Temperature control method, temperature control device, electronic equipment and computer program product

Abstract

The invention discloses a temperature control method, a temperature control device, electronic equipment and a computer program product. The method comprises the steps that the heat intensity of each load area in the data center is acquired; according to the heat intensity in each load area, the area heat emergency coefficient of each load area is evaluated; the air conditioning equipment of the data center is pooled, an air conditioning resource pool is obtained, and the scheduling weight of each air conditioning equipment relative to the target area is recorded in the air conditioning resource pool; a temperature control strategy for temperature control over the data center is generated according to the area thermal emergency coefficients of the multiple load areas and the dispatching weights of the multiple air conditioner devices for the target area, and the temperature control strategy at least represents that the target area is subjected to temperature control under the condition that the thermal emergency coefficients of the target area are met. The proportion of the refrigeration task amount allocated to each air conditioner device relative to the total task amount of the air conditioner resource pool. The technical problem that a traditional temperature control system is low in temperature control precision is solved.

Description

Technical Field

[0001] This invention relates to the field of temperature control, and more specifically, to a temperature control method, apparatus, electronic device, and computer program product. Background Technology

[0002] With the continuous expansion of data center (IDC) scale and the sustained growth in computing power demand, the operating density of IT equipment within the server room has increased significantly. This has led to increasingly complex heat load distribution, making it significantly more difficult for temperature control systems to balance energy saving and stability. Currently, mainstream data centers generally employ air conditioning systems that combine various cooling methods, such as air cooling, in-row air conditioning (CRAC), and rack-mounted cooling components, to cool the equipment. These systems typically collect server room temperature data through ambient temperature sensors to achieve constant temperature control.

[0003] However, traditional temperature control systems mostly rely on static configuration for scheduling and control, depending primarily on single temperature sensing information (such as supply / return air temperature) and controlling the air conditioner's on / off state, fan speed, or cooling intensity according to preset strategies. This approach ignores the dynamic impact of the actual load on the IT equipment on heat intensity, making it difficult to achieve precise adjustments under dynamically changing heat and cold load distribution. This can easily lead to redundant air conditioning resources, decreased energy efficiency, or the formation of localized hotspots.

[0004] Therefore, current mainstream air conditioning control solutions mainly rely on the ambient temperature of the computer room as the sole indicator, and employ preset start-stop strategies or uniform setpoint control methods. These methods struggle to adapt to varying load demands and micro-regional thermal differences, generally exhibiting problems such as response lag, uneven heat and cold distribution, resource redundancy, and low energy efficiency. In particular, heat islands can easily form in high-load localized areas, affecting equipment stability and lifespan.

[0005] There is currently no effective solution to the problem of low temperature control accuracy in traditional temperature control systems. Summary of the Invention

[0006] This invention provides a temperature control method, apparatus, electronic device, and computer program product to at least solve the technical problem of low temperature control accuracy in traditional temperature control systems.

[0007] According to one aspect of the present invention, a temperature control method is provided, comprising: acquiring the heat intensity of each load area within a data center, wherein the load area is an area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; evaluating a regional thermal urgency coefficient for each load area based on the heat intensity within each load area, wherein the heat intensity is positively correlated with the regional thermal urgency coefficient, and the regional thermal urgency coefficient represents the immediate demand for cooling resources in the load area; and pooling the air conditioning equipment of the data center to obtain an air conditioning resource pool, wherein the air conditioning resource pool records the information of each air conditioner... The scheduling weight of the equipment relative to the target area, where the target area is any of the load areas, is positively correlated with the proportion of cooling resources generated by the air conditioning equipment that are effectively allocated to the target area. Based on the regional thermal urgency coefficients of the multiple load areas and the scheduling weights of the multiple air conditioning equipment to the target area, a temperature control strategy for temperature control of the data center is generated. The temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning equipment relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met. The cooling task is positively correlated with the scheduling weight.

[0008] Optionally, obtaining the heat intensity of each load area within the data center includes: collecting operational status data of the data service equipment within each load area and scene data of the load area, wherein the operational status data includes at least: the utilization rate of computing resources in the data service equipment and the power consumption of the data service equipment, both of which are positively correlated with the load; the scene data includes at least: the inlet air temperature of the load area, the outlet air temperature of the load area, and the local wind speed of the load area; and using a pre-trained load-heat intensity collaborative sensing model to analyze the operational status data and scene data of each load area to obtain the heat intensity of each load area, wherein the load-heat intensity collaborative sensing model is used to determine the heat intensity based on the weighted result of the utilization rate, the power consumption, the temperature difference between the inlet air temperature and the outlet air temperature, and the local wind speed, combined with a residual term, wherein the first weight coefficients corresponding to the utilization rate, the power consumption, the temperature difference, and the local wind speed, and the residual term are pre-trained.

[0009] Optionally, before analyzing the operating status data and the scene data using a pre-trained load-heat intensity collaborative sensing model to obtain the heat intensity of each load area, the method further includes: collecting operating status data of the data service equipment in each load area and thermal physical data of each load area during multiple sampling periods, wherein the thermal physical data includes at least: the scene data and environmental data for calculating the heat intensity; determining the operating status data, the scene data, and the heat intensity calculated based on the environmental data within the same sampling period as a set of training data; and training the load-heat intensity collaborative sensing model based on multiple sets of training data.

[0010] Optionally, after obtaining the heat intensity of each load area within the data center, the method further includes: performing thermal clustering on the multiple load areas based on the heat intensity to obtain clustering results, wherein the clustering results include at least one clustering region, and two adjacent load areas whose Euclidean distance between the heat intensities is less than a preset distance threshold are clustered in the same clustering region; and performing temperature control according to an air conditioning control strategy pre-set for each clustering region.

[0011] Optionally, evaluating the regional thermal emergency coefficient of each load region based on the heat intensity within each load region includes: detecting the regional state of each load region, wherein the regional state includes at least: the temperature rise rate of the load region and an alarm state represented by a binary variable; and using a pre-trained regional thermal emergency coefficient model to analyze the heat intensity and the regional state of each load region to obtain the regional thermal emergency coefficient of each load region, wherein the regional thermal emergency coefficient model is used to determine the regional thermal emergency coefficient based on a weighted result of the heat intensity, the temperature rise rate, and the alarm state, and the second weight coefficients corresponding to the heat intensity, the temperature rise rate, and the alarm state are pre-trained.

[0012] Optionally, pooling the air conditioning equipment in the data center to obtain an air conditioning resource pool includes: detecting the performance data of each air conditioning device, wherein the performance data includes at least: the energy efficiency value of each air conditioning device, the response latency of the air conditioning device, and the cooling coupling coefficient of the air conditioning device to the target area, wherein the cooling coupling coefficient represents the ability of the cooling resources generated by the air conditioning device to effectively act on the target area; and using a pre-trained scheduling weight prediction model to analyze the energy efficiency value, the response latency, and the cooling coupling coefficient of each air conditioning device to determine the scheduling weight of each air conditioning device relative to the target area, wherein the scheduling weight prediction model is: i represents the load area, and k represents the air conditioning equipment. The energy efficiency value is... The response delay is given when i is the target region. The cold coupling coefficient is... The scheduling weight is used as the basis for generating the air conditioning resource pool based on the scheduling weight of each air conditioning device relative to different target areas.

[0013] Optionally, generating a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of the multiple load areas and the scheduling weights of the multiple air conditioning devices for the target area includes: determining a first ratio by the ratio of the regional thermal urgency coefficient of the target area to the regional thermal urgency coefficients of all load areas in the data center; determining a second ratio by the ratio of the scheduling weight of the target device relative to the target area to the scheduling weight of all air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one of the air conditioning devices in the air conditioning resource pool; determining a temperature control sub-strategy for the target device to control the temperature of the target area based on the product of the first ratio and the second ratio; and determining the temperature control strategy based on the multiple temperature control sub-strategies for different target areas for different target devices.

[0014] According to another aspect of the present invention, a temperature control device is also provided, comprising: an acquisition module, configured to acquire the heat intensity of each load area within a data center, wherein the load area is an area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; an evaluation module, configured to evaluate a regional thermal urgency coefficient for each load area based on the heat intensity within each load area, wherein the heat intensity is positively correlated with the regional thermal urgency coefficient, and the regional thermal urgency coefficient represents the immediate demand for cooling resources in the load area; and a pooling module, configured to pool the air conditioning equipment of the data center to obtain a unified air conditioning resource pool, wherein the air conditioning resource pool The system records the scheduling weight of each air conditioning device relative to a target area, where the target area is any of the load areas. The scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning devices that are effectively allocated to the target area. The generation module is used to generate a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of the multiple load areas and the scheduling weights of the multiple air conditioning devices for the target area. The temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met. The cooling task is positively correlated with the scheduling weight.

[0015] According to another aspect of the present invention, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to execute the above-described temperature control method through the computer program.

[0016] According to another aspect of the present invention, a computer program product is also provided, including computer instructions, which, when executed by a processor, implement the steps of the temperature control method described above.

[0017] In the embodiments described above, since the load of data service equipment is positively correlated with heat intensity, obtaining the heat intensity of each load area within the data center can accurately reflect the heat generation under the operating state of the equipment. The area's thermal urgency coefficient is calculated based on the heat intensity, directly reflecting the immediate demand for cooling resources in the load area, making resource allocation more precise. Subsequently, all air conditioning equipment in the data center is uniformly included in the air conditioning resource pool, and the scheduling weight of each device for the target area is recorded. The scheduling weight reflects the proportion of cooling resources generated by the air conditioning equipment effectively allocated to the target area, ensuring efficient utilization of cooling resources. Then, based on the area thermal urgency coefficient and scheduling weight, a temperature control strategy is generated, clarifying the proportion of cooling tasks allocated to each air conditioning device relative to the total tasks in the air conditioning resource pool, while meeting the thermal urgency coefficient of the target area. This achieves dynamic and intelligent allocation of cooling resources. By evaluating the area thermal urgency coefficient of each load area and the scheduling weight of the air conditioning equipment to generate a temperature control strategy, unreasonable allocation of data center cooling resources is avoided. This not only improves the cooling efficiency of the data center but also reduces energy consumption, meeting the goal of green energy saving and achieving the technical effect of improving temperature control accuracy. This solves the technical problem of low temperature control accuracy in traditional temperature control systems. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0019] Figure 1 This is a flowchart of a temperature control method according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of an air conditioning resource elastic scheduling method based on load heat intensity sensing according to an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of an elastic scheduling sequence of air conditioning resources based on load heat intensity sensing according to an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram of a temperature control device according to an embodiment of the present invention;

[0023] Figure 5 This is a structural block diagram of a computer terminal according to an embodiment of the present invention. Detailed Implementation

[0024] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] First, some nouns or terms that appear in the description of the embodiments of this application shall be interpreted as follows:

[0027] Load-Thermal Intensity: refers to the intensity of heat generated per unit of load by an IT device within a certain time period. It is usually derived from parameters such as power consumption, current, and temperature gradient.

