A method and system for controlling coolant flow in an immersion liquid cooling system

Abstract

The present invention provides a method and system for controlling the flow of coolant in an immersion liquid cooling system, which obtains a set temperature of the liquid cooling system, obtains a power range of a liquid cooling distribution unit and a speed range of a cooling water pump based on the set temperature of the liquid cooling system, defines a first random process based on a Monte Carlo algorithm and obtains a first random liquid cooling sample, defines a second random process based on a random forest algorithm and obtains a second random liquid cooling sample, executes a particle swarm optimization algorithm on the first random liquid cooling sample to obtain a first optimized liquid cooling control parameter, executes a random forest algorithm on the second random liquid cooling sample to obtain a second optimized liquid cooling control parameter, performs stability judgment based on a root locus diagram of a second-order transfer function of liquid cooling, and obtains an optimized liquid cooling control parameter or a fused liquid cooling control parameter; and imports the optimized liquid cooling control parameter or the fused liquid cooling control parameter into a liquid cooling control module, thereby improving the cooling effect of the immersion liquid cooling system on servers or IT equipment.

Description

Technical Field

[0001] The present invention relates to the field of liquid cooling technology, and in particular to a method and system for controlling the flow of coolant in an immersion liquid cooling system. Background Art

[0002] In the digital age, the computing power of data centers, artificial intelligence server clusters, and high-performance computing equipment is growing exponentially, and the heat generated during equipment operation is also increasing sharply. Traditional air cooling methods have limited heat dissipation efficiency and can no longer meet the heat dissipation needs of high-heat-density equipment. Immersion liquid cooling systems stand out due to their heat dissipation performance.

[0003] In the field of traditional immersion liquid cooling system control, existing technologies often use flow regulation strategies based on PID control or single optimization algorithms, such as genetic algorithms and gradient descent methods. Although these can achieve basic temperature control functions, they have limitations.

[0004] First, a single optimization algorithm is prone to falling into local optimality, especially in high-dimensional parameter spaces with multi-variable coupling, such as the joint optimization of the power of the liquid cooling distribution unit and the speed of the cooling water pump, which is difficult to take into account. Second, traditional coolant flow control methods often use a single parameter setting, which will lead to temperature control lag and energy consumption surge. For this reason, the present invention provides a fusion hybrid algorithm that can be used to perform a control method for collaborative cooling effect optimization. Summary of the Invention

[0005] In view of the above-mentioned problems, in combination with the first aspect of the present invention, the objectives and functions of the coolant flow control method and system of an immersion liquid cooling system of the present invention are achieved by the following specific technical means:

[0006] A method for controlling the flow of coolant in an immersion liquid cooling system, comprising:

[0007] S1: Acquire a set temperature of the liquid cooling system, and acquire a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system;

[0008] S2: defining a first random process of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on a Monte Carlo algorithm, obtaining a first random liquid cooling sample based on the first random process, defining a second random process of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on a random forest algorithm, and obtaining a second random liquid cooling sample based on the second random process;

[0009] S3: executing a particle swarm optimization algorithm on the first random liquid cooling sample, performing repeated simulation updates based on the particle swarm optimization algorithm, and obtaining first optimized liquid cooling control parameters;

[0010] S4: Execute a random forest algorithm on the second random liquid cooling sample, perform repeated simulation updates based on the random forest algorithm, and obtain second optimized liquid cooling control parameters;

[0011] S5: establishing a liquid cooling second-order transfer function, drawing a root locus diagram of the liquid cooling second-order transfer function, performing stability judgment based on the root locus diagram of the liquid cooling second-order transfer function, and obtaining optimized liquid cooling control parameters or integrated liquid cooling control parameters based on the stability judgment;

[0012] S6: Importing the optimized liquid cooling control parameters or the fused liquid cooling control parameters into a liquid cooling control module, through which the power of the liquid cooling distribution unit and the speed of the cooling water pump can be controlled in real time.

[0013] As a further solution of the present invention, the stability judgment is performed based on the root locus diagram of the liquid cooling second-order transfer function, and the optimized liquid cooling control parameters or the integrated liquid cooling control parameters are obtained based on the stability judgment, including:

[0014] Importing the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter into a liquid cooling second-order transfer function, and finding positions of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter on a root locus diagram of the liquid cooling transfer function;

[0015] If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the second or third quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in a stable state, performing a fusion operation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and obtaining the fused liquid cooling control parameter based on the fusion operation;

[0016] If a set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in the first quadrant or the fourth quadrant of the root locus diagram, then the set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in an unstable state, and another set of parameters in a stable state is obtained as the optimized liquid cooling control parameters;

[0017] If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the first quadrant or the fourth quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in an unstable state, and step S2 is executed again.

[0018] As a further solution of the present invention, performing a fusion operation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and obtaining the fused liquid cooling control parameter based on the fusion operation, includes:

[0019] The fusion operation is represented by weighted summing of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter. The calculation formula of the weighted summing is expressed as:

[0020] ;

[0021] in, Expressed as fusion liquid cooling control parameters, It represents the first optimized liquid cooling control parameter, It is represented as the second optimized liquid cooling control parameter, It is expressed as the calculation weight of the first optimized liquid cooling control parameter, It represents the calculation weight of the second optimized liquid cooling control parameter;

[0022] When performing the calculation, a constraint condition is introduced, which is expressed as:

[0023] ,and .

