An intelligent temperature control energy-saving method for linkage of a machine room server cluster and a refrigeration air conditioner

Abstract

The present application relates to air conditioning technology field, specifically is a kind of intelligent temperature control energy-saving method of machine room server cluster and refrigeration air conditioner linkage, comprising: after receiving temperature control instruction information, the mapping relationship of server cluster in machine room and air conditioning equipment is established;According to the physical partition of server cluster, the working temperature and load rate of server cluster are called, and local temperature trend characteristics are determined;Based on the temperature change value of local temperature trend characteristics, temperature change curve is constructed, and the trend correlation characteristics under the running state of air conditioner are determined based on the difference of temperature rise at each physical partition;Based on the trend correlation characteristics of each physical partition, the target running state of each air conditioner under current temperature control instruction information is determined;Based on the target running state corresponding to temperature control instruction information, temperature control instruction information is fed back, and the update cycle of temperature control instruction information is adjusted.The efficiency and scene adaptability of temperature regulation are improved.

Description

Technical Field

[0001] This invention relates to the field of air conditioning technology, specifically to an intelligent temperature control and energy-saving method that links a server cluster in a computer room with a cooling and air conditioning system. Background Technology

[0002] In the field of industrial air conditioning, the refrigeration system of computer room air conditioning is usually relatively simple, controlled by one or more condenser fans according to preset temperature thresholds. However, these methods tend to be based on single command control, which can easily overlook abnormal temperature rises in some areas. In addition, in scenarios with multiple condenser fans for rapid cooling, if the abnormal locations are not cooled, it can easily lead to the problem of cooling loss.

[0003] For example, Chinese Patent Publication No. CN116989429A discloses an energy-saving method and system for a central air conditioning system based on intelligent control. This includes: setting a temperature control association initialization trigger command and establishing a user database to record user temperature control behavior information for each room; when a temperature control association initialization trigger command is received, outputting a preset multi-room association temperature adjustment relationship filling request, receiving feedback, and deleting historical data from the user database; establishing personalized control logic for the central air conditioning system based on the multi-room association temperature adjustment relationship data obtained from the feedback; and when the amount of data in the user database reaches a preset user personalization prediction condition, predicting user air conditioning usage behavior based on the data in the user database and modifying the multi-room association temperature adjustment relationship data.

[0004] For example, Chinese Patent Publication No. CN116105316A discloses an energy-saving control method, device, energy-saving control equipment, and medium for a central air conditioning system. The method includes: receiving a specific start command, the location information of each temperature control point in its area, and the energy generated per unit time; for each temperature control point, based on the correspondence between the area where the temperature control point is located and the preset correspondence between the air outlet and the unit area, obtaining the effective air outlet corresponding to the temperature control point and its location information; based on the heat generated per unit time corresponding to the temperature control point, obtaining the working time of the effective air outlet; and based on the location information of the temperature control point in its area and the location information of the effective air outlet corresponding to the temperature control point, obtaining the airflow direction of the effective air outlet; and operating the central air conditioning system based on the specific start command, each effective air outlet, the working time of each effective air outlet, and the airflow direction.

[0005] Existing technologies determine the predicted results of personalized temperature control for multiple rooms by checking the frequency of air conditioner temperature adjustment; or select effective air outlets by measuring the heat generated at the temperature control point per unit time and utilize the working time of the effective air outlets to achieve energy-saving control. However, existing technologies tend to focus on temperature verification in a single area and do not consider the differences in temperature rise in different areas. Consequently, when using effective air outlet control, it is difficult to respond to different air conditioner operating scenarios, reducing the adaptability of air conditioner temperature control to different scenarios. Summary of the Invention

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an intelligent temperature control and energy-saving method for linking a server cluster in a computer room with a cooling air conditioner, comprising: after receiving any temperature control command information, establishing a mapping relationship between the server cluster and the air conditioning equipment in the computer room.

[0007] Based on the mapping relationship between server clusters and air conditioning equipment, the operating temperature and load rate of the server cluster are retrieved, and the local temperature trend characteristics are determined according to the physical partitions of the server cluster.

