CN111144543A - Data center air conditioner tail end temperature control method, device and medium - Google Patents

Data center air conditioner tail end temperature control method, device and medium Download PDF

Info

Publication number
CN111144543A
CN111144543A CN201911398281.XA CN201911398281A CN111144543A CN 111144543 A CN111144543 A CN 111144543A CN 201911398281 A CN201911398281 A CN 201911398281A CN 111144543 A CN111144543 A CN 111144543A
Authority
CN
China
Prior art keywords
temperature
weight
candidate
prediction model
return air
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201911398281.XA
Other languages
Chinese (zh)
Inventor
侯晓雯
李程贵
王旭光
吕楠
刘天伟
刘志强
杨培艳
侯瑞强
马宇晴
张建雪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Mobile Communications Group Co Ltd
China Mobile Group Inner Mongolia Co Ltd
Original Assignee
China Mobile Communications Group Co Ltd
China Mobile Group Inner Mongolia Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Mobile Communications Group Co Ltd, China Mobile Group Inner Mongolia Co Ltd filed Critical China Mobile Communications Group Co Ltd
Priority to CN201911398281.XA priority Critical patent/CN111144543A/en
Publication of CN111144543A publication Critical patent/CN111144543A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
    • F24F11/63—Electronic processing
    • F24F11/64—Electronic processing using pre-stored data
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/70—Control systems characterised by their outputs; Constructional details thereof
    • F24F11/80—Control systems characterised by their outputs; Constructional details thereof for controlling the temperature of the supplied air
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Mechanical Engineering (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Fuzzy Systems (AREA)
  • Air Conditioning Control Device (AREA)

Abstract

The embodiment of the invention provides a method, a device and a medium for controlling the temperature of a tail end of a data center air conditioner, which are used for realizing intelligent and accurate control of the temperature of a data center air conditioner system, enabling all air conditioning equipment to work cooperatively and consistently, improving the running reliability of the tail end air conditioner and reducing energy consumption. The method for controlling the temperature of the tail end of the air conditioner of the data center comprises the following steps: acquiring air supply temperature and IT equipment power consumption of each air conditioning equipment of a data center; predicting the acquired air supply temperature of each air-conditioning device and the power consumption of the IT device by using a temperature prediction model to obtain a return air temperature predicted value of each air-conditioning device, wherein weight parameters of the temperature prediction model are trained by using a neural network according to power consumption temperature sample data and are selected from N groups of weight combinations, and the N groups of weight combinations are determined by using an ant colony optimization algorithm according to weight definition domains; and adjusting and respectively adjusting the return air temperature of each air conditioning equipment according to the obtained return air temperature predicted value of each air conditioning equipment.

Description

Data center air conditioner tail end temperature control method, device and medium
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a method, a device and a medium for controlling the temperature of the tail end of an air conditioner
Background
The service objects faced by the air conditioning system equipped in the large data center are mainly IT equipment such as various servers and storages, the CPU (central processing unit) and the GPU (graphic processing unit) of the air conditioning system have large heat productivity during working and higher temperature and humidity requirements, machine rooms in international standard Tier III and above often adopt a novel air conditioner terminal to provide rack-level and inter-row-level nearby refrigeration, the air conditioners are distributed on the front door or the back plate of a machine cabinet, each machine cabinet is used as an independent refrigeration unit, and the refrigeration efficiency is higher for rack air supply, as shown in figure 1. As the novel air-conditioning tail end of the Data center is randomly configured, 100 and 200 air conditioners are distributed in each IDC (Internet Data center) machine room and are independently controlled.
The novel air conditioner control method is mainly a supply/return air temperature control method, and controls the supply/return air temperature value by using the temperature value acquired by a temperature sensor at the supply/return air side of the unit, compares the supply/return air temperature value with a target temperature value set by the unit, and controls the capacity output of the unit and the action of other components as required by the calculated cold quantity requirement. A schematic of the supply/return air temperature control logic is shown in fig. 2.