[0024] As a further solution of the present invention, establishing a liquid cooling second-order transfer function and drawing a root locus diagram of the liquid cooling second-order transfer function include:

[0025] When drawing the root locus diagram of the liquid-cooled second-order transfer function, a plane coordinate system is established, in which the X-axis is represented as a real number and the Y-axis is represented as a complex number. When the root locus diagram of the liquid-cooled second-order transfer function is in the second or third quadrant of the plane coordinate system, it is in a stable state. When the root locus diagram of the liquid-cooled second-order transfer function is in the first or fourth quadrant of the plane coordinate system, it is in an unstable state.

[0026] As a further solution of the present invention, the first random process of defining the liquid cooling distribution unit power and the cooling water pump speed based on the Monte Carlo algorithm and obtaining the first random liquid cooling sample based on the first random process include:

[0027] Dividing the liquid cooling distribution unit power interval and the cooling water pump speed interval into ten liquid cooling random intervals respectively, wherein the random probability of each liquid cooling random interval is equal, and the random probability of each liquid cooling random interval is ten percent;

[0028] Obtain two extreme value intervals within the liquid cooling random interval, where the extreme value intervals are represented by the lowest and highest extreme value intervals in the liquid cooling random interval, and widen the boundaries of the two extreme value intervals;

[0029] For the lowest extreme value interval, the minimum value of the extreme value interval is reduced by 10%, and for the highest extreme value interval, the maximum value of the extreme value interval is increased by 10%;

[0030] Random sampling is performed based on a liquid cooling random interval corresponding to the power interval of the liquid cooling distribution unit and the speed interval of the cooling water pump to obtain a first random liquid cooling sample;

[0031] When performing random sampling, the sampling density of each liquid-cooling random interval is equal, and the distance between each first random liquid-cooling sample is greater than a preset threshold.

[0032] As a further solution of the present invention, the second random process of defining the power of the liquid cooling distribution unit and the cooling water pump speed based on the random forest algorithm, and obtaining a second random liquid cooling sample based on the second random process, includes:

[0033] Obtain key characteristics of liquid cooling, where the key characteristics of liquid cooling are represented by physical operating characteristics of the immersion liquid cooling system during operation. Randomly generate random samples within the power range of the liquid cooling distribution unit and the speed range of the cooling water pump based on the random forest algorithm, import the random samples into the random forest algorithm, perform decision tree splitting based on the key characteristics of liquid cooling, and obtain a second random liquid cooling sample.

[0034] As a further solution of the present invention, executing a particle swarm optimization algorithm on the first random liquid cooling sample, performing repeated simulation updates based on the particle swarm optimization algorithm, and obtaining first optimized liquid cooling control parameters include:

[0035] Establishing a two-dimensional space, mapping the first random liquid-cooling sample into the two-dimensional space, using the first random liquid-cooling sample as a liquid-cooling particle group, and using the random liquid-cooling distribution unit power and the random cooling water pump speed included in the first random liquid-cooling sample as liquid-cooling particles;

[0036] Initialize the liquid-cooled particles and randomly mutate 20% of them. The amplitude of random mutation is within the range of

[0037] The initial positions of the liquid-cooling particles are simulated and updated based on the particle swarm optimization algorithm, and 30% of the optimal liquid-cooling particles are retained for the next round of simulation update in each simulation update;

[0038] When the particle swarm optimization algorithm reaches the optimal solution, the first optimized liquid cooling control parameter is output.

[0039] As a further solution of the present invention, executing a random forest algorithm on the second random liquid cooling sample, performing repeated simulation updates based on the random forest algorithm, and obtaining second optimized liquid cooling control parameters include:

[0040] A liquid cooling prediction model is constructed using the deep regression tree in the random forest algorithm, the second random liquid cooling sample is imported into the liquid cooling prediction model, and a simulation update is performed based on the liquid cooling prediction model. When the liquid cooling prediction model reaches the optimal solution, the second optimized liquid cooling control parameter is output.

[0041] As a further solution of the present invention, obtaining a set temperature of the liquid cooling system and obtaining a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system include:

[0042] Physical condition restrictions are imposed on the obtained liquid cooling distribution unit power range and cooling water pump speed range. The physical condition restrictions are expressed as eliminating the liquid cooling distribution unit power that exceeds the actual operating power of the liquid cooling distribution unit within the liquid cooling distribution unit power range, and eliminating the cooling water pump speed that exceeds the actual operating speed of the cooling water pump within the cooling water pump speed range.

[0043] A coolant flow control system for an immersion liquid cooling system, comprising:

[0044] An acquisition module, the acquisition module is used to obtain a set temperature of the liquid cooling system, and can obtain a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system;

[0045] a first solution module, wherein the first solution module may define a first random process of the liquid cooling distribution unit power and the cooling water pump speed based on a Monte Carlo algorithm, obtain a first random liquid cooling sample based on the first random process, and may also obtain a first optimized liquid cooling control parameter based on a particle swarm optimization algorithm;

[0046] a second solution module, wherein the second solution module may define a second random process of the liquid cooling distribution unit power and the cooling water pump speed based on a random forest algorithm, obtain a second random liquid cooling sample based on the second random process, and further obtain a second optimized liquid cooling control parameter based on the random forest algorithm;

[0047] a stabilization module, the stabilization module being used to establish a liquid cooling second-order transfer function, perform stability judgment based on a root locus diagram of the liquid cooling second-order transfer function, and further obtain optimized liquid cooling control parameters or integrated liquid cooling control parameters based on the stability judgment;

[0048] A liquid cooling control module can perform real-time control of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on optimized liquid cooling control parameters or integrated liquid cooling control parameters.

