can be inferred that power changes may be related to insufficient heat dissipation. This comprehensive state-to-state correspondence helps to gain a deeper understanding of the device's operating patterns.

[0098] For example, to determine the final reference data set, the above analysis results can be integrated into a structured dataset containing information such as the power, energy consumption distribution, and efficiency characteristics of the device under different load conditions. Suppose a server has an energy efficiency of only 70% at 90% load, but reaches 85% efficiency at 60% load. Then, its operating load can be adjusted based on the reference data set to improve overall efficiency.

[0099] For example, in practical implementation, the above methods can be further optimized from a data visualization perspective, using charts to display the correlation between load rate and power changes, intuitively presenting the equipment status. This approach helps managers quickly grasp the equipment's operating status and improve decision-making efficiency. Through the above multi-dimensional analysis and processing, the level of precision in data center equipment monitoring is improved, data utilization is significantly increased, and reliable support is provided for equipment management.

[0100] refer to Figure 5 S22 includes:

[0101] S221, Based on the energy conversion efficiency coefficient, and combining the time dimension and the equipment dimension, the historical operation records are classified and organized to obtain the feature content related to power demand, and a third basic data set is obtained.

[0102] S222, By combining the third basic dataset with the data time series characteristics, the power demand for future periods is analyzed and processed to obtain the predicted power demand distribution results;

[0103] S223, perform data verification based on the power demand distribution results and power supply capacity limits. If the limits are exceeded, trigger the load balancing mechanism to obtain the adjusted power allocation strategy.

[0104] S224. Based on the power allocation strategy, and combining the time dimension and the device dimension, a prediction matrix containing multi-dimensional information is constructed to obtain the power consumption prediction matrix.

[0105] For example, in the scenario of monitoring data center equipment, the application of energy conversion efficiency coefficient (ECC) can be analyzed by starting with historical operating records and combining information from both time and equipment dimensions. By categorizing and organizing historical operating records, features related to power demand can be extracted. Suppose that in the past month's operating data for a certain server, the ECC was 0.75 during peak hours and 0.85 during off-peak hours. Combining this with time-dimensional data reveals that efficiency fluctuations are closely related to the operating time period. Classifying this data by time period and equipment type forms an initial dataset, laying the foundation for subsequent predictions.

[0106] For example, when forecasting power demand based on an initial dataset and its time-series characteristics, the distribution of power demand in future periods is analyzed. Suppose that during the upcoming holiday peak season, based on historical data, the power demand of a server cluster is predicted to reach 5000 watts, an increase of 20% compared to normal times. This forecasting process comprehensively considers historical operating patterns and time-related characteristics to ensure that the results closely reflect actual demand.

[0107] Assuming the total power supply capacity of the data center is limited to 8000 watts, while the predicted power demand is 8500 watts, exceeding the preset threshold by 500 watts, the system will automatically trigger a load balancing mechanism to adjust the power allocation strategy, transferring some of the load to backup equipment or operating during off-peak hours to avoid the risk of insufficient power supply.

[0108] For example, to generate power allocation strategies, a multi-dimensional prediction matrix is ​​constructed by combining time and device-level information. Assuming that after adjustment, a server's load during peak hours is limited to 70%, with a power allocation of 3000 watts, while some tasks are scheduled for off-peak hours in the early morning, the generated prediction matrix not only includes power data but also covers device operating status and time distribution information, providing managers with a comprehensive reference.

[0109] For example, the data content can be further refined during the formation of the final power consumption prediction data. Suppose the prediction matrix shows that a device's average power consumption over the next week is 2500 watts, with potential short-term peaks during specific periods. By comparing this data with historical records, countermeasures can be developed in advance, such as adjusting equipment operating schedules or increasing cooling support, thereby ensuring stable equipment operation. This integration of multi-dimensional information helps improve the accuracy and practicality of predictions, providing strong support for data center management.