[0028] Elastic Scheduling refers to a scheduling strategy that dynamically adjusts resource allocation based on the current system state and predicted demand, making resource utilization more flexible and efficient.

[0029] Cooling Resource Pool: refers to a collection that abstracts different types of air conditioning equipment (such as CRAC, air valves, cold plates, etc.) into a unified control unit and performs unified orchestration and scheduling based on state characteristics.

[0030] Thermal modeling: Using sensor data and physical parameters to construct a mathematical model of the heat distribution and conduction characteristics within a specific space, in order to predict temperature change trends.

[0031] Hot-Cold Zone: Refers to localized high-temperature or low-temperature areas within the IDC interior space due to factors such as load differences and airflow layout. These are micro-space units that require different treatment.

[0032] Power Usage Effectiveness (PUE): An important indicator of data center energy efficiency, it is equal to the ratio of total energy consumption to the energy consumption of IT equipment, and is used to measure the operating efficiency and energy-saving level of data centers.

[0033] Edge inference refers to deploying lightweight algorithm models at edge nodes (such as air conditioner controllers and sensor terminals) to independently complete inference calculations and control decisions, reducing reliance on a central hub.

[0034] Feedback Loop Control: The control system adjusts the control strategy based on the feedback information of the execution results, so that the output results gradually approach the target state, thereby achieving dynamic optimization.

[0035] According to an embodiment of the present invention, a temperature control method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0036] Figure 1 This is a flowchart of a temperature control method according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0037] Step S102: Obtain the heat intensity of each load area in the data center, where the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity.

[0038] Step S104: Based on the heat intensity within each load area, evaluate the regional thermal emergency coefficient of each load area, where heat intensity is positively correlated with the regional thermal emergency coefficient, and the regional thermal emergency coefficient represents the immediate demand of the load area for cooling resources.

[0039] Step S106: Pool the air conditioning equipment in the data center to obtain an air conditioning resource pool. The air conditioning resource pool records the scheduling weight of each air conditioning device relative to the target area. The target area is any load area. The scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning devices that are effectively allocated to the target area.

[0040] Step S108: Based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area, a temperature control strategy for temperature control of the data center is generated. The temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met. The cooling task is positively correlated with the scheduling weight.

[0041] In the embodiments described above, since the load of data service equipment is positively correlated with heat intensity, obtaining the heat intensity of each load area within the data center can accurately reflect the heat generation under the operating state of the equipment. The area's thermal urgency coefficient is calculated based on the heat intensity, directly reflecting the immediate demand for cooling resources in the load area, making resource allocation more precise. Subsequently, all air conditioning equipment in the data center is uniformly included in the air conditioning resource pool, and the scheduling weight of each device for the target area is recorded. The scheduling weight reflects the proportion of cooling resources generated by the air conditioning equipment effectively allocated to the target area, ensuring efficient utilization of cooling resources. Then, based on the area thermal urgency coefficient and scheduling weight, a temperature control strategy is generated, clarifying the proportion of cooling tasks allocated to each air conditioning device relative to the total tasks in the air conditioning resource pool, while meeting the thermal urgency coefficient of the target area. This achieves dynamic and intelligent allocation of cooling resources. By evaluating the area thermal urgency coefficient of each load area and the scheduling weight of the air conditioning equipment to generate a temperature control strategy, unreasonable allocation of data center cooling resources is avoided. This not only improves the cooling efficiency of the data center but also reduces energy consumption, meeting the goal of green energy saving and achieving the technical effect of improving temperature control accuracy. This solves the technical problem of low temperature control accuracy in traditional temperature control systems.

[0042] In step S102 above, a data center refers to a facility that centrally stores, processes, exchanges, and distributes large amounts of data and information. It is a place equipped with computing devices (such as servers), storage devices, network communication equipment, and other necessary auxiliary equipment (such as air conditioning, uninterruptible power supplies, fire protection systems, etc.), which together form a highly integrated IT environment.

[0043] In step S102 above, data service equipment refers to hardware equipment used to provide data processing, storage, communication and other data services in a data center or network environment. The higher the load undertaken by the data service equipment, the more heat it generates, and the higher the heat intensity of the load area where the data service equipment is deployed.

[0044] Optionally, the data service equipment includes, but is not limited to, the following:

[0045] Server: A high-performance computer used to run applications and services, such as web servers and database servers.

[0046] Storage equipment includes disk arrays, solid-state drives, tape drives, etc., used to store data and information.

[0047] Networking equipment, such as routers, switches, and firewalls, is used to build and manage network communications.

[0048] Security equipment includes intrusion detection systems, security agents, encryption devices, etc., used to ensure data security.

[0049] Load Balancer: Distributes network traffic to multiple servers to improve website response speed and server availability.

[0050] Backup and recovery equipment, such as backup software and tape libraries, is used for data backup and disaster recovery.

[0051] In step S102 above, thermal intensity typically refers to the amount of heat passing through a unit area per unit time under specific conditions; it is a physical quantity that measures the rate of heat transfer. The calculation of thermal intensity is usually based on fundamental principles of heat transfer, such as Fourier's law (conduction), Newton's law of cooling (convection), and the Stefan-Boltzmann law (radiation). In practical applications, the measurement and analysis of thermal intensity can help optimize thermal management systems, improve energy conversion efficiency, reduce heat loss, and evaluate the performance of materials and structures under thermal conditions.

[0052] In step S104 above, the regional thermal emergency coefficient of each load area is evaluated based on the thermal intensity, temperature rise rate and alarm status within the same load area. The temperature rise rate represents the rate at which the temperature of the load area increases, and the alarm status represents whether the temperature of the load area exceeds the set maximum temperature, which is represented by a Boolean value.

[0053] In step S108 above, the temperature control strategy is based on the regional thermal emergency coefficient of multiple load areas and the balancing strategy of each air conditioning device in the air conditioning resource pool for multiple load areas. It can be expressed as the proportion of cooling tasks that each air conditioning device in the air conditioning resource pool should undertake.

[0054] In step S108 above, the target area can be the load area with the highest regional thermal urgency coefficient, and the target equipment can be the air conditioning equipment that provides the most effective cooling resources to the target area. For example, the air conditioning equipment with the highest scheduling weight relative to the target area can prioritize the target area and the target equipment. Then, the analysis is conducted on the target equipment providing cooling resources to the target area, and the operating conditions of other air conditioning equipment for other load areas are determined based on the radiation of the target equipment to other load areas. Furthermore, the analysis is conducted on the other air conditioning equipment providing cooling resources to other load areas, and the operating conditions of the target equipment for the target area are adjusted based on the radiation of other air conditioning equipment to the target area.

[0055] It should be noted that within a data center, the heat intensity of each load area is generated by the data service equipment deployed within that load area. Therefore, the operational status data of the data service equipment in each load area can be used for calculation.

[0056] As an optional embodiment, obtaining the thermal intensity of each load area within the data center includes: collecting operational status data of data service devices and scenario data of the load area within each load area. The operational status data includes at least the utilization rate of computing resources in the data service devices and the power consumption of the data service devices, both of which are positively correlated with the load. The scenario data includes at least the inlet air temperature, outlet air temperature, and local wind speed of the load area. A pre-trained load-thermal intensity collaborative sensing model is used to analyze the operational status data and scenario data of each load area to obtain the thermal intensity of each load area. The load-thermal intensity collaborative sensing model is used to determine the thermal intensity based on a weighted result of utilization rate, power consumption, the temperature difference between inlet and outlet air temperatures, and local wind speed, combined with a residual term. The first weight coefficients corresponding to utilization rate, power consumption, temperature difference, and local wind speed, and the residual term are pre-trained.

[0057] The embodiments described above in this application combine the operating status data of data service equipment and the scene data of load areas, and use a load-heat intensity collaborative sensing model for in-depth analysis to effectively predict the heat intensity of each load area within the data center. The heat intensity analysis method not only considers the heat load generated by equipment operation, but also the influence of environmental factors on heat diffusion, thereby achieving a more comprehensive and accurate heat intensity prediction. Based on the accurate predicted heat intensity, the cooling resources of the data center can be allocated more precisely, which not only improves the cooling efficiency of the data center, but also reduces energy consumption, meets the goal of green energy saving, and achieves the technical effect of improving temperature control accuracy, thereby solving the technical problem of low temperature control accuracy in traditional temperature control systems.

[0058] The embodiments described above in this application, through the comprehensive application of weighted parameters and residual terms, enable the load-heat intensity collaborative sensing model to adapt to the specific conditions of different load areas, thereby improving the accuracy and reliability of prediction. By using the accurately predicted heat intensity, data center managers can promptly identify areas with abnormal heat intensity, optimize the layout and operation strategy of the cooling system, thereby improving overall energy efficiency, reducing energy waste, and ensuring the stable operation of data center equipment.

[0059] It should be noted that although the heat in each load area within a data center is generated by the data service equipment, the process from the data service equipment assuming the load to generating heat does not occur instantaneously. Therefore, by using the load-heat intensity collaborative sensing model, the heat intensity of each load area can be predicted in advance. Furthermore, the predicted heat intensity can be used to schedule the cooling resources of the data center in advance, achieving more precise temperature control.