[0049] Based on the above aspects, the embodiment of the present application first obtains the power range of the liquid cooling distribution unit and the cooling water pump speed range by setting the temperature of the liquid cooling system, defines a first random process based on the Monte Carlo algorithm and obtains a first random liquid cooling sample, defines a second random process based on the random forest algorithm and obtains a second random liquid cooling sample, then executes a particle swarm optimization algorithm on the first random liquid cooling sample, executes a random forest algorithm on the second random liquid cooling sample, and performs repeated simulation updates to obtain the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and then establishes a liquid cooling second-order transfer function, and obtains the optimized liquid cooling control parameter or the fused liquid cooling control parameter through stability judgment. Finally, the liquid cooling control module can perform real-time control on the power of the liquid cooling distribution unit and the speed of the cooling water pump based on optimizing the liquid cooling control parameters or integrating the liquid cooling control parameters, thereby controlling the coolant flow of the immersion liquid cooling system. By applying the hybrid algorithm to the control of the immersion liquid cooling system, it can avoid a single algorithm falling into a local optimal solution, resulting in poor control of the coolant flow of the immersion liquid cooling system. By optimizing the liquid cooling control parameters or integrating the liquid cooling control parameters, the power of the liquid cooling distribution unit and the speed of the cooling water pump are controlled in real time, avoiding the limitations of a single parameter setting. This method can improve the cooling effect of the immersion liquid cooling system on servers or IT equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The present invention provides a flow chart of an implementation of a method for controlling the flow of cooling liquid in an immersion liquid cooling system.

[0051] Figure 2 It is a schematic diagram of the execution flow of stability judgment in a coolant flow control method for an immersion liquid cooling system provided by an embodiment of the present invention.

[0052] Figure 3 This is a flow chart of a method for controlling the flow of cooling liquid in an immersion liquid cooling system provided by an embodiment of the present invention.

[0053] Figure 4 Schematic diagram of a cooling liquid flow control system for an immersion liquid cooling system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0054] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the technical solutions of the present invention, but are not intended to limit the scope of protection of the present invention.

[0055] Example:

[0056] As attached Figure 1 、 Figure 2 、 Figure 3 、 Figure 4 As shown:

[0057] An embodiment of the present invention provides a method for controlling the flow of coolant in an immersion liquid cooling system, which is applicable to the immersion liquid cooling system and includes the following steps:

[0058] An embodiment of the present invention provides a cooling liquid flow control system for an immersion liquid cooling system, which is suitable for liquid cooling control and includes:

[0059] Step S1: obtaining a set temperature of a liquid cooling system, and obtaining a power range of a liquid cooling distribution unit and a speed range of a cooling water pump based on the set temperature of the liquid cooling system.

[0060] Specifically, physical condition restrictions are imposed on the acquired liquid cooling distribution unit power range and cooling water pump speed range. The physical condition restrictions are expressed as eliminating the liquid cooling distribution unit power that exceeds the actual operating power of the liquid cooling distribution unit within the liquid cooling distribution unit power range, and eliminating the cooling water pump speed that exceeds the actual operating speed of the cooling water pump within the cooling water pump speed range.

[0061] Furthermore, after the user sets the cooling temperature of the immersion liquid cooling cabinet through the liquid cooling system, when obtaining the power range of the liquid cooling distribution unit and the speed range of the cooling water pump, the power and speed within the power range of the liquid cooling distribution unit and the speed range of the cooling water pump may accidentally exceed the actual operating power range of the liquid cooling distribution unit or the actual operating speed range of the cooling water pump due to calculation errors and other reasons. If physical condition restrictions are not introduced, incorrect control parameters will be obtained in subsequent algorithm processing, resulting in input into the liquid cooling control module. When the liquid cooling distribution unit and the cooling water pump are controlled by the liquid cooling control module, the liquid cooling distribution unit or the cooling water pump cannot achieve the expected optimization control effect. By introducing physical condition restrictions, the power and speed within the power range of the liquid cooling distribution unit and the speed range of the cooling water pump are always within the actual operating power range of the liquid cooling distribution unit or the actual operating speed range of the cooling water pump, providing a reliable data basis for subsequent algorithm processing.

[0062] Step S2: define a first random process of the liquid cooling distribution unit power and the cooling water pump speed based on the Monte Carlo algorithm, define a second random process of the liquid cooling distribution unit power and the cooling water pump speed based on the random forest algorithm, obtain a first random liquid cooling sample based on the first random process, and obtain a second random liquid cooling sample based on the second random process.

[0063] In this embodiment, the first random process includes:

[0064] In step S21 , the power interval of the liquid cooling distribution unit and the speed interval of the cooling water pump are divided into ten liquid cooling random intervals respectively, and the random probability of each liquid cooling random interval is equal, and the random probability of each liquid cooling random interval is ten percent.

