Cooperative control optimization method suitable for near-field fan wall micromodule

Patent metadata

Field	Value
Publication number	US-12449150-B1
Application number	US-202519227423-A
Country	US
Kind code	B1
Filing date	Jun 3, 2025
Priority date	Jul 5, 2024
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

A cooperative control optimization method suitable for a near-field fan wall micromodule, comprising the following steps: step 10 : obtaining IT system scheduling data and air-conditioning system data of the near-field fan wall micromodule within a preset period of time, and separately establishing a micromodule server real-time power model, an air-conditioning system power model, and a rapid server temperature prediction model; and step 20 : performing cooperative control on deployment of virtual machines and an air-conditioning system by using an optimization method of combining real-time joint optimization and timed decoupling optimization based on the micromodule server real-time power model, the air-conditioning system power model, and the rapid server temperature prediction model. The present invention provides a cooperative control optimization method suitable for a near-field fan wall micromodule, to resolve the technical problem that local hot spots and overcooling are prone to occur during energy-saving scheduling of a near-field fan wall micromodule.

First claim

What is claimed is:

1. A cooperative control optimization method suitable for a near-field fan wall micromodule, comprising the following steps:
step 10 : obtaining information technology (IT) system scheduling data and air-conditioning system data of the near-field fan wall micromodule within a preset period of time, and separately establishing a micromodule server real-time power model, an air-conditioning system power model, and a rapid server temperature prediction model; and
step 20 : performing cooperative control on deployment of virtual machines and an air-conditioning system by using an optimization method of combining real-time joint optimization and timed decoupling optimization based on the micromodule server real-time power model, the air-conditioning system power model, and the rapid server temperature prediction model,
wherein in the step 20 , the performing cooperative control on deployment of the virtual machines and the air-conditioning system by using the optimization method of combining the real-time joint optimization and the timed decoupling optimization comprises:
step 21 : when a cloud platform receives a deployment request for a new virtual machine, performing joint optimization on a deployment location of the new virtual machine and a fan frequency of each of cooling terminals, to achieve a lowest overall power of servers and each of the cooling terminals under deployment of an individual virtual machine; and
step 22 : at each of given moments, first performing whole migration on the virtual machines running on each of servers, so that a total power of the servers is the lowest;
then, performing optimization on a coil evaporation temperature and the fan frequency of each of the cooling terminals of the air-conditioning system, to achieve a lowest overall power of the air-conditioning system.

2. The cooperative control optimization method according to claim 1 , wherein an IT system scheduling data comprises central processing unit (CPU) resource requirements of the virtual machines, mapping relationships between the virtual machines and servers, a CPU utilization rate of each of the servers, a real-time power of each of the servers, and a chip temperature of each of the servers; and the air-conditioning system data comprises the fan frequency of each of cooling terminals, a coil evaporation temperature, an outdoor unit condensing temperature, a load ratio, and a total power of the air-conditioning system.

3. The cooperative control optimization method according to claim 2 , wherein in the step 10 , the micromodule server real-time power model is established according to the IT system scheduling data; and
the step 10 comprises:
step 101 : selecting power data of each of the servers at different chip temperatures and CPU utilization rates from the IT system scheduling data;
step 102 : establishing a server static power matrix according to power data of the servers in a state in which the CPU utilization rate is 0;
step 103 : fitting a conversion coefficient between the CPU utilization rate and a dynamic power of each of the servers according to data of variation of the real-time power of each of the servers with the CPU utilization rate by using a multiple linear regression method, to form a conversion coefficient matrix of the CPU utilization rates and dynamic powers;
step 104 : fitting an expression of a leakage power of each of the servers by using the multiple linear regression method according to different temperature data of a chip of each of the servers at a same CPU utilization rate, to form a server leakage power matrix; and
step 105 : combining the server static power matrix, the conversion coefficient matrix of the CPU utilization rates and the dynamic powers, and the server leakage power matrix, to obtain the micromodule server real-time power model expressed as 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),

wherein P PM (t) represents a real-time power matrix for the servers, P SC (t) represents the server static power matrix, u vm (t) represents a CPU resource requirement matrix of the virtual machines, MM VM→PM (A(t),F(t)) represents a mapping matrix between the virtual machines and the servers, A(t) represents a real-time deployment matrix of the virtual machines, F(t) represents a virtual machine migration matrix, u vm (t)*MM VM→PM (A(t),F(t)) represents the CPU utilization rates of all servers, f h represents the conversion coefficient matrix of the CPU utilization rates and the dynamic powers, and D(T PM (t)) represents the server leakage power matrix.

4. The cooperative control optimization method according to claim 2 , wherein in the step 10 , the air-conditioning system power model expressed as formula (2) is established according to the air-conditioning system data:
P FC ( t )= P CP ( T co ( t ), T ev ( t ), PLR ( t ))+Σ P FX ( FP ( t ))  formula (2),

wherein P FC (t) represents the total power of the air-conditioning system, P CP (T co (t),T ev (t),PLR(t)) represents an outdoor unit power, T co (t) represents the coil evaporation temperature, T ev (t) represents the outdoor unit condensing temperature, PLR(t) represents a load ratio, FP (t) represents a fan frequency matrix of each of the cooling terminals, and P FX (FP(t)) represents a power matrix for each of the cooling terminals.

5. The cooperative control optimization method according to claim 2 , wherein in the step 10 , the rapid server temperature prediction model is established according to the IT system scheduling data and the air-conditioning system data; and
the step 10 comprises:
step 131 : selecting a mapping relationship between the virtual machines and the servers, CPU resource requirements of the virtual machines, and the chip temperature of each of the servers from the IT system scheduling data, and selecting the fan frequency and the coil evaporation temperature of each of the cooling terminals from the air-conditioning system data, to form a modeling data set; and
step 132 : training a BP neural network model based on the modeling data set by using the fan frequency of each of the cooling terminals, the coil evaporation temperature, and the CPU utilization rate of each of the servers as input variables of the BP neural network model, and using the chip temperature of each of the servers as an output variable of the BP neural network model, to obtain the rapid server temperature prediction model.

6. The cooperative control optimization method according to claim 1 , wherein the step 21 comprises:
step 211 : after the cloud platform receives the deployment request for the new virtual machine, traversing remaining CPU resources of all active servers according to CPU resource requirements of the new virtual machine, to search for a server available for deployment;
step 212 : if there is a server available for deployment, using a location of the server available for deployment as an optimization variable for the deployment of the new virtual machine, and performing step 213 ; if remaining CPU resources of all of the active servers are not suitable for the deployment of the new virtual machine, randomly switching a dormant server to an active state, deploying the new virtual machine, and performing step 214 ;
step 213 : performing joint optimization on the deployment location of the new virtual machine and the fan frequency of each of the cooling terminals by using a genetic algorithm by using, as an optimization objective, a sum of the total power of servers and the total power of cooling terminals being minimum after the deployment, and by using, as optimization constraints, a sum of resources used by all of the virtual machines running on each of

Assignees

• China Constr Ind & Energy Eng Group Co Ltd
• Nanjing University Of Technology

Inventors

• Wang Ligang
• Li Depeng
• Deng Tao
• Liu Wei
• Lin Yumao
• Li Meng
• Huang Qing
• Xu Yizhe
• Yan Chengchu

Classifications

• F24F11/46 Primary

  Improving electric energy efficiency or saving · CPC title

• F24F11/63 Primary

  Electronic processing · CPC title

• H05K7/20745

  within rooms for removing heat from cabinets, e.g. by air conditioning device · CPC title

• G05D23/1917

  using digital means · CPC title