Intelligent liquid cooling control method and system for data center

Abstract

The invention provides an intelligent liquid cooling control method and system for a data center, and belongs to the technical field of liquid cooling, and the method comprises the following steps: S101, obtaining an optimization constraint condition, and determining a flow parameter interval, a rotating speed parameter interval and a cooling parameter interval based on the optimization constraint condition; s102, acquiring an initial antibody population based on an artificial immune algorithm; s103, acquiring an affinity function and affinity, screening the plurality of random antibody feature vectors based on the affinity, and retaining a plurality of preferred antibody feature vectors; s104, performing cloning and mutation operation on the plurality of preferable antibody feature vectors to obtain a plurality of preferable antibody feature vector copies and a plurality of variant antibody feature vectors; s105, establishing a new antibody population, carrying out iterative optimization and adopting a mutation confrontation strategy; and S106, when the maximum number of iteration optimization times is reached, an optimization feature vector with the highest affinity is exported and imported into a liquid cooling control module, and the liquid cooling control module controls a cooling unit to be adjusted to corresponding parameters so as to save energy consumption.

Description

Technical Field

[0001] The present invention belongs to the field of liquid cooling technology, and more specifically, particularly relates to an intelligent liquid cooling control method and system for a data center. Background Art

[0002] When the IT equipment in the data control center is running, the large amount of heat it generates will affect the operating performance of the IT equipment if it cannot be effectively dissipated. Liquid cooling is an efficient heat dissipation method. Its principle is to remove and dissipate the heat of the IT equipment through the continuous circulation of the coolant. However, when using liquid cooling, the energy consumption is relatively large. Traditional liquid cooling control often only focuses on the performance of the heat dissipation effect, and pays relatively little attention to energy consumption. For example, under the premise of ensuring the heat dissipation capacity, a higher coolant flow rate and water pump speed, as well as a higher power cooling distribution unit, are often used, resulting in a higher overall energy consumption of liquid cooling. Summary of the invention

[0003] In order to solve the above technical problems, the present invention provides a data center intelligent liquid cooling control method and system to solve the technical problem in the prior art that traditional liquid cooling control often has high energy consumption while ensuring the heat dissipation effect.

[0004] The purpose and effect of the intelligent liquid cooling control method and system for a data center of the present invention are achieved by the following specific technical means: A data center intelligent liquid cooling control method comprises the following steps: S101: Acquire a set temperature parameter, where the set temperature parameter is a set cooling temperature, acquire an optimization constraint condition based on the set temperature parameter, determine a flow parameter interval, a speed parameter interval, and a cooling parameter interval based on the optimization constraint condition, and the optimization constraint condition constrains the flow parameter interval, the speed parameter interval, and the cooling parameter interval; S102: Based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, a plurality of flow parameters, a plurality of speed parameters and a plurality of cooling parameters are obtained, wherein the flow parameter is a coolant flow rate of a flow unit, the speed parameter is a speed of a cooling water pump, and the cooling parameter is a power of a liquid cooling distribution unit; and an initial antibody population is obtained based on an artificial immune algorithm, wherein the initial antibody population is an individual set of a plurality of random antibody feature vectors; S103: obtaining an affinity function based on an artificial immune algorithm, importing the initial antibody population into the affinity function, obtaining a plurality of affinities, wherein the plurality of affinities respectively correspond to a plurality of random antibody feature vectors, screening the plurality of random antibody feature vectors based on the plurality of affinities and through a tournament selection strategy, and screening and retaining a plurality of preferred antibody feature vectors; S104: performing a cloning operation on the plurality of preferred antibody feature vectors, obtaining a plurality of preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the plurality of preferred antibody feature vector copies to the plurality of antibody feature vectors is four to one, and performing a mutation operation on the plurality of preferred antibody feature vector copies to obtain a plurality of mutated antibody feature vectors; S105: establishing a new antibody population based on the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors, and the multiple preferred antibody feature vectors, wherein the new antibody population is an individual collection of the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors, and the multiple preferred antibody feature vectors, establishing memory cells, performing iterative optimization based on the new antibody population, and performing a mutation countermeasure strategy during the iterative optimization process; S106: When the maximum number of iterative optimization times is reached, an optimized antibody population is derived, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, and multiple optimized affinities are obtained, the multiple optimized affinities correspond to the multiple optimized feature vectors, an optimized feature vector with the highest optimized affinity is derived, the optimized feature vector includes an optimized flow parameter, an optimized speed parameter and an optimized cooling parameter, and the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter are imported into the liquid cooling control module.

[0005] As a further solution of the present invention, iterative optimization is performed based on a new antibody population, and a mutation resistance strategy is performed during the iterative optimization process, and memory cells are established based on the mutation resistance strategy, wherein the mutation resistance strategy includes the following steps: S201: Establishing a memory cell, establishing a mutation constraint condition based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, wherein the mutation constraint condition can be expressed as: ; In the formula, is the constraint condition of the flow parameter contained in the variant antibody feature vector, is the constraint condition of the rotation speed parameter contained in the variant antibody feature vector, is the constraint condition of the cooling parameter contained in the variant antibody feature vector, is the flow parameter contained in the variant antibody feature vector, is the rotation speed parameter contained in the variant antibody feature vector, is the cooling parameter contained in the variant antibody feature vector; S202: judging the plurality of variant antibody feature vectors generated in the iterative optimization process according to the mutation constraint condition, and deleting the variant antibody feature vectors that have mutated; S203: Importing the deleted variant antibody feature vector into the memory cell, and storing the deleted variant antibody feature vector through the memory cell. The memory cell can be represented as: ; In the formula, for the memory cell, is the number of mutant antibody feature vectors that have undergone mutations, For the Feature vectors of mutant antibodies with mutations; S204: When performing the subsequent iterative optimization, the multiple variant antibody feature vectors generated in each iterative optimization process are compared with the memory cells, and the variant antibody feature vectors identical to those in the memory cells are deleted. After deletion, steps S202 and S203 are executed.