[0065] It can be understood that the form adopted is the uniform distribution method in the Monte Carlo algorithm. The probability density of each point uniformly distributed in the specified interval is constant, and completely random sample coverage can be achieved without complex parameter adjustment, thereby ensuring efficiency and fairness in the subsequent random sampling process. A similar uniform sampling method is also used in the subsequent random sampling process, which can also reduce the amount of calculation in the overall process and increase the processing speed. Uniform distribution can effectively avoid sampling bias and provide basic support for subsequent algorithm random simulation.

[0066] Step S22 : obtaining two extreme value intervals within the liquid cooling random interval, wherein the extreme value intervals are represented by the lowest and highest extreme value intervals in the liquid cooling random interval, and widening the boundaries of the two extreme value intervals.

[0067] Specifically, for the lowest extreme value interval, the minimum value of the extreme value interval is reduced by 10%, and for the highest extreme value interval, the maximum value of the extreme value interval is increased by 10%.

[0068] It is understandable that after being divided into liquid-cooled random intervals, the boundaries of the two extreme value intervals of the liquid-cooled random interval need to be widened to provide redundant intervals for subsequent algorithm processing and retain a safety margin.

[0069] Step S23 , performing random sampling based on a liquid cooling random interval corresponding to the liquid cooling distribution unit power interval and the cooling water pump speed interval, to obtain a first random liquid cooling sample.

[0070] Furthermore, when performing random sampling, the sampling density of each liquid-cooling random interval is equal, and the distance between each first random liquid-cooling sample is greater than a preset threshold.

[0071] It can be understood that when performing random sampling, uniform sampling similar to the uniform distribution mentioned above is adopted, so that the number of first random liquid-cooled samples sampled in each liquid-cooled random interval is equal, ensuring that random sample sampling can be achieved in each liquid-cooled random interval.

[0072] Furthermore, keeping the distance between each first random liquid-cooled sample greater than a preset threshold can maintain the minimum distance between the first random liquid-cooled samples, preventing random sampling from falling into local sampling. If it falls into local sampling, the difference between the first random liquid-cooled samples is small, and in the subsequent algorithm processing process, it is easy for the algorithm to fall into the local optimal solution. By keeping the distance at the minimum preset threshold, the randomness and discreteness between the first random liquid-cooled samples can be ensured. The preset threshold can be adjusted based on the number of samples set by the user.

[0073] In this embodiment, the second random process is represented by obtaining the key characteristics of liquid cooling, and the key characteristics of liquid cooling are represented by the physical operating characteristics of the immersion liquid cooling system during operation. Based on the random forest algorithm, random samples are randomly generated within the power range of the liquid cooling distribution unit and the speed range of the cooling water pump. The random samples are imported into the random forest algorithm, and the decision tree splitting is performed based on the key characteristics of liquid cooling to obtain the second random liquid cooling sample.

[0074] It can be understood that feature sampling based on the random forest algorithm is a data-driven sampling method based on ensemble learning and feature importance analysis. Its core lies in identifying sensitive areas and guiding the generation of parameter samples by constructing a multidimensional feature space and a decision tree network. In the specific implementation, by obtaining the key features of liquid cooling, the key features of liquid cooling can be expressed as parameters of feature categories such as thermodynamics, fluid mechanics, equipment status, and environmental parameters. The specific physical meaning can be expressed as physical meanings such as heat dissipation requirements, flow state, equipment health, and heat exchange boundary conditions. The decision tree split is constructed through the key features of liquid cooling, and then the randomly generated samples can be imported into the decision tree split. The randomly generated random samples are screened through the decision tree split to obtain the second random liquid cooling sample.

[0075] Step S3: executing a particle swarm optimization algorithm on the first random liquid cooling sample, performing repeated simulation updates based on the particle swarm optimization algorithm, and obtaining first optimized liquid cooling control parameters.

[0076] In this embodiment, step S3 includes:

[0077] Step S31: establish a two-dimensional space, map the first random liquid-cooling sample into the two-dimensional space, use the first random liquid-cooling sample as a liquid-cooling particle group, and use the random liquid-cooling distribution unit power and random cooling water pump speed contained in the first random liquid-cooling sample as liquid-cooling particles.

[0078] Step S32: Initialize the liquid-cooled particles and randomly mutate 20% of them. The amplitude of random mutation is within the range.

[0079] In step S33 , the initial positions of the liquid-cooling particles are simulated and updated based on the particle swarm optimization algorithm. 30% of the optimal liquid-cooling particles are retained for the next round of simulation update in each simulation update.

[0080] Step S34: When the particle swarm optimization algorithm reaches the optimal solution, the first optimized liquid cooling control parameter is output.

[0081] In this embodiment, the particle swarm optimization algorithm uses the power of the liquid cooling distribution unit and the cooling water pump speed as decision variables, constructs an initialized particle population through a first random sample, and each particle position corresponds to a combination of the power of the liquid cooling distribution unit and the cooling water pump speed. The velocity vector determines its search direction and step size, and introduces a mutation mechanism to randomly mutate 20% of the liquid cooling particles. This can be simply understood as randomly resetting the initial positions of 20% of the liquid cooling particles to escape the local optimal solution. Subsequently, the objective function of the particle swarm optimization algorithm can be simulated and updated to solve the optimal solution.