[0008] Based on the temperature change values ​​in the local temperature trend characteristics, a temperature change curve is constructed, and the trend correlation characteristics under the air conditioning operation state are determined according to the differences in temperature rise at each physical zone.

[0009] Based on the trend correlation characteristics of each physical partition, scenario transformation analysis is performed on the server cluster of each physical partition to determine the target operating status of each air conditioner under the current temperature control command information.

[0010] Based on the target operating status corresponding to the temperature control command information, the temperature control command information is fed back, and the update cycle of the temperature control command information is adjusted.

[0011] The beneficial effects of this invention are as follows: First, by establishing a mapping relationship between server clusters and air conditioning equipment, determining local temperature trend characteristics by physical partitions, determining trend correlation characteristics under air conditioning operation, determining the target operation state of the air conditioning and the update cycle of feedback temperature control commands, this invention focuses the current analysis process on the correlation of each region. Through scenario adaptation under trend correlation, while ensuring the safety of server operation, it improves the scenario adaptability under temperature control and avoids the problems of increased energy consumption during off-peak periods and slow response during peak periods.

[0012] Second, this invention initializes the target partition using the server cluster specified by the temperature control command, and regenerates and issues temperature control commands based on the operating status of the servers and air conditioners in the target partition. Then, it obtains the command type and coverage, extracts the basic temperature range according to the device density of the physical partition, corrects the temperature range in combination with the server load rate, and then matches the command type to determine the final temperature value range. Based on this, it constructs local temperature trend features, ensuring that subsequent data has a clear value range space, avoiding the impact of abnormal and sudden data on the current temperature control, and improving the accuracy of data configuration.

[0013] Third, this invention locks in the temperature rise period, retrieves a time window centered on the temperature change time point, and employs differentiated correlation mining methods for the normal / abnormal operating states of the air conditioner. It combines temperature change values ​​with air conditioner operating status identifiers into a state vector, performs correlation analysis through clustering, and finally outputs trend correlation features based on the temperature change slope. This enables correlation analysis across multiple scenarios, providing a data foundation for subsequent temperature control.

[0014] IV. This invention identifies temperature control anomaly types using trend correlation features, divides risk zones based on average / local maximum temperature rise rates, and determines the air conditioning adjustment direction through global optimization based on anomaly type and risk zone. After priority ranking, the target operating state of the air conditioning is determined. Global optimization achieves the optimal allocation of cooling capacity in the computer room, while priority ranking ensures priority control of high-risk, high-load areas, balancing temperature control safety and energy efficiency. Finally, feedback information after temperature control command execution completes zoned control, improving the iterative effect and accuracy of temperature control. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0016] Figure 1 This is a flowchart illustrating an intelligent temperature control and energy-saving method that links a server cluster in a data center with a cooling and air conditioning system.

[0017] Figure 2 This is a flowchart illustrating step S2 of an intelligent temperature control and energy-saving method that links a server cluster in a data center with a cooling and air conditioning system.

[0018] Figure 3 This is a flowchart illustrating step S3 of an intelligent temperature control and energy-saving method that links a server cluster in a data center with a cooling and air conditioning system.

[0019] Figure 4 This is a flowchart illustrating step S4 of an intelligent temperature control and energy-saving method that links a server cluster in a data center with a cooling and air conditioning system. Detailed Implementation

[0020] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Where specific techniques or conditions are not specified in the embodiments, they shall be performed in accordance with the techniques or conditions described in the literature in the art or in accordance with the product manual.

[0021] See Figure 1 A smart temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner includes: S1, after receiving any temperature control command information, establishing a mapping relationship between the server cluster and the air conditioning equipment in the data center.

[0022] S2, based on the mapping relationship between the server cluster and the air conditioning equipment, retrieves the operating temperature and load rate of the server cluster, and determines the local temperature trend characteristics according to the physical partition of the server cluster.

[0023] S3, based on the temperature change value in the local temperature trend characteristics, constructs a temperature change curve, and determines the trend correlation characteristics under the air conditioning operation state according to the difference in temperature rise at each physical zone.

