Collaborative control optimization method suitable for near-field wind wall micro-module

Abstract

A collaborative control optimization method suitable for a near-field wind wall micro-module, comprising the following steps: step 10, acquiring IT system scheduling data and air conditioning system data of the near-field wind wall micro-module within a preset time period, and respectively establishing a micro-module server real-time power consumption model, an air conditioning system power consumption model and a server temperature rapid prediction model; and step 20, on the basis of the micro-module server real-time power consumption model, the air conditioning system power consumption model and the server temperature rapid prediction model, using an optimization method combining real-time coupling and periodic decoupling to perform collaborative control on virtual machine deployment and an air conditioning system. The method can solve the technical problem of the near-field wind wall micro-module being prone to local hotspots and overcooling during energy-efficient scheduling, thereby improving the overall energy efficiency and operational stability of the near-field wind wall micro-module.

Description

A cooperative control optimization method suitable for near-field wind wall micro modules TECHNICAL FIELD

[0001] The application belongs to the technical field of data center air cooling control, and particularly relates to a cooperative control optimization method suitable for near-field wind wall micro modules. BACKGROUND

[0002] The near-field wind wall air conditioner is one of the data center air cooling technologies with the highest heat exchange efficiency at present, and is widely used in various high-power-density data centers. However, in actual engineering, the energy-saving scheduling of the IT system and the cooling system of the near-field wind wall micro module is executed separately, that is, the cloud platform of the IT system usually adopts the scheduling strategy of the least active server, and the near-field wind wall air conditioner adopts the feedback control strategy based on the supply / return air temperature (usually the fan frequency is controlled according to the supply air temperature, and the evaporating temperature of the coil is controlled according to the return air temperature). There is a lack of effective cooperation between the two, which leads to problems such as local hot spots and excessive cooling, resulting in the energy consumption of the IT system and the cooling system being this and that, and greatly affecting the safety and stability of the data center.

[0003] In order to improve the precise cooling capacity of the near-field wind wall air conditioner, an effective way is to predict the chip temperature of the server according to the real-time power change of the server in the micro module, so as to regulate and control the end air volume and the evaporating temperature of the coil of the near-field wind wall air conditioner. However, due to the characteristics of fluid transportation and heat capacity, the cooling system represented by the near-field wind wall air conditioner has significant time lag and regulation accuracy problems during the cooling process, and it is difficult to dynamically match the heat dissipation change of each server. Therefore, this technology cannot fundamentally solve the cold and hot mismatch problem of the data center.

[0004] SUMMARY

[0005] The application provides a cooperative control optimization method suitable for near-field wind wall micro modules, which solves the technical problem that local hot spots and excessive cooling easily occur in the energy-saving scheduling of the near-field wind wall micro module, so as to improve the overall energy efficiency and operation stability of the near-field wind wall micro module.

[0006] To solve the above technical problems, the application adopts the following technical scheme:

[0007] The application provides a cooperative control optimization method suitable for near-field wind wall micro modules, which includes the following steps:

[0008] Step 10, obtaining the IT system scheduling data and air conditioner system data of the near-field wind wall micro module in a preset time period, and respectively establishing a micro module server real-time power consumption model, an air conditioner system power consumption model and a server temperature rapid prediction model;

[0009] Step 20, based on the micro-module server real-time power consumption model, the air conditioning system power consumption model and the server temperature fast prediction model, the optimization method of real-time joint and timing decoupling is adopted to cooperatively control the virtual machine deployment and the air conditioning system.

[0010] As a further improvement of the application, the IT system scheduling data includes virtual machine CPU resource demand, virtual machine and server mapping relationship, server CPU usage, server real-time power consumption and server chip temperature; the air conditioning system data includes fan frequency of each terminal module, coil evaporating temperature, outdoor unit condensing temperature, load rate and total power consumption of the air conditioning system.