[0060] As an optional example, the load-heat intensity co-sensing model is as follows: Where i is the load region. For the first Region in time The thermal intensity (unit: W / m²). To improve the utilization of devices such as CPU / GPU Power consumption (in W). Intake air temperature, The outlet air temperature, This refers to the temperature rise (i.e., the temperature difference). For local wind speeds, As the first weighting coefficient, This is the residual term.

[0061] It should be noted that the load of data service equipment can be measured by the utilization rate of CPU / GPU devices and the power consumption of the data service equipment. In other words, the utilization rate of CPU / GPU devices and the power consumption of the data service equipment are quantitative indicators of the load.

[0062] Optionally, the collection points for operational status data and scenario data include: sensor arrays deployed in cabinets, power modules, and inter-row areas.

[0063] Optionally, the operational status data and scenario data can be collected time series. Before using a pre-trained load-heat intensity collaborative sensing model to analyze the operational status data and scenario data of each load area, the operational status data and scenario data can be preprocessed. The specific preprocessing process includes at least: forming a unified time series sample set through time synchronization processing and missing value imputation.

[0064] It should be noted that the load-heat intensity collaborative sensing model can be built by combining multidimensional input data as training data and using methods such as multiple linear regression or random forest regression. Its core idea is to predict the heat output (i.e., heat intensity) of data service equipment or load area per unit time, using load indicators as input.

[0065] As an optional embodiment, before using a pre-trained load-heat intensity collaborative sensing model to analyze operating status data and scene data to obtain the heat intensity of each load area, the method further includes: collecting operating status data of data service devices in each load area and thermal physical data of each load area during multiple sampling periods, wherein the thermal physical data includes at least: scene data and environmental data for calculating heat intensity; determining the operating status data, scene data, and heat intensity calculated based on environmental data within the same sampling period as a set of training data; and training the load-heat intensity collaborative sensing model based on multiple sets of training data.

[0066] In the embodiments described above, operational status data of data service devices within each load area and thermophysical data of each load area are collected during multiple sampling periods. The thermophysical data includes at least scene data and environmental data for calculating heat intensity, such as temperature, humidity, and device power consumption. The operational status data, scene data, and heat intensity calculated based on environmental data within the same sampling period are defined as a set of training data. By collecting multiple sets of such training data, different operational states and environmental conditions can be covered, thereby constructing a comprehensive training set. The load-heat intensity collaborative sensing model is trained based on multiple sets of training data in the training set, enabling the model to learn the complex relationship between operational status, scene, and heat intensity, and to achieve accurate prediction of heat intensity. This not only improves the training efficiency of the model but also enhances its adaptability and robustness in practical applications, ensuring that the model has excellent stability and accuracy when facing complex and ever-changing load area heat intensity prediction tasks. It can better cope with the heat intensity prediction needs under different scenarios, achieve accurate prediction of heat intensity in the load area, and, based on the accurate prediction of heat intensity, can schedule the cooling resources of the data center in advance to achieve more precise temperature control, solving the technical problem of low temperature control accuracy in traditional temperature control systems.

[0067] Optionally, the collection of training data can be further optimized by adjusting the sampling frequency and adding data types to improve the model's predictive performance.

[0068] As an alternative example, as operational data accumulates, an incremental learning approach is used to periodically train and adaptively update the load-heat intensity co-sensing model, ensuring that it still has high prediction accuracy under different load characteristics.

[0069] As an alternative example, within a provincial-level ultra-large data center pilot zone, three high-density computing areas were selected, deploying a total of 60 temperature sensors, 10 wind speed monitoring points, and data acquisition modules connected to server power meters, continuously collecting operational data for 21 days. The load-heat intensity collaborative sensing model was trained using multiple linear regression, with the prediction target being heat intensity. .

[0070] The experimental results show that the established load-heat intensity collaborative sensing model has an average absolute error of ±5.6 W / m² on the test set, which is about 28% more accurate than the method that only relies on temperature trend prediction. After the model is deployed, it can output the high heat load prediction area (i.e., the load area with high heat intensity) 5 minutes in advance before the air conditioning scheduling strategy, providing a decision-making feedforward basis for flexible scheduling.

[0071] As an optional embodiment, after obtaining the heat intensity of each load area in the data center, the method further includes: performing thermal clustering on multiple load areas based on the heat intensity to obtain clustering results, wherein the clustering results include at least one clustering region, and two adjacent load areas whose Euclidean distance between heat intensities is less than a preset distance threshold are clustered in the same clustering region; and performing temperature control according to an air conditioning control strategy preset for each clustering region.

[0072] In the embodiments described above, the loads of various data service devices within a data center differ, resulting in varying heat intensities in different load areas. Consequently, different air conditioning control strategies are required for load areas with varying heat intensities. Therefore, by calculating the Euclidean distance between load areas based on their heat intensities, adjacent areas with a distance below a preset threshold are clustered into the same cluster area, forming at least one cluster area. This thermal clustering enables refined management of heat sources within the data center. By setting corresponding air conditioning control strategies based on the characteristics of different cluster areas, not only is the efficiency and accuracy of temperature regulation improved, but it also effectively avoids equipment performance degradation and failure risks caused by localized overheating. This optimizes the energy consumption and heat dissipation performance of the entire data center, ensures stable equipment operation, extends equipment lifespan, achieves more precise temperature control, and solves the technical problem of low temperature control accuracy in traditional temperature control systems.

[0073] It should be noted that, in addition to Euclidean distance-based clustering methods, other clustering algorithms (such as hierarchical clustering, k-means clustering, etc.) can be used to adapt to the layout characteristics and thermal management needs of different data centers.

[0074] Optionally, by adjusting the preset distance threshold or introducing more dimensions of thermal characteristic parameters (such as dynamic changes of heat sources, heat conduction paths, etc.), the thermal clustering effect can be further optimized, and the level of intelligence in temperature control can be improved.

[0075] As an alternative example, the thermal clustering formula based on the Euclidean distance method is expressed as: ,like Then the region and They are classified into the same cluster region, among which, and For adjacent load areas, The Euclidean distance between thermal intensities. This is a preset distance threshold.

[0076] As an optional example, the thermal clustering process includes: obtaining the thermal intensity of the entire internal space of the IDC in each time period. The system constructs two-dimensional or three-dimensional heat distribution maps; it employs spatial clustering algorithms (such as K-Means and DBSCAN) to group areas with similar heat intensities into the same partition, typically clustering them into two main classes: high-load zones and low-load zones.

[0077] As an alternative example, clustering regions include: hot regions, cold regions, and neutral regions.

[0078] Optionally, the air conditioning control strategy for hot zones may include setting a lower target temperature, increasing airflow, and increasing the frequency of air conditioning.

[0079] Optionally, the air conditioning control strategy for the cold zone setting includes: moderately increasing the supply air temperature, reducing the frequency, or shutting down redundant equipment.

[0080] Optionally, the air conditioning control strategy for the neutral zone setting includes: Neutral zone: Maintain the basic fan speed and set temperature without active adjustment.

[0081] As an alternative example, within a data center, air conditioning equipment is logically grouped (Group A, B, C), and then mapped according to the coverage of hot and cold zones and capacity adaptability, setting the operating frequency, air volume, and response threshold for each group.

[0082] Optionally, a feedback adjustment mechanism can be used to evaluate the effectiveness of the strategy by using temperature sensor feedback. If a certain hot zone is found to still exceed the standard, the strategy will adaptively increase the air conditioning intensity of that zone to achieve closed-loop control.

[0083] As an alternative example, in a national-level IDC data center, a row of 10 parallel server racks was selected, and a heat intensity prediction and sensing system was deployed. 45 temperature measurement points were sampled, and the space was divided into 2 hot zones (corresponding to high-density computing server areas) and 1 cold zone (the area with lower load near the entrance) using the K-Means clustering algorithm.

[0084] Optionally, the air conditioning control strategy response is as follows:

[0085] The air conditioning outlet temperature in the hot zone was lowered from 22°C to 18°C, while the air volume was increased by 25%.

[0086] The air volume in the cold zone is reduced by 10%, and both air conditioning units are put into low-power standby mode.

[0087] The supply air temperature in the neutral zone remains constant at 22°C.

[0088] Two weeks after implementation, the PUE in the area decreased by 0.041, the average local temperature difference decreased by about 1.5°C, and users reported that the equipment's operating temperature control was more stable.

[0089] As an optional embodiment, evaluating the regional thermal emergency coefficient of each load area based on the heat intensity within each load area includes: detecting the regional state of each load area, wherein the regional state includes at least: the temperature rise rate of the load area and the alarm state represented by binary variables; and using a pre-trained regional thermal emergency coefficient model to analyze the heat intensity and regional state of each load area to obtain the regional thermal emergency coefficient of each load area, wherein the regional thermal emergency coefficient model is used to determine the regional thermal emergency coefficient based on the weighted result of heat intensity, temperature rise rate and alarm state, and the second weight coefficients corresponding to heat intensity, temperature rise rate and alarm state are pre-trained.

[0090] In the embodiments described above, the regional status of each load area is detected, which at least includes the temperature rise rate and alarm status of the load area. The alarm status is represented by a binary Boolean value to accurately indicate its presence or absence. A pre-trained regional thermal urgency coefficient model is employed. This model, based on a weighted sum of heat intensity, temperature rise rate, and alarm status, comprehensively analyzes the thermal characteristics of each load area to determine its regional thermal urgency coefficient. The second weighting coefficients corresponding to heat intensity, temperature rise rate, and alarm status are dynamically adjusted based on a large amount of thermal data during model training, ensuring the accuracy and reliability of the evaluation results. Based on the regional thermal urgency coefficient model, the thermal conditions of each load area of ​​the equipment can be monitored and evaluated in real time, enabling timely and effective measures to avoid the risk of thermal runaway, improve the thermal management performance and operational stability of the equipment, achieve more precise temperature control, and solve the technical problem of low temperature control accuracy in traditional temperature control systems.

