Data center temperature intelligent adjusting method and system

Abstract

The invention discloses a data center temperature intelligent adjusting method and system. The method comprises the following steps that thermodynamic diagram data, environment data and operation parameters of a data center are obtained through a sensor; establishing a temperature prediction model; acquiring a cabinet temperature change curve in a time period by using the temperature prediction model, and determining a temperature adjustment strategy according to the cabinet temperature change curve by using distributed calculation; cooling heat information is obtained according to the temperature adjusting strategy, heat demand information is obtained from a heat recovery system, a heat recovery channel is determined according to the cooling heat information and the heat demand information, and heat is recovered; a cabinet temperature change curve is obtained by using the temperature prediction model, so that a temperature adjustment strategy can be formulated in time, hardware safety and operation stability of a data center are guaranteed, energy consumption input for temperature adjustment is reduced, heat recovery is performed on air conditioner chilled water and cooling liquid, cooling cost is reduced, secondary utilization of energy is realized, and energy conservation and emission reduction are promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature regulation in a data center, and in particular to a method and system for intelligently regulating the temperature in a data center. Background Art

[0002] In recent years, with the large-scale application of big data, AI, etc., the development of data centers has become more rapid, and the construction of data centers of various levels has increased day by day. At the same time, how to deal with the heat generated when the data center processes data and how to adjust the temperature of the data center have become an important issue in data center applications; Reasonable adjustment of the temperature of the data center can avoid hardware damage and effectively extend the service life of the hardware.

[0003] At present, a Chinese invention with a publication number of CN115344073A discloses a method and system for intelligent temperature regulation of a data center, wherein a temperature sample architecture is determined by various temperature samples batch-calculated by a temperature sample debugging thread; then the key temperature tags corresponding to the temperature description of the device are mined to obtain temperature anomaly expressions; finally, the temperature anomaly expressions are loaded into the temperature sample debugging thread to determine the temperature intelligent control strategy. Although it can effectively reduce costs and improve the accuracy and reliability of the temperature intelligent control strategy, it does not perform temperature prediction for the temperature of the data center, which is not conducive to timely taking temperature strategies for adjustment, and does not formulate corresponding temperature regulation strategies for data centers with abnormal temperatures, but performs temperature regulation for the entire data center, which still results in energy waste, and the absorbed heat of the data center is not reused, resulting in energy waste. Summary of the invention

[0004] The technical problem solved by the present invention is that the prior art does not predict the temperature of a data center, which is not conducive to timely taking temperature strategies for adjustment, and does not formulate corresponding temperature adjustment strategies for data centers with abnormal temperatures. Instead, the temperature of the data center is adjusted as a whole, which still results in energy waste, and the absorbed heat of the data center is not reused, resulting in energy waste.

[0005] In order to solve the above technical problems, in a first aspect, the present invention provides a data center temperature intelligent adjustment method, comprising the following steps:

[0006] Step S1, using sensors to obtain thermal map data, environmental data and operating parameters of the data center;

[0007] Step S2, obtaining a first cabinet center temperature of each cabinet according to the thermal map data, and obtaining a historical cabinet center temperature, historical environmental data, and historical operating parameters, and establishing a temperature prediction model according to the historical cabinet center temperature, the historical environmental data, and the historical operating parameters;

[0008] Step S3, using the temperature prediction model to obtain a cabinet temperature change curve within a time period, and using distributed computing to determine a temperature adjustment strategy according to the cabinet temperature change curve, wherein the temperature adjustment strategy includes a natural temperature adjustment strategy, an air conditioning temperature adjustment strategy, and a liquid temperature adjustment strategy;

[0009] Step S4, obtaining cooling heat information according to the temperature adjustment strategy, and obtaining heat demand information from the heat recovery system, determining a heat recovery channel according to the cooling heat information and the heat demand information, and recovering heat;

[0010] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0011] The steps S1 and S2 specifically include the following steps:

[0012] Using an infrared detector to obtain infrared thermal imaging of the data center, using image processing technology to obtain a grayscale image of the data center based on the infrared thermal imaging, extracting the brightness value of the grayscale image of the data center, and converting the brightness value into a temperature value based on the brightness-temperature correlation relationship, dividing the grayscale image of the data center according to the cabinet position distribution, obtaining the image area and area number of each cabinet, using a temperature sensor and a humidity sensor to obtain environmental data, and extracting the operating parameters of the data center, the environmental data;