[0006] As a further solution of the present invention, judging a plurality of the variant antibody feature vectors in the iterative optimization process by using the mutation constraint condition includes: Comparing the plurality of mutation feature vectors with the mutation constraint conditions; If the parameters in the mutated feature vector make , and If any of is 0, the variant antibody feature vector is deemed to be a mutated variant antibody feature vector and is deleted; If the parameters in the mutated feature vector make , and If both are 1, the variant antibody feature vector is deemed to be a normal variant antibody feature vector and is retained.

[0007] As a further solution of the present invention, based on the plurality of affinities, a plurality of the random antibody feature vectors are screened through a tournament selection strategy to select and retain a plurality of preferred antibody feature vectors, including the following steps: S1031: Setting Tournament Size Parameters , the tournament size parameter represents the number of said random antibody feature vectors for each tournament comparison; S1032: Extracting based on the tournament scale parameter The random antibody feature vectors extracted The random antibody feature vectors are not repeated; S1033: Based on comparing the affinities corresponding to the random antibody feature vectors, selecting the random antibody feature vector with the highest affinity as the “winner” and retaining it, and the “winner” is the preferred antibody feature vector; S1034: Repeat steps S1032 and S1033 until the maximum number of repetitions is reached. The maximum number of repetitions is calculated as follows: ; In the formula, is the maximum number of repetitions, is the number of random antibody feature vectors, is the tournament size parameter.

[0008] As a further solution of the present invention, performing a mutation operation on a plurality of copies of the preferred antibody feature vectors to obtain a plurality of mutated antibody feature vectors includes: The variation operation is to proportionally adjust the flow parameter, the speed parameter and the cooling parameter contained in the plurality of copies of the preferred antibody feature vectors, and the magnitude of the proportional adjustment is respectively , and Adjust within the range of It is represented as the proportional amplitude range for proportionally adjusting the flow parameter, It is represented as the proportional amplitude range for proportional adjustment of the speed parameter, It is represented as a proportional amplitude range for proportionally adjusting the cooling parameter, and a plurality of variant antibody feature vectors are generated based on the mutation operation, and a ratio of a plurality of copies of the preferred antibody feature vector to a plurality of the variant antibody feature vectors is two to one.

[0009] As a further solution of the present invention, when the flow parameter, the speed parameter and the cooling parameter contained in the plurality of preferred antibody feature vector copies are proportionally adjusted by mutation operation, the magnitude of the proportional adjustment is respectively , and Randomly selected within the range.

[0010] As a further solution of the present invention, a plurality of flow parameters, a plurality of speed parameters and a plurality of cooling parameters are obtained based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, and an initial antibody population is obtained based on an artificial immune algorithm, wherein the initial antibody population is an individual set of a plurality of random antibody feature vectors, including: A plurality of the random antibody feature vectors are established based on a plurality of the flow parameters, a plurality of the rotation speed parameters and a plurality of the cooling parameters. The random antibody feature vectors can be expressed as: , where For the The antibody feature vectors, For the The flow parameters, For the The speed parameter, For the The cooling parameters, wherein , and random.

[0011] As a further solution of the present invention, the plurality of flow parameters, the plurality of rotation speed parameters and the plurality of cooling parameters are all presented in the form of real numbers.

[0012] An intelligent liquid cooling control system for a data center, comprising: An initial module, the initial module includes a setting module, a constraint module and an acquisition module; The setting module, through which the user can set the cooling temperature, after the user completes the setting, the setting module can obtain the setting temperature parameters, the setting parameters; The constraint module is used to obtain the optimization constraint condition, obtain the optimization constraint condition based on the set temperature parameter, obtain the flow parameter interval, the speed parameter interval and the cooling parameter interval based on the optimization constraint condition, and the optimization constraint condition constrains the flow parameter interval, the speed parameter interval and the cooling parameter interval; The acquisition module is used to acquire multiple flow parameters, multiple speed parameters and multiple cooling parameters, the flow parameter is the coolant flow of the flow unit, the speed parameter is the speed of the cooling water pump, and the cooling parameter is the power of the liquid cooling distribution unit; An optimization module, wherein the optimization module includes an immune module and a memory module; The immune module is used to establish an artificial immune algorithm, and obtain an initial antibody population based on the artificial immune algorithm, wherein the initial antibody population is an individual set of multiple random antibody feature vectors; Obtaining an affinity function based on the artificial immune algorithm, importing the initial antibody population into the affinity function to obtain multiple affinities, wherein the multiple affinities correspond to multiple random antibody feature vectors respectively, and screening the multiple random antibody feature vectors based on the multiple affinities and through a tournament selection strategy to screen and retain multiple preferred antibody feature vectors; performing a cloning operation on the plurality of the preferred antibody feature vectors, obtaining a plurality of preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the plurality of the preferred antibody feature vector copies to the plurality of the antibody feature vectors is four to one, and performing a mutation operation on the plurality of the preferred antibody feature vector copies to obtain a plurality of mutated antibody feature vectors; Establishing a new antibody population based on the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors and the multiple preferred antibody feature vectors, wherein the new antibody population is an individual set of the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors and the multiple preferred antibody feature vectors, and performing iterative optimization based on the new antibody population; When the maximum number of iterative optimization times is reached, an optimized antibody population is derived, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, multiple optimized affinities are obtained, the multiple optimized affinities correspond to the multiple optimized feature vectors, an optimized feature vector with the highest optimized affinity is derived, the optimized feature vector includes an optimized flow parameter, an optimized speed parameter and an optimized cooling parameter, the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter are imported into a liquid cooling control module; The memory module is used to establish memory cells and execute mutation countermeasure strategy when the artificial immune algorithm is iteratively optimized; A cooling unit, comprising a cooling water pump, a flow unit and a liquid cooling distribution unit, through which the cooling water pump, the flow unit and the liquid cooling distribution unit can be used to liquid cool external IT equipment; The liquid cooling control module is used to receive the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter outputted from the optimization module, and control the cooling unit to adjust to the corresponding parameters to save energy consumption.