[0091] Optionally, by further optimizing the setting of the second weighting coefficient, the generalization ability and applicability of the model can be enhanced to adapt to the thermal characteristics and application scenarios of different devices. For example, for high-power electronic devices, it may be necessary to increase the weight of thermal intensity to more accurately reflect their thermal emergency conditions; while for temperature-sensitive precision instruments, increasing the weight of temperature rise rate and alarm status may be more important to ensure more refined thermal status monitoring and management.

[0092] As an optional implementation, the formula for the regional thermal emergency coefficient model is expressed as follows: ,in, Let i be the regional thermal emergency coefficient for the load region. The thermal intensity of the load area i is expressed in W / m². Let i be the rate of temperature rise in the load region i; Indicates whether load region i is in an alarm state (binary variable); This is the second weighting coefficient.

[0093] Optionally, the regional thermal emergency coefficient model can be pre-trained using machine learning, and the parameters obtained through training include at least a second weighting coefficient.

[0094] As an optional embodiment, pooling the air conditioning equipment in the data center to obtain an air conditioning resource pool includes: detecting the performance data of each air conditioning device, wherein the performance data includes at least: the energy efficiency value of each air conditioning device, the response latency of the air conditioning device, and the cooling coupling coefficient of the air conditioning device to the target area, the cooling coupling coefficient representing the ability of the cooling resources generated by the air conditioning device to effectively act on the target area; using a pre-trained scheduling weight prediction model, analyzing the energy efficiency value, response latency, and cooling coupling coefficient of each air conditioning device, and determining the scheduling weight of each air conditioning device relative to the target area, wherein the scheduling weight prediction model is: i represents the load area, and k represents the air conditioning equipment. Energy efficiency value, To address latency, when i is the target region, This is the cold coupling coefficient. Assign scheduling weights; generate an air conditioning resource pool based on the scheduling weight of each air conditioning unit relative to different target areas.

[0095] The embodiments described above integrate the air conditioning equipment of the data center into the management process of the air conditioning resource pool. By using a pre-trained scheduling weight prediction model, the scheduling weight of each air conditioning device for the target area is quantified. Based on this quantification result, the generation of the air conditioning resource pool is no longer just a simple collection of devices, but an intelligent allocation of resources based on the scheduling weight of each air conditioning device in different target areas. This ensures the effective utilization and precise scheduling of cooling resources, thereby improving the energy efficiency and environmental control accuracy of the data center. It makes the management of air conditioning equipment more flexible and efficient, and can automatically adjust the cooling strategy according to real-time demand, reducing energy consumption while optimizing heat dissipation. This helps the data center achieve its energy conservation and emission reduction goals, and achieves more precise temperature control, solving the technical problem of low temperature control accuracy in traditional temperature control systems.

[0096] Optionally, the scheduling weight prediction model can be pre-trained using machine learning, and the parameters obtained through training include at least the cold coupling coefficient.

[0097] Optionally, the scheduling weight can adopt different parameter combinations and calculation logic to adapt to specific data center layouts and air conditioning equipment characteristics, but the core is still to achieve dynamic management and optimized configuration of the resource pool through performance data detection and scheduling weight prediction.

[0098] As an optional embodiment, generating a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area includes: determining a first ratio by the ratio of the regional thermal urgency coefficient of the target area to the regional thermal urgency coefficients of all load areas in the data center; determining a second ratio by the ratio of the scheduling weight of the target device relative to the target area to the scheduling weight of all air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one of the air conditioning devices in the air conditioning resource pool; determining a temperature control sub-strategy for the target device to control the temperature of the target area based on the product of the first ratio and the second ratio; and determining a temperature control strategy based on multiple temperature control sub-strategies for different target areas for different target devices.

[0099] The embodiments described above in this application determine a temperature control sub-strategy for the target device to control the temperature of the target area based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area. Furthermore, based on the multiple temperature control sub-strategies for different target areas by different target devices, a comprehensive temperature control strategy is generated by integrating them. This strategy can dynamically adjust the allocation of cooling resources according to the thermal urgency of each area and the scheduling capacity of the air conditioning devices, thereby improving energy efficiency, ensuring temperature stability in the data center load area, avoiding equipment failure due to local overheating, improving the stability and security of the data center, achieving more precise temperature control, and solving the technical problem of low temperature control accuracy in traditional temperature control systems.

[0100] Optionally, by adding additional heat source prediction models, the scheduling strategy of air conditioning equipment can be further optimized to achieve more refined temperature control. Of course, adjusting the operating mode of air conditioning equipment is not limited to increasing or decreasing cooling power; it can also include adjusting the airflow direction and speed to adapt to the heat dissipation needs of different types of load areas, ensuring efficient cooling of the entire data center. Through the implementation of the above solutions, data centers can achieve intelligent temperature management and energy consumption optimization, improve operational efficiency, and reduce operating costs.

[0101] Optionally, the method for determining the first proportion includes: ,in, This represents the first proportion of the load region i. Let be the regional thermal urgency factor for load zone i, and m be the number of load zones within the data center. This represents the regional thermal emergency factor for any load area within the data center.

[0102] Alternatively, the second proportion may be determined as follows: Where i is the target area and n is the number of air conditioning units in the air conditioning resource pool. Let k be the scheduling weight of the target device k relative to the target region i.

[0103] Optionally, for the target area With target equipment The allocation ratio of the temperature control sub-strategy is as follows: .

[0104] As a specific implementation method, the above temperature control method includes the following steps:

[0105] Step S1: Register all controllable air conditioning equipment (such as CRAC, In-Row, ceiling-mounted, etc.) in a unified air conditioning resource pool, denoted as a set. The cooling capacity, operating status, energy efficiency ratio, and other parameters of each air conditioning unit are used to define its scheduling weight.

[0106] Step S2, assess the urgency of regional heat load to obtain the regional heat urgency coefficient, specifically including:

[0107] For all load areas Based on its current heat intensity Temperature rise rate Alarm status, etc., to calculate the regional thermal emergency coefficient. This indicates the immediate demand for cooling resources in the region.

[0108] Optionally, the formula for calculating the regional thermal emergency factor (i.e., the regional thermal emergency factor model) is as follows: ,in: Thermal intensity, unit: W / m²; The rate of temperature rise; Is the area in an alarm state? (Binary variable) Adjustment coefficient (also known as the second weighting coefficient).

[0109] Step S3, scheduling priority allocation mechanism, specifically includes: calculating the scheduling priority of each load region. (That is, the first proportion), whose value serves as the primary basis for resource allocation: , indicating region The proportion of all current thermal load response resources occupied.

[0110] Step S4: Determine the scheduling weights based on the air conditioning equipment capacity adaptation function, wherein each air conditioning unit is weighted according to its efficiency value. Response latency The cooling coupling coefficient of the current location to the region Generate scheduling weights The scheduling weight prediction model is .

[0111] Step S5, final proportion calculation, i.e., determining the temperature control strategy, specifically includes: for the region With air conditioning The allocation ratio represented by the temperature control strategy is: Based on this, the air volume, cooling frequency, or start / stop status of the air conditioner can be controlled to achieve flexible pooling and allocation.

[0112] As an alternative example, in a multi-region IDC environment, six different models of air conditioning units (three CRAC units, two ceiling-mounted units, and one in-row unit) are deployed. The system collects load data and temperature dynamics for four key hot zones to generate a regional thermal emergency coefficient. The scheduling priority is generated based on the above calculation mechanism. In conjunction with the physical location and equipment efficiency of the air conditioners, the proportion of cooling tasks that each air conditioner should undertake is calculated.

[0113] For example, hot zone Z1 has a high urgency level and accounts for a large proportion. The corresponding energy is allocated to the ceiling-mounted air conditioner with the highest efficiency. highest proportion The system automatically increases its air output power to the maximum limit, while other devices provide proportional assistance.

[0114] After a week of trial operation, the system's average response time improved by about 40%, temperature control efficiency improved by 25%, and cooling capacity redundancy decreased significantly.

[0115] As an optional embodiment, the above temperature control method can be executed jointly by edge devices and the cloud.

[0116] It should be noted that in the current intelligent scheduling process of IDC air conditioning systems, cloud models are typically used for global analysis and calculation. However, in scenarios with frequent changes in heat load and high real-time requirements, cloud solutions suffer from high response latency, large bandwidth consumption, and insufficient data security. Especially in data centers with multiple partitions and nodes, if all data needs to be uploaded to the cloud before issuing instructions for each scheduling, it can easily create a system bottleneck and restrict the efficiency of air conditioning elastic scheduling response.

[0117] Optionally, a compressed version of the inference model can be deployed on the edge gateway (i.e., the edge device of the data center) near the physical air conditioning unit and the heat zone sensing source to quickly assess the heat intensity of local load areas within the data center and generate preliminary scheduling suggestions without relying on cloud center (e.g., cloud) decisions. The cloud acts as a secondary control center, intervening only when there are edge inference conflicts or policy fusion. This architecture improves response speed and significantly reduces central resource consumption, exhibiting good scalability and feasibility.

[0118] As an optional embodiment, the edge inference model in the edge device is designed as follows: based on the correlation between hot load characteristics and scheduling strategy, a small number of highly correlated feature variables are extracted to construct a compressed model, which is deployed on the edge control node. The model structure is a lightweight neural network or logistic regression module.

[0119] Optionally, the input feature vector of the edge inference model is: ,in, Thermal intensity per unit time; Temperature fluctuations within the time window; This indicates the current operating status of the air conditioning equipment (such as the fan speed setting). This is a local device alarm flag (using Boolean values, that is, using binary variables to represent the alarm status).

[0120] Optionally, the edge inference model can control variables and then obtain Y based on the superposition of variables and H.

[0121] Optionally, the decision vector output by the edge inference model is: ,in, To recommend the air conditioning operating level; This is the wind speed adjustment ratio; Suggestions for switch on / off status.

[0122] It should be noted that each edge node (i.e., edge device) is deployed on an air conditioning management unit (such as a data center CRAC controller, PDU) or an edge gateway (Raspberry Pi, industrial box), collects local sensor data, and completes model inference and policy execution on its own.