[0110] refer to Figure 6 S23 includes:

[0111] S231, Based on the power consumption prediction matrix, the predicted power data of each device is classified and processed to obtain the power demand distribution information related to the operation of the device, and a classified power demand set is obtained.

[0112] S232, Based on the power demand set combined with the preset supply limit and power requirements, verify the matching of the power demand of each device with the constraints. If the predicted power of one device exceeds the power supply limit, adjust the power allocation ratio of that device and determine the adjusted power demand list.

[0113] S233, using the power demand list, combined with the objective function and the energy consumption cost data, process the power allocation scheme of each device to obtain a preliminary allocation result that minimizes cost;

[0114] S234. Based on the preliminary allocation results and the actual operating requirements of the equipment, an optimized power allocation scheme for each device is generated to obtain the optimal power allocation scheme.

[0115] For example, in the scenario of data center equipment management, the application of power consumption prediction matrices involves classifying the predicted power data. Suppose a data center contains multiple server clusters, and the prediction matrix shows that some devices have higher power demands during specific time periods, while other devices have relatively stable power requirements. This data can be classified by device type and time period to form power demand distribution information, resulting in a categorized power demand set. For example, the predicted power demand for server A might be 4000 watts during peak periods and 2000 watts during off-peak periods.

[0116] For example, the verification process for power demand sets analyzes whether the power demands of each device meet the constraints, based on preset supply limits and power requirements. Suppose the total supply limit of the data center is 10,000 watts, and a server cluster's predicted peak power reaches 11,000 watts, exceeding the limit by 1,000 watts. In this case, the system will identify the over-limit devices and adjust their power allocation ratio, for example, reducing the cluster's power allocation ratio from 100% to 90%, thus generating an adjusted power demand list to ensure that the total power demand is controlled within the supply limit.

[0117] For example, in the process of optimizing power allocation schemes based on power demand lists, the power allocation schemes for each device are processed by combining the objective function and energy cost data. Assuming the objective function is cost minimization-oriented, if the operating cost of a certain device is higher during peak hours and lower during off-peak hours, the system will tend to shift some load to off-peak hours. The initial allocation results might show that server B's power allocation during peak hours is adjusted from 3000 watts to 2000 watts, with the remaining load scheduled for the early morning hours.

[0118] For example, further optimization of the initial allocation results, combined with the actual operational needs of the equipment, generates final energy allocation data. Suppose server C's power is initially limited to 2500 watts, but its actual operational needs require at least 2800 watts. The system will adjust the allocation ratio of other devices according to priority to ensure that server C's operational needs are met. The final generated energy allocation data not only includes optimized power schemes for each device but also time distribution information, providing managers with a comprehensive reference.

[0119] For example, in the above process, each step of classification, verification, adjustment, and optimized allocation is closely integrated with the actual operating characteristics of the data center equipment to ensure that the power allocation scheme meets both supply constraints and equipment needs. This approach helps improve resource utilization efficiency, reduce operating costs, and ensure stable equipment operation during peak periods, providing reliable support for data center management.

[0120] In step S3, a real-time energy scheduling system is established based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and perform continuous monitoring. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision mechanism is obtained.

[0121] In the field of data center equipment, energy configuration status refers to a comprehensive description of the distribution, usage, and real-time operating status of electrical energy within a data center or server room. It involves the deployment and monitoring of key elements such as power systems, power distribution units, equipment load, and backup power, aiming to ensure that energy efficiently, stably, and securely supports the operation of data center equipment.

[0122] refer to Figure 7 S3 includes:

[0123] S31, continuously collect the actual power consumption of each device, determine whether the actual power consumption exceeds the preset range, if so, automatically adjust the operating parameters, and generate an adjusted parameter configuration scheme in combination with the requirements of intelligent management;

[0124] S32, according to the adjusted parameter configuration scheme, accurately allocate power to each device, update the energy configuration data feedback in real time, and obtain a power allocation list that matches the current device status;

[0125] S33, compare the power allocation list with the real-time monitoring data to determine the power control strategy for each device and obtain the energy configuration status.