[0011] As a further improvement of the application, in step 10, according to the IT system scheduling data, the micro-module server real-time power consumption model is established;

[0012] Specifically, it includes:

[0013] Step 101, the power consumption data of each server under different chip temperatures and CPU usage is screened from the IT system scheduling data;

[0014] Step 102, according to the power consumption data of each server under the condition of CPU usage being 0, a server static power consumption matrix is established;

[0015] Step 103, according to the data of the change of real-time power consumption of each server with CPU usage, the conversion coefficient of CPU usage and dynamic power consumption is fitted by using multiple linear regression method, and a conversion coefficient matrix of CPU usage and dynamic power consumption is formed;

[0016] Step 104, according to the data of different temperatures of the chip of each server under the same CPU usage, the expression of the leakage power consumption of each server is fitted by using multiple linear regression method, and a server leakage power consumption matrix is formed;

[0017] Step 105, the server static power consumption matrix, the conversion coefficient matrix of CPU usage and running power consumption and the server leakage power consumption matrix are combined to obtain the micro-module server real-time power consumption model shown in formula (1): P PM (t)=P SC (t)+u VM (t)*MM VM→PM (A(t),F(t))*f h +D(T PM (t)) Formula (1)

[0018] In the formula, P PM (t) represents the server real-time power consumption matrix, P SC (t) represents the server static power consumption matrix, u VM(t) represents a CPU resource requirement matrix of a virtual machine, MM VM→PM (A(t),F(t)) represents a mapping matrix between a virtual machine and a server, A(t) represents a virtual machine real-time deployment matrix, F(t) represents a virtual machine migration matrix, u VM (t)*MM VM→PM (A(t),F(t)) represents a CPU usage rate of a server, f h represents a conversion coefficient matrix of a CPU usage rate and dynamic power consumption, D(T PM (t)) represents a server leakage power consumption matrix.

[0019] As a further improvement of the application, in step 10, an air conditioning system power consumption model shown in formula (2) is established according to air conditioning system data: P FC (t) = P CP (T co (t), T ev (t), PLR(t)) + ∑P FX (FP(t)) Formula (2)

[0020] In the formula, P FC (t) represents total power consumption of an air conditioning system, P CP (T co (t), T ev (t), PLR(t)) represents power consumption of an outdoor unit, T co (t) represents evaporator temperature of a coil, T ev (t) represents condenser temperature of an outdoor unit, PLR(t) represents load rate, FP(t) represents a fan frequency matrix of each terminal module, P FX (FP(t)) represents a power consumption matrix of each terminal module.

[0021] As a further improvement of the application, in step 10, a server temperature rapid prediction model is established according to IT system scheduling data and air conditioning system data;

[0022] Specifically, it comprises:

[0023] Step 131, the mapping relationship between a virtual machine and a server, virtual machine CPU resource requirements, and the temperature of each server chip are screened out from the IT system scheduling data, and the fan frequency of each terminal module and the evaporator temperature of a coil are screened out from the air conditioning system data to form a modeling data set;

[0024] Step 132, based on the modeling data set, the fan frequency of each terminal module, the evaporator temperature of a coil, and the CPU usage rate of each server are taken as input variables of a BP neural network model, and the temperature of each server chip is taken as an output variable of the BP neural network model, the BP neural network model is trained to obtain a server temperature rapid prediction model.

[0025] As a further improvement of the application, in step 20, the optimization method combining real-time joint and timing decoupling is used to cooperatively control the virtual machine deployment and the air conditioning system, including:

[0026] Step 21, when the cloud platform receives a new virtual machine deployment request, the deployment position of the new virtual machine and the fan frequency of each terminal module are jointly optimized to achieve the lowest overall power consumption of the server and each terminal module under single virtual machine deployment;

[0027] Step 22, at a given time, first, the virtual machines running on each server are migrated as a whole to minimize the total power consumption of the server; then, the evaporator temperature of the air conditioning system and the fan frequency of each terminal module are optimized to achieve the lowest overall power consumption of the air conditioning system.

[0028] As a further improvement of the application, step 21 specifically includes:

[0029] Step 211, after the cloud platform receives a new virtual machine deployment request, the remaining CPU resources of all active servers are traversed according to the CPU resource requirements of the new virtual machine to find a server that can be deployed;

[0030] Step 212, if there is a server that can be deployed, the position of the server that can be deployed is used as an optimization variable for the deployment of the new virtual machine, and step 213 is executed; if the remaining CPU resources of all active servers cannot deploy the new virtual machine, a dormant server is randomly activated, the new virtual machine is deployed, and step 214 is executed;