[0123] Optionally, the inference frequency of edge devices is dynamically adjusted according to load changes, typically once every 30 seconds, to ensure real-time response.

[0124] Optionally, edge devices and the cloud adopt an edge-cloud collaborative fusion mechanism. When the output of the edge node (i.e., the edge device) deviates from the expected control value of the cloud by more than a set threshold, If the load in a local area fluctuates significantly, the system will activate the cloud intervention mode and perform secondary optimization scheduling by integrating multi-edge data.

[0125] As an optional implementation, the formula for determining the control value deviation between the edge device and the cloud is: ,in, For control values ​​or temperature control strategies output by edge devices, This refers to the control values ​​or temperature control strategies output from the cloud.

[0126] As an alternative example, in a medium-sized IDC test area, six edge gateway nodes were deployed between the air conditioning control loop and the environmental sensor on top of the cabinet. The edge model used a compressed LSTM neural network, with a training time of 3 hours. After deployment to each edge device, it occupied less than 15MB of memory and had an inference time of less than 200ms.

[0127] During the week-long trial run, the frequency of cloud model calls decreased by more than 70%; the average latency for responding to changes in thermal load was reduced to 0.8 seconds; air conditioning control errors decreased by 21%; and the prediction accuracy of the edge model remained stable at over 93%. This approach validates the technical advantages of deploying edge intelligence in the elastic scheduling of IDC air conditioning, achieving high response, low power consumption, and weak dependency.

[0128] It should be noted that after each inference, the edge node (i.e., the edge device) sends the suggested strategy to the air conditioner controller, calls the interface to adjust the operating status, and records the execution results for subsequent feedback and learning.

[0129] As an optional implementation, after generating a temperature control strategy for controlling the temperature of the data center, an energy efficiency optimization mechanism based on a feedback closed loop of hot and cold load adjustment can be used.

[0130] Optionally, based on the energy efficiency optimization mechanism of feedback closed loop for cooling and heating load regulation, a real-time closed-loop regulation model is constructed with energy efficiency evaluation indicators (such as PUE), regional temperature fluctuation range and start-up and shutdown frequency of key equipment as multi-objective constraints. Feedback factors are introduced into the control system to quantify the actual execution results of the control strategy, continuously adjust the strategy parameters, realize intelligent self-optimization of the air conditioning system, improve energy efficiency utilization, and reduce energy consumption fluctuations and maintenance costs.

[0131] As an optional example, the energy efficiency optimization mechanism based on the feedback closed loop of cooling and heating load regulation specifically includes: constructing an energy efficiency feedback control objective function; constructing a dynamic closed-loop feedback link; and introducing an operational fluctuation penalty factor.

[0132] Optionally, constructing the energy efficiency feedback control objective function includes: modeling the impact of heating and cooling regulation on the system's PUE, temperature stability, and switching frequency, and setting the control objective function as follows: ,in, For control strategy combinations (including temperature setpoint, fan speed, compressor cooling cycle, etc.). For a moment Data center power usage effectiveness (Power Usage Effectiveness) values. The standard deviation of temperature in each thermal zone reflects the degree of fluctuation in ambient temperature. This refers to the number of times critical equipment (such as compressors, fans, etc.) is switched per unit of time. The target weighting factor is dynamically set according to the operational strategy.

[0133] Optionally, constructing a dynamic closed-loop feedback link includes: after each control strategy execution, collecting corresponding energy efficiency data and environmental parameters, scoring the strategy execution effect, and updating the control parameter database, wherein the feedback formula is: ,in, For the first Adjustment amount for each strategy parameter; The objective function value; The learning rate (adjustment step size) is used to adjust the strategy in the next cycle, forming an adaptive closed-loop optimization loop with energy efficiency and temperature stability as the core.

[0134] Optionally, the introduction of an operational fluctuation penalty factor includes: defining an operational fluctuation penalty term to suppress the impact of frequent switching on equipment lifespan. ,in, For a moment The strategy combination value; Let be the cost function for policy jitter; This is the penalty intensity factor.

[0135] Alternatively, the objective function can be expanded to: .

[0136] As an optional example, in a real-world deployment, a certain IDC region integrated this mechanism into its air conditioning group control system, using the following settings: Prioritize reducing PUE; To ensure stable temperature fluctuations; To control the switching frequency of equipment, the system collects real-time data on air conditioning energy consumption, room temperature distribution, and equipment operating status every 15 minutes. All feedback updates are completed in real time under the cloud-edge hybrid architecture. After two months of continuous operation and optimization, the following actual results were achieved: the average PUE decreased by approximately 0.03 (e.g., from 1.46 to 1.43); the average temperature difference fluctuation in the data center was controlled within ±1.2℃; the start-stop frequency of air conditioning compressors and fans decreased by approximately 18%, significantly extending the service life of the equipment; and the number of system alarm events decreased by 11%, indicating enhanced operational stability.

[0137] The above embodiments of this application demonstrate that the feedback optimization mechanism effectively improves the utilization efficiency of cooling resources and realizes the self-evolution of equipment operation strategies, proving its strong practical application value in large-scale data center scenarios.

[0138] The embodiments described above provide a dynamic heat intensity sensing mechanism that integrates IT load information to achieve feedforward prediction and proactive response of cooling strategies; construct a differentiated management strategy for hot and cold zones to improve the scheduling accuracy of air conditioning resources; establish an air conditioning resource pooling and elastic allocation system to achieve on-demand cooling and energy saving; introduce edge intelligent control to alleviate the burden on central control and improve system robustness; and achieve self-learning and long-term optimal adjustment of air conditioning strategies through an energy efficiency feedback closed loop. In summary, this application not only helps improve the overall temperature control intelligence level of IDC systems but also effectively reduces operating energy consumption, ensures equipment thermal stability, and provides key support for the construction of green data centers.

[0139] It should be noted that with the development of high-density deployment of data centers (IDC) and frequent fluctuations in computing load, traditional temperature control systems face many challenges. On the one hand, the dynamic correlation between IT equipment load changes and heat distribution is increasing, and the hot and cold loads exhibit high spatial heterogeneity and temporal uncertainty. On the other hand, as one of the main energy sources in data centers, the optimization of the operating efficiency and response mechanism of cooling equipment directly affects the overall power efficiency ratio (PUE) and operating costs.

[0140] This invention provides an optional embodiment of an air conditioning system elastic scheduling and intelligent temperature control method based on IT load heat intensity perception. This method constructs a load-heat correlation model, integrates business operation indicators and thermophysical characteristics to achieve accurate prediction of heat intensity; further, it combines dynamic identification of hot and cold zones and abstraction of air conditioning resource pools to promote on-demand allocation and differentiated control of cooling resources; at the same time, it introduces an edge intelligent inference mechanism to improve system response speed and autonomy, and constructs an energy efficiency feedback closed loop to achieve continuous optimization of air conditioning strategies.

[0141] Figure 2 This is a schematic diagram of an air conditioning resource elastic scheduling method based on load heat intensity sensing according to an embodiment of the present invention, as shown below. Figure 2 As shown, the temperature control method provided in this application can be implemented through an air conditioning resource elastic scheduling method based on load heat intensity sensing, specifically including:

[0142] Step S201: Collect IDC data on load and temperature.

[0143] Step S202: Construct a dynamic sensing model of thermal intensity, that is, construct a load-thermal intensity collaborative sensing model.

[0144] It should be noted that in IDC data centers, the operating load of IT equipment significantly affects the heat generated, and the localized accumulation of heat directly affects the efficiency of air conditioning resource control. Traditional temperature control systems mainly rely on ambient temperature for regulation, ignoring the direct driving effect of IT load and equipment power consumption on heat distribution, thus resulting in problems such as response lag and uneven heat distribution.

[0145] To address the aforementioned issues, this application constructs a load-heat intensity collaborative sensing model. This model dynamically collects operational status data (such as CPU / GPU utilization, current, voltage, and operating power) and thermophysical data (such as infrared thermal imaging, temperature sensors, and airflow velocity) from IT equipment. It then utilizes regression analysis and machine learning to model heat generation patterns, achieving a precise mapping from "load input" to "heat output." This model can provide heat intensity prediction support for subsequent elastic scheduling of air conditioning resources.

[0146] Step S203: Divide the heat intensity area labels.

[0147] It should be noted that as the load density within data centers (IDCs) increases, the problem of localized "hot spots" becomes increasingly prominent. Traditional air conditioning systems often employ a uniform air supply or fixed temperature control mode, lacking differentiated control for areas with varying heat intensities. This leads to insufficient cooling in high-heat areas and energy waste in low-heat areas, affecting equipment stability and reducing power efficiency (PUE). Therefore, by constructing a hot / cold zone identification and zonal control mechanism, the air conditioning system can achieve differentiated group control responses for different micro-regions, thereby improving temperature control accuracy and energy management efficiency.

[0148] Optionally, load areas within the data center can be dynamically divided based on heat intensity distribution to achieve hot and cold zone partitioning, and differentiated air conditioning group control strategies can be generated based on the characteristics of each zone. Through key steps such as spatial heat clustering, strategy mapping, and weight control, a collaborative mechanism of hot and cold zone distribution perception and precise response control can be achieved, further promoting the evolution of the IDC intelligent temperature control system towards refinement.

[0149] Step S204: Based on the air conditioning resource pooling and allocation framework of elastic scheduling priority, a cooling scheduling strategy is generated.

[0150] It should be noted that in traditional IDC data centers, air conditioning equipment is mostly deployed according to physical partitions or zones, with each air conditioner corresponding to a fixed working area, lacking flexible unified resource scheduling capabilities. This "static allocation" method is insufficient in response to dynamic load changes, sudden heat accumulation, or energy consumption optimization needs, often resulting in cooling redundancy or untimely temperature control in hot areas.