[0126] For example, in data center equipment management scenarios, to meet the need for real-time monitoring, the power consumption information of each device is continuously acquired. Suppose a data center contains multiple server clusters, and the actual power consumption data of each device is recorded every minute. If it is found that the power consumption of a certain server suddenly increases from the expected 3000 watts to 3800 watts during a certain period, while the preset tolerance range is ±10%, meaning 3300 watts is the upper limit, then the system determines that the device's power consumption exceeds the tolerance range and generates a preliminary judgment result. This method can promptly capture abnormal states and provide a basis for subsequent adjustments.

[0127] Assuming the server experiences abnormal power consumption due to excessive load, the system analyzes device status data and decides to reduce its operating frequency or limit some non-critical tasks, generating an adjusted parameter configuration scheme to keep power consumption below 3200 watts. This temporary operating mode ensures basic device functionality while avoiding the risk of overload.

[0128] For example, in implementing precise power allocation, dynamic power distribution is performed on the equipment based on the adjusted parameter configuration scheme. Assuming the total power supply limit of the data center is 12,000 watts and the current total demand is 11,800 watts, the system updates the energy configuration data in real time, reducing the power of servers exceeding the tolerance from 3,800 watts to 3,200 watts, while simultaneously allocating the saved 600 watts to other equipment with demand, forming a new power allocation list. This dynamic balancing strategy ensures the rational use of overall power resources.

[0129] For example, comparing the power allocation list with real-time monitoring data can further optimize energy configuration. If the comparison reveals that a device's power consumption is still close to its tolerance limit after adjustments, the system will analyze its load trends using historical data to determine whether to further reduce the power allocation ratio or transfer some tasks to other devices, ultimately determining the power control strategy for each device. This approach continuously optimizes the configuration state and adapts to dynamic changes in device operation.

[0130] For example, during the final configuration state generation process under dynamic balancing, the system comprehensively considers the real-time status and historical data of all devices. Suppose a server cluster has an adjusted power allocation of 2500 watts, but monitoring data shows that its workload is about to increase. The system will reserve a 300-watt power buffer in advance to ensure no anomalies occur during peak load periods. This proactive adjustment strategy effectively improves the stability of device operation.

[0131] For example, from an overall perspective, the above-mentioned links are closely integrated with the actual operating characteristics of the data center equipment, forming a complete closed-loop management process from data acquisition to parameter adjustment, and then to power distribution and optimized configuration. This approach can optimize energy efficiency as much as possible while ensuring the normal operation of the equipment, providing reliable support for data center management.

[0132] refer to Figure 8 S3 includes:

[0133] S34. If abnormal fluctuations are detected, analyze the trend of energy consumption changes and, in conjunction with the preset threshold range, determine whether the conditions for initiating emergency regulation are met.

[0134] S35, if satisfied, extract allocation strategy information that matches the current fluctuation characteristics from the historical experience database, generate an adjusted parameter scheme, obtain updated configuration data and feed it back to the monitoring process, determine whether the preset stability standard has been reached, and obtain the final operating state.

[0135] For example, in a data center energy management scenario, to meet the need for continuous monitoring, energy consumption information of each device is extracted in real time. Assuming there are multiple server clusters in the data center, the operating status data of each device is recorded hourly, including key indicators such as voltage, current, and power. Preliminary analysis reveals that the energy consumption of a certain server fluctuates significantly, rising from 2000 watts to 2600 watts within a short period. The system marks this as a potential risk point. This method can quickly identify anomalies, providing a basis for subsequent processing.

[0136] For example, preliminary results of abnormal fluctuations will be used to conduct a more detailed analysis of energy consumption trends. Assuming the preset energy consumption fluctuation threshold range is ±15%, meaning the server's upper limit is 2300 watts, and the current 2600 watts exceeds this range, the system determines that the conditions for initiating emergency adjustments are met. This detailed analysis helps clarify whether fluctuations require intervention, avoiding misjudgments or over-adjustments.