[0024] S4, based on the trend correlation characteristics of each physical partition, performs scenario transformation analysis on the server cluster of each physical partition to determine the target operating status of each air conditioner under the current temperature control command information.

[0025] S5, based on the target operating status corresponding to the temperature control command information, provides feedback on the temperature control command information and adjusts the update cycle of the temperature control command information.

[0026] When any temperature control command information is received in step S1, the implementation method further includes: obtaining the location set of the server cluster in the computer room, and using the server cluster specified by the temperature control command information in the location set as the initial target partition for temperature control.

[0027] Based on the initialization of the server running status and air conditioner running status of the target partition, temperature control command information is regenerated and issued to complete the update of temperature control command information and data mapping.

[0028] By deploying temperature and load sensors on each server node, the system collects real-time data on the operating temperature, CPU load, and other specific operating statuses of each server, and summarizes the data according to the physical partitions of the server cluster (such as rack groups, which directly represent the spatial location of the server cluster). At the same time, it records data such as the operating identifier (running / standby / fault), operating parameters (fan speed, cooling power, outlet air temperature, outlet air direction, etc.), and basic parameters (running time, service life, equipment response time, average cooling efficiency, etc.) of each air conditioner, which serves as the current operating status of the air conditioning equipment in each server cluster of the data center.

[0029] Upon receiving any temperature control command, it is necessary to re-determine the data that the current temperature control command needs to regulate based on the server and air conditioner corresponding to the current temperature control command, combined with the server's operating status and the air conditioner's operating status, and regenerate the specific command information to be issued. The server cluster that needs to be temperature controlled is then bound to the air conditioner to form a configured mapping relationship.

[0030] In one embodiment of the present invention, in step S2, the current monitoring and adjustment control range is determined according to the instruction type of any temperature control instruction information. The instruction types include different types such as initialization instruction, abnormal adjustment instruction, and timed calibration instruction.

[0031] like Figure 2 As shown, when determining the local temperature trend characteristics in step S2, the implementation method further includes: S21, obtaining the instruction type and instruction coverage of the current temperature control instruction information at the moment the temperature control instruction information is received; the instruction coverage can be divided into coverage of a single physical partition / the entire computer room, etc., to determine the control range corresponding to the temperature control instruction.

[0032] S22 extracts the operating temperature and load rate of the server cluster within the instruction coverage area, and retrieves the preset temperature range of each physical partition according to the device density controlled by each physical partition.

[0033] When obtaining the device density of each physical partition control, the system tends to identify the number of servers corresponding to the execution of initialization commands, abnormal adjustment commands, and timed calibration commands. Based on the different number of servers, the system further verifies the range of their operating temperatures, thereby enabling the setting of temperature ranges for high-density server areas and low-density server areas.

[0034] It should be noted that this temperature range setting is used for early warning and verification of rapid server overheating, to prevent some servers from working under excessive load, which would reduce the efficiency of temperature control.

[0035] Since the server load rate is directly related to the operating temperature, the load rate obtained here is only used as a boundary condition for verifying the temperature range, in order to complete the anomaly comparison of the temperature range.

[0036] In step S22, when retrieving the preset temperature range for each physical partition, the implementation method further includes: extracting the base temperature range at the corresponding location from the database based on the device density of the current physical partition; the base temperature range represents the base temperature for the corresponding number of devices and can be directly extracted from the database according to the current device density. It should be noted that the temperature range being checked will analyze whether the operating temperature of each device in the physical partition conforms to the range under the current temperature control command, thereby generating the partition temperature matrix that needs to be output.

[0037] The base temperature range is adjusted by combining the load rate to determine the adjusted temperature range. When the load rate is introduced for adjustment, the higher the load rate, the smaller the temperature range interval. That is, the configured theoretical temperature range needs to be smaller to identify the temperature points where abnormal temperature rise occurs under high load and combine them into the partition temperature matrix of the corresponding partition.