[0013] Compared with the prior art, the present invention has the following beneficial effects: First, obtain the set temperature parameters, and obtain the optimization constraints based on the set temperature parameters, and obtain the flow parameter interval, the speed parameter interval and the cooling parameter interval based on the optimization constraints, so as to constrain the acquisition of the flow parameters, the speed parameters and the cooling parameters to ensure the heat dissipation effect of the liquid cooling, and then obtain multiple flow parameters, multiple speed parameters and multiple cooling parameters based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, and obtain the initial antibody population based on the artificial immune algorithm. The initial antibody population is an individual set of multiple random antibody feature vectors. After that, obtain the affinity function, import the initial antibody population into the affinity function, obtain multiple affinities, screen and retain multiple preferred antibody feature vectors through the tournament selection strategy, clone and mutate the multiple preferred antibody features, so as to obtain multiple copies of the preferred antibody feature vector and multiple variant antibody feature vectors, and multiple variant antibody features. A new antibody population is established based on the feature vector, multiple copies of the preferred antibody feature vector and multiple preferred antibody feature vectors, memory cells are established, iterative optimization is performed based on the new antibody population, mutation confrontation strategy is performed in the process of iterative optimization, and the efficiency and stability of the algorithm can be improved by the mutation confrontation strategy. Finally, when the maximum number of iterative optimization times is reached, the optimized antibody population is exported, the optimized antibody population is imported into the affinity function, multiple optimized affinities are obtained, and an optimized feature vector with the highest optimized affinity is exported from the optimized antibody population. The optimized feature vector includes optimized flow parameters, optimized speed parameters and optimized cooling parameters, thereby completing the optimization of flow parameters, speed parameters and cooling parameters, importing the optimized flow parameters, optimized speed parameters and optimized cooling parameters into the liquid cooling control module, and controlling the cooling unit to adjust to the corresponding operating parameters through the liquid cooling control module, thereby reducing energy consumption while ensuring the heat dissipation effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flow chart of steps of a data center intelligent liquid cooling control method and system of the present invention; Figure 2 It is a flow chart of the mutation countermeasure strategy in step S105 of a data center intelligent liquid cooling control method and system of the present invention; Figure 3 It is a flow chart of the tournament selection strategy in step S103 of a data center intelligent liquid cooling control method and system of the present invention; Figure 4 It is a structural block diagram of an initial module in a data center intelligent liquid cooling control method and system of the present invention; Figure 5 It is a structural block diagram of a data center intelligent liquid cooling control method and a cooling unit in the system of the present invention; Figure 6 It is a structural block diagram of an optimization module in a data center intelligent liquid cooling control method and system of the present invention. DETAILED DESCRIPTION

[0015] The following is a further detailed description of the embodiments of the present invention in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the technical solutions of the present invention, but cannot be used to limit the scope of protection of the present invention. Example

[0016] As attached Figure 1 To Attachment Figure 6 As shown: The present invention provides a data center intelligent liquid cooling control method, comprising the following steps: S101: Obtain a set temperature parameter, where the set temperature parameter is a set cooling temperature, obtain an optimization constraint condition based on the set temperature parameter, determine a flow parameter interval, a speed parameter interval, and a cooling parameter interval based on the optimization constraint condition, and constrain the flow parameter interval, the speed parameter interval, and the cooling parameter interval under the optimization constraint condition; It is understandable that when cooling external IT equipment, the heat dissipation effect of the external IT equipment must be ensured first. The user can set the cooling temperature through the setting module in the initial module. The cooling temperature is the operating temperature to be maintained when the external IT equipment is running. After the user setting is completed, the initial module can obtain the set temperature parameters corresponding to the cooling temperature. By optimizing the constraints, the flow parameters, speed parameters and cooling parameters can be constrained and restricted, thereby determining the flow parameter range, speed parameter range and cooling parameter range, and clarifying the value range of the flow parameter, speed parameter and cooling parameter. The purpose is to enable liquid cooling to maintain the set cooling temperature during operation, thereby ensuring the normal operation and stable performance of the external IT equipment.