[0137] For example, after determining the state of the triggering conditions, the system extracts allocation strategy information with similar fluctuation characteristics from a historical experience database. Assuming the database records past cases where energy consumption was controlled by reducing the operating frequency under similar circumstances, the system generates an adjustment plan based on this, reducing the server's frequency from full load by 20%, with an expected energy consumption reduction to below 2200 watts. This adaptive decision-making can flexibly respond to actual conditions, ensuring the rationality of the configuration data.

[0138] For example, updated configuration data can dynamically adjust the energy configuration status. Assuming the data center's total power capacity is 10,000 watts and the current total demand is 9,800 watts, the system can reduce the server's power from 2,600 watts to 2,200 watts, leaving 400 watts available for other devices. Simultaneously, the adjustment information is fed back to the monitoring process to determine if a stable standard has been reached. This dynamic adjustment mechanism effectively balances resource allocation.

[0139] For example, during the final confirmation of operational status, the system continuously monitors the effects of adjustments. Assuming the server's power consumption stabilizes at 2150 watts after adjustments, and no new fluctuations occur in the overall data center operation, the system determines that the preset stability standard has been met. This closed-loop feedback mechanism can promptly verify the effects of adjustments, ensuring long-term operational reliability.

[0140] For example, from another perspective, the system can also predict potential risks based on historical data to address complex fluctuations. Suppose a server's energy consumption is stable after adjustments, but historical data shows frequent surges in workload during specific periods; the system will then reserve a 200-watt buffer capacity for it in advance. This proactive strategy further enhances the level of management precision.

[0141] For example, the system also periodically updates the strategy information in the database to meet the objectives of the management mechanism. If a better frequency regulation ratio is accumulated through multiple adjustments, the system will prioritize this strategy and apply it to subsequent fluctuation handling. This continuous optimization approach can continuously improve the adaptability and efficiency of energy allocation.

[0142] refer to Figure 9 The method further includes:

[0143] A closed-loop feedback control system is established based on the adaptive energy consumption management decision-making mechanism. The deviation between actual energy consumption and expected targets is compared, the weight parameters of the prediction model are dynamically adjusted, and a continuously optimized energy management and control scheme is determined.

[0144] Specifically:

[0145] According to the energy consumption management decision mechanism, actual energy consumption data is obtained, compared with the expected target, and it is determined whether it exceeds the preset threshold range to obtain the initial deviation result. If it exceeds the threshold, real-time information is collected and combined with historical operating data to obtain the adjusted status record.

[0146] Based on the adjusted status records, a matching scheme is extracted from the pre-established database to determine the updated weight value;

[0147] The energy management scheme is dynamically adjusted using the updated weight values. To achieve continuous optimization, the adjusted scheme data is fed back to the closed-loop process to determine whether it meets the preset conditions and obtain the final configuration result.

[0148] For example, in the scenario of data center energy management, the implementation of energy consumption management decision-making mechanisms can be achieved by obtaining actual energy consumption data from equipment operation records. Suppose the daily energy consumption records of a server cluster show an average power of 3000 watts, while the preset expected target is 2800 watts, revealing a difference of 200 watts. If the preset threshold range is ±100 watts, the initial deviation result indicates that the range has been exceeded. This comparison method provides a basis for subsequent adjustments.

[0149] For example, in cases where thresholds are exceeded, real-time information on device status is collected. Suppose the server's load suddenly increases; combining this with historical data on energy consumption under similar loads, the system corrects its status estimates, discovering that actual energy consumption may be affected by temporary tasks. The adjusted status record shows the fluctuations as short-term phenomena. This correction helps to more accurately determine whether intervention is needed.

[0150] For example, after obtaining the adjusted status records, the current configuration of the prediction weights is analyzed. Suppose the current weights are biased towards historical averages, but the current deviation indicates a sudden load. The system extracts a matching scheme from the database and decides to adjust the weights to a mode that focuses more on real-time data. The updated weight values ​​better reflect the current state. This dynamic adjustment ensures the relevance of the predictions.