[0038] Its load rate can affect the heat dissipation of the server per unit area, highlighting the impact of load rate on overall temperature at different numbers of devices.

[0039] When correcting the baseline temperature range, linear correction is preferred. The current baseline temperature range and load rate are converted into vectors and matched with historical data. Cosine similarity is used to select the temperature range with the highest similarity. Alternatively, linear regression can be used to adjust the upper and lower limits of the current baseline temperature range to complete the temperature range correction.

[0040] The corrected temperature range is matched with the command type, and the matched data is considered the output temperature range. When matching the temperature range with the command type later, the theoretical temperature required by the corresponding command type can be viewed. For example, the initialization command only checks whether it is within the normal range to ensure relatively stable operation during temperature control; the abnormal adjustment command checks whether it is within the warning range, quickly identifies any abnormalities, and triggers a temperature control alarm; the timed calibration command checks for any abnormalities to achieve regular data updates and synchronization.

[0041] Furthermore, when the corrected temperature range matches the instruction type, its temperature range is also the theoretical temperature range. It is only used to check whether the current temperature range corresponds to the theoretical temperature required by the corresponding instruction. If they do not correspond, the basic temperature range is corrected again based on the load rate until the current output theoretical temperature meets the temperature range required under the combination of specific instruction type + load rate + device density. For example, the initialization command can be set to a temperature range of 20-25℃, 18-27℃, and 15-30℃ according to the normal range, warning range, and emergency range; the abnormal adjustment command can be set to a temperature range of 22-24℃, 20-26℃, and 18-28℃ according to the normal range, warning range, and emergency range; the timed calibration command can be set to a temperature range of 21-26℃, 19-28℃, and 17-30℃ according to the normal range, warning range, and emergency range. The temperature range after load rate correction must fall within the temperature range defined by the corresponding command type. After matching the data to the current command type according to the normal range, warning range, or emergency range corresponding to the current temperature value, multiple sets of theoretical temperature ranges with conditional constraints are obtained. Then, the currently collected working temperature is checked to complete the temperature range processing.

[0042] Although there is obvious overlap in the value ranges of the above-mentioned normal range, warning range, and emergency range, these three ranges are essentially extended temperature check ranges. For example, the warning range is 2°C higher than the upper and lower limits of the normal range, and the emergency range is 3°C higher than the warning range. Their values ​​are mostly selected according to the requirements of the temperature control command information issued at the moment, and the corresponding data extension is selected to monitor the data that reaches its upper and lower limits.

[0043] S23, if the currently collected operating temperature is within the preset temperature range, construct the local temperature trend characteristics of the corresponding physical partition based on the time point at which each physical partition receives the temperature control command information. If the real-time data exceeds the currently set range, it indicates that there is an abnormal situation in the currently acquired data, and an alarm needs to be triggered immediately.

[0044] The output local temperature trend features may include values ​​such as partition ID, number of servers, average operating temperature (°C), maximum operating temperature (°C), average load rate (%), maximum load rate (%), and data integrity (%). These values ​​are used to reveal temperature ranges in each physical partition that do not conform to the load rate + instruction type + device density criteria, and to consider locally excessively high or low temperature values ​​as the current output local temperature trend features. Data integrity here is only used to verify the integrity of data collection at each server location, preventing missing data.

[0045] Furthermore, in step S3, the temperature rise will be analyzed based on the temperature change curves at each physical partition to identify the differences in temperature rise in different areas, and then the trend of temperature change will be decomposed. The temperature change curve will be constructed by taking the working temperature collected at each physical partition, using its timestamp as the horizontal axis and the working temperature of the server at the corresponding location as the vertical axis.

[0046] In one embodiment of the present invention, such as Figure 3 As shown, the implementation method of determining the trend correlation features under the air conditioner operation status in step S3 includes: S31, taking the location corresponding to the local temperature trend feature in the current physical partition as the analysis object, determining the temperature change value between adjacent times, and filtering the temperature rise period when the temperature change value is greater than 0; each temperature rise period starts from the time point when the temperature change value is greater than 0 and ends at the time point when the temperature change value of three consecutive points is less than or equal to 0, and extracts multiple temperature periods.