[0151] To address this, this application proposes a pooled allocation framework for air conditioning resources based on elastic scheduling priorities. This framework abstracts multiple air conditioning devices into a unified resource pool and, through heat load zoning and scheduling priority mechanisms, dynamically allocates cooling tasks to each air conditioning device on demand, according to time, and by priority. The framework introduces parameters such as "equipment efficiency weight," "regional thermal urgency coefficient," and "scheduling proportion factor," and uses a linear allocation model for elastic allocation, significantly improving the utilization efficiency and intelligent control level of cooling resources.

[0152] Step S205: Collect feedback data in real time.

[0153] It should be noted that in traditional IDC data center air conditioning resource scheduling systems, the adjustment of heating and cooling loads mostly relies on preset static strategies, lacking dynamic feedback mechanisms and energy efficiency self-adaptation capabilities. This leads to the following problems in actual operation: difficulty in continuously optimizing energy efficiency (such as PUE), insufficient environmental temperature stability, and energy waste and reduced lifespan due to frequent equipment switching. By collecting feedback data in real time and evaluating and adjusting the scheduling parameters in the temperature control strategy obtained according to the above temperature control method based on the feedback data, more precise temperature control can be achieved.

[0154] Step S206: Evaluate the effectiveness of the strategy (e.g., energy efficiency, control stability).

[0155] Step S207: Determine whether the optimization target has been achieved. If yes, proceed to step S208; otherwise, proceed to step S209.

[0156] Step S208: Save the current strategy.

[0157] Step S209: Adjust the scheduling parameters and re-execute.

[0158] The embodiments described above provide a zoned cooling identification method based on dynamic modeling of load heat intensity. This method can dynamically perceive the influence of different regional loads on temperature control resources based on real-time IT load intensity and its historical thermal response data, forming refined heat intensity labels, and accordingly making differentiated allocation of air conditioning strategies. This technology breaks the traditional logic of "equal cooling supply" from the source, realizing accurate identification of heat load and hierarchical management of cooling control strategies.

[0159] The above embodiments of this application realize a flexible allocation mechanism for air conditioning cooling resources. It innovatively proposes that the cooling resources of the air conditioning system be dynamically adjusted according to the principle of "heat intensity-driven flexible scheduling". It can automatically adapt the air volume, temperature control zone load distribution and air conditioning power level when the computer room operating status changes, effectively avoid excessive energy redundancy configuration and significantly improve the load response flexibility of the cooling system.

[0160] The above embodiments of this application provide a joint control mechanism based on a cold-heat coupling model (used to analyze regional thermal emergency coefficient and scheduling weight). It introduces a heat exchange mathematical model that links cold and heat resources, fully considers the heat migration law between cold and heat sources, and realizes a joint control strategy for multiple physical quantities such as cold air output power, air supply angle, and air supply channel velocity. This achieves the goal of minimizing global energy dissipation while ensuring temperature control effect.

[0161] The embodiments described above in this application provide a feedback closed-loop and energy efficiency optimization control mechanism for adjusting cooling and heating loads, and construct a feedback optimization mechanism for scheduling cooling and heating resources. After each round of strategy execution, key operating parameters (such as PUE, fan operating status, and regional temperature difference) are automatically collected, and the effectiveness of the strategy is dynamically scored and iteratively updated, realizing a self-learning and self-adjusting "closed-loop temperature control system" to avoid static and rigid strategies.

[0162] The above embodiments of this application improve the intelligent evaluation model of air conditioning energy consumption and load matching. This model proposes a set of energy consumption evaluation index systems based on the degree of adaptation between load heat intensity and cooling behavior. Through comprehensive evaluation of regional heat load, air conditioning output and actual cooling effect, it realizes the quantitative "matching degree" calculation between cooling resource input and actual business load, and provides data support for operation and maintenance optimization and intelligent cooling control.

[0163] Figure 3 This is a schematic diagram of an elastic scheduling sequence of air conditioning resources based on load heat intensity sensing according to an embodiment of the present invention, as shown below. Figure 3 As shown, the user initiates a scheduling process with the monitoring system. The monitoring system provides load (such as operational status data) and temperature data (such as scene data) to the thermal intensity modeling module. The thermal intensity modeling module uses a load-thermal intensity collaborative sensing model to calculate thermal intensity. The thermal intensity modeling module can partition the load areas within the data center based on thermal intensity and output the thermal intensity partitioning results to the monitoring system, such as determining cluster partitions. Then, the monitoring system adds labels to each load area according to thermal intensity and provides the thermal intensity labels of each load area to the policy generation engine. The policy generation engine uses the aforementioned temperature control method to generate a temperature control policy and issues elastic scheduling instructions to the air conditioning control system. After the air conditioning control system performs temperature control according to the temperature control policy, it can return to the monitoring system to execute the air conditioning resource configuration. Subsequently, the monitoring system provides temperature / energy consumption / load feedback to the feedback evaluation module. The feedback evaluation module can generate feedback policy optimization suggestions based on the feedback information and send these suggestions to the policy generation engine to further optimize the temperature control policy generated using the aforementioned temperature control method. After receiving the feedback policy optimization suggestions, the policy generation engine can return the scheduling execution status and energy efficiency results to the user, thus providing feedback to the user.

[0164] The embodiments described above in this application, through intelligent sensing and on-demand cooling mechanisms, avoid energy waste caused by the imbalance between cooling and heating supply and demand in traditional modes. The system's PUE decreases steadily over the long term, effectively supporting the green and low-carbon operation goals of data centers and significantly improving the energy efficiency of the air conditioning system. Relying on the load heat intensity zoning control mechanism, it can respond to the heating status of each area in real time, enabling rapid deployment of cooling resources on demand, greatly reducing extreme situations such as hot spots and overcooling, improving the stability of business operations, and enhancing the stability and regional balance of the room's ambient temperature. It avoids frequent equipment start-ups and shutdowns and full-load operation, alleviates equipment operating pressure through flexible scheduling, extends its maintenance cycle, and reduces the failure rate, thereby reducing maintenance costs while ensuring performance and improving the life cycle and operational flexibility of air conditioning equipment. It forms a highly adaptable and robust closed-loop cooling control system with self-learning and adjustment capabilities, continuously adapting to changes in business load and environment, and improving the cooling stability and strategy flexibility of the IDC data center under complex operating conditions. It provides an expanded foundation for multi-energy coordinated scheduling. Its modular design allows it to be linked with other energy-saving solutions (such as air-cooled-liquid-cooled coordination and renewable energy load transfer), laying a technical foundation for subsequent large-scale intelligent energy consumption optimization.

[0165] According to an embodiment of the present invention, a temperature control device embodiment is also provided. It should be noted that the temperature control device can be used to execute the temperature control method in the embodiment of the present invention, and the temperature control method in the embodiment of the present invention can be executed in the temperature control device.

[0166] Figure 4 This is a schematic diagram of a temperature control device according to an embodiment of the present invention, such as... Figure 4 As shown, the device may include: an acquisition module 42, used to acquire the heat intensity of each load area within the data center, wherein the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; an evaluation module 44, used to evaluate the regional thermal urgency coefficient of each load area based on the heat intensity within each load area, wherein the heat intensity is positively correlated with the regional thermal urgency coefficient, and the regional thermal urgency coefficient represents the immediate demand of the load area for cooling resources; a pooling module 46, used to pool the air conditioning equipment of the data center to obtain a unified air conditioning resource pool, wherein the air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area, the target area being any load area, and the scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning equipment effectively allocated to the target area; and a generation module 48, used to generate a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area, wherein the temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool under the condition of satisfying the thermal urgency coefficient of the target area, and the cooling task is positively correlated with the scheduling weight.

[0167] It should be noted that the initiating module 72 in this embodiment can be used to execute step S102 in this application embodiment, the opening module 74 in this embodiment can be used to execute step S104 in this application embodiment, the sending module 76 in this embodiment can be used to execute step S106 in this application embodiment, and the first closing module 78 in this embodiment can be used to execute step S108 in this application embodiment. The examples and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments.

[0168] In the embodiments described above, since the load of data service equipment is positively correlated with heat intensity, obtaining the heat intensity of each load area within the data center can accurately reflect the heat generation under the operating state of the equipment. The area's thermal urgency coefficient is calculated based on the heat intensity, directly reflecting the immediate demand for cooling resources in the load area, making resource allocation more precise. Subsequently, all air conditioning equipment in the data center is uniformly included in the air conditioning resource pool, and the scheduling weight of each device for the target area is recorded. The scheduling weight reflects the proportion of cooling resources generated by the air conditioning equipment effectively allocated to the target area, ensuring efficient utilization of cooling resources. Then, based on the area thermal urgency coefficient and scheduling weight, a temperature control strategy is generated, clarifying the proportion of cooling tasks allocated to each air conditioning device relative to the total tasks in the air conditioning resource pool, while meeting the thermal urgency coefficient of the target area. This achieves dynamic and intelligent allocation of cooling resources. By evaluating the area thermal urgency coefficient of each load area and the scheduling weight of the air conditioning equipment to generate a temperature control strategy, unreasonable allocation of data center cooling resources is avoided. This not only improves the cooling efficiency of the data center but also reduces energy consumption, meeting the goal of green energy saving and achieving the technical effect of improving temperature control accuracy. This solves the technical problem of low temperature control accuracy in traditional temperature control systems.

[0169] As an optional embodiment, the acquisition module includes: a collection unit, used to collect the operating status data of the data service equipment and the scene data of the load area in each load area, wherein the operating status data includes at least the utilization rate of computing resources in the data service equipment and the power consumption of the data service equipment, both of which are positively correlated with the load; the scene data includes at least the inlet air temperature of the load area, the outlet air temperature of the load area, and the local wind speed of the load area; and a first analysis unit, used to analyze the operating status data and scene data of each load area using a pre-trained load-heat intensity collaborative sensing model to obtain the heat intensity of each load area, wherein the load-heat intensity collaborative sensing model is used to determine the heat intensity based on the weighted result of utilization rate, power consumption, temperature difference between inlet and outlet air temperature, and local wind speed, and in combination with the residual term, wherein the first weight coefficients corresponding to utilization rate, power consumption, temperature difference, and local wind speed, and the residual term are pre-trained.