[0031] Step 213, taking the sum of the real-time total power consumption of the server after deployment and the total power consumption of the terminal module as the optimization objective, taking the sum of the resources used by all virtual machines running on each server not exceeding the total rated resources of the server and each server chip temperature not exceeding the preset temperature threshold as the optimization constraint, using a genetic algorithm to jointly optimize the deployment position of the new virtual machine and the fan frequency of each terminal module, the optimal real-time deployment position of the virtual machine and the optimal fan frequency of each terminal module are obtained; the cloud platform implements the deployment of the new virtual machine according to the optimal real-time deployment position of the virtual machine, and the air conditioning system controls the fan frequency of each terminal module according to the optimal fan frequency of each terminal module;

[0032] Step 214, taking the sum of the power consumption of each terminal module after the deployment of the new virtual machine as the optimization objective, taking the constraint that each server chip temperature cannot exceed the preset temperature threshold, using a genetic algorithm to optimize the fan frequency of each terminal module, the optimal fan frequency of each terminal module is obtained; the air conditioning system controls the fan frequency of each terminal module according to the optimal fan frequency of each terminal module.

[0033] As a further improvement of the application, the step 22 specifically comprises:

[0034] Step 221, at a given time, calculating the minimum number of active server required according to the number of all virtual machines running on the server and the sum of their CPU resource requirements;

[0035] Step 222, according to the minimum number of active server required, selecting servers as active servers in the next period in the order of rack number and server number on each rack until the number of selected servers reaches the minimum number of active server;

[0036] Step 223, according to the load balancing strategy, migrating all existing virtual machines to the selected active servers; hibernating idle servers;

[0037] Step 224, taking the total power consumption of the air conditioning system as the optimization goal and taking the temperature of each server chip not exceeding the preset temperature threshold as the constraint, using genetic algorithm to optimize the evaporator temperature of the coil and the fan frequency of each end module to obtain the optimal evaporator temperature of the coil and the optimal fan frequency of each end module;

[0038] Step 225, according to the optimal evaporator temperature of the coil, the air conditioning system controls the evaporator temperature of the coil, and according to the optimal fan frequency of each end module, the fan frequency of each end module is controlled.

[0039] Compared with the prior art, the technical scheme of the application has the following beneficial effects:

[0040] The application provides a cooperative control optimization method suitable for a near-field wind wall micro module. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed to be used in the embodiments of the application will be briefly introduced as follows. Obviously, the drawings described below are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0042] Fig. 1 is a flow chart of the cooperative control optimization method of the embodiments of the application.

[0043] Fig. 2 is a flow chart of the real-time joint optimization of the virtual machine deployment and the air conditioning system in the method of the embodiments of the application.

[0044] Fig. 3 is a flow chart of the decoupling optimization of the whole virtual machine migration and the air conditioning system in the method of the embodiments of the application.

[0045] Fig. 4 is a structural schematic diagram of the near-field wind wall micro module suitable for the method of the embodiments of the application.

[0046] In the drawings, the near-field wind wall micro module 1, the rack 11, the server 12, the end module 21, the fan 22, the evaporator coil 23, the refrigerant pipeline 24 and the outdoor unit 25. DETAILED DESCRIPTION

[0047] The technical solutions of the present application will be described in detail below with reference to the drawings.

[0048] The embodiment of the present application provides a kind of suitable for the cooperative control optimization method of near field wind wall micro module, it is applied to as shown in Figure 4 near field wind wall micro module 1.In near field wind wall micro module 1, there are N racks 11, each rack is equipped with M servers 12.Near field wind wall has N end modules 21 corresponding to N racks 11, and the fan 22 in single end module can only be uniformly frequency conversion, and evaporator coil 23 is connected with external machine 25 (including condenser, compressor and throttling device) by refrigerant pipeline 24.

[0049] The embodiment of the present application provides a kind of suitable for the cooperative control optimization method of near field wind wall micro module, as shown in Figure 1, including the following steps:

[0050] Step 10, obtains the IT system scheduling data and air conditioning system data of near field wind wall micro module in preset time period;According to IT system scheduling data, the micro module server real-time power consumption model of near field wind wall micro module is established;According to air conditioning system data, the air conditioning system power consumption model of near field wind wall micro module is established;According to IT system scheduling data and air conditioning system data, the server temperature rapid prediction model of near field wind wall micro module is established.