[0013] Extract the temperature value of the cabinet center as the first cabinet center temperature, and obtain the historical cabinet center temperature, historical environmental data and historical operating parameters from the data acquisition module, and use machine learning technology to train the model according to the historical cabinet center temperature, historical environmental data and historical operating data. The historical cabinet center temperature, the historical environmental data and the historical operating data are used as model inputs, and the cabinet center temperature is used as the model output. When the model fit reaches the expected fitting threshold, the training is stopped to obtain the temperature prediction model;

[0014] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0015] The step S3 specifically comprises the following steps:

[0016] Step S301, using the temperature prediction model and the center temperature of the first cabinet to obtain a cabinet temperature change curve within a time period, and obtain the temperature peak value of each cabinet temperature change curve, if the temperature peak value is greater than or equal to a first expected threshold, the corresponding cabinet is defined as an abnormal cabinet;

[0017] Step S302, if the temperature peak is higher than the outdoor ambient temperature, obtain the internal and external difference between the temperature peak and the outdoor ambient temperature, if the internal and external difference is greater than the second expected threshold, open the natural ventilation channel, if the temperature peak is lower than or equal to the outdoor ambient temperature, and the temperature peak is less than the third expected threshold, start the air conditioning temperature control strategy;

[0018] Step S303: define the air conditioner in the data center as an air conditioner agent, and obtain the state space S (tran 1 , tran 2 , tran 3 , ... tran n ), the tran n is the cabinet temperature in the air-conditioning area, and obtains the action space a of the air-conditioning agent i (f i , t i ), the f i For air conditioning i The air supply speed, t i For air conditioning i The air supply temperature is obtained, and the air conditioning agent in the area where the abnormal cabinet is located is obtained according to the state space S, and the action space of the air conditioning agent is determined according to the temperature peak value and temperature change curve of the abnormal cabinet;

[0019] Step S304, obtaining a reward function and a penalty function of the action space, obtaining the sum of the action spaces of all air conditioners in the data center, obtaining a value network of the sum of the action spaces using the penalty function and the reward function, evaluating a reward value of the sum of the action spaces according to the value network, adjusting the sum of the action spaces according to the reward value, and obtaining an air conditioning temperature adjustment strategy according to the action space with the largest reward value;

[0020] Step S305: if the temperature peak value is greater than or equal to the third expected threshold, start the liquid temperature control strategy of the corresponding cabinet;

[0021] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0022] The step S4 specifically comprises the following steps:

[0023] Step S401, predicting air conditioner cooling heat information and coolant cooling heat information according to the temperature adjustment strategy, air conditioner efficiency and coolant efficiency, wherein the air conditioner cooling heat information includes air conditioner number, cooling time, chilled water initial temperature, chilled water recovery temperature and chilled water absorbed heat, and the coolant heat includes liquid device number, liquid cooling time, liquid initial temperature, liquid recovery temperature and liquid absorbed heat;

[0024] Step S402, matching a heating device in the heating system according to the air conditioner cooling heat information and the coolant cooling heat information, and using the heating device to recover the air conditioner chilled water and the cooling liquid of the liquid cooling device;

[0025] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0026] Determining the action space of the air-conditioning agent according to the temperature peak value and temperature change curve of the abnormal cabinet includes:

[0027] Acquire historical temperature control data, wherein the historical temperature control data includes historical air-conditioning operation data and historical cabinet cooling data, perform correlation analysis on the historical air-conditioning operation data and the historical cabinet cooling temperature, acquire an air-conditioning cooling function, acquire a first strategy network according to the abnormal cabinet number using the air-conditioning cooling function, perform reward training on the first strategy network using a reward function, acquire a reward value of the first strategy function, punish the first strategy network using a penalty function, acquire a penalty value of the first strategy network, adjust the first strategy network according to the penalty value, acquire a second strategy network, and acquire strategy values ​​of each strategy network using a value network, define the strategy network with the highest strategy value as an air-conditioning temperature control strategy, the air-conditioning temperature control strategy includes an action space of each air-conditioning agent, and the air-conditioning cooling function is used to represent the air-conditioning operation data a i (f i ,t i ,T i ,E i ) and each cabinet cooling data Com j (T j-begin ,T j-end ), where f i is the air supply speed of the air conditioner, t i is the air supply temperature of the air conditioner, T i is the running time of the air conditioner, E i is the air conditioning energy consumption, T j-begin is the initial temperature of the cabinet, T j-end Cooling temperature for the cabinet.