[0082] It can be understood that when generating and simulating the first random sample, the Monte Carlo algorithm and the particle swarm optimization algorithm are connected and mixed to improve sample diversity and solution speed. The Monte Carlo algorithm provides a diverse initial solution distribution for the particle swarm through global random sampling, avoiding the premature convergence problem caused by the aggregation of the initial particle population in the traditional particle swarm algorithm. At the same time, its large-scale exploration capability enables the particle swarm optimization algorithm to maintain sample diversity during the simulation update process and prevent the loss of potential optimal solutions. The particle swarm optimization algorithm uses a group collaboration mechanism to perform targeted accelerated optimization of the candidate solutions generated by Monte Carlo. By utilizing the dual guidance of individual historical optimality and group optimality, the random search is transformed into an efficient gradient approach, improving the speed and accuracy of convergence. The combination of the two not only retains the robustness of the Monte Carlo algorithm for multidimensional, nonlinear, and uncertain problems, but also compensates for the blind computational redundancy of the Monte Carlo algorithm through the local refined search of the particle swarm optimization algorithm. In the application scenario of the immersion liquid cooling system of this embodiment, the problem of being trapped in the local optimal solution is reduced while reducing the overall computing power.

[0083] Step S4: executing a random forest algorithm on the second random liquid cooling sample, performing repeated simulation updates based on the random forest algorithm, and obtaining second optimized liquid cooling control parameters.

[0084] Specifically, a liquid cooling prediction model is constructed using a deep regression tree in a random forest algorithm, the second random liquid cooling sample is imported into the liquid cooling prediction model, and a simulation update is performed based on the liquid cooling prediction model. When the liquid cooling prediction model reaches the optimal solution, the second optimized liquid cooling control parameter is output.

[0085] Specifically, the deep regression tree in the random forest algorithm constructs a liquid cooling prediction model by integrating multiple deep regression trees. After inputting the second random liquid cooling sample, that is, inputting the parameter combination of the liquid cooling distribution unit power and the cooling water pump speed, it predicts the performance of energy efficiency, stability and response speed based on multi-dimensional features such as temperature gradient and equipment efficiency attenuation rate. In actual use, generalization models such as Bagging strategy and feature random shielding can also be added to enhance the liquid cooling prediction model. After predicting the performance, the multi-objective optimization problem can be converted into a single-objective sequence, and multiple groups of optimal solutions can be generated after gradual relaxation. Then, the feasibility of multiple groups of optimal solutions is quickly evaluated through update simulation of the liquid cooling prediction model to achieve the global optimal solution.

[0086] It can be understood that this embodiment obtains the first optimized liquid cooling control parameters through the Monte Carlo algorithm and the particle swarm optimization algorithm, and obtains the second optimized liquid cooling control parameters through the random forest algorithm, providing a basis for subsequent stability judgment. When both meet the stability requirements, the first optimized liquid cooling control parameters can be combined with the second optimized liquid cooling control parameters to obtain a better cooling effect. If only one of the groups meets the stability requirements, it is used directly.

[0087] Step S5: establishing a liquid cooling second-order transfer function, drawing a root locus diagram of the liquid cooling second-order transfer function, and performing stability judgment based on the root locus diagram of the liquid cooling second-order transfer function.

[0088] Furthermore, when establishing the liquid-cooled second-order transfer function, the thermodynamic equation can be Laplace transformed to obtain a second-order transfer function containing two inertia links and one integral link. The constants in the liquid-cooled second-order transfer function can be identified through a frequency sweep experiment. When constructing the liquid-cooled second-order transfer function, nonlinear compensation can be added to the liquid-cooled second-order transfer function to describe the flow state mutation in the immersion liquid cooling system and characterize the phase characteristics near the end flow critical point. After the liquid-cooled second-order transfer function, the root locus diagram of the liquid-cooled second-order transfer function can be drawn based on the existing root locus drawing method.

[0089] Specifically, when drawing the root locus diagram of the liquid-cooled second-order transfer function, a plane coordinate system is established, in which the X-axis is represented as a real number and the Y-axis is represented as a complex number. When the root locus diagram of the liquid-cooled second-order transfer function is in the second or third quadrant of the plane coordinate system, it is in a stable state. When the root locus diagram of the liquid-cooled second-order transfer function is in the first or fourth quadrant of the plane coordinate system, it is in an unstable state.

[0090] It can be understood that the root locus diagram is a graphical tool used for analysis in control systems. It can intuitively display the motion trajectory of the system's closed-loop poles when they change, thereby judging the stability and dynamic performance of the system. In the root locus diagram, the poles move along a specific path. When the poles are located in the left half of the complex plane, the system is stable. The closer the poles are to the imaginary axis, the slower but smoother the response. The farther away from the imaginary axis, the faster the response. If the poles move to the right half plane, the system will be unstable.

[0091] In this embodiment, step S5 includes:

[0092] In step S51, the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are imported into the liquid cooling second-order transfer function, and the positions of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are found on the root locus diagram of the liquid cooling transfer function.

[0093] Step S52: If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the second or third quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in a stable state, and a fusion operation is performed on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and the fused liquid cooling control parameter is obtained based on the fusion operation.

[0094] In this embodiment, if the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the second or third quadrant of the root locus diagram, it means that the two parameters are in a stable state, and the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter can be used.