[0051] Step 20, based on micro module server real-time power consumption model, air conditioning system power consumption model and server temperature rapid prediction model, using real-time joint and timing decoupling combined optimization method, the virtual machine deployment and air conditioning system are cooperatively controlled.

[0052] In the above step 10, IT system scheduling data is obtained from scheduling data stored in cloud platform.IT system scheduling data includes virtual machine CPU resource demand, mapping relationship between virtual machine and server, CPU usage of each server, real-time power consumption of each server and chip temperature of each server.Air conditioning system data is obtained from the management system of near field wind wall air conditioning system.Air conditioning system data includes fan frequency of each end module, evaporator temperature of coil, condenser temperature of external machine, load rate and total power consumption of air conditioning system.Preset time period is preferably not less than 24 hours.

[0053] In the above step 10, assuming that each server is isomorphic, according to IT system scheduling data, matrix modeling method is used to establish micro module server real-time power consumption model.

[0054] Specifically, it includes:

[0055] Step 101, filter the power consumption data of each server under different chip temperature and CPU usage from IT system scheduling data.

[0056] Step 102, according to the power consumption data of each server in the CPU usage rate of 0 state, the server static power consumption matrix is established.

[0057] Step 103, according to the data of the change of the real-time power consumption of each server with the CPU usage rate, the conversion coefficient of the CPU usage rate and the dynamic power consumption (the real-time power consumption minus the static power consumption under the standard working condition of the server is the dynamic power consumption) is fitted by using the multiple linear regression method, and the conversion coefficient matrix of the CPU usage rate and the dynamic power consumption is constructed.

[0058] Step 104, according to the data of the different temperatures of the chip of each server under the same CPU usage rate, the expression of the leakage power consumption of the server is fitted by using the multiple linear regression method, and the leakage power consumption matrix of the server is constructed.

[0059] Step 105, the server static power consumption matrix, the conversion coefficient matrix of the CPU usage rate and the running power consumption and the leakage power consumption matrix of the server are combined to obtain the real-time power consumption model of the micro-module server shown in formula (1): PM (t)=P SC (t)+u VM (t)*MM VM→PM (A(t),F(t))*f h +D(T PM (t)) Formula (1)

[0060] In the formula, P PM (t) represents the real-time power consumption matrix of the server, and each element in the matrix P PM (t) represents the real-time power consumption model of each server, that is, each element corresponds to each server in the near-field wind wall micro-module, and one element represents the power consumption model of one server; P SC (t) represents the server static power consumption matrix, u VM (t) represents the CPU resource demand matrix of the virtual machine. MM VM→PM (A(t),F(t)) represents the mapping matrix between the virtual machine and the server, which is a function of the real-time deployment matrix A(t) of the virtual machine and the migration matrix F(t) of the virtual machine, and can be represented in the form of a (0, 1) matrix. For example, indicates that the No. 1 virtual machine is deployed on the No. 1 server, and the No. 2 virtual machine is deployed on the No. 2 server. u VM (t)*MM VM→PM (A(t),F(t)) represents the CPU usage rate of the server, f h represents the conversion coefficient matrix of the CPU usage rate and the dynamic power consumption, and D(T PM (t)) represents the leakage power consumption matrix of the server.

[0061] The real-time power consumption of the micro-module server is composed of static power consumption, dynamic power consumption and leakage power consumption, wherein the static power consumption is a fixed value related to the server performance, the dynamic power consumption is a function of the CPU usage rate of the server, and the leakage power consumption is a function of the CPU chip temperature of the server.

[0062] In step 10, the air conditioning system power consumption model shown in formula (2) is established according to the air conditioning system data: P FC (t)=P CP (T co (t),T ev (t),PLR(t))+∑P FX (FP(t)) Formula (2)

[0063] In formula (2), P FC (t) represents the total power consumption of the air conditioning system; P CP (T co (t),T ev (t),PLR(t)) represents the outdoor unit power consumption, which is a function of the evaporator temperature T co (t) of the coil, the condenser temperature T ev (t) of the outdoor unit and the load rate PLR(t); P FX (FP(t)) represents the power consumption matrix of each terminal module, which is a function of the fan frequency matrix FP(t) of each terminal module.

[0064] In step 10, the BP neural network model is established, the fan frequency of each terminal module, the evaporator temperature of the coil and the CPU usage rate of each server are taken as input variables, and the chip temperature of each server is taken as output variable, so as to build the dynamic mapping relationship between the fan frequency of each terminal module, the evaporator temperature of the coil, the CPU usage rate of each server and the chip temperature of each server.