[0028] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0029] The penalty function is expressed as follows:

[0030]

[0031] Among them, P i is the action space penalty value of air conditioner number i, is the penalty coefficient, Tj-end is the cabinet cooling temperature, T j-begin is the initial temperature of the cabinet, λ is a real number greater than zero;

[0032] The reward function is expressed as follows:

[0033] R i =-[βP i +(1-β)*E i *ω]

[0034] Among them, R i is the action space reward value of air conditioner number i, β is the penalty value weight, E i is the energy consumption value of air conditioner number i, ω is the reward coefficient;

[0035] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0036] The operating parameters include the cabinet power-on time, cabinet running time, CPU usage, memory usage and I / O request quantity;

[0037] The environmental data includes outdoor ambient temperature and outdoor ambient humidity;

[0038] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0039] The historical air conditioner operation data includes the operating air conditioner number, the operating air conditioner air supply temperature, the operating air conditioner air supply speed and the air conditioner operation time, and the historical cabinet cooling data includes the cabinet number, the cabinet initial temperature and the cabinet cooling temperature;

[0040] As a preferred solution of the data center temperature intelligent adjustment method described in the present invention, wherein:

[0041] Using the value network to obtain the strategic value of each strategic network includes:

[0042] Obtaining a reward value and a penalty value of each air conditioner in each strategy network, performing weighted calculation on the reward value and the penalty value, and obtaining a strategy value of each strategy network;

[0043] Adjusting the first strategy network according to the penalty value includes:

[0044] A first mapping relationship is obtained through the penalty value, wherein the first mapping relationship includes a correspondence between the penalty value and the action space of the corresponding air conditioner, a first adjustment strategy is obtained according to the first mapping relationship, and the first strategy network is adjusted according to the first adjustment strategy to obtain a second strategy network.

[0045] In a second aspect, the present invention provides a data center temperature intelligent adjustment system, including a data acquisition module, a temperature prediction module, a temperature adjustment strategy module and a waste heat recovery module;

[0046] The data acquisition module is used to acquire thermal map data, environmental data and operating parameters of the data center;

[0047] The temperature prediction module is used to obtain historical cabinet center temperature, historical environmental data and historical operating parameters, and establish a temperature prediction model according to the historical cabinet center temperature, the historical environmental data and the historical operating parameters;

[0048] The temperature adjustment strategy module is used to determine the temperature adjustment strategy according to the cabinet temperature change curve by using distributed computing;

[0049] The waste heat recovery module is used to obtain cooling heat information according to the temperature adjustment strategy and obtain heat demand information from the heat recovery system.

[0050] Beneficial effects of the present invention: The present invention establishes a temperature prediction model based on historical cabinet center temperature, historical environmental data and historical operating parameters, and uses the temperature prediction model to obtain the cabinet temperature change curve within a time period, which is conducive to timely formulating a temperature adjustment strategy according to the cabinet temperature change curve to ensure the hardware safety and operating stability of the data center.

[0051] Distributed computing is used to develop temperature control strategies for data centers, and targeted temperature control is performed on abnormal cabinets to achieve intelligent and targeted temperature control, which can effectively reduce energy consumption for temperature control, reduce energy consumption, and reduce cooling costs.

[0052] According to the temperature adjustment strategy, air conditioning efficiency and coolant efficiency, the air conditioning cooling heat information and coolant cooling heat information are obtained, and the air conditioning cooling heat information and coolant cooling heat information are used to match the heating equipment nearby, so as to recycle and reuse the air conditioning cooling water and cooling liquid, thereby reducing the waste heat waste in the data center, realizing the secondary utilization of energy, and promoting the realization of energy conservation and emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A basic flow chart of a method for intelligently adjusting temperature in a data center provided by an embodiment of the present invention.

[0054] Figure 2 A basic flow chart of a data center temperature intelligent adjustment system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0056] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a data center temperature intelligent adjustment method, comprising the following steps:

[0057] Step S1, using sensors to obtain thermal map data, environmental data and operating parameters of the data center;

[0058] Step S2, obtaining a first cabinet center temperature of each cabinet according to the thermal map data, and obtaining a historical cabinet center temperature, historical environmental data, and historical operating parameters, and establishing a temperature prediction model according to the historical cabinet center temperature, the historical environmental data, and the historical operating parameters;

[0059] Step S3, using the temperature prediction model to obtain a cabinet temperature change curve within a time period, and using distributed computing to determine a temperature adjustment strategy according to the cabinet temperature change curve, wherein the temperature adjustment strategy includes a natural temperature adjustment strategy, an air conditioning temperature adjustment strategy, and a liquid temperature adjustment strategy;

[0060] Step S4, obtaining cooling heat information according to the temperature adjustment strategy, and obtaining heat demand information from the heat recovery system, determining a heat recovery channel according to the cooling heat information and the heat demand information, and recovering heat.