[0095] Specifically, the fusion operation is represented by performing a weighted summation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter. The calculation formula of the weighted summation is expressed as:

[0096] ;

[0097] in, Expressed as fusion liquid cooling control parameters, It represents the first optimized liquid cooling control parameter, It is represented as the second optimized liquid cooling control parameter, It is expressed as the calculation weight of the first optimized liquid cooling control parameter, It represents the calculation weight of the second optimized liquid cooling control parameter;

[0098] When performing the calculation, a constraint condition is introduced, which is expressed as:

[0099] ,and .

[0100] It can be understood that the sum of the calculation weight of the first optimized liquid cooling control parameter and the calculation weight of the second optimized liquid cooling control parameter should be equal to 1. If the sum of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter is not 1, the fused liquid cooling control parameter generated by the fusion operation is an incorrect parameter. After importing it into the liquid cooling control module, it cannot achieve the expected effect. The first optimized liquid cooling control parameter generated by simulation by the Monte Carlo algorithm and the particle swarm algorithm is better than the second optimized control parameter generated by simulation by the random forest algorithm and the random forest algorithm. Therefore, the calculation weight of the first optimized liquid cooling control parameter is greater than the calculation weight of the second optimized control parameter, and the values ​​of the two should be relatively close, otherwise it will cause a certain optimized liquid cooling control parameter to be in an absolutely dominant state. For example, if the calculation weight of the first optimized liquid cooling control parameter is 0.65, the calculation weight of the second optimized liquid cooling control parameter is 0.35.

[0101] Step S53: If a group of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in the first quadrant or the fourth quadrant in the root locus diagram, then the group of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in an unstable state, and another group of parameters in a stable state are obtained as the optimized liquid cooling control parameters.

[0102] In this embodiment, if the first optimized liquid cooling control parameter or the second optimized liquid cooling control parameter is in the first quadrant or the fourth quadrant in the root locus diagram, it means that one set of parameters is not in a stable state, and this set of parameters needs to be eliminated, and the remaining set of stable parameters can be used.

[0103] In step S54, if the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the first quadrant or the fourth quadrant in the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in an unstable state, and step S2 is executed again.

[0104] In this embodiment, if the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the first quadrant or the fourth quadrant of the root locus diagram, it means that both sets of parameters are not in a stable state and both sets of parameters cannot be used. At this time, step S2 needs to be re-executed and random sampling is performed again through the Monte Carlo algorithm and the random forest algorithm.

[0105] Step S6: Based on the stability judgment, the optimized liquid cooling control parameters or the fused liquid cooling control parameters are obtained, and the optimized liquid cooling control parameters or the fused liquid cooling control parameters are imported into the liquid cooling control module. The liquid cooling control module can be used to control the power of the liquid cooling distribution unit and the speed of the cooling water pump in real time.

[0106] It can be understood that the optimized liquid cooling control parameters or the fused liquid cooling control parameters are obtained through step S5 in this embodiment, and the optimized liquid cooling control parameters or the fused liquid cooling control parameters are imported into the liquid cooling control module. The liquid cooling control module can adjust the liquid cooling distribution unit power and the cooling water pump speed in real time based on the above-mentioned optimized liquid cooling control parameters or the fused liquid cooling control parameters, thereby improving the cooling effect of the immersion liquid cooling system on the IT equipment or servers working in the liquid cooling cabinet, so that the IT equipment or servers are in a stable temperature environment for a long time when processing high-density and high-load computing tasks.

[0107] An embodiment of the present invention provides a cooling liquid flow control system for an immersion liquid cooling system, comprising:

[0108] The acquisition module is used to acquire the set temperature of the liquid cooling system, and can acquire the power range of the liquid cooling distribution unit and the cooling water pump speed range based on the set temperature of the liquid cooling system.

[0109] The first solution module can define a first random process of the liquid cooling distribution unit power and the cooling water pump speed based on the Monte Carlo algorithm, obtain a first random liquid cooling sample based on the first random process, and can also obtain a first optimized liquid cooling control parameter based on the particle swarm optimization algorithm.

[0110] The second solution module can define a second random process of the liquid cooling distribution unit power and the cooling water pump speed based on the random forest algorithm, obtain a second random liquid cooling sample based on the second random process, and can also obtain a second optimized liquid cooling control parameter based on the random forest algorithm.

[0111] A stabilization module is used to establish a liquid cooling second-order transfer function, perform stability judgment based on a root locus diagram of the liquid cooling second-order transfer function, and also obtain optimized liquid cooling control parameters or integrated liquid cooling control parameters based on the stability judgment.

[0112] A liquid cooling control module can perform real-time control of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on optimized liquid cooling control parameters or integrated liquid cooling control parameters.

[0113] The specific usage and function of this embodiment are as follows:

[0114] First, the power range of the liquid cooling distribution unit and the speed range of the cooling water pump are obtained by setting the temperature of the liquid cooling system. The first random process is defined based on the Monte Carlo algorithm and the first random liquid cooling sample is obtained. The second random process is defined based on the random forest algorithm and the second random liquid cooling sample is obtained. Then, the particle swarm optimization algorithm is executed on the first random liquid cooling sample, and the random forest algorithm is executed on the second random liquid cooling sample. Repeated simulation updates are performed to obtain the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters. Then, the second-order transfer function of liquid cooling is established, and the optimized liquid cooling control parameters or the fused liquid cooling control parameters are obtained through stability judgment. Finally, the liquid cooling control parameters are adjusted. The control module can perform real-time control of the liquid cooling distribution unit power and the cooling water pump speed based on optimizing the liquid cooling control parameters or integrating the liquid cooling control parameters, thereby regulating the coolant flow of the immersion liquid cooling system. By applying the hybrid algorithm to the control of the immersion liquid cooling system, it is avoided that a single algorithm falls into a local optimal solution, resulting in poor control of the coolant flow of the immersion liquid cooling system. By optimizing the liquid cooling control parameters or integrating the liquid cooling control parameters, the liquid cooling distribution unit power and the cooling water pump speed are controlled in real time, avoiding the limitations of a single parameter setting. This method can improve the cooling effect of the immersion liquid cooling system on servers or IT equipment.