The cold requirement is related to the supply/return air temperature, the temperature set point, the temperature dead zone and the temperature proportional band, namely: the cold quantity requirement is f (air supply temperature, temperature set point, temperature dead zone, temperature proportional band), and the specific calculation formula is as follows:
Figure BDA0002346877640000011
wherein:
the proportion belt: and the temperature interval accords with the use condition of each device in the machine room and can be controlled.
Temperature dead zone: near the temperature set point, the temperature in the machine room can be approximately considered to reach the temperature interval required by the setting, the temperature interval is divided into a positive dead zone and a negative dead zone, the size of the dead zone can be set according to the control precision of the actual scene temperature, and the maximum value of the dead zone is +/-3 ℃. Fig. 3 is a schematic view of the temperature dead zone.
Although the existing data center machine room is equipped with a novel air conditioner terminal device and has a reasonable airflow organization, the high-density and low-density mixed deployment of each large data center server occurs later, and hundreds of machine room air conditioners in the same machine room run independently and are not connected with each other, in this case, the following problems may occur: the air conditioning units may run competitively, some air conditioning units are refrigerating and some air conditioning units are heating, and when the cold redundancy of the machine room is large, the fans of the redundant units still work for 24 hours continuously, so that energy consumption is wasted; when one air conditioner in the machine room fails or a local area of the machine room is overheated, the air conditioners in other areas cannot respond correspondingly in time.
Disclosure of Invention
The embodiment of the invention provides a method, a device and a medium for controlling the temperature of a tail end of an air conditioner, which are used for realizing intelligent and precise control of the temperature of a data center air conditioning system, enabling all air conditioning equipment to work cooperatively and consistently, improving the running reliability of the tail end air conditioner and reducing energy consumption.
In a first aspect, a method for controlling the temperature of the tail end of an air conditioner in a data center is provided, which comprises the following steps:
acquiring air supply temperature and IT equipment power consumption of each air conditioning equipment of a data center;
predicting the acquired air supply temperature of each air-conditioning device and the power consumption of the IT device by using a temperature prediction model to obtain a return air temperature predicted value of each air-conditioning device, wherein weight parameters of the temperature prediction model are trained by using a neural network according to power consumption temperature sample data and are selected from N groups of weight combinations, the N groups of weight combinations are determined by using an ant colony optimization algorithm according to weight definition domains, and N is an integer greater than 1;
and adjusting and respectively adjusting the return air temperature of each air conditioning equipment according to the obtained return air temperature predicted value of each air conditioning equipment.
In one embodiment, the ant colony optimization algorithm is used to determine the N sets of weight combinations according to the following procedures according to the weight definition domain:
uniformly dividing the weight definition domain into a plurality of sub-regions, wherein the boundary point of each sub-region is a candidate weight;
releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once, and recording corresponding subregion identification;
determining M sub-regions passed by each ant to form M groups of reference weights for training the temperature prediction model according to the recorded sub-region identifications, wherein M is an integer greater than 1;
determining the power consumption temperature sample data, training by using a plurality of groups of reference weights determined by the ants as candidate weight parameters of the temperature prediction model respectively, and outputting a first return air temperature candidate predicted value of the air conditioning equipment;
and selecting N groups of reference weights according to a first error between the first return air temperature candidate predicted value and a first return air temperature expected value and the sequence of the corresponding first errors from small to large to obtain N groups of weight combinations.
In one embodiment, after the traversal of each ant is finished, the method further comprises:
and updating the pheromone corresponding to each candidate weight.
In one embodiment, the pheromone corresponding to each candidate weight is updated according to the following formula:
τ(t+1)＝ρ(t)+Δτ(t)
wherein:
τ (t +1) represents pheromone corresponding to the candidate weight at the time of t + 1;
ρ represents a preset pheromone residual coefficient;
Δ τ (t) represents the total amount of pheromone increment on the candidate weight by all ants during traversal at time t.
In one embodiment, according to the power consumption temperature sample data, training by using a neural network according to the following procedures to obtain the temperature prediction model:
aiming at each group of weight values contained in the N groups of weight value combinations, training the temperature prediction model by using the group of weight values as initial values of weight value parameters;
respectively inputting the power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by a temperature prediction model corresponding to each group of weights;
and determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to a second error between the second return air temperature prediction candidate value and a second return air temperature expected value.