[0017] Furthermore, in order to maintain the cooling temperature set by the user when the external IT equipment is running, the heat balance equation and the temperature calculation model can be used to establish an association with the cooling water pump speed, liquid cooling flow rate and cooling distribution unit power, so as to obtain the optimization constraints. For example, to ensure that the temperature of the external IT equipment is maintained at 40°C, the coolant flow rate, coolant flow rate and cooling distribution unit power can be reversed according to the heat balance equation and the temperature calculation model. The coolant flow rate is proportional to the cooling water pump speed, so as to obtain the value range of the flow parameter, speed parameter and cooling parameter.

[0018] S102: obtaining a plurality of flow parameters, a plurality of speed parameters, and a plurality of cooling parameters based on the flow parameter interval, the speed parameter interval, and the cooling parameter interval, wherein the flow parameter is the coolant flow of the flow unit, the speed parameter is the speed of the cooling water pump, and the cooling parameter is the power of the liquid cooling distribution unit, and obtaining an initial antibody population based on an artificial immune algorithm, wherein the initial antibody population is an individual set of a plurality of random antibody feature vectors; Furthermore, the random antibody feature vector can be expressed as: , where For the Antibody feature vectors, For the flow parameters, For the Speed ​​parameters, For the cooling parameters, among which, , and Random, a random antibody feature vector represents a parameter configuration in liquid cooling control. The parameters contained in each random antibody feature vector are within the corresponding parameter range. Such a random antibody feature vector represents an operating configuration in liquid cooling control, that is, an antibody in the artificial immune algorithm. When taking values ​​for each random antibody feature vector, they are taken in a random manner. There are large differences in parameter values ​​between randomly generated individuals, which can ensure that the initial antibody population has a high diversity. For example, the cooling parameter may be within the cooling parameter range. The parameters are close to the minimum value, maximum value and any position in the middle. The same is true for the speed parameter and flow parameter. This avoids limitations. If the initial antibody population lacks high diversity, it is easy for the algorithm to fall into a local optimal solution and cannot be further optimized.

[0019] Furthermore, the plurality of flow parameters, the plurality of rotational speed parameters and the plurality of cooling parameters are all presented in the form of real numbers, and the presentation in the form of real numbers facilitates iterative calculation of the algorithm.

[0020] S103: Obtain an affinity function based on an artificial immune algorithm, import the initial antibody population into the affinity function, obtain multiple affinities, the multiple affinities respectively correspond to multiple random antibody feature vectors, screen the multiple random antibody feature vectors based on the multiple affinities and through a tournament selection strategy, and screen and retain multiple preferred antibody feature vectors.

[0021] It can be understood that affinity determines the quality of each random antibody feature vector during the algorithm iteration process. The higher the affinity corresponding to the random antibody feature vector, the better the random feature vector is. Conversely, the lower the affinity corresponding to the random antibody feature vector, the worse the random feature vector is. When obtaining the affinity function based on the artificial immune algorithm, the objective function can be selected to establish the affinity function. The objective function can be regarded as a representation of the total energy consumption of liquid cooling.

[0022] Furthermore, when considering the total energy consumption of liquid cooling, the energy consumption of the cooling water pump, the liquid cooling distribution unit, and the coolant flow rate all affect the total energy consumption of liquid cooling. The energy consumption of the cooling water pump can be expressed as: ; In the formula, Indicates the energy consumption of the cooling water pump, Indicates the coefficient related to the cooling water pump's own characteristics, is the speed parameter.

[0023] Furthermore, the objective function can be expressed as: ; In the formula, represents the objective function, Expressed as a proportional coefficient related to the characteristics of the flow unit itself, is the cooling parameter, is the flow parameter.

[0024] Specifically, the affinity function can be established through the objective function, and the affinity function can be expressed as: ; In the formula, is an affinity function, and a plurality of affinities respectively corresponding to the plurality of random antibody feature vectors are obtained by importing a plurality of random antibody feature matrices contained in the initial antibody population into the affinity function.

[0025] Among them, based on multiple affinities and through a tournament selection strategy, multiple random antibody feature vectors are screened to select and retain multiple preferred antibody feature vectors, including the following steps: S1031: Setting Tournament Size Parameters , the tournament size parameter Represents the number of random antibody feature vectors for each tournament comparison.

[0026] Furthermore, the core of the tournament selection strategy is to select relatively superior individuals from the antibody population by simulating the mechanism of competition selection in reality, that is, random antibody feature vectors with higher affinity, so that they can enter cloning and mutation operations, and then pass on the superior individuals to guide the population to optimize in a better direction. The tournament selection strategy is relatively simple to implement and does not require complex probability calculations and other processes. It can quickly screen out superior individuals through multiple simple small-scale comparisons. The computational cost is relatively low, and it can also operate more efficiently when dealing with larger populations.

[0027] S1032: Extraction based on tournament scale parameters Random antibody feature vectors, extracted The random antibody feature vectors are not repeated.

[0028] S1033: Based on The affinities corresponding to the random antibody feature vectors are compared, and the random antibody feature vector with the highest affinity is selected as the "winner" and retained. The "winner" is the preferred antibody feature vector.

[0029] For example, if , it means that 3 random antibody feature vectors will be randomly selected from the initial antibody population, and the 3 random antibody feature vectors are not repeated during random selection, ensuring that the random antibody feature vectors participating in the comparison are different each time, avoiding unreasonable comparisons, and comparing the corresponding affinity of the 3 random antibody feature vectors. The random antibody feature vector with the highest affinity is selected, and the random antibody feature vector is marked as the "winner", which is retained and identified as the preferred antibody feature vector.