[0115] In addition, an embodiment of the present invention also provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method in the above embodiment.

[0116] The following is a detailed introduction to the various components of electronic equipment:

[0117] The term "processor" is used to refer to the control center of an electronic device and can be a single processor or a collective term for multiple processing elements. For example, a processor can be one or more central processing units (CPUs), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs) or one or more field programmable gate arrays (FPGAs).

[0118] The processor can execute various functions of the electronic device by running or executing software programs stored in the memory and calling data stored in the memory.

[0119] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment and will not be repeated here.

[0120] The memory may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, a random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only (CD-ROM), or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. The memory may be integrated with the processor or exist independently and be coupled to the processor via an interface circuit of the electronic device, and this is not specifically limited in the embodiments of the present invention.

[0121] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wireless communication (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0122] It should be understood that the term "and / or" as used herein simply describes an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent the existence of A alone, the existence of both A and B, or the existence of B alone. A and B can be singular or plural. Furthermore, the character " / " as used herein generally indicates an "or" relationship between the related objects, but it may also indicate an "and / or" relationship. For specific understanding, please refer to the context.

[0123] 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.

[0124] 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 they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for controlling the flow of coolant in an immersion liquid cooling system, characterized in that: The method comprises: S1: Acquire a set temperature of the liquid cooling system, and acquire a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system; S2: defining a first random process of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on a Monte Carlo algorithm, obtaining a first random liquid cooling sample based on the first random process, defining a second random process of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on a random forest algorithm, and obtaining a second random liquid cooling sample based on the second random process; S3: executing a particle swarm optimization algorithm on the first random liquid cooling sample, performing repeated simulation updates based on the particle swarm optimization algorithm, and obtaining first optimized liquid cooling control parameters; S4: Execute a random forest algorithm on the second random liquid cooling sample, perform repeated simulation updates based on the random forest algorithm, and obtain second optimized liquid cooling control parameters; S5: establishing a liquid cooling second-order transfer function, drawing a root locus diagram of the liquid cooling second-order transfer function, performing stability judgment based on the root locus diagram of the liquid cooling second-order transfer function, and obtaining optimized liquid cooling control parameters or integrated liquid cooling control parameters based on the stability judgment; Importing the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter into a liquid cooling second-order transfer function, and finding positions of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter on a root locus diagram of the liquid cooling transfer function; If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the second or third quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in a stable state, performing a fusion operation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and obtaining the fused liquid cooling control parameter based on the fusion operation; If a set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in the first quadrant or the fourth quadrant of the root locus diagram, then the set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in an unstable state, and another set of parameters in a stable state is obtained as the optimized liquid cooling control parameters; If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the first quadrant or the fourth quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in an unstable state, and step S2 is executed again; S6: Importing the optimized liquid cooling control parameters or the fused liquid cooling control parameters into a liquid cooling control module, through which the power of the liquid cooling distribution unit and the speed of the cooling water pump can be controlled in real time.

2. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: The performing a fusion operation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and obtaining the fused liquid cooling control parameter based on the fusion operation, includes: The fusion operation is represented by weighted summing of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter. The calculation formula of the weighted summing is expressed as: ； in, Expressed as fusion liquid cooling control parameters, It represents the first optimized liquid cooling control parameter, It is represented as the second optimized liquid cooling control parameter, It is expressed as the calculation weight of the first optimized liquid cooling control parameter, It represents the calculation weight of the second optimized liquid cooling control parameter; When performing the calculation, a constraint condition is introduced, which is expressed as: ,and .

3. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: The step of establishing a liquid cooling second-order transfer function and drawing a root locus diagram of the liquid cooling second-order transfer function includes: When drawing the root locus diagram of the liquid-cooled second-order transfer function, a plane coordinate system is established, in which the X-axis is represented as a real number and the Y-axis is represented as a complex number. When the root locus diagram of the liquid-cooled second-order transfer function is in the second or third quadrant of the plane coordinate system, it is in a stable state. When the root locus diagram of the liquid-cooled second-order transfer function is in the first or fourth quadrant of the plane coordinate system, it is in an unstable state.

4. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: Defining a first random process of the liquid cooling distribution unit power and the cooling water pump speed based on the Monte Carlo algorithm, and obtaining a first random liquid cooling sample based on the first random process, includes: Dividing the liquid cooling distribution unit power interval and the cooling water pump speed interval into ten liquid cooling random intervals respectively, wherein the random probability of each liquid cooling random interval is equal, and the random probability of each liquid cooling random interval is ten percent; Obtain two extreme value intervals within the liquid cooling random interval, where the extreme value intervals are represented by the lowest and highest extreme value intervals in the liquid cooling random interval, and widen the boundaries of the two extreme value intervals; For the lowest extreme value interval, the minimum value of the extreme value interval is reduced by 10%, and for the highest extreme value interval, the maximum value of the extreme value interval is increased by 10%; Random sampling is performed based on a liquid cooling random interval corresponding to the power interval of the liquid cooling distribution unit and the speed interval of the cooling water pump to obtain a first random liquid cooling sample; When performing random sampling, the sampling density of each liquid-cooling random interval is equal, and the distance between each first random liquid-cooling sample is greater than a preset threshold.

5. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: Defining a second random process of the liquid cooling distribution unit power and the cooling water pump speed based on the random forest algorithm, and obtaining a second random liquid cooling sample based on the second random process, includes: Obtain key characteristics of liquid cooling, where the key characteristics of liquid cooling are represented by physical operating characteristics of the immersion liquid cooling system during operation. Randomly generate random samples within the power range of the liquid cooling distribution unit and the speed range of the cooling water pump based on the random forest algorithm, import the random samples into the random forest algorithm, perform decision tree splitting based on the key characteristics of liquid cooling, and obtain a second random liquid cooling sample.

6. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: The performing of a particle swarm optimization algorithm on the first random liquid cooling sample, performing repeated simulation updates based on the particle swarm optimization algorithm, and obtaining first optimized liquid cooling control parameters includes: Establishing a two-dimensional space, mapping the first random liquid-cooling sample into the two-dimensional space, using the first random liquid-cooling sample as a liquid-cooling particle group, and using the random liquid-cooling distribution unit power and the random cooling water pump speed included in the first random liquid-cooling sample as liquid-cooling particles; Initialize the liquid-cooled particles and randomly mutate 20% of them. The amplitude of random mutation is within the range of The initial positions of the liquid-cooling particles are simulated and updated based on the particle swarm optimization algorithm, and 30% of the optimal liquid-cooling particles are retained for the next round of simulation update in each simulation update; When the particle swarm optimization algorithm reaches the optimal solution, the first optimized liquid cooling control parameter is output.

7. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: The performing of a random forest algorithm on the second random liquid cooling sample, performing repeated simulation updates based on the random forest algorithm, and obtaining a second optimized liquid cooling control parameter includes: A liquid cooling prediction model is constructed using the deep regression tree in the random forest algorithm, the second random liquid cooling sample is imported into the liquid cooling prediction model, and a simulation update is performed based on the liquid cooling prediction model. When the liquid cooling prediction model reaches the optimal solution, the second optimized liquid cooling control parameter is output.

8. The method for controlling the flow of coolant in an immersion liquid cooling system according to claim 1, wherein: The step of obtaining a set temperature of the liquid cooling system and obtaining a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system includes: Physical condition restrictions are imposed on the obtained liquid cooling distribution unit power range and cooling water pump speed range. The physical condition restrictions are expressed as eliminating the liquid cooling distribution unit power that exceeds the actual operating power of the liquid cooling distribution unit within the liquid cooling distribution unit power range, and eliminating the cooling water pump speed that exceeds the actual operating speed of the cooling water pump within the cooling water pump speed range.

9. A cooling liquid flow control system for an immersion liquid cooling system, applying the method according to any one of claims 1 to 8, characterized in that: include: An acquisition module, the acquisition module is used to obtain a set temperature of the liquid cooling system, and can obtain a power range of the liquid cooling distribution unit and a speed range of the cooling water pump based on the set temperature of the liquid cooling system; a first solution module, wherein the first solution module may define a first random process of the liquid cooling distribution unit power and the cooling water pump speed based on a Monte Carlo algorithm, obtain a first random liquid cooling sample based on the first random process, and may also obtain a first optimized liquid cooling control parameter based on a particle swarm optimization algorithm; a second solution module, wherein the second solution module may define a second random process of the liquid cooling distribution unit power and the cooling water pump speed based on a random forest algorithm, obtain a second random liquid cooling sample based on the second random process, and further obtain a second optimized liquid cooling control parameter based on the random forest algorithm; a stabilization module, the stabilization module being used to establish a liquid cooling second-order transfer function, perform stability judgment based on a root locus diagram of the liquid cooling second-order transfer function, and further obtain optimized liquid cooling control parameters or integrated liquid cooling control parameters based on the stability judgment; Importing the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter into a liquid cooling second-order transfer function, and finding positions of the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter on a root locus diagram of the liquid cooling transfer function; If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the second or third quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in a stable state, performing a fusion operation on the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter, and obtaining the fused liquid cooling control parameter based on the fusion operation; If a set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in the first quadrant or the fourth quadrant of the root locus diagram, then the set of parameters among the first optimized liquid cooling control parameters and the second optimized liquid cooling control parameters are in an unstable state, and another set of parameters in a stable state is obtained as the optimized liquid cooling control parameters; If the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in the first quadrant or the fourth quadrant of the root locus diagram, then the first optimized liquid cooling control parameter and the second optimized liquid cooling control parameter are both in an unstable state, and step S2 is executed again; A liquid cooling control module can perform real-time control of the power of the liquid cooling distribution unit and the speed of the cooling water pump based on optimized liquid cooling control parameters or integrated liquid cooling control parameters.

Images