[0065] Specifically, it includes:

[0066] In step 131, the mapping relationship between the virtual machine and the server, the CPU resource demand of the virtual machine and the chip temperature of each server are screened from the IT system scheduling data, the fan frequency of each terminal module and the evaporator temperature of the coil are screened from the air conditioning system data, and the modeling data set is constructed. The modeling data set is divided into a training set and a test set according to a ratio of 4:1.

[0067] Step 132, the fan frequency of each end module, the coil evaporation temperature and the CPU usage rate of each server (calculated according to the mapping relationship between the virtual machine and the server and the CPU resource demand of the virtual machine) are taken as the input variables of the BP neural network model, and the chip temperature of each server is taken as the output variable of the neural network model. The training set is used to train the BP neural network model, and the number of layers, the number of neurons in each layer and other hyperparameters of the model are optimized during the training process.

[0068] The accuracy of the trained BP neural network model is tested by using the test set. If the model error meets the requirements (for example, less than 1%), the model is considered qualified, and a server temperature fast prediction model is obtained.

[0069] In the above step 20, a real-time joint and timed decoupling combined optimization method is used to cooperatively control the virtual machine deployment and the air conditioning system, including:

[0070] Step 21, when the cloud platform receives a new virtual machine deployment request, the virtual machine deployment position and the fan frequency of each end module are jointly optimized to achieve the lowest overall power consumption of the server and each end module under single virtual machine deployment.

[0071] Step 22, at each given time, the virtual machines running on each server are migrated as a whole to minimize the total power consumption of the server. On this basis, the coil evaporation temperature of the air conditioning system and the fan frequency of each end module are optimized to achieve the lowest overall power consumption of the air conditioning system. Preferably, the time period between two adjacent given times is 3-15 minutes.

[0072] Preferably, as shown in FIG. 2, step 21 specifically includes:

[0073] Step 211, after the cloud platform receives the deployment request of the new virtual machine, the remaining CPU resources of all active servers are traversed according to the CPU resource demand of the new virtual machine to find a deployable server.

[0074] Step 212, if there is a deployable server, the position of the deployable server is taken as the optimization variable of the new virtual machine deployment, and step 213 is executed. If the remaining CPU resources of all active servers cannot deploy the new virtual machine, a dormant server is randomly turned into an active state, the new virtual machine is deployed, and step 214 is executed.

[0075] Step 213, to achieve the optimization goal of the sum of the total power consumption of all servers and the total power consumption of all end modules being the lowest in real time after deployment, with the constraint conditions of the sum of the used resources of all virtual machines running on each server not exceeding the total amount of rated resources of the server, and the temperature of each server chip not exceeding the preset temperature threshold, the joint optimization of the deployment position of the new virtual machine and the fan frequency of each end module is performed by using a genetic algorithm to obtain the optimal virtual machine real-time deployment position and the optimal fan frequency of each end module.

[0076] As follows:

[0077] Optimization goal: min P total,PM+FX (t) = ∑P PM (t) + ∑P FX (FP(t))

[0078] Optimization constraint: T PM (t) ≤ T max,PM and ∑uVMi in PMk(t) ≤ C PM

[0079] Optimization variable: {A(t), FP(t)}

[0080] In the formula, ∑P PM (t) represents the total real-time power consumption of all servers, ∑P FX (FP(t)) represents the total power consumption of all end modules, T PM (t) represents the temperature of each server chip, T max,PM represents the preset temperature threshold, uVMi in PMk(t) represents the sum of the used resources of all virtual machines running on each server, and C PM represents the total amount of rated resources of the server.

[0081] The deployment of the new virtual machine is implemented according to the optimal virtual machine real-time deployment position by the cloud platform, and the fan frequency of each end module is controlled according to the optimal fan frequency of each end module by the air conditioning system.

[0082] Step 214, to achieve the optimization goal of the sum of the power consumption of each end module being the lowest after the deployment of the new virtual machine, with the constraint condition of the temperature of each server chip not exceeding the preset temperature threshold, the optimization of the fan frequency of each end module is performed by using a genetic algorithm to obtain the optimal fan frequency of each end module.