In one embodiment, the neural network is constructed according to the following formula:
Figure BDA0002346877640000031
wherein n is1The number of neurons in the hidden layer;
n is the number of neurons in the input layer;
m is the number of neurons in the output layer;
a is a preset constant.
In a second aspect, a data center air conditioner terminal temperature control device is provided, comprising:
the system comprises an acquisition unit, a processing unit and a control unit, wherein the acquisition unit is used for acquiring the air supply temperature and the IT equipment power consumption of each air conditioning equipment of a data center;
the prediction unit is used for predicting the acquired air supply temperature of each air-conditioning device and the power consumption of the IT device by using a temperature prediction model to obtain a return air temperature prediction value of each air-conditioning device, wherein weight parameters of the temperature prediction model are trained by using a neural network according to power consumption temperature sample data and are selected from N groups of weight combinations, the N groups of weight combinations are determined by using an ant colony optimization algorithm according to weight definition domains, and N is an integer greater than 1;
and the adjusting unit is used for adjusting and respectively adjusting the return air temperature of each air conditioning equipment according to the obtained return air temperature predicted value of each air conditioning equipment.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
the dividing unit is used for uniformly dividing the weight definition domain into a plurality of sub-domains, and the boundary point of each sub-domain is a candidate weight;
the control unit is used for releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once and recording corresponding subregion identification;
the first determining unit is used for determining M sub-regions which are penetrated by each ant to form M groups of reference weights for training the temperature prediction model according to the recorded sub-region identifications, wherein M is an integer greater than 1;
the first training unit is used for training a plurality of groups of reference weights determined by the ants according to the power consumption temperature sample data and respectively serving as candidate weight parameters of the temperature prediction model, and outputting a first return air temperature candidate prediction value of the air conditioning equipment;
and the selection unit is used for selecting N groups of reference weights according to a first error between the first return air temperature candidate predicted value and a first return air temperature expected value and according to the sequence of the corresponding first errors from small to large to obtain the N groups of weight combinations.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
and the updating unit is used for updating the pheromone corresponding to each candidate weight after each ant finishes traversing.
In one embodiment, the updating unit may update the pheromone corresponding to each candidate weight value according to the following formula:
τ(t+1)＝ρ(t)+Δτ(t)
wherein:
τ (t +1) represents pheromone corresponding to the candidate weight at the time of t + 1;
ρ represents a preset pheromone residual coefficient;
Δ τ (t) represents the total amount of pheromone increment on the candidate weight by all ants during traversal at time t.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
a second determining unit, configured to train the temperature prediction model by using each set of weight values included in the N sets of weight value combinations as an initial value of a weight value parameter;
the second training unit is used for respectively inputting the power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by the temperature prediction model corresponding to each group of weights;
and the third determining unit is used for determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to the second error between the second return air temperature prediction candidate value and the second return air temperature expected value.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
a construction unit configured to construct the neural network according to the following formula:
Figure BDA0002346877640000051
wherein n is1The number of neurons in the hidden layer;
n is the number of neurons in the input layer;
m is the number of neurons in the output layer;
a is a preset constant.
In a third aspect, an embodiment of the present invention provides a computing apparatus, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which when executed by the processor, implement the method of the first aspect of the embodiments described above.
In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which computer program instructions are stored, which, when executed by a processor, implement the method of the first aspect in the foregoing embodiments.