[0030] S1034: Repeat steps S1032 and S1033 until the maximum number of repetitions is reached. The maximum number of repetitions is calculated as follows: ; In the formula, is the maximum number of repetitions, is the number of random antibody feature vectors, is the tournament size parameter.

[0031] S104: performing a cloning operation on the multiple preferred antibody feature vectors, obtaining multiple preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the multiple preferred antibody feature vector copies to the multiple antibody feature vectors is four to one, and performing a mutation operation on the multiple preferred antibody feature vector copies to obtain multiple mutated antibody feature vectors.

[0032] It can be understood that the cloning operation is to copy multiple preferred antibody feature vectors, copy the preferred antibody feature vectors into preferred antibody feature vector copies, and the copy ratio is four to one, that is, each preferred antibody feature vector is copied into four preferred antibody feature vector copies.

[0033] It can be understood that the mutation operation is to proportionally adjust the flow parameters, speed parameters and cooling parameters contained in the multiple copies of the preferred antibody feature vectors, and the magnitude of the proportional adjustment is respectively , and Adjust within the range of It is expressed as the proportional amplitude range for proportional adjustment of flow parameters, It is expressed as the proportional amplitude range for proportional adjustment of the speed parameter. It is represented as a proportional amplitude range for proportionally adjusting the cooling parameter, and multiple mutant antibody feature vectors are generated based on the mutation operation, and the ratio of the multiple preferred antibody feature vector copies to the multiple mutant antibody feature vectors is two to one.

[0034] Further, when the flow parameters, speed parameters and cooling parameters contained in the plurality of copies of the preferred antibody feature vectors are proportionally adjusted by mutation operation, the magnitude of the proportional adjustment is respectively , and Randomly select within the range of , when proportionally adjusting the flow parameters, speed parameters and cooling parameters contained in multiple copies of the preferred antibody feature vector, , and The range can be set to about.

[0035] S105: establishing a new antibody population based on multiple variant antibody feature vectors, multiple copies of preferred antibody feature vectors, and multiple preferred antibody feature vectors, where the new antibody population is an individual collection of multiple variant antibody feature vectors, multiple copies of preferred antibody feature vectors, and multiple preferred antibody feature vectors, establishing memory cells, performing iterative optimization based on the new antibody population, and performing a mutation countermeasure strategy during the iterative optimization process.

[0036] It can be understood that the newly generated antibody feature vectors are obtained through cloning and mutation operations as well as the tournament selection strategy. The tournament selection strategy optimizes the random antibody feature vectors to obtain the preferred antibody feature vectors. The cloning operation enables the preferred antibody feature vectors to proliferate and exist in the form of copies of the preferred antibody feature vectors, increasing their proportion in the population. The mutation operation obtains the mutated antibody feature vectors based on these antibody feature vector copies, introducing a certain degree of randomness and diversity.

[0037] It can be understood that the new antibody population, as the starting point of the next round of iterative optimization, integrates the optimization results of the previous round. After entering the next round of iterative optimization, the new antibody population will perform steps S103, S104 and S105 again and continue to perform iterative optimization. As the number of iterative optimizations increases, the antibody feature vectors in the antibody population gradually approach the optimal solution.

[0038] Furthermore, when the next round of iterative optimization is performed, when the new antibody population performs step S104, it is necessary to ensure that the total number of the generated preferred antibody feature vectors, preferred antibody feature vector copies, and variant antibody feature vectors is the same as that of the previous round of iterative optimization.

[0039] For example, if the total number of preferred antibody feature vectors, preferred antibody feature vector copies, and variant antibody feature vectors contained in the new antibody population established in the previous round of iterative optimization is 80, and in the next round of iterative optimization, 20 preferred antibody feature vectors are retained after screening by the tournament selection strategy, then the total number of preferred antibody feature vectors, preferred antibody feature vector copies, and variant antibody feature vectors needs to be increased to 80 through cloning and mutation operations, thereby ensuring that the total number of preferred antibody feature vectors, preferred antibody feature vector copies, and variant antibody feature vectors contained in the new antibody population established in the next round of iterative optimization is the same as the total number in the previous round of iterative optimization.

[0040] Among them, the mutation resistance strategy includes the following steps: S201: Establish memory cells, and establish mutation constraints based on the flow parameter range, the speed parameter range, and the cooling parameter range. The mutation constraints can be expressed as: ; In the formula, is the constraint condition of the flow parameter contained in the mutant antibody feature vector, is the constraint condition of the speed parameter contained in the variant antibody feature vector, is the constraint condition of the cooling parameter contained in the variant antibody feature vector, is the flow parameter contained in the variant antibody feature vector, is the speed parameter contained in the variant antibody feature vector, is the cooling parameter contained in the variant antibody feature vector.

[0041] in, is the minimum value of the flow parameter interval, is the maximum value of the flow parameter interval, is the minimum value of the speed parameter range, is the maximum value of the speed parameter range, is the minimum value of the cooling parameter interval, is the maximum value of the cooling parameter range.

[0042] It can be understood that when a mutation operation is performed on multiple copies of the preferred antibody feature vector, the parameters contained in the preferred antibody feature vector will be proportionally adjusted. When the proportional adjustment is performed, the parameters contained in the preferred antibody feature vector may exceed the flow parameter interval, the speed parameter interval and the cooling parameter interval. The mutated antibody feature vector containing parameters exceeding the flow parameter interval, the speed parameter interval and the cooling parameter interval will be marked as a mutated antibody feature vector.