[0047] S32, for each temperature rise period, synchronizes the operating status of the air conditioning equipment based on the mapping relationship between the server cluster and the air conditioning equipment, and mines the temperature-scenario correlation. The essence of the correlation it mines is to obtain the correlation between the current temperature rise and the air conditioning cooling under the air conditioning operating scenario, and regards the data with abnormal temperature rise in the correlation as the subject of processing, so as to prevent the existence of uncovered server equipment under temperature control.

[0048] When mining the temperature-scene correlation, the implementation method also includes: taking the time point when the temperature changes as the center, retrieving the time window when the temperature changes, and extracting the operating status identifier of the air conditioning equipment within the time window; this time window is extended forward and backward by a fixed time length, such as 5 minutes or 15 minutes, based on the time point when the temperature changes, and the operating status of the air conditioning equipment within the corresponding time is recorded.

[0049] This operating status identifier indicates whether the air conditioning equipment is currently running, in standby mode, or experiencing a fault, as well as the specific values ​​of its operating parameters. This identifier serves as the basis for the current correlation analysis.

[0050] When the operating status is marked as normal, the time interval between the last temperature control command information and the current temperature control command information is obtained, and combined with the currently collected temperature change value, as the output temperature-scene correlation relationship. It can be seen that the local feature output in step S2 is the data value of excessive temperature rise. If the air conditioning equipment is in normal operation and its parameters are within the range of the current scene, if its temperature change is still too high, it is necessary to check the time interval between two consecutive temperature control commands to determine the subsequent feedback processing measures, and use the corresponding data as the output trend correlation feature to fully reveal the current operating status of the equipment.

[0051] When the operating status indicator is not in normal operating condition, a correlation analysis is performed between the temperature change value and the operating status indicator, and the data after the correlation analysis is regarded as the output temperature-scene correlation relationship.

[0052] If the device is not in normal operating condition, it indicates that it is in standby / fault mode. Its specific operating parameters may not respond to temperature control commands. It is necessary to verify the correlation patterns that appear when the temperature rises and use these correlation patterns as the current output data to form the basis for subsequent adjustments to the control strategy. For example, correlation analysis involves clustering temperature change values ​​with operating status indicators to determine whether the combination of current temperature and air conditioning status is reasonable, such as a high load rate + high temperature rise pattern, or a low load rate + low temperature rise pattern.

[0053] When performing correlation analysis between temperature change values ​​and operating status identifiers, the implementation method also includes: combining temperature change values ​​and operating status identifiers into a state vector, where all values ​​in the state vector are standardized data to facilitate subsequent cluster analysis.

[0054] Clustering is performed using the eigenvalues ​​of the state vectors to obtain multiple clusters. Data within each cluster is considered as data after correlation analysis. The clustering method can be k-means clustering, grouping similar scenarios into one class, with each class represented by a cluster center, showing the relative temperature changes under different operating conditions. During clustering, the k-value is determined using the elbow rule, and the clustering termination condition is that the change in cluster centers is less than 0.001 after 10 consecutive iterations. The Euclidean distance calculated from the state vectors is used as the clustering index; thus completing the clustering process.

[0055] S33 synchronizes the temperature-scene correlation to the corresponding physical partition, and outputs trend correlation features by combining the slope corresponding to the temperature change value.

[0056] The trend correlation features output at this time represent the standard deviation, maximum value, minimum value, temperature slope, air conditioning related parameters, correlation and other values ​​of temperature in each scenario. These values ​​represent the abnormal distribution of temperature values.

[0057] Preferably, when performing scene change analysis, since the input trend correlation features represent data combinations under specific correlations, there may be an abnormal situation where the air conditioning equipment is running normally but the server collects an excessively high temperature. It is necessary to determine the current target operating state in order to determine the content form of scene change processing after the temperature control command is executed, in order to complete the temperature regulation at local locations of each server cluster.