[0061] In this embodiment, a temperature prediction model is established based on the historical cabinet center temperature, historical environmental data and historical operating parameters. The temperature prediction model is used to obtain the cabinet temperature change curve within a time period, which is conducive to timely formulating temperature adjustment strategies according to the cabinet temperature change curve to ensure the hardware security and operation stability of the data center.

[0062] Distributed computing is used to develop temperature control strategies for data centers, and targeted temperature control is performed on abnormal cabinets to achieve intelligent and targeted temperature control, which can effectively reduce energy consumption for temperature control, reduce energy consumption, and reduce cooling costs.

[0063] According to the temperature adjustment strategy, air conditioning efficiency and coolant efficiency, the air conditioning cooling heat information and coolant cooling heat information are obtained, and the air conditioning cooling heat information and coolant cooling heat information are used to match the heating equipment nearby, so as to recycle and reuse the air conditioning cooling water and cooling liquid, thereby reducing the waste heat waste in the data center, realizing the secondary utilization of energy, and promoting the realization of energy conservation and emission reduction.

[0064] The steps S1 and S2 specifically include the following steps:

[0065] Using an infrared detector to obtain infrared thermal imaging of the data center, using image processing technology to obtain a grayscale image of the data center based on the infrared thermal imaging, extracting the brightness value of the grayscale image of the data center, and converting the brightness value into a temperature value based on the brightness-temperature correlation relationship, dividing the grayscale image of the data center according to the cabinet position distribution, obtaining the image area and area number of each cabinet, using a temperature sensor and a humidity sensor to obtain environmental data, and extracting the operating parameters of the data center, the environmental data;

[0066] The temperature value of the cabinet center is extracted as the first cabinet center temperature, and the historical cabinet center temperature, historical environmental data and historical operating parameters are obtained from the data acquisition module. The model is trained using machine learning technology according to the historical cabinet center temperature, historical environmental data and historical operating data. The historical cabinet center temperature, the historical environmental data and the historical operating data are used as model inputs, and the cabinet center temperature is used as model output. When the model fit reaches the expected fitting threshold, the training is stopped to obtain the temperature prediction model.

[0067] In this embodiment, the infrared thermal imaging means that the infrared specific band signal of the thermal radiation of the object is converted into images and graphics that can be distinguished by human vision;

[0068] In this embodiment, the grayscale image represents an image in which each pixel has only one sampled color;

[0069] In this embodiment, the fitting expected threshold is 96%.

[0070] In this embodiment, a temperature prediction model is established based on the historical cabinet center temperature, historical environmental data and historical operating parameters, and the temperature prediction model is used to obtain the cabinet temperature change curve within a time period, which is conducive to timely formulating temperature adjustment strategies according to the cabinet temperature change curve to ensure the hardware security and operation stability of the data center.

[0071] The step S3 specifically comprises the following steps:

[0072] Step S301, using the temperature prediction model and the center temperature of the first cabinet to obtain a cabinet temperature change curve within a time period, and obtain the temperature peak value of each cabinet temperature change curve, if the temperature peak value is greater than or equal to a first expected threshold, the corresponding cabinet is defined as an abnormal cabinet;

[0073] Step S302, if the temperature peak is higher than the outdoor ambient temperature, obtain the internal and external difference between the temperature peak and the outdoor ambient temperature, if the internal and external difference is greater than the second expected threshold, open the natural ventilation channel, if the temperature peak is lower than or equal to the outdoor ambient temperature, and the temperature peak is less than the third expected threshold, start the air conditioning temperature control strategy;

[0074] Step S303: define the air conditioner in the data center as an air conditioner agent, and obtain the state space S (tran 1 , tran 2 , tran 3 , ... tran n ), the tran n is the cabinet temperature in the air-conditioning area, and obtains the action space a of the air-conditioning agent i (f i , t i ), the f i For air conditioning i The air supply speed, t i For air conditioning i The air supply temperature is obtained, and the air conditioning agent in the area where the abnormal cabinet is located is obtained according to the state space S, and the action space of the air conditioning agent is determined according to the temperature peak value and temperature change curve of the abnormal cabinet;