[0043] It can be understood that memory cells are used to store feature vectors of mutated variant antibodies.

[0044] S202: multiple mutant antibody feature vectors generated during the iterative optimization process are judged according to mutation constraints, and mutant antibody feature vectors that have mutated are deleted.

[0045] S203: Importing the deleted variant antibody feature vector into the memory cell, and storing the deleted variant antibody feature vector through the memory cell. The memory cell can be represented as: ; In the formula, For memory cells, is the number of mutant antibody feature vectors that have undergone mutations, For the The feature vector of the mutant antibodies with mutations.

[0046] S204: When performing subsequent iterative optimization, multiple variant antibody feature vectors generated in each iterative optimization process are compared with the memory cells, and the variant antibody feature vectors identical to those in the memory cells are deleted. After deletion, steps S202 and S203 are executed.

[0047] It can be understood that when the first iterative optimization is performed, memory cells are established and the deleted variant antibody feature vectors are imported into the memory cells. In each subsequent round of iterative optimization, the generated variant antibody feature vectors are compared with the memory cells, and the variant antibody feature vectors that are identical to those in the memory cells are deleted. After deletion, steps S202 and S203 are executed, and the mutated variant antibody feature vectors are deleted through steps S202 and S203, and the deleted mutated variant antibody feature vectors are imported into the memory cells, and this cycle is repeated.

[0048] Specifically, through this mutation adversarial strategy, the efficiency and stability of the algorithm can be improved, and the optimal parameter configuration can be found more quickly.

[0049] Furthermore, multiple mutant antibody feature vectors are judged during the iterative optimization process through mutation constraints, including: Compare multiple mutation feature vectors with mutation constraints; If the parameters in the mutated eigenvector are such that , and If any of is 0, the variant antibody feature vector is considered to be a mutated variant antibody feature vector and is deleted; If the parameters in the mutated eigenvector are such that , and If both are 1, the variant antibody feature vector is considered to be a normal variant antibody feature vector and is retained.

[0050] S106: When the maximum number of iterative optimization times is reached, the optimized antibody population is exported, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, and multiple optimized affinities are obtained. The multiple optimized affinities correspond to multiple optimized feature vectors, and an optimized feature vector with the highest optimized affinity is exported. The optimized feature vector includes optimized flow parameters, optimized speed parameters and optimized cooling parameters. The optimized flow parameters, optimized speed parameters and optimized cooling parameters are imported into the liquid cooling control module.

[0051] Furthermore, the optimized flow parameter is the optimized coolant flow, the optimized speed parameter is the optimized speed of the cooling water pump, and the optimized cooling parameter is the optimized power of the cooling distribution unit. These parameters are imported into the liquid cooling control module, and the liquid cooling control module controls and adjusts the cooling unit based on these parameters to save energy.

[0052] An intelligent liquid cooling control system for a data center, comprising: Initial module, the initial module includes a setting module, a constraint module and an acquisition module; Setting module, the user can set the cooling temperature through the setting module. After the user completes the setting, the setting module can obtain the setting temperature parameters and set the parameters; A constraint module is used to obtain optimization constraint conditions, obtain the optimization constraint conditions based on the set temperature parameters, obtain the flow parameter interval, the speed parameter interval and the cooling parameter interval based on the optimization constraint conditions, and the optimization constraint conditions constrain the flow parameter interval, the speed parameter interval and the cooling parameter interval; An acquisition module, the acquisition module is used to acquire multiple flow parameters, multiple speed parameters and multiple cooling parameters, the flow parameter is the coolant flow of the flow unit, the speed parameter is the speed of the cooling water pump, and the cooling parameter is the power of the liquid cooling distribution unit; Optimization module, the optimization module includes an immune module and a memory module; An immune module is used to establish an artificial immune algorithm and obtain an initial antibody population based on the artificial immune algorithm. The initial antibody population is an individual set of multiple random antibody feature vectors. Obtaining an affinity function based on an artificial immune algorithm, importing an initial antibody population into the affinity function, obtaining multiple affinities, wherein the multiple affinities correspond to multiple random antibody feature vectors respectively, screening the multiple random antibody feature vectors based on the multiple affinities and through a tournament selection strategy, and screening and retaining multiple preferred antibody feature vectors; Performing a cloning operation on multiple preferred antibody feature vectors, obtaining multiple preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the multiple preferred antibody feature vector copies to the multiple antibody feature vectors is four to one, and performing a mutation operation on the multiple preferred antibody feature vector copies to obtain multiple mutated antibody feature vectors; Establishing a new antibody population based on multiple variant antibody feature vectors, multiple copies of preferred antibody feature vectors, and multiple preferred antibody feature vectors, wherein the new antibody population is an individual set of multiple variant antibody feature vectors, multiple copies of preferred antibody feature vectors, and multiple preferred antibody feature vectors, and performing iterative optimization based on the new antibody population; When the maximum number of iterative optimization times is reached, an optimized antibody population is derived, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, multiple optimized affinities are obtained, the multiple optimized affinities correspond to multiple optimized feature vectors, an optimized feature vector with the highest optimized affinity is derived, the optimized feature vector includes optimized flow parameters, optimized speed parameters and optimized cooling parameters, the optimized flow parameters, optimized speed parameters and optimized cooling parameters are imported into the liquid cooling control module; The memory module is used to build memory cells and execute mutation countermeasure strategies when the artificial immune algorithm is iteratively optimized; A cooling unit, including a cooling water pump, a flow unit and a liquid cooling distribution unit, through which the external IT equipment can be liquid cooled; The liquid cooling control module is used to receive the optimized flow parameters, optimized speed parameters and optimized cooling parameters output by the optimization module, and control the cooling unit to adjust to the corresponding parameters to save energy consumption.