[0058] In one embodiment of the present invention, such as Figure 4 As shown, when determining the target operating state of each air conditioner under the current temperature control command information in step S4, the implementation method also includes: S41, using trend correlation features as input, determining the abnormal type during temperature regulation. The abnormal type includes air conditioner control lag, local heat island, uneven distribution of cooling capacity, etc. These abnormal types can be obtained through data correlation analysis and normal state statistics. For example, in step S32, there is a situation of air conditioner control lag when operating normally, and there are abnormal situations such as local heat island and uneven distribution of cooling capacity when operating abnormally.

[0059] Air conditioning control lag indicates that the air conditioner has adjusted the cooling capacity / fan speed according to the instructions, but the corresponding trend of temperature change rate in each zone is delayed by more than 10 minutes (e.g., the temperature rise rate does not decrease within 10 minutes after the cooling is turned on).

[0060] Localized heat islands are areas where the average temperature is normal, but the temperature rise rate of a localized high-load server location is ≥2℃ / 10min, and this location is not effectively covered by air conditioning.

[0061] Uneven cooling capacity distribution is defined as a temperature difference ≥3℃ between adjacent physical zones, and a cooling capacity / fan speed of the air conditioner in the high-load zone being lower than that in the low-load zone, or a temperature difference ≥2℃ between different cabinets within the same zone. In addition to the above explanation, the anomaly type identified can be adjusted according to its specific scenario. After checking the corresponding constraints from the database, the relative temperature identified in the trend correlation features can be processed.

[0062] S42 determines the direction of air conditioning adjustment based on the rate distribution of temperature change values ​​in different physical zones.

[0063] When determining the direction of air conditioning adjustment, the method includes: calculating the average temperature rise rate and the local maximum temperature rise rate of each physical zone, classifying them according to the average temperature rise rate and the local maximum temperature rise rate to generate multiple risk zones; for example, high-risk zones are judged by the values ​​of the average temperature rise rate and the local maximum temperature rise rate, with an average temperature rise rate ≥ 1℃ / 10min or a local maximum temperature rise rate ≥ 2℃ / 10min; medium-risk zones and low-risk zones are judged only by the average temperature rise rate, such as the average temperature rise rate of the medium-risk zone being between 0.5℃ and 1℃, and the average temperature rise rate of the low-risk zone being less than 0.5℃; in this way, multiple risk zones are divided.

[0064] For all risk areas, global optimization is performed according to the anomaly type corresponding to each risk area to determine the updated parameters for each risk area, and the updated parameters are regarded as the output air conditioning adjustment direction.

[0065] During global optimization, a constrained particle swarm optimization algorithm is used, with a population size of 30 and an iteration count of 100. The objective function is defined as minimizing the total cooling energy consumption of the computer room, and the constraints are set according to the temperature range of each risk zone, such as high-risk zone temperature ≤24℃, medium-risk zone temperature ≤25℃, low-risk zone temperature ≤26℃, and air conditioner fan speed adjustment step size ≤20% / time. The iteration termination condition is defined as the objective function change being less than 1% after 20 consecutive iterations, or reaching the maximum number of iterations. The parameters updated for the current risk zone are obtained from the final output parameters after the particle swarm iteration.

[0066] Once the risk zones are identified, the air conditioning control lag can be matched with all risk zones. Its adjustment direction will focus on shortening the response time, such as increasing the fan speed adjustment step from 10% to 20%, increasing the operation time and parameter configuration of some areas, and prioritizing the adjustment of high-risk areas.

[0067] If the current anomaly type is air conditioning control lag, and the current anomaly type is obtained through the average temperature rise rate, the average temperature rise rate is used as the control basis to determine the cooling capacity of the temperature control command information at the corresponding location. The parameters related to the cooling capacity are regarded as updated parameters. For example, parameters that directly affect the cooling capacity of the corresponding area, such as fan speed, are selected. After these parameters are increased proportionally, the configuration of the area with air conditioning control lag is completed.