[0170] As an optional embodiment, the device further includes: a data acquisition submodule, used to acquire, within multiple sampling periods, the operating status data of the data service equipment in each load area and the thermal physical data of each load area before analyzing the operating status data and scene data using the pre-trained load-heat intensity collaborative sensing model to obtain the thermal intensity of each load area, wherein the thermal physical data includes at least: scene data and environmental data for calculating the thermal intensity; a determination submodule, used to determine the operating status data, scene data, and thermal intensity calculated based on the environmental data within the same sampling period as a set of training data; and a training submodule, used to train the load-heat intensity collaborative sensing model based on multiple sets of training data.

[0171] As an optional embodiment, the apparatus further includes: a clustering submodule, configured to perform thermal clustering on multiple load areas based on thermal intensity after obtaining the thermal intensity of each load area within the data center, to obtain clustering results, wherein the clustering results include at least one temperature partition clustering area, and two adjacent load areas whose Euclidean distance between thermal intensities is less than a preset distance threshold are clustered in the same temperature partition clustering area; and a control submodule, configured to perform temperature control according to an air conditioning control strategy pre-set for each temperature partition clustering area.

[0172] As an optional embodiment, the evaluation module includes: a first detection unit for detecting the regional state of each load region, wherein the regional state includes at least: the temperature rise rate of the load region and the alarm state represented by binary variables; and a second analysis unit for analyzing the thermal intensity and regional state of each load region using a pre-trained regional thermal emergency coefficient model to obtain the regional thermal emergency coefficient of each load region, wherein the regional thermal emergency coefficient model is used to determine the regional thermal emergency coefficient based on the weighted result of thermal intensity, temperature rise rate and alarm state, and the second weight coefficients corresponding to thermal intensity, temperature rise rate and alarm state are pre-trained.

[0173] As an optional embodiment, the pooling module includes: a second detection unit for detecting performance data of each air conditioning device, wherein the performance data includes at least: the energy efficiency value of each air conditioning device, the response latency of the air conditioning device, and the cooling coupling coefficient of the air conditioning device to the target area, the cooling coupling coefficient representing the ability of the cooling resources generated by the air conditioning device to effectively act on the target area; and a second analysis unit for analyzing the energy efficiency value, response latency, and cooling coupling coefficient of each air conditioning device using a pre-trained scheduling weight prediction model to determine the scheduling weight of each air conditioning device relative to the target area, wherein the scheduling weight prediction model is: i represents the load area, and k represents the air conditioning equipment. Energy efficiency value, To address latency, when i is the target region, This is the cold coupling coefficient. The scheduling weight is used for generating the air conditioning resource pool based on the scheduling weight of each air conditioning device relative to different target areas.

[0174] As an optional embodiment, the generation module includes: a first determining unit, configured to determine a first ratio by comparing the regional thermal urgency coefficient of the target area with the regional thermal urgency coefficients of all load areas within the data center; a second determining unit, configured to determine a second ratio by comparing the scheduling weight of the target device relative to the target area with the scheduling weight of all air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one air conditioning device in the air conditioning resource pool; a third determining unit, configured to determine a temperature control sub-strategy for the target device to control the temperature of the target area based on the product of the first ratio and the second ratio; and a fourth determining unit, configured to determine a temperature control strategy based on multiple temperature control sub-strategies for different target areas for different target devices.

[0175] Embodiments of the present invention can provide an electronic device, which can be a computer terminal, and the computer terminal can be any one of a group of computer terminal devices. Optionally, in this embodiment, the computer terminal can also be replaced by a mobile terminal or other terminal device.

[0176] Optionally, in this embodiment, the computer terminal may be located in at least one of a plurality of network devices in a computer network.

[0177] In this embodiment, the computer terminal described above can execute program code for the following steps in the temperature control method: obtaining the heat intensity of each load area within the data center, wherein the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; evaluating the regional thermal urgency coefficient of each load area based on the heat intensity within each load area, wherein the heat intensity is positively correlated with the regional thermal urgency coefficient, and the regional thermal urgency coefficient represents the immediate demand of the load area for cooling resources; pooling the air conditioning equipment in the data center to obtain an air conditioning resource pool, wherein the air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area, the target area being any load area, and the scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning equipment effectively allocated to the target area; generating a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area, wherein the temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met, and the cooling task is positively correlated with the scheduling weight.

[0178] Figure 5 This is a structural block diagram of a computer terminal according to an embodiment of the present invention, such as... Figure 5 As shown, the computer terminal 50 may include one or more (only one is shown in the figure) processors 52 and memory 54.

[0179] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the temperature control method and device in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned temperature control method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal 50 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0180] The processor can invoke information and applications stored in memory via a transmission device to perform the following steps: Obtain the heat intensity of each load area within the data center, where a load area is an area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; Based on the heat intensity within each load area, assess the regional thermal urgency coefficient of each load area, where heat intensity is positively correlated with the regional thermal urgency coefficient, which represents the immediate demand for cooling resources in the load area; Pool the air conditioning equipment in the data center to obtain an air conditioning resource pool, where the air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area, where the target area is any load area, and the scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning devices that are effectively allocated to the target area; Based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area, generate a temperature control strategy for temperature control of the data center, where the temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met, and the cooling task is positively correlated with the scheduling weight.

[0181] Optionally, the processor may also execute program code for the following steps: collecting operational status data of data service devices and scene data of the load area within each load area, wherein the operational status data includes at least the utilization rate of computing resources in the data service devices and the power consumption of the data service devices, both of which are positively correlated with the load; the scene data includes at least the inlet air temperature, outlet air temperature, and local wind speed of the load area; and using a pre-trained load-heat intensity collaborative sensing model to analyze the operational status data and scene data of each load area to obtain the heat intensity of each load area, wherein the load-heat intensity collaborative sensing model is used to determine the heat intensity based on the weighted results of utilization rate, power consumption, temperature difference between inlet and outlet air temperatures, and local wind speed, combined with the residual term, wherein the first weight coefficients corresponding to utilization rate, power consumption, temperature difference, and local wind speed, and the residual term are pre-trained.

[0182] Optionally, the processor may also execute program code for the following steps: collecting operational status data of data service devices in each load area and thermal physical data of each load area during multiple sampling periods, wherein the thermal physical data includes at least: scene data and environmental data for calculating thermal intensity; determining the operational status data, scene data, and thermal intensity calculated based on the environmental data in the same sampling period as a set of training data; and training the load-thermal intensity collaborative sensing model based on multiple sets of training data.

[0183] Optionally, the processor may also execute program code for the following steps: performing thermal clustering on multiple load areas based on thermal intensity to obtain clustering results, wherein the clustering results include at least one clustering region, and two adjacent load areas whose Euclidean distance between thermal intensities is less than a preset distance threshold are clustered in the same clustering region; and performing temperature control according to an air conditioning control strategy preset for each clustering region.

[0184] Optionally, the processor may also execute program code that performs the following steps: detects the region status of each load region, wherein the region status includes at least: the temperature rise rate of the load region and the alarm status represented by binary variables; and uses a pre-trained region thermal emergency coefficient model to analyze the thermal intensity and region status of each load region to obtain the region thermal emergency coefficient of each load region, wherein the region thermal emergency coefficient model is used to determine the region thermal emergency coefficient based on the weighted result of thermal intensity, temperature rise rate and alarm status, and the second weight coefficients corresponding to thermal intensity, temperature rise rate and alarm status are pre-trained.

[0185] Optionally, the processor may also execute program code for the following steps: detecting performance data for each air conditioning unit, wherein the performance data includes at least: the energy efficiency value of each air conditioning unit, the response latency of the air conditioning unit, and the cooling coupling coefficient of the air conditioning unit to the target area, the cooling coupling coefficient representing the ability of the cooling resources generated by the air conditioning unit to effectively act on the target area; using a pre-trained scheduling weight prediction model, analyzing the energy efficiency value, response latency, and cooling coupling coefficient of each air conditioning unit, and determining the scheduling weight of each air conditioning unit relative to the target area, wherein the scheduling weight prediction model is: i represents the load area, and k represents the air conditioning equipment. Energy efficiency value, To address latency, when i is the target region, This is the cold coupling coefficient. Assign scheduling weights; generate an air conditioning resource pool based on the scheduling weight of each air conditioning unit relative to different target areas.

[0186] Optionally, the processor may also execute program code for the following steps: determining a first ratio by comparing the regional thermal urgency coefficient of the target area with the regional thermal urgency coefficients of all load areas within the data center; determining a second ratio by comparing the scheduling weight of the target device relative to the target area with the scheduling weight of all air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one of the air conditioning devices in the air conditioning resource pool; determining a temperature control sub-strategy for the target device to control the temperature of the target area based on the product of the first ratio and the second ratio; and determining a temperature control strategy based on multiple temperature control sub-strategies for different target areas for different target devices.

[0187] Those skilled in the art will understand that Figure 5 The structure shown is for illustrative purposes only. The computer terminal can also be a smartphone (such as an Android phone, an iOS phone, etc.), a tablet computer, a mobile internet device (MID), a PAD, and other terminal devices. Figure 5 This does not limit the structure of the aforementioned electronic device. For example, computer terminal 50 may also include components that are more advanced than those described above. Figure 5 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 5 The different configurations shown.

[0188] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a computer program instructing the hardware related to the terminal device. The computer program can be stored in a non-volatile medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc.

[0189] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the temperature control method provided in the above embodiments.

[0190] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0191] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: obtaining the heat intensity of each load area within the data center, wherein the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; evaluating the regional thermal urgency coefficient of each load area based on the heat intensity within each load area, wherein the heat intensity is positively correlated with the regional thermal urgency coefficient, and the regional thermal urgency coefficient represents the immediate demand of the load area for cooling resources; pooling the air conditioning equipment in the data center to obtain an air conditioning resource pool, wherein the air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area, the target area being any load area, and the scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning devices effectively allocated to the target area; generating a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of multiple load areas and the scheduling weights of multiple air conditioning devices for the target area, wherein the temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target area is met, and the cooling task is positively correlated with the scheduling weight.