[0075] Step S304, obtaining a reward function and a penalty function of the action space, obtaining the sum of the action spaces of all air conditioners in the data center, obtaining a value network of the sum of the action spaces using the penalty function and the reward function, evaluating a reward value of the sum of the action spaces according to the value network, adjusting the sum of the action spaces according to the reward value, and obtaining an air conditioning temperature adjustment strategy according to the action space with the largest reward value;

[0076] Step S305: if the temperature peak value is greater than or equal to the third expected threshold, the liquid temperature control strategy of the corresponding cabinet is started.

[0077] In this embodiment, distributed computing is used to formulate a temperature control strategy for the data center based on the cabinet temperature change curve, and targeted temperature control is performed on abnormal cabinets, thereby realizing intelligent and targeted temperature control, which can effectively reduce the energy consumption input for temperature control, reduce energy consumption, and reduce cooling costs.

[0078] The step S4 specifically comprises the following steps:

[0079] Step S401, predicting air conditioner cooling heat information and coolant cooling heat information according to the temperature adjustment strategy, air conditioner efficiency and coolant efficiency, wherein the air conditioner cooling heat information includes air conditioner number, cooling time, chilled water initial temperature, chilled water recovery temperature and chilled water absorbed heat, and the coolant heat includes liquid device number, liquid cooling time, liquid initial temperature, liquid recovery temperature and liquid absorbed heat;

[0080] Step S402, matching heating equipment in the heating system according to the air conditioner cooling heat information and the coolant cooling heat information, and using the heating equipment to recover the air conditioner chilled water and the cooling liquid of the liquid cooling device.

[0081] In this embodiment, the air conditioning cooling heat information and the coolant cooling heat information are obtained according to the temperature adjustment strategy, the air conditioning efficiency and the coolant efficiency, and the air conditioning cooling heat information and the coolant cooling heat information are used to match the heating equipment nearby, so as to recycle and reuse the air conditioning cooling water and the cooling liquid, thereby reducing the waste heat waste of the data center, realizing the secondary utilization of energy, and promoting the realization of energy conservation and emission reduction.

[0082] Determining the action space of the air-conditioning agent according to the temperature peak value and temperature change curve of the abnormal cabinet includes:

[0083] Acquire historical temperature control data, wherein the historical temperature control data includes historical air-conditioning operation data and historical cabinet cooling data, perform correlation analysis on the historical air-conditioning operation data and the historical cabinet cooling temperature, acquire an air-conditioning cooling function, acquire a first strategy network according to the abnormal cabinet number using the air-conditioning cooling function, perform reward training on the first strategy network using a reward function, acquire a reward value of the first strategy function, punish the first strategy network using a penalty function, acquire a penalty value of the first strategy network, adjust the first strategy network according to the penalty value, acquire a second strategy network, and acquire strategy values ​​of each strategy network using a value network, define the strategy network with the highest strategy value as an air-conditioning temperature control strategy, the air-conditioning temperature control strategy includes an action space of each air-conditioning agent, and the air-conditioning cooling function is used to represent the air-conditioning operation data a i (f i ,t i ,T i ,E i ) and each cabinet cooling data Com j (T j-begin ,T j-end ), where f i is the air supply speed of the air conditioner, t i is the air supply temperature of the air conditioner, T i is the running time of the air conditioner, E i is the air conditioning energy consumption, T j-beginis the initial temperature of the cabinet, T j-end Cooling temperature for the cabinet.

[0084] The strategy network is an air conditioning temperature control strategy composed of the action spaces of each air conditioner.

[0085] In this embodiment, the action space of the air-conditioning intelligent body is determined according to the temperature peak value and temperature change curve of the abnormal cabinet, the optimal air-conditioning temperature control strategy is obtained, and the temperature of the abnormal cabinet is adjusted in a targeted manner according to the air-conditioning temperature control strategy, thereby realizing intelligent and targeted temperature control, which can effectively reduce the energy consumption input for temperature control, reduce energy consumption, and reduce cooling costs.

[0086] The penalty function is expressed as follows:

[0087]

[0088] Among them, P i is the action space penalty value of air conditioner number i, is the penalty coefficient, T j-end is the cabinet cooling temperature, T j-begin is the initial temperature of the cabinet, λ is a real number greater than zero;

[0089] The reward function is expressed as follows:

[0090] R i =-[βP i +(1-β)*E i *ω]

[0091] Among them, R i is the action space reward value of air conditioner number i, β is the penalty value weight, E i is the energy consumption value of air conditioner numbered i, and ω is the reward coefficient.