[0068] If the current anomaly type is obtained through the local maximum temperature rise rate, the time interval between two consecutive temperature control commands is called for the local maximum temperature rise rate. This time interval is used as the updated parameter to adjust the frequency of temperature control adjustment in the corresponding area. The time interval can be reduced proportionally according to the proportion of the local maximum temperature rise rate exceeding the normal temperature, so as to quickly monitor the stable adjustment of the corresponding area.

[0069] The local heat island will be directly matched with high- and medium-risk areas. The adjustments will be made to the air outlet direction and swing angle to adjust the cooling situation of the corresponding risk areas and simultaneously increase the cooling capacity.

[0070] If the current anomaly type is a localized heat island, the air outlet direction and swing angle corresponding to the average heating rate are used as the updated parameters to adjust the area where the localized heat island occurs.

[0071] Uneven cooling capacity distribution will directly match the high and medium risk zones, while simultaneously redistributing the cooling capacity in adjacent physical zones. The cooling capacity in high and medium risk zones will be increased by 10-20% proportionally, while the cooling capacity in low risk zones can be reduced by 5-10% while ensuring that their operating temperature remains within the normal range. This completes the configuration of the air conditioning adjustment direction.

[0072] If the current anomaly type is uneven cooling capacity distribution, the number of risk zones corresponding to the current anomaly type is used as the verification standard, and the relative spatial interval of each risk zone is used as the updated parameter to determine the spatial interval between risk zones after each temperature control command is issued. In this case, the scenario of uneven cooling capacity distribution will focus on the form of spatial distribution, so as to facilitate the adjustment of subsequent temperature control command strategies.

[0073] S43, prioritize the air conditioning adjustment direction to determine the target operating state of the air conditioning equipment.

[0074] In step S43, the parameters that need to be adjusted in each risk zone are prioritized, such as high-risk zone - medium-risk zone - low-risk zone. The parameters that need to be updated within the same physical partition are summarized, and the target operating status of the air conditioning equipment is determined step by step according to the order of different risk levels.

[0075] In one embodiment of the present invention, when adjusting the operating status of the air conditioners in each physical zone in step S5, the implementation method further includes: according to the target operating status corresponding to the temperature control command information, synchronizing the feedback information after the temperature control command information is executed, and adjusting each physical zone according to the time point of the feedback information output, until the operating temperature of each server cluster tends to be within the normal range; selecting the average time interval during the adjustment of each physical zone as the update cycle of the temperature control command information.

[0076] In step S5, the above steps will be repeated until the output data does not contain any abnormal temperature data. The average time interval for data updates in each physical partition will be taken as the update cycle when the current temperature control command is issued, and the data update statistics at the corresponding location will be completed.

[0077] The base value for the update cycle can be set to 10 minutes, with an upper limit of 30 minutes and a lower limit of 1 minute after adjustment. Furthermore, when belonging to different risk zones, the adjustment ratio for each update cycle can be determined based on the different representations of each risk zone. For example, the update cycle for high-risk zones is shortened proportionally to the proportion exceeding the normal operating range, down to a minimum of 1 minute; the update cycle for medium-risk zones is set to 5 minutes; and the update cycle for low-risk zones can be extended to 15-30 minutes. The adjustment step for each update cycle is 1 minute per cycle to avoid frequent fluctuations in the cycle.

[0078] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention, which are still covered within the protection scope of the present invention.