[0083] As follows:

[0084] Optimization goal: min P total,FX (t) = ∑P FX (FP(t))

[0085] Optimization constraint: T PM (t) ≤ Tmax,PM

[0086] Optimization variables: {FP(t)}.

[0087] The fan frequency of each end module is controlled by the air conditioning system according to the optimal fan frequency of each end module.

[0088] In the method of the embodiment, when real-time optimization is performed, a server available for deployment of a new virtual machine is first found from active servers, and if there is one, the deployment position of the new virtual machine and the fan frequency of each end module are jointly optimized, so that the total energy consumption of the server and each end module is minimized; if there is none, a hibernating server is randomly started for deployment, and the fan frequency of each end module is optimized, so that the total energy consumption of each end module is minimized. The real-time matching degree between server heat dissipation and air conditioning end cooling is improved, so as to reduce the probability of occurrence of local hot spots and supercooling in the micro module.

[0089] Preferably, as shown in FIG. 3, step 22 specifically comprises:

[0090] Step 221, at a given time, the minimum number of active servers required is calculated according to the number of all virtual machines running on the server and the sum of CPU resource requirements thereof.

[0091] Step 222, according to the minimum number of active servers required, servers on each rack are selected as active servers in the next period in the order of rack number and from bottom to top, until the number of selected servers reaches the minimum number of active servers. In view of the fact that the heat dissipation effect of the lower part of each rack is the best, when the virtual machine is migrated in the embodiment, the server is selected for deployment from bottom to top in a single rack.

[0092] Step 223, according to the load balancing strategy, the existing virtual machines are evenly migrated to the selected active servers. The idle servers are hibernated.

[0093] Step 224, taking the total power consumption of the air conditioning system as the optimization objective and taking the condition that the temperature of each server chip does not exceed a preset temperature threshold as a constraint condition, the coil evaporating temperature and the fan frequency of each end module are optimized by using a genetic algorithm, so as to obtain the optimal coil evaporating temperature and the optimal fan frequency of each end module.

[0094] The following formula is used:

[0095] Optimization objective: min P FC (t)=P CP (T co (t),T ev (t))+∑P FX (FP(t))

[0096] Optimization constraint: TPM (t)≤T max,PM

[0097] Optimization variables: {T ev (t), FP(t)}.

[0098] Step 225, the evaporating temperature of the coil is regulated according to the optimal evaporating temperature of the coil, and the fan frequency of each terminal module is controlled according to the optimal fan frequency of each terminal module.

[0099] In the method of the embodiment, in the timing optimization, the virtual machines are migrated as a whole, and the idle server after migration is hibernated, so that the positions of the active servers in the micro module are more concentrated, and the total power consumption of the server is minimized. Then, on this basis, the evaporating temperature of the coil and the fan frequency of each terminal module are optimized, so that the total power consumption of the air conditioning system is minimized. Thus, the timing high matching between the server heat dissipation and the air conditioning cooling is promoted, so as to further improve the utilization efficiency of the cooling capacity.

[0100] In the method of the embodiment, when there is a new virtual machine deployment request, the virtual machine deployment and the air conditioning system are controlled in real time to optimize the deployment, so as to improve the real-time matching degree between the server heat dissipation in the near-field air wall micro module and the air conditioning terminal cooling, so as to reduce the probability of local hot spots and supercooling in the micro module, thereby realizing the minimum total power consumption of the server and the air conditioning terminal under single virtual machine deployment. However, if only real-time deployment optimization is performed, the active servers are easily dispersed, that is, the virtual machine deployment of a single rack server is dispersed to multiple rack servers. Although the total power consumption of the server does not change, due to the limited fine adjustment ability of the near-field air wall air conditioning terminal, the dispersion of the active servers easily causes waste of cooling capacity.

[0101] By controlling the virtual machine deployment and the air conditioning system in timing decoupling optimization at each given time, the virtual machines running on each server are migrated as a whole to promote the centralized adjustment of the active servers, so as to avoid waste of cooling capacity and minimize the total power consumption of the server. On this basis, the evaporating temperature of the coil and the fan frequency of each terminal module of the air conditioning system are optimized to minimize the total power consumption of the air conditioning system and promote the timing high matching between the server heat dissipation and the air conditioning cooling, thereby further improving the utilization efficiency of the cooling capacity. However, if only timing decoupling optimization is performed, the frequent migration of the virtual machines easily affects the service quality, and the frequent adjustment of the operating conditions of the compressor easily causes a significant decline in the service life, so the time interval of the timing decoupling optimization should not be too short, and a long time interval obviously cannot meet the real-time deployment requirements of the virtual machines.