[0092] In this embodiment, the reward function and penalty function of the air conditioning temperature control strategy are obtained, which provides specific function support for obtaining the reward value and penalty value of the strategy network, and ensures the reliability and accuracy of evaluating and adjusting the strategy network based on the penalty value and reward value.

[0093] The operating parameters include the cabinet power-on time, cabinet running time, CPU usage, memory usage and I / O request quantity;

[0094] The environmental data includes outdoor environmental temperature and outdoor environmental humidity.

[0095] In this embodiment, obtaining operating parameters and environmental data provides specific and detailed data support for establishing a temperature prediction model, and is conducive to real-time understanding of the operating conditions of the data center and the impact of environmental data on the operating conditions of the data center.

[0096] The historical air conditioner operation data includes the operating air conditioner number, the operating air conditioner supply air temperature, the operating air conditioner supply air speed and the air conditioner operation time, and the historical cabinet cooling data includes the cabinet number, the cabinet initial temperature and the cabinet cooling temperature.

[0097] In this embodiment, the air conditioning cooling function is obtained by acquiring historical air conditioning operation data and historical cabinet cooling data, which provides detailed and specific data support for obtaining the air conditioning cooling function, which is conducive to understanding the temperature cooling effect of each air conditioner on each cabinet according to the air conditioning cooling function and formulating a strategy network.

[0098] Using the value network to obtain the strategic value of each strategic network includes:

[0099] Obtaining a reward value and a penalty value of each air conditioner in each strategy network, performing weighted calculation on the reward value and the penalty value, and obtaining a strategy value of each strategy network;

[0100] Adjusting the first strategy network according to the penalty value includes:

[0101] A first mapping relationship is obtained through the penalty value, wherein the first mapping relationship includes a correspondence between the penalty value and the action space of the corresponding air conditioner, a first adjustment strategy is obtained according to the first mapping relationship, and the first strategy network is adjusted according to the first adjustment strategy to obtain a second strategy network.

[0102] In this embodiment, the value network is used to obtain the strategy value, and the strategy network is adjusted according to the penalty value, which is conducive to obtaining the optimal strategy network. While performing targeted temperature adjustment, the energy consumption input for temperature adjustment is effectively reduced, energy consumption is reduced, and cooling costs are reduced.

[0103] Example 2, reference Figure 2 , which is another embodiment of the present invention, which is different from the first embodiment in that it provides a data center temperature intelligent adjustment system, characterized in that it includes a data acquisition module, a temperature prediction module, a temperature adjustment strategy module and a waste heat recovery module;

[0104] The data acquisition module is used to acquire thermal map data, environmental data and operating parameters of the data center;

[0105] The temperature prediction module is used to obtain historical cabinet center temperature, historical environmental data and historical operating parameters, and establish a temperature prediction model according to the historical cabinet center temperature, the historical environmental data and the historical operating parameters;

[0106] The temperature adjustment strategy module is used to determine the temperature adjustment strategy according to the cabinet temperature change curve by using distributed computing;

[0107] The waste heat recovery module is used to obtain cooling heat information according to the temperature adjustment strategy and obtain heat demand information from the heat recovery system.

[0108] The present invention establishes a temperature prediction model based on historical cabinet center temperature, historical environmental data and historical operating parameters, and uses the temperature prediction model to obtain the cabinet temperature change curve within a time period, which is conducive to timely formulating a temperature adjustment strategy according to the cabinet temperature change curve to ensure the hardware safety and operation stability of the data center.

[0109] Distributed computing is used to develop temperature control strategies for data centers, and targeted temperature control is performed on abnormal cabinets to achieve intelligent and targeted temperature control, which can effectively reduce energy consumption for temperature control, reduce energy consumption, and reduce cooling costs.

[0110] According to the temperature adjustment strategy, air conditioning efficiency and coolant efficiency, the air conditioning cooling heat information and coolant cooling heat information are obtained, and the air conditioning cooling heat information and coolant cooling heat information are used to match the heating equipment nearby, so as to recycle and reuse the air conditioning cooling water and cooling liquid, thereby reducing the waste heat waste in the data center, realizing the secondary utilization of energy, and promoting the realization of energy conservation and emission reduction.