Claims

1. A smart temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, characterized in that, include: Upon receiving any temperature control command, a mapping relationship is established between the server cluster and the air conditioning equipment in the computer room; Based on the mapping relationship between server clusters and air conditioning equipment, the operating temperature and load rate of the server cluster are retrieved, and the local temperature trend characteristics are determined according to the physical partitions of the server cluster. Based on the analysis of temperature change values ​​in local temperature trend characteristics, temperature change curves are constructed, and trend correlation characteristics under air conditioning operation are determined according to the differences in temperature rise at each physical zone. Based on the trend correlation characteristics of each physical partition, scenario transformation analysis is performed on the server cluster of each physical partition to determine the target operating status of each air conditioner under the current temperature control command information. Based on the target operating status corresponding to the temperature control command information, the temperature control command information is fed back, and the update cycle of the temperature control command information is adjusted. The methods for determining trend correlation characteristics under air conditioning operating conditions include: The analysis focuses on the location corresponding to the local temperature trend characteristics within the current physical partition, determining the temperature change value between adjacent moments, and filtering out temperature rise periods with a temperature change value greater than 0. For each temperature rise period, based on the mapping relationship between the server cluster and the air conditioning equipment, the operating status of the air conditioning equipment is synchronized, and the temperature-scene correlation is explored; The temperature-scene correlation is synchronized to the corresponding physical partition, and combined with the slope corresponding to the temperature change value, the output is a trend correlation feature. When mining the temperature-scene correlation, the implementation methods also include: Taking the time point when the temperature change occurs as the center, retrieve the time window when the temperature change occurs, and extract the operating status identifier of the air conditioning equipment within the time window; When the operating status is marked as normal, the time interval between the last temperature control command information and the current temperature control command information is obtained, and combined with the currently collected temperature change value, it is used as the output temperature-scene correlation relationship; When the operating status indicator is not in normal operating condition, a correlation analysis is performed between the temperature change value and the operating status indicator, and the data after the correlation analysis is regarded as the output temperature-scene correlation relationship.

2. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 1, is characterized in that... When receiving any temperature control command information, the implementation methods also include: Obtain the location set of the server clusters in the computer room, and use the server cluster specified by the temperature control command information in the location set as the initial target partition for temperature control; Based on the initialization status of the target partition's server and air conditioner, temperature control command information is regenerated and issued.

3. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 1, is characterized in that... When determining local temperature trend characteristics, the methods also include: Based on the moment the temperature control command information is received, obtain the command type and command coverage of the current temperature control command information; Extract the operating temperature and load rate of the server cluster within the command coverage area, and retrieve the preset temperature range of each physical partition according to the device density controlled by each physical partition; If the current operating temperature is within the preset temperature range, the local temperature trend characteristics of the corresponding physical partition are constructed according to the time point at which the temperature control command information is received in each physical partition.

4. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 3, is characterized in that... When retrieving the preset temperature range for each physical partition, the implementation methods also include: Based on the device density of the current physical partition, extract the base temperature range for the corresponding location from the database; The base temperature range is adjusted by combining the load rate to determine the adjusted temperature range; The corrected temperature range is matched with the instruction type, and the matched data is regarded as the output temperature range.

5. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 1, is characterized in that... When performing correlation analysis between temperature change values ​​and operating status indicators, the implementation methods also include: Temperature change values ​​and operating status identifiers are combined into a state vector. Clustering is performed using the feature values ​​of the state vector to obtain multiple clusters. Data within each cluster is considered as data after correlation analysis.

6. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 1, is characterized in that... When determining the target operating status of each air conditioner under the current temperature control command information, the implementation methods also include: Using trend correlation features as input, the anomaly type during temperature regulation can be determined; The direction of air conditioning adjustment is determined by the rate distribution of temperature change values ​​in different physical zones; Prioritize the air conditioning adjustment directions to determine the target operating state of the air conditioning equipment.

7. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 6, is characterized in that... When determining the direction of air conditioner adjustment, the methods include: Calculate the average temperature rise rate and local maximum temperature rise rate of each physical zone, and classify them according to the average temperature rise rate and local maximum temperature rise rate to generate multiple risk zones; For all risk areas, global optimization is performed according to the anomaly type corresponding to each risk area to determine the updated parameters for each risk area, and the updated parameters are regarded as the output air conditioning adjustment direction.

8. The intelligent temperature control and energy-saving method for linking a server cluster in a data center with a cooling air conditioner, as described in claim 1, is characterized in that... When adjusting the operating status of air conditioners in each physical zone, the implementation methods also include: Based on the target operating status corresponding to the temperature control command information, the feedback information after the temperature control command information is executed is synchronized, and the physical partitions are adjusted according to the time point of the feedback information output until the operating temperature of each server cluster tends to be within the normal range; the average time interval during the adjustment of each physical partition is selected as the update cycle of the temperature control command information.

Images