[0151] For example, the energy management scheme can be dynamically adjusted based on the updated weight values. Assuming the total power supply capacity of the data center is 10,000 watts and the current total demand is 9,500 watts, the system will reallocate the excess 200 watts of power, reducing the server's operating frequency, and the expected energy consumption will drop to 2,850 watts. This adjustment method effectively balances the overall resource allocation.

[0152] For example, to achieve continuous optimization, the adjusted solution data is fed back into the closed-loop process. Assuming the server energy consumption remains stable within the target range after adjustment, and the overall data center operation is unaffected, the system determines that the preset conditions are met, and the final configuration result is confirmed. This closed-loop mechanism ensures the reliability of the adjustments.

[0153] For example, from another perspective, the system can also supplement and adjust based on the specific characteristics of energy consumption deviations using historical solutions in the database. For instance, if reducing fan speed could have reduced energy consumption in similar situations in the past, the system can use this as a backup plan, applying it flexibly in conjunction with the current weight values. This multi-solution approach enhances the adaptability of management.

[0154] For example, in actual operation, the process of correcting state estimates can incorporate additional information such as equipment runtime. If a server has been running continuously for longer than expected, the system will further analyze whether its energy consumption deviation is related to hardware aging, and thus adjust the state records. This refined analysis provides a more comprehensive basis for subsequent decision-making.

[0155] Reference Figure 10 This invention provides a cloud computing-based data center energy-saving control system, comprising:

[0156] The detection end is used to acquire real-time data and historical power consumption changes of the equipment in the computer room, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain the equipment operating parameter matrix.

[0157] On the processing end, the device load is obtained according to the device operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the device is obtained according to the comparison result, and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained, and the optimal power allocation scheme of each device is determined according to the power consumption prediction matrix.

[0158] The decision-making end is used to establish a real-time energy scheduling system based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and perform continuous monitoring. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision-making mechanism is obtained.

[0159] It should be noted that the cloud-based data center energy-saving control system provided in this embodiment of the invention is used to execute all the process steps of the cloud-based data center energy-saving control method in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0160] This invention also provides a terminal device. The terminal device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a cloud-based data center energy-saving control program. When the processor executes the computer program, it implements the steps described in the various cloud-based data center energy-saving control method embodiments above, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above system embodiments.

[0161] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0162] The terminal device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The terminal device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of terminal devices and do not constitute a limitation on the terminal device. It may include more or fewer components than described above, or a combination of certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, etc.

[0163] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0164] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0165] Wherein, if the modules / units integrated in the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0166] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0167] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A cloud computing-based energy-saving control method for computer rooms, characterized in that, include: Acquire real-time data and historical power consumption patterns of data center equipment, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain a matrix of equipment operating parameters. The equipment load is obtained according to the equipment operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the equipment is obtained according to the comparison result, and the power demand of each equipment is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained, and the optimal power allocation scheme of each equipment is determined according to the power consumption prediction matrix. A real-time energy scheduling system is established based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and continuously monitor it. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision mechanism is obtained.

2. The cloud computing-based data center energy-saving control method according to claim 1, characterized in that, The process involves acquiring real-time data and historical power consumption patterns of the data center equipment, establishing a basic data set including power peak, average, and fluctuation amplitudes, and obtaining a matrix of equipment operating parameters, including: The power consumption, temperature distribution and load status of the equipment in the computer room are collected in real time through a multi-dimensional sensor network to obtain raw data streams covering multiple operating conditions. The raw data streams are then stored in a pre-established database in a time series format to obtain structured equipment operation records. Based on the structured equipment operation records, the historical change trajectory is segmented, and the power peak, average power and fluctuation amplitude in each time period are calculated to form the first basic data. If the fluctuation amplitude exceeds the preset fluctuation threshold, the first basic data is marked as abnormal, and the operating parameters under abnormal conditions are obtained by combining the load status and temperature distribution data. The basic data set marked with anomalies is compared with the operating condition coverage requirements, and the equipment operating parameter matrix is ​​organized for subsequent monitoring.