[0111] It should be understood by those skilled in the art that the embodiments of the present invention can be provided as methods, systems or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, disk or optical disk. These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A data center temperature intelligent adjustment method, characterized in that: The following steps are involved: Step S1, using sensors to obtain thermal map data, environmental data and operating parameters of the data center; Step S2, obtaining a first cabinet center temperature of each cabinet according to the thermal map data, and obtaining a historical cabinet center temperature, historical environmental data, and historical operating parameters, and establishing a temperature prediction model according to the historical cabinet center temperature, the historical environmental data, and the historical operating parameters; Step S3, using the temperature prediction model to obtain a cabinet temperature change curve within a time period, and using distributed computing to determine a temperature adjustment strategy according to the cabinet temperature change curve, wherein the temperature adjustment strategy includes a natural temperature adjustment strategy, an air conditioning temperature adjustment strategy, and a liquid temperature adjustment strategy; Step S4, obtaining cooling heat information according to the temperature adjustment strategy, and obtaining heat demand information from the heat recovery system, determining a heat recovery channel according to the cooling heat information and the heat demand information, and recovering heat.

2. A data center temperature intelligent adjustment method according to claim 1, characterized in that: The steps S1 and S2 specifically include the following steps: Using an infrared detector to obtain infrared thermal imaging of the data center, using image processing technology to obtain a grayscale image of the data center based on the infrared thermal imaging, extracting the brightness value of the grayscale image of the data center, and converting the brightness value into a temperature value based on the brightness-temperature correlation relationship, dividing the grayscale image of the data center according to the cabinet position distribution, obtaining the image area and area number of each cabinet, using a temperature sensor and a humidity sensor to obtain environmental data, and extracting the operating parameters of the data center, the environmental data; The temperature value of the cabinet center is extracted as the first cabinet center temperature, and the historical cabinet center temperature, historical environmental data and historical operating parameters are obtained from the data acquisition module. The model is trained using machine learning technology according to the historical cabinet center temperature, historical environmental data and historical operating data. The historical cabinet center temperature, the historical environmental data and the historical operating data are used as model inputs, and the cabinet center temperature is used as model output. When the model fit reaches the expected fitting threshold, the training is stopped to obtain the temperature prediction model.

3. A data center temperature intelligent adjustment method according to claim 1, characterized in that: The step S3 specifically comprises the following steps: Step S301, using the temperature prediction model and the center temperature of the first cabinet to obtain a cabinet temperature change curve within a time period, and obtain the temperature peak value of each cabinet temperature change curve, if the temperature peak value is greater than or equal to a first expected threshold, the corresponding cabinet is defined as an abnormal cabinet; Step S302, if the temperature peak is higher than the outdoor ambient temperature, obtain the internal and external difference between the temperature peak and the outdoor ambient temperature, if the internal and external difference is greater than the second expected threshold, open the natural ventilation channel, if the temperature peak is lower than or equal to the outdoor ambient temperature, and the temperature peak is less than the third expected threshold, start the air conditioning temperature control strategy; Step S303: define the air conditioner of the data center as an air conditioner agent, and obtain the state space S (tran1, tran2, tran3, ... tran n ), the tran n is the cabinet temperature in the air-conditioning area, and obtains the action space a of the air-conditioning agent i (f i , t i ), the f i For air conditioning i The air supply speed, t i For air conditioning i The air supply temperature is obtained, and the air conditioning agent in the area where the abnormal cabinet is located is obtained according to the state space S, and the action space of the air conditioning agent is determined according to the temperature peak value and temperature change curve of the abnormal cabinet; Step S304, obtaining a reward function and a penalty function of the action space, obtaining the sum of the action spaces of all air conditioners in the data center, obtaining a value network of the sum of the action spaces using the penalty function and the reward function, evaluating a reward value of the sum of the action spaces according to the value network, adjusting the sum of the action spaces according to the reward value, and obtaining an air conditioning temperature adjustment strategy according to the action space with the largest reward value; Step S305: if the temperature peak value is greater than or equal to the third expected threshold, the liquid temperature control strategy of the corresponding cabinet is started.