[0192] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: collecting operating status data of data service devices and scene data of the load area in each load area, wherein the operating status data includes at least: the utilization rate of computing resources in the data service devices and the power consumption of the data service devices, both of which are positively correlated with the load; the scene data includes at least: the inlet air temperature of the load area, the outlet air temperature of the load area, and the local wind speed of the load area; using a pre-trained load-heat intensity collaborative sensing model, analyzing the operating status data and scene data of each load area to obtain the heat intensity of each load area, wherein the load-heat intensity collaborative sensing model is used to determine the heat intensity based on the weighted result of utilization rate, power consumption, temperature difference between inlet and outlet air temperature, and local wind speed, and combined with the residual term, wherein the first weight coefficients corresponding to utilization rate, power consumption, temperature difference, and local wind speed, and the residual term are pre-trained.

[0193] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: collecting operating status data of data service devices in each load area and thermal physical data of each load area during multiple sampling periods, wherein the thermal physical data includes at least: scene data and environmental data for calculating thermal intensity; determining the operating status data, scene data, and thermal intensity calculated based on the environmental data in the same sampling period as a set of training data; and training the load-thermal intensity collaborative sensing model based on multiple sets of training data.

[0194] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: thermally clustering multiple load regions based on thermal intensity to obtain clustering results, wherein the clustering results include at least one clustering region, and two adjacent load regions whose Euclidean distance between thermal intensities is less than a preset distance threshold are clustered in the same clustering region; temperature control is performed according to an air conditioning control strategy preset for each clustering region.

[0195] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: detecting the region status of each load region, wherein the region status includes at least: the temperature rise rate of the load region and the alarm status represented by binary variables; using a pre-trained region thermal emergency coefficient model, analyzing the thermal intensity and region status of each load region to obtain the region thermal emergency coefficient of each load region, wherein the region thermal emergency coefficient model is used to determine the region thermal emergency coefficient based on the weighted result of thermal intensity, temperature rise rate and alarm status, and the second weight coefficients corresponding to thermal intensity, temperature rise rate and alarm status are pre-trained.

[0196] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: detecting performance data of each air conditioning device, wherein the performance data includes at least: the energy efficiency value of each air conditioning device, the response latency of the air conditioning device, and the cooling coupling coefficient of the air conditioning device to the target area, the cooling coupling coefficient representing the ability of the cooling resources generated by the air conditioning device to effectively act on the target area; using a pre-trained scheduling weight prediction model, analyzing the energy efficiency value, response latency, and cooling coupling coefficient of each air conditioning device, and determining the scheduling weight of each air conditioning device relative to the target area, wherein the scheduling weight prediction model is: i represents the load area, and k represents the air conditioning equipment. Energy efficiency value, To address latency, when i is the target region, This is the cold coupling coefficient. Assign scheduling weights; generate an air conditioning resource pool based on the scheduling weight of each air conditioning unit relative to different target areas.

[0197] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining a first ratio by comparing the regional thermal urgency coefficient of the target area with the regional thermal urgency coefficients of all load areas within the data center; determining a second ratio by comparing the scheduling weight of the target device relative to the target area with the scheduling weight of all air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one of the air conditioning devices in the air conditioning resource pool; determining a temperature control sub-strategy for the target device to control the temperature of the target area based on the product of the first ratio and the second ratio; and determining a temperature control strategy based on multiple temperature control sub-strategies for different target areas for different target devices.

[0198] Embodiments of the present invention also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it implements the steps of the temperature control method provided in the above embodiments.

[0199] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0200] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0201] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0202] The units described as separate components may or may not be physically separate. 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 units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0203] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0204] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a non-volatile storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned non-volatile storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0205] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A temperature control method, characterized in that, include: The heat intensity of each load area within the data center is obtained, wherein the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; Based on the heat intensity in each load area, an area thermal urgency factor for each load area is evaluated, wherein the heat intensity is positively correlated with the area thermal urgency factor, and the area thermal urgency factor represents the immediate demand for cooling resources in the load area. The air conditioning equipment in the data center is pooled to obtain an air conditioning resource pool. The air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area. The target area is any load area. The scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning device that are effectively allocated to the target area. Based on the regional thermal urgency coefficients of the multiple load regions and the scheduling weights of the multiple air conditioning devices for the target region, a temperature control strategy for temperature control of the data center is generated. The temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target region is met. The cooling task is positively correlated with the scheduling weight.

2. The method according to claim 1, characterized in that, Obtaining the heat intensity of each load zone within the data center includes: The system collects operational status data of the data service devices and scene data of the load areas within each load area. The operational status data includes at least the utilization rate of computing resources in the data service devices and the power consumption of the data service devices, both of which are positively correlated with the load. The scene data includes at least the inlet air temperature, outlet air temperature, and local wind speed of the load areas. A pre-trained load-heat intensity collaborative sensing model is used to analyze the operating status data and scene data of each load area to obtain the heat intensity of each load area. The load-heat intensity collaborative sensing model is used to determine the heat intensity based on the weighted result of the utilization rate, the power consumption, the temperature difference between the inlet air temperature and the outlet air temperature, and the local wind speed, combined with the residual term. The first weight coefficients corresponding to the utilization rate, the power consumption, the temperature difference, and the local wind speed, as well as the residual term, are pre-trained.

3. The method according to claim 2, characterized in that, Before analyzing the operating status data and the scene data using a pre-trained load-heat intensity co-sensing model to obtain the heat intensity of each load region, the method further includes: During multiple sampling periods, the operating status data of the data service equipment in each load area and the thermophysical data of each load area are collected respectively, wherein the thermophysical data includes at least: the scene data and environmental data for calculating the heat intensity; The operating status data, the scene data, and the heat intensity calculated based on the environmental data within the same sampling period are determined as a set of training data; The load-heat intensity collaborative sensing model is trained based on multiple sets of training data.

4. The method according to claim 1, characterized in that, After obtaining the heat intensity of each load area within the data center, the method further includes: Based on the heat intensity, multiple load regions are thermally clustered to obtain clustering results, wherein the clustering results include at least one clustering region, and two adjacent load regions whose Euclidean distance between the heat intensities is less than a preset distance threshold are clustered in the same clustering region; Temperature control is performed according to the air conditioning control strategy pre-set for each cluster region.

5. The method according to claim 1, characterized in that, The assessment of the regional thermal urgency factor for each of the load regions, based on the thermal intensity within each load region, includes: Detect the region status of each of the load regions, wherein the region status includes at least: the temperature rise rate of the load region and the alarm status represented by a binary variable; A pre-trained regional thermal emergency coefficient model is used to analyze the thermal intensity and regional state of each load region to obtain the regional thermal emergency coefficient of each load region. The regional thermal emergency coefficient model is used to determine the regional thermal emergency coefficient based on the weighted result of the thermal intensity, the temperature rise rate and the alarm state. The second weight coefficients corresponding to the thermal intensity, the temperature rise rate and the alarm state are pre-trained.

6. The method according to claim 1, characterized in that, The air conditioning equipment in the data center is pooled to obtain an air conditioning resource pool, which includes: The performance data of each of the air conditioning devices is detected, wherein the performance data includes at least: the energy efficiency value of each air conditioning device, the response delay of the air conditioning device, and the cooling coupling coefficient of the air conditioning device to the target area, wherein the cooling coupling coefficient represents the ability of the cooling resources generated by the air conditioning device to effectively act on the target area; Using a pre-trained scheduling weight prediction model, the energy efficiency value, response delay, and cold coupling coefficient of each air conditioning unit are analyzed to determine the scheduling weight of each air conditioning unit relative to the target area. The scheduling weight prediction model is as follows: i represents the load area, and k represents the air conditioning equipment. The energy efficiency value is... The response delay is given when i is the target region. The cold coupling coefficient is... The scheduling weight; The air conditioning resource pool is generated based on the scheduling weight of each air conditioning device relative to different target areas.

7. The method according to claim 1, characterized in that, Based on the regional thermal urgency coefficients of the multiple load areas and the scheduling weights of the multiple air conditioning devices for the target area, a temperature control strategy for temperature control of the data center is generated, including: The ratio of the regional thermal emergency coefficient of the target area to the regional thermal emergency coefficient of all load areas within the data center is used to determine a first ratio. The second ratio is determined by the ratio of the scheduling weight of the target device relative to the target area to the scheduling weight of all the air conditioning devices in the air conditioning resource pool relative to the target area, wherein the target device is any one of the air conditioning devices in the air conditioning resource pool; Based on the product of the first ratio and the second ratio, a temperature control sub-strategy is determined for the target device to control the temperature of the target area. The temperature control strategy is determined based on the multiple temperature control sub-strategies for different target areas, according to the different target devices.

8. A temperature control device, characterized in that, include: The acquisition module is used to acquire the heat intensity of each load area within the data center, wherein the load area is the area in the data center where data service equipment is deployed, and the load of the data service equipment is positively correlated with the heat intensity; An evaluation module is used to evaluate the regional thermal emergency coefficient of each load area based on the heat intensity in each load area, wherein the heat intensity is positively correlated with the regional thermal emergency coefficient, and the regional thermal emergency coefficient represents the immediate demand of the load area for cooling resources. The pooling module is used to pool the air conditioning equipment of the data center to obtain a unified air conditioning resource pool. The air conditioning resource pool records the scheduling weight of each air conditioning device relative to a target area. The target area is any load area. The scheduling weight is positively correlated with the proportion of cooling resources generated by the air conditioning device that are effectively allocated to the target area. A generation module is used to generate a temperature control strategy for temperature control of the data center based on the regional thermal urgency coefficients of the multiple load regions and the scheduling weights of the multiple air conditioning devices for the target region, wherein the temperature control strategy at least represents the proportion of the cooling task allocated to each air conditioning device relative to the total task of the air conditioning resource pool, provided that the thermal urgency coefficient of the target region is met, and the cooling task is positively correlated with the scheduling weight.

9. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to execute the temperature control method according to any one of claims 1 to 7 through the computer program.

10. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the temperature control method according to any one of claims 1 to 7.

Images