In the method, the device and the medium for controlling the temperature of the tail end of the air conditioner of the data center, provided by the embodiment of the invention, the ant colony optimization algorithm is utilized to optimize the initial weight combination of the temperature prediction model to obtain N sets of weight combinations, the temperature prediction model is trained based on the N sets of weight combinations, and a set of weight with the smallest error between the corresponding predicted value and the expected value is selected as the weight parameter of the temperature prediction model.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments of the present invention will be briefly described below, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 shows a schematic diagram of a network architecture in the prior art in which an OLT schedules on a BRAS through an OTN device;
FIG. 2 shows a schematic of the prior art supply/return air temperature control logic;
FIG. 3 illustrates a prior art temperature dead band diagram;
FIG. 4 is a schematic flow chart illustrating the determination of N sets of weight combinations of a BP neural network by using an ant colony algorithm according to an embodiment of the present invention;
FIG. 5 shows a schematic diagram of a neural network architecture according to an embodiment of the present invention;
FIG. 6 shows a schematic flow chart of training a temperature prediction model using a BP neural network algorithm according to an embodiment of the present invention;
FIG. 7 illustrates a schematic flow chart of training a temperature prediction model using an ACO-BP neural network according to an embodiment of the present invention;
FIG. 8 shows a schematic structural diagram of a controller according to an embodiment of the present invention;
FIG. 9 is a schematic flow chart illustrating an implementation of a data center air conditioner terminal temperature control method according to an embodiment of the invention;
FIG. 10 shows a schematic diagram of a data center air conditioning terminal temperature control apparatus according to an embodiment of the present invention;
FIG. 11 illustrates a schematic structural diagram of a computing device according to an embodiment of the present invention.
Detailed Description
Features and exemplary embodiments of various aspects of the present invention will be described in detail below, and in order to make objects, technical solutions and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not to be construed as limiting the invention. It will be apparent to one skilled in the art that the present invention may be practiced without some of these specific details. The following description of the embodiments is merely intended to provide a better understanding of the present invention by illustrating examples of the present invention.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
In order to improve the heat dissipation efficiency of a data center, reduce energy consumption and effectively reduce the electric energy consumption of an IDC machine room air conditioning system under the condition of determining the normal operation of IT equipment, the embodiment of the invention utilizes a BP (Back propagation) neural network to realize the intelligent and precise control of a data center air conditioning tail end system, so that air conditioners can orderly, consistently and efficiently work in a coordinated manner, and the purposes of improving the reliability of the tail end air conditioners and reducing the energy consumption are achieved.
However, since the BP neural network is prone to be involved in local infinitesimal complement, in view of this, an embodiment of the present invention provides an ACO (ant colony optimization algorithm) -BP neural network training method. The neural network training process can be regarded as an optimization problem, namely a group of optimal weight parameters is found, so that the error between an output result and an expected result under the weight is minimum, and the ant colony optimization algorithm becomes a better choice for searching the M (M is an integer greater than 1) groups of optimal weight parameters. The global optimization capability of the ant colony optimization algorithm can be utilized to provide N (N is an integer greater than 1) groups of optimal initial weight combinations for the neural network, so that the defects that the BP algorithm is easy to fall into local optimization and is sensitive to initial value setting are overcome; furthermore, the weight is further finely adjusted by utilizing the BP algorithm gradient downhole principle, and a real global optimum point is searched, so that the quantization error caused by the division of the definition domain and the complement of the single ant colony algorithm training network which consumes too long time are overcome.
Based on the above, in the embodiment of the invention, the basic principle of BP neural network model control is optimized by using an ant colony optimization algorithm, a temperature prediction model of an IDC machine room air conditioning system is established, and neural network prediction control is performed on the system, specifically, the IT equipment power consumption and the air conditioner air supply temperature value are set as input parameters of the temperature prediction model, the return air temperature value is set as an output parameter of the temperature prediction model, and the neural network model with a better mapping corresponding relation can be obtained finally through continuous training and connection of power consumption temperature sample data on the temperature prediction model. During specific implementation, the trained temperature prediction model can be further iterated to be a controller of a neural network, so that the actual return air temperature output by the air conditioner tail end system is infinitely close to the expected value of the system, and ideal group control of the temperature of the data center air conditioner system is more effectively realized.
As shown in fig. 4, it is a schematic flow chart of determining N sets of weight combinations of the BP neural network by using the ant colony algorithm according to the weight definition domain, and includes the following steps:
and S41, uniformly dividing the weight definition domain into a plurality of sub-regions, wherein the boundary point of each sub-region is a candidate weight.
In specific implementation, all weights are uniformly divided into weight intervals [ W ]min,Wmax]For spn equal parts, the point of each sub-region boundary is a candidate weight.
And S42, releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once, and recording the corresponding subregion identification.