4. The method for intelligently adjusting temperature of a data center according to claim 1, characterized in that: The step S4 specifically comprises the following steps: Step S401, predicting air conditioner cooling heat information and coolant cooling heat information according to the temperature adjustment strategy, air conditioner efficiency and coolant efficiency, wherein the air conditioner cooling heat information includes air conditioner number, cooling time, chilled water initial temperature, chilled water recovery temperature and chilled water absorbed heat, and the coolant heat includes liquid device number, liquid cooling time, liquid initial temperature, liquid recovery temperature and liquid absorbed heat; Step S402, matching heating equipment in the heating system according to the air conditioner cooling heat information and the coolant cooling heat information, and using the heating equipment to recover the air conditioner chilled water and the cooling liquid of the liquid cooling device.

5. A data center temperature intelligent adjustment method as claimed in claim 3, characterized in that: Determining the action space of the air-conditioning agent according to the temperature peak value and temperature change curve of the abnormal cabinet includes: Acquire historical temperature control data, wherein the historical temperature control data includes historical air-conditioning operation data and historical cabinet cooling data, perform correlation analysis on the historical air-conditioning operation data and the historical cabinet cooling temperature, acquire an air-conditioning cooling function, acquire a first strategy network according to the abnormal cabinet number using the air-conditioning cooling function, perform reward training on the first strategy network using a reward function, acquire a reward value of the first strategy function, punish the first strategy network using a penalty function, acquire a penalty value of the first strategy network, adjust the first strategy network according to the penalty value, acquire a second strategy network, and acquire strategy values ​​of each strategy network using a value network, define the strategy network with the highest strategy value as an air-conditioning temperature control strategy, the air-conditioning temperature control strategy includes an action space of each air-conditioning agent, and the air-conditioning cooling function is used to represent the air-conditioning operation data a i (f i ,t i ,T i ,E i ) and each cabinet cooling data Com j (T j-begin ,T j-end ), where f i is the air supply speed of the air conditioner, t i is the air supply temperature of the air conditioner, T i is the running time of the air conditioner, E i is the air conditioning energy consumption, T j-begin is the initial temperature of the cabinet, T j-end Cooling temperature for the cabinet.

6. A data center temperature intelligent adjustment method as claimed in claim 3, characterized in that: The penalty function is expressed as follows: Among them, P i is the action space penalty value of air conditioner number i, is the penalty coefficient, T j-end is the cabinet cooling temperature, T j-begin is the initial temperature of the cabinet, λ is a real number greater than zero; The reward function is expressed as follows: R i =-[βP i +(1-β)*E i *oh] Among them, R i is the action space reward value of air conditioner number i, β is the penalty value weight, E i is the energy consumption value of air conditioner numbered i, and ω is the reward coefficient.

7. A data center temperature intelligent adjustment method according to claim 1, characterized in that: The operating parameters include the cabinet power-on time, cabinet running time, CPU usage, memory usage and I / O request quantity; The environmental data includes outdoor environmental temperature and outdoor environmental humidity.

8. A data center temperature intelligent adjustment method as claimed in claim 5, characterized in that: The historical air conditioner operation data includes the operating air conditioner number, the operating air conditioner supply air temperature, the operating air conditioner supply air speed and the air conditioner operation time, and the historical cabinet cooling data includes the cabinet number, the cabinet initial temperature and the cabinet cooling temperature.

9. A data center temperature intelligent adjustment method as claimed in claim 3, characterized in that: Using the value network to obtain the strategic value of each strategic network includes: Obtaining a reward value and a penalty value of each air conditioner in each strategy network, performing weighted calculation on the reward value and the penalty value, and obtaining a strategy value of each strategy network; Adjusting the first strategy network according to the penalty value includes: A first mapping relationship is obtained through the penalty value, wherein the first mapping relationship includes a correspondence between the penalty value and the action space of the corresponding air conditioner, a first adjustment strategy is obtained according to the first mapping relationship, and the first strategy network is adjusted according to the first adjustment strategy to obtain a second strategy network.

10. A data center temperature intelligent adjustment system, characterized in that: It includes data acquisition module, temperature prediction module, temperature adjustment strategy module and waste heat recovery module; The data acquisition module is used to acquire thermal map data, environmental data and operating parameters of the data center; The temperature prediction module is used to obtain historical cabinet center temperature, historical environmental data and historical operating parameters, and establish a temperature prediction model according to the historical cabinet center temperature, the historical environmental data and the historical operating parameters; The temperature adjustment strategy module is used to determine the temperature adjustment strategy according to the cabinet temperature change curve by using distributed computing; The waste heat recovery module is used to obtain cooling heat information according to the temperature adjustment strategy and obtain heat demand information from the heat recovery system.

Images