[0053] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer age, server or data center by a wireless (such as infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0054] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there may be three relationships. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone, where A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0055] It should be understood that in the embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0056] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A data center intelligent liquid cooling control method, characterized in that: The following steps are included: S101: Acquire a set temperature parameter, where the set temperature parameter is a set cooling temperature, acquire an optimization constraint condition based on the set temperature parameter, determine a flow parameter interval, a speed parameter interval, and a cooling parameter interval based on the optimization constraint condition, and the optimization constraint condition constrains the flow parameter interval, the speed parameter interval, and the cooling parameter interval; S102: Based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, a plurality of flow parameters, a plurality of speed parameters and a plurality of cooling parameters are obtained, wherein the flow parameter is a coolant flow rate of a flow unit, the speed parameter is a speed of a cooling water pump, and the cooling parameter is a power of a liquid cooling distribution unit; and an initial antibody population is obtained based on an artificial immune algorithm, wherein the initial antibody population is an individual set of a plurality of random antibody feature vectors; S103: obtaining an affinity function based on an artificial immune algorithm, importing the initial antibody population into the affinity function, obtaining a plurality of affinities, wherein the plurality of affinities respectively correspond to a plurality of random antibody feature vectors, screening the plurality of random antibody feature vectors based on the plurality of affinities and through a tournament selection strategy, and screening and retaining a plurality of preferred antibody feature vectors; S104: performing a cloning operation on the plurality of preferred antibody feature vectors, obtaining a plurality of preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the plurality of preferred antibody feature vector copies to the plurality of antibody feature vectors is four to one, and performing a mutation operation on the plurality of preferred antibody feature vector copies to obtain a plurality of mutated antibody feature vectors; S105: establishing a new antibody population based on the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors, and the multiple preferred antibody feature vectors, wherein the new antibody population is an individual collection of the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors, and the multiple preferred antibody feature vectors, establishing memory cells, performing iterative optimization based on the new antibody population, and performing a mutation countermeasure strategy during the iterative optimization process; S106: When the maximum number of iterative optimization times is reached, an optimized antibody population is derived, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, and multiple optimized affinities are obtained, the multiple optimized affinities correspond to the multiple optimized feature vectors, an optimized feature vector with the highest optimized affinity is derived, the optimized feature vector includes an optimized flow parameter, an optimized speed parameter and an optimized cooling parameter, and the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter are imported into the liquid cooling control module.

2. According to claim 1, a data center intelligent liquid cooling control method is characterized in that , performing iterative optimization based on the new antibody population, and performing a mutation countermeasure strategy during the iterative optimization process, and establishing memory cells based on the mutation countermeasure strategy, wherein the mutation countermeasure strategy includes the following steps: S201: Establishing a memory cell, establishing a mutation constraint condition based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, wherein the mutation constraint condition can be expressed as: ; In the formula, is the constraint condition of the flow parameter contained in the variant antibody feature vector, is the constraint condition of the rotation speed parameter contained in the variant antibody feature vector, is the constraint condition of the cooling parameter contained in the variant antibody feature vector, is the flow parameter contained in the variant antibody feature vector, is the rotation speed parameter contained in the variant antibody feature vector, is the cooling parameter contained in the variant antibody feature vector; S202: judging the plurality of variant antibody feature vectors generated in the iterative optimization process according to the mutation constraint condition, and deleting the variant antibody feature vectors that have mutated; S203: Importing the deleted variant antibody feature vector into the memory cell, and storing the deleted variant antibody feature vector through the memory cell. The memory cell can be represented as: ； In the formula, for the memory cell, is the number of mutant antibody feature vectors that have undergone mutations, For the Feature vectors of mutant antibodies with mutations; S204: When performing the subsequent iterative optimization, the multiple variant antibody feature vectors generated in each iterative optimization process are compared with the memory cells, and the variant antibody feature vectors identical to those in the memory cells are deleted. After deletion, steps S202 and S203 are executed.

3. A data center intelligent liquid cooling control method according to claim 2, characterized in that: The multiple variant antibody feature vectors in the iterative optimization process are judged by the mutation constraint condition, including: Comparing the plurality of mutation feature vectors with the mutation constraint conditions; If the parameters in the mutated feature vector make , and If any of is 0, the variant antibody feature vector is deemed to be a mutated variant antibody feature vector and is deleted; If the parameters in the mutated feature vector make , and If both are 1, the variant antibody feature vector is deemed to be a normal variant antibody feature vector and is retained.

4. The intelligent liquid cooling control method for a data center according to claim 1, characterized in that: Based on the plurality of affinities and through a tournament selection strategy, the plurality of random antibody feature vectors are screened to retain a plurality of preferred antibody feature vectors, comprising the following steps: S1031: Setting Tournament Size Parameters , the tournament size parameter represents the number of said random antibody feature vectors for each tournament comparison; S1032: Extracting based on the tournament scale parameter The random antibody feature vectors extracted The random antibody feature vectors are not repeated; S1033: Based on comparing the affinities corresponding to the random antibody feature vectors, selecting the random antibody feature vector with the highest affinity as the "winner", retaining it, and the "winner" is the preferred antibody feature vector; S1034: Repeat steps S1032 and S1033 until the maximum number of repetitions is reached. The maximum number of repetitions is calculated as follows: ； In the formula, is the maximum number of repetitions, is the number of random antibody feature vectors, is the tournament size parameter.