3. The cloud computing-based data center energy-saving control method according to claim 1, characterized in that, The process of obtaining the device load based on the device operating parameter matrix and comparing it with a preset load threshold, obtaining the energy conversion efficiency coefficient of the device based on the comparison result, predicting and calculating the power demand of each device, and if the predicted power exceeds the current power supply capacity, obtaining a power consumption prediction matrix and determining the optimal power allocation scheme for each device based on the power consumption prediction matrix includes: Based on the equipment operating parameter matrix, key feature vectors are extracted, and a multivariate regression model between equipment status and energy consumption level is constructed. If the equipment load rate exceeds the preset load threshold, it is marked as a high energy consumption state. The energy conversion efficiency coefficient of various types of equipment under different load conditions is determined. The power consumption prediction model of the device is established by the energy conversion efficiency coefficient and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the load balancing mechanism is triggered to obtain a power consumption prediction matrix that includes time and device dimensions. An energy allocation optimization model is constructed based on the power consumption prediction matrix. The objective function is set as minimizing the overall energy consumption cost. The constraints include the minimum operating power requirement of the equipment and the upper limit of power supply. The optimal power allocation scheme for each device is determined by calculation.

4. The cloud computing-based data center energy-saving control method according to claim 3, characterized in that, Based on the equipment operating parameter matrix, key feature vectors are extracted to construct a multiple regression model between equipment status and energy consumption level. If the equipment load rate exceeds a preset load threshold, it is marked as a high-energy-consuming state. The energy conversion efficiency coefficients of various types of equipment under different load conditions are determined, including: The power changes and operating parameters in the equipment operating parameter matrix are standardized and organized to obtain the data content corresponding to the load. The second basic dataset is then merged and compared with the load threshold. If the energy consumption exceeds the limit, the corresponding device status will be marked as the high energy consumption state, and the marked status data group will be determined. Data under different load conditions are processed in layers to obtain the energy consumption distribution corresponding to the device status, and a set of classified efficiency data is obtained. The efficiency data set is correlated with the power change to obtain the energy conversion efficiency coefficient.

5. The cloud computing-based data center energy-saving control method according to claim 3, characterized in that, The process involves establishing a device power consumption prediction model using the energy conversion efficiency coefficient and predicting the power demand of each device. If the predicted power exceeds the current power supply capacity, a load balancing mechanism is triggered, resulting in a power consumption prediction matrix that includes both time and device dimensions. Based on the energy conversion efficiency coefficient, and combining the time and equipment dimensions, the historical operation records are classified and organized to obtain the feature content related to power demand, thus obtaining the third basic data set. By combining the third basic dataset with the temporal characteristics of the data, the power demand for future periods is analyzed and processed to obtain the predicted power demand distribution results. Based on the power demand distribution results and combined with the power supply capacity limit, data verification is performed. If the limit is exceeded, the load balancing mechanism is triggered to obtain the adjusted power allocation strategy. Based on the power allocation strategy, and combining the time dimension and the device dimension, a prediction matrix containing multi-dimensional information is constructed to obtain the power consumption prediction matrix.

6. The cloud computing-based data center energy-saving control method according to claim 3, characterized in that, The energy allocation optimization model is constructed based on the power consumption prediction matrix, with the objective function set as minimizing the overall energy consumption cost. Constraints include minimum operating power requirements for equipment and a power supply ceiling. The optimal power allocation scheme for each device is determined through calculation, including: Based on the power consumption prediction matrix, the predicted power data of each device is classified and processed to obtain the power demand distribution information related to the operation of the device, and a classified power demand set is obtained. Based on the power demand set combined with the preset supply limit and power requirements, the matching of the power demand of each device with the constraints is verified. If the predicted power of one device exceeds the power supply limit, the power allocation ratio of that device is adjusted, and the adjusted power demand list is determined. Using the power demand list, combined with the objective function and the energy consumption cost data, the power allocation scheme for each device is processed to obtain a preliminary allocation result that minimizes cost. Based on the preliminary allocation results and the actual operating requirements of the equipment, an optimized power allocation scheme for each device is generated, resulting in the optimal power allocation scheme.