In this step, m ants are randomly placed at n places, and each point has the same number of pheromones at the initial moment. Each weight has a corresponding pheromone table, as shown in table 1: the weight value needing to be optimized for the ith time is Wi(ii) a Scale value of division aiConsidered as a point; τ (i) is aiThe corresponding pheromone; the number of spn candidates is divided into (spn-1) parts. And (3) the ants pass through each candidate weight of each sub-region only once, the corresponding sub-region identification is recorded, and the boundary point of the sub-region through which each ant passes is the selected numerical value.
TABLE 1
Reference numerals 1 2 …… spn+1
Dividing scale a1 a2 …… a(spn+1)
Pheromone τ(1) τ(2) …… τ(spn+1)
In specific implementation, after releasing m ants, wherein the ant k is from one sub-region to the next sub-region based on the following probability formula:
Figure BDA0002346877640000091
the ant records the label of the sub-region passed by, i.e. selects a numerical value for the weight, and records in the function tabuk. After the ants select values for the ownership value parameters, the ants complete one traversal, and all the recorded values form the ownership value parameters of the temperature prediction model.
S43, determining M sub-regions through which each ant passes to form M groups of reference weights for training the temperature prediction model.
I.e. a set of weights for the BP neural network is made up of a combination of these sub-regions.
And S44, training by using M groups of reference weights determined by a plurality of ants as candidate weight parameters of the temperature prediction model respectively according to the power consumption temperature sample data, and outputting a first return air temperature candidate prediction value of the air conditioning equipment.
S45, determining a first error between the first return air temperature candidate predicted value and the first return air temperature expected value, and selecting N groups of reference weights according to the sequence of the corresponding first errors from small to large to obtain the N groups of weight combinations.
In specific implementation, if the pheromone on the path is increased continuously and the residual information is excessive, the residual information submerges the heuristic information, so after each ant finishes traversing, the residual information on the path can be updated, that is, the pheromone corresponding to each candidate weight is updated, specifically, at the time t +1, the information amount on the path (i, j), that is, the corresponding candidate weight, can be updated and adjusted according to the following formula, wherein Δ τ is calculated according to the formula.
τ(t+1)＝ρ(t)+Δτ(t)
Wherein τ (t +1) represents pheromone corresponding to the candidate weight at the time of t + 1; rho represents pheromone residual coefficient, the degree of individual interaction among ants is reflected, the value of rho is [0,1] to prevent excessive accumulation of pheromones, and delta tau (t) represents the increment total amount of pheromones on the paths (i, j) of all ants in the circulation.
Figure BDA0002346877640000101
Wherein, Q is a constant, which represents the total pheromone amount and has certain influence on the convergence rate of the ant colony algorithm, the smaller the Q value becomes, the slower the convergence rate of the ant colony algorithm is, otherwise, the faster the convergence rate of the ant colony algorithm is; eAThe total length of the path of the ant k in the path finding is recorded, the convergence rate of the ant colony algorithm is influenced to a certain extent, and EAThe smaller the value becomes, the better the convergence speed of the algorithm is, and the bestsolution represents the optimal solution determined after the traversal of ants is finished.
Through the process, N groups of better weight combinations can be determined by using the ant colony optimization algorithm, and based on the determined N groups of weight combinations, the temperature prediction model is trained by using the BP neural network algorithm so as to determine the weight parameters of the temperature prediction model.
The BP neural network mainly uses a gradient steepest descent method, and continuously adjusts the weight and the threshold value of the network by using the principle of back propagation so as to ensure that the sum of squares of errors in the neural network reaches the minimum, and the weight of the network is corrected by the back propagation of the errors. The topology structure of the BP neural network model mainly includes an input layer (input layer), a hidden layer (hidden layer), and an output layer (output layer) of the network, and the specific neural network structure is shown in fig. 5.
As can be seen from fig. 5, the input layer has i neurons, and each neuron corresponds to one input; the hidden layer has j neurons; the input layer is provided with k neurons, and each neuron corresponds to one output; w is aijAnd wjkNetwork weights from the input layer to the hidden layer and from the hidden layer to the output layer are represented, respectively.