[0102] The method of the embodiment cooperates real-time deployment optimization with timing decoupling optimization, and through the time sequence cooperation of the two, the technical problem that local hot spots and excessive cooling easily occur in the energy-saving scheduling of the near-field wind wall micro module can be solved, so that the overall energy efficiency and operation stability of the near-field wind wall micro module are improved.

[0103] In the method of the present application, during the running of the server, the virtual machine is automatically released (real-time shutdown) after completing the task. If all virtual machines on a server are released, the server is regarded as an idle server, and will be automatically hibernated (real-time power consumption is converted to 0). The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited by the above specific embodiments, and the above specific embodiments and descriptions in the specification are only for further illustration of the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A method for coordinated control optimization suitable for near field wind wall micro modules, characterized in that, The method comprises the following steps: Step 10, obtaining IT system scheduling data and air conditioning system data of the near-field wind wall micro module within a preset time period, and establishing a micro module server real-time power consumption model, an air conditioning system power consumption model and a server temperature rapid prediction model; Step 20, based on the micro module server real-time power consumption model, the air conditioning system power consumption model and the server temperature rapid prediction model, using a real-time joint and timing decoupling combined optimization method, the virtual machine deployment and the air conditioning system are cooperatively controlled.

2. The collaborative control optimization method of claim 1, wherein, The IT system scheduling data comprises virtual machine CPU resource demand, mapping relationship between virtual machines and servers, CPU usage rate of each server, real-time power consumption of each server and chip temperature of each server; the air conditioning system data comprises fan frequency of each terminal module, evaporating temperature of a coil, condensing temperature of an outdoor unit, load rate and total power consumption of the air conditioning system.

3. The collaborative control optimization method of claim 2, wherein, In the step 10, the micro module server real-time power consumption model is established according to the IT system scheduling data; Specifically comprising: Step 101, screening power consumption data of each server under different chip temperatures and CPU usage rates from the IT system scheduling data; Step 102, establishing a server static power consumption matrix according to the power consumption data of each server under the condition that the CPU usage rate is 0; Step 103, using a multiple linear regression method to fit the conversion coefficient of the CPU usage rate and the dynamic power consumption of each server according to the data of the real-time power consumption of each server changing with the CPU usage rate, and constructing a conversion coefficient matrix of the CPU usage rate and the dynamic power consumption; Step 104, using a multiple linear regression method to fit the expression of the leakage power consumption of each server according to the data of the different temperatures of the chip of each server under the same CPU usage rate, and constructing a server leakage power consumption matrix; Step 105, combining the server static power consumption matrix, the conversion coefficient matrix of the CPU usage rate and the running power consumption and the server leakage power consumption matrix to obtain the micro module server real-time power consumption model shown in formula (1): P PM (t) = P SC (t) + u VM (t) * MM VM→PM (A(t), F(t)) * f h + D(T PM (t)) Equation (1) where P PM (t) represents the server real-time power consumption matrix, P SC (t) represents the server static power consumption matrix, u VM (t) represents the virtual machine CPU resource requirement matrix, MM VM→PM (A(t),F(t)) represents the mapping matrix between the virtual machine and the server, A(t) represents the virtual machine real-time deployment matrix, F(t) represents the virtual machine migration matrix, u VM (t)*MM VM→PM (A(t),F(t)) represents the CPU usage rate of the server, f h represents the conversion coefficient matrix of the CPU usage rate and the dynamic power consumption, D(T PM (t) represents the server leakage power consumption matrix.

4. The collaborative control optimization method of claim 2, wherein, In step 10, according to the air conditioning system data, an air conditioning system power consumption model shown in equation (2) is established: P FC (t) = P CP (T co (t), T ev (t), PLR(t)) + ∑P FX (FP(t)) Equation (2) where P FC (t) represents the total power consumption of the air conditioning system, P CP (t) represents the total power consumption of the air conditioning system, P co (t) represents the total power consumption of the air conditioning system, P ev (t) represents the total power consumption of the air conditioning system, P co (t) represents the total power consumption of the air conditioning system, P ev (t) represents the total power consumption of the air conditioning system, P Load rate, FP(t) denotes the fan frequency matrix of each end module, P FX (FP(t)) denotes the power consumption matrix of each end module.