5. The intelligent liquid cooling control method for a data center according to claim 1, characterized in that: Performing a mutation operation on the multiple copies of the preferred antibody feature vectors to obtain multiple mutated antibody feature vectors includes: The variation operation is to proportionally adjust the flow parameter, the speed parameter and the cooling parameter contained in the plurality of copies of the preferred antibody feature vectors, and the magnitude of the proportional adjustment is respectively , and Adjust within the range of It is represented as the proportional amplitude range for proportionally adjusting the flow parameter, It is represented as the proportional amplitude range for proportional adjustment of the speed parameter, It is represented as a proportional amplitude range for proportionally adjusting the cooling parameter, and a plurality of variant antibody feature vectors are generated based on the mutation operation, and a ratio of a plurality of copies of the preferred antibody feature vector to a plurality of the variant antibody feature vectors is two to one.

6. The intelligent liquid cooling control method for a data center according to claim 5, characterized in that: When the flow parameter, the speed parameter and the cooling parameter included in the plurality of preferred antibody feature vector copies are proportionally adjusted by mutation operation, the magnitude of the proportional adjustment is respectively , and Randomly selected within the range.

7. The intelligent liquid cooling control method for a data center according to claim 1, characterized in that: Based on the flow parameter interval, the speed parameter interval and the cooling parameter interval, a plurality of flow parameters, a plurality of speed parameters and a plurality of cooling parameters are obtained; based on an artificial immune algorithm, an initial antibody population is obtained, and the initial antibody population is an individual set of a plurality of random antibody feature vectors, including: A plurality of the random antibody feature vectors are established based on a plurality of the flow parameters, a plurality of the rotation speed parameters and a plurality of the cooling parameters. The random antibody feature vectors can be expressed as: , where For the The antibody feature vectors, For the The flow parameters, For the The speed parameter, For the The cooling parameters, wherein , and random.

8. The intelligent liquid cooling control method for a data center according to claim 6, characterized in that: The plurality of flow parameters, the plurality of rotation speed parameters, and the plurality of cooling parameters are all presented in the form of real numbers.

9. An intelligent liquid cooling control system for a data center, characterized in that: include: An initial module, the initial module includes a setting module, a constraint module and an acquisition module; The setting module, through which the user can set the cooling temperature, after the user completes the setting, the setting module can obtain the setting temperature parameters, the setting parameters; The constraint module is used to obtain the optimization constraint condition, obtain the optimization constraint condition based on the set temperature parameter, obtain the flow parameter interval, the speed parameter interval and the cooling parameter interval based on the optimization constraint condition, and the optimization constraint condition constrains the flow parameter interval, the speed parameter interval and the cooling parameter interval; The acquisition module is used to acquire multiple flow parameters, multiple speed parameters and multiple cooling parameters, the flow parameter is the coolant flow of the flow unit, the speed parameter is the speed of the cooling water pump, and the cooling parameter is the power of the liquid cooling distribution unit; An optimization module, wherein the optimization module includes an immune module and a memory module; The immune module is used to establish an artificial immune algorithm, and obtain an initial antibody population based on the artificial immune algorithm, wherein the initial antibody population is an individual set of multiple random antibody feature vectors; Obtaining an affinity function based on the artificial immune algorithm, importing the initial antibody population into the affinity function to obtain multiple affinities, wherein the multiple affinities correspond to multiple random antibody feature vectors respectively, and screening the multiple random antibody feature vectors based on the multiple affinities and through a tournament selection strategy to screen and retain multiple preferred antibody feature vectors; performing a cloning operation on the plurality of the preferred antibody feature vectors, obtaining a plurality of preferred antibody feature vector copies based on the cloning operation, wherein the ratio of the plurality of the preferred antibody feature vector copies to the plurality of the antibody feature vectors is four to one, and performing a mutation operation on the plurality of the preferred antibody feature vector copies to obtain a plurality of mutated antibody feature vectors; Establishing a new antibody population based on the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors and the multiple preferred antibody feature vectors, wherein the new antibody population is an individual set of the multiple variant antibody feature vectors, the multiple copies of the preferred antibody feature vectors and the multiple preferred antibody feature vectors, and performing iterative optimization based on the new antibody population; When the maximum number of iterative optimization times is reached, an optimized antibody population is derived, the optimized antibody population is an individual set of multiple optimized feature vectors, the optimized antibody population is imported into the affinity function, multiple optimized affinities are obtained, the multiple optimized affinities correspond to the multiple optimized feature vectors, an optimized feature vector with the highest optimized affinity is derived, the optimized feature vector includes an optimized flow parameter, an optimized speed parameter and an optimized cooling parameter, the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter are imported into a liquid cooling control module; The memory module is used to establish memory cells and execute mutation countermeasure strategy when the artificial immune algorithm is iteratively optimized; A cooling unit, comprising a cooling water pump, a flow unit and a liquid cooling distribution unit, through which the external IT equipment can be liquid cooled; The liquid cooling control module is used to receive the optimized flow parameter, the optimized speed parameter and the optimized cooling parameter outputted from the optimization module, and control the cooling unit to adjust to the corresponding parameters to save energy consumption.

Images