7. The cloud computing-based data center energy-saving control method according to any one of claims 1-6, characterized in that, The real-time energy scheduling system established according to the optimal power allocation scheme, if the actual power consumption of one of the devices exceeds a preset range, adjusts the operating parameters of the equipment in the computer room to obtain the energy configuration status, including: The system continuously collects the actual power consumption of each device and determines whether the actual power consumption exceeds the preset range. If so, it automatically adjusts the operating parameters and generates an adjusted parameter configuration scheme in accordance with the requirements of intelligent management. Based on the adjusted parameter configuration scheme, each device is precisely allocated, the energy configuration data feedback is updated in real time, and a power allocation list that matches the current device status is obtained. By comparing the power allocation list with the real-time monitoring data, the power control strategy for each device is determined, and the energy configuration status is obtained.

8. The cloud computing-based data center energy-saving control method according to any one of claims 1-6, characterized in that, The mechanism for updating energy allocation parameters and obtaining an adaptive energy management decision-making mechanism upon detecting abnormal fluctuations includes: If abnormal fluctuations are detected, the trend of energy consumption changes is analyzed, and in combination with the preset threshold range, it is determined whether the conditions for initiating emergency regulation are met, and the specific state of the triggering conditions is determined. Based on the specific state of the triggering conditions, the allocation strategy information that matches the current fluctuation characteristics is extracted from the historical experience database, the adjusted parameter scheme is generated, the updated configuration data is obtained and fed back to the monitoring process, it is determined whether the preset stability standard has been reached, and the final operating state is obtained.

9. The cloud computing-based data center energy-saving control method according to any one of claims 1-6, characterized in that, The method further includes: A closed-loop feedback control system is established based on the adaptive energy consumption management decision-making mechanism. The deviation between actual energy consumption and expected targets is compared, the weight parameters of the prediction model are dynamically adjusted, and a continuously optimized energy management and control scheme is determined. Specifically: According to the energy consumption management decision mechanism, actual energy consumption data is obtained, compared with the expected target, and it is determined whether it exceeds the preset threshold range to obtain the initial deviation result. If it exceeds the threshold, real-time information is collected and combined with historical operating data to obtain the adjusted status record. Based on the adjusted status records, a matching scheme is extracted from the pre-established database to determine the updated weight value; The energy management scheme is dynamically adjusted using the updated weight values. To achieve continuous optimization, the adjusted scheme data is fed back to the closed-loop process to determine whether it meets the preset conditions and obtain the final configuration result.

10. A cloud computing-based energy-saving control system for computer rooms, characterized in that, include: The detection end is used to acquire real-time data and historical power consumption changes of the equipment in the computer room, establish a basic data set including power peak, average, and fluctuation amplitude, and obtain the equipment operating parameter matrix. On the processing end, the device load is obtained according to the device operating parameter matrix and compared with a preset load threshold. The energy conversion efficiency coefficient of the device is obtained according to the comparison result, and the power demand of each device is predicted and calculated. If the predicted power exceeds the current power supply capacity, the power consumption prediction matrix is ​​obtained, and the optimal power allocation scheme of each device is determined according to the power consumption prediction matrix. The decision-making end is used to establish a real-time energy scheduling system based on the optimal power allocation scheme. If the actual power consumption of one of the devices exceeds the preset range, the operating parameters of the equipment in the computer room are adjusted to obtain the energy configuration status and perform continuous monitoring. If abnormal fluctuations are detected, the energy allocation parameters are updated and an adaptive energy consumption management decision-making mechanism is obtained.

Images