5. The collaborative control optimization method of claim 2, wherein, In the step 10, the server temperature rapid prediction model is established according to the IT system scheduling data and the air conditioning system data; Specifically comprising: Step 131, screening the mapping relationship between virtual machines and servers, virtual machine CPU resource demand and chip temperature of each server from the IT system scheduling data, and screening fan frequency of each terminal module and evaporating temperature of a coil from the air conditioning system data to construct a modeling data set; Step 132, based on the modeling data set, taking the fan frequency of each terminal module, the evaporating temperature of a coil and the CPU usage rate of each server as input variables of a BP neural network model, taking the chip temperature of each server as an output variable of the BP neural network model, training the BP neural network model to obtain the server temperature rapid prediction model.

6. The collaborative control optimization method of claim 1, wherein, In the step 20, the real-time joint and timing decoupling combined optimization method is used to cooperatively control the virtual machine deployment and the air conditioning system, comprising: Step 21, when the cloud platform receives a new virtual machine deployment request, the deployment position of the new virtual machine and the fan frequency of each terminal module are jointly optimized to realize the lowest overall power consumption of the server and each terminal module under single virtual machine deployment; Step 22, at each given time, firstly, the virtual machines running on each server are migrated in whole to minimize the total power consumption of the servers; then the evaporating temperature of the coil of the air conditioning system and the fan frequency of each end module are optimized to minimize the total power consumption of the air conditioning system.

7. The collaborative control optimization method of claim 6, wherein, The step 21 specifically comprises: Step 211, after the cloud platform receives the deployment request of the new virtual machine, the remaining CPU resources of all active servers are traversed according to the CPU resource demand of the new virtual machine to find a deployable server; Step 212, if there is a deployable server, the location of the deployable server is taken as an optimization variable of the deployment of the new virtual machine, and step 213 is executed; if the remaining CPU resources of all active servers cannot deploy the new virtual machine, a hibernating server is randomly switched to active state, the new virtual machine is deployed, and step 214 is executed; Step 213, taking the sum of the real-time total power consumption of the servers and the total power consumption of the end modules after deployment as the optimization objective, taking the sum of the usage resources of all virtual machines running on each server not exceeding the total rated resources of the server and each server chip temperature not exceeding the preset temperature threshold as the optimization constraint, the genetic algorithm is used to jointly optimize the deployment location of the new virtual machine and the fan frequency of each end module to obtain the optimal real-time deployment location of the virtual machine and the optimal fan frequency of each end module; the cloud platform implements the deployment of the new virtual machine according to the optimal real-time deployment location of the virtual machine, and the air conditioning system controls the fan frequency of each end module according to the optimal fan frequency of each end module; Step 214, taking the sum of the power consumption of each end module after the deployment of the new virtual machine as the optimization objective, taking the constraint that each server chip temperature cannot exceed the preset temperature threshold, the genetic algorithm is used to optimize the fan frequency of each end module to obtain the optimal fan frequency of each end module; the air conditioning system controls the fan frequency of each end module according to the optimal fan frequency of each end module.

8. The collaborative control optimization method of claim 6, wherein, The step 22 specifically comprises: Step 221, at a given time, the minimum number of active servers required is calculated according to the number of all virtual machines running on the server and the sum of their CPU resource demands; Step 222, according to the minimum number of active servers required, servers are selected as active servers in the next period in the order of rack number and server number on each rack until the number of selected servers reaches the minimum number of active servers; Step 223, according to the load balancing strategy, the existing virtual machines are evenly migrated to the selected active servers; the idle servers are hibernated; Step 224, taking the minimum total power consumption of the air conditioning system as the optimization objective, taking the constraint that each server chip temperature cannot exceed the preset temperature threshold, the genetic algorithm is used to optimize the evaporating temperature of the coil and the fan frequency of each end module to obtain the optimal evaporating temperature of the coil and the optimal fan frequency of each end module; Step 225, the air conditioning system controls the evaporating temperature of the coil according to the optimal evaporating temperature of the coil, and controls the fan frequency of each end module according to the optimal fan frequency of each end module.

Images