The input and output of the input layer are respectively:
Figure BDA0002346877640000111
wherein I represents the input of the neuron and O represents the output of the neuron; superscript (1) represents the first layer of neurons, and so on; m represents the number of input variables.
The input and output of the hidden layer are respectively:
Figure BDA0002346877640000112
wherein f (-) is the hidden layer excitation function, and N is the number of hidden layer neurons.
The input and output of the output layer are respectively:
Figure BDA0002346877640000113
wherein g (-) is the output layer excitation function, and Q is the number of output layer neurons.
The error performance indicator function of the network is defined as:
Figure BDA0002346877640000114
in the formula (I), the compound is shown in the specification,
Figure BDA0002346877640000115
and
Figure BDA0002346877640000116
expected and actual values, respectively, for output layer neuron k, then ekThe error for output layer neuron k. If the calculated outputs (actual outputs) of the k neurons of the output layer all match their corresponding k expected outputs (i.e., e)kLess than or equal to the allowable error epsilon), the learning process of the neural network is ended; otherwise, the network adjusts the weight w of the network by the error back propagation processjkAnd wij。
In the training process, the error back propagation adopts a gradient descent method to adjust the weight, namely
wjk(k+1)＝wjk(k)+Δwjk
Figure BDA0002346877640000121
Figure BDA0002346877640000122
wij(k+1)＝wij(k)+Δwij
Figure BDA0002346877640000123
Figure BDA0002346877640000124
The above formula is to adjust the weight wjk、wijThe formula (2) is calculated. And continuously adjusting delta w when the error is propagated reversely, thereby achieving the purpose of modifying the weight of each layer. The initial weight of the BP algorithm is a better N groups of weight combinations searched by the ant colony algorithm, and the difference between the output of the temperature prediction model and the expected output at the moment is calculated: and the error is reversely propagated from the output layer back to the front-end input layer, and the weight is adjusted again until the condition is met.
Based on this, in the embodiment of the present invention, the temperature prediction model may be trained according to the flow shown in fig. 5 by using a BP neural network algorithm:
s61, aiming at each group of weight values contained in the N groups of weight value combinations, the group of weight values are used as initial values of weight value parameters to train the temperature prediction model.
And S62, respectively inputting power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by the temperature prediction model corresponding to each group of weights.
And S63, determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to the second error between the second return air temperature prediction candidate value and the second return air temperature expected value.
In summary, in the embodiment of the present invention, the temperature prediction model may be trained according to the process shown in fig. 7:
and S71, initializing parameters.
In this step, the ant colony optimization algorithm is carried outThe parameters are initialized. Specifically, the weight value interval [ Wmin,Wmax]Uniformly dividing the weight into spn equal parts, establishing a pheromone table shown in table 1 for each candidate weight, and setting the initial value of the pheromone as tau0Pheromone volatility coefficient rho, pheromone increment intensity Q, and the maximum iteration number of the ant colony optimization algorithm can be set to be NACOThe learning rate of the BP neural network algorithm is set to η, and the maximum iteration number of the neural network algorithm is NBPThe training error exit condition is E0And the optimal solution reserves parameters such as the number N and the like.
S72, judging whether the ant colony optimization algorithm meets the end condition, if so, executing the step S76, otherwise, executing the step S73.
And S73, releasing m ants.
Specifically, each ant reaches from one point to another according to the probability formula described above.
And S74, searching the optimal weight combination.
S75, the pheromone is updated, and step S72 is executed.
And S76, outputting N sets of preferred weight combinations.
And S77, selecting the combination of the i groups of weights as an initial value to train the temperature prediction model.
And S78, submitting network input, and calculating the output of the hidden layer and the output layer.
And S79, calculating the difference between the expected output and the actual output.
S110, updating the value of i
In this step, the value of i is updated to i + 1.
And S711, judging whether i is larger than N, if so, executing step S712, and if not, executing step S713.
And S712, outputting a group of weight combinations with the best prediction results, and ending the process.
S713, turning to reverse invasive treatment, and calculating error signals of each layer.
And S714, adjusting the weights of the network output layer and the hidden layer, and executing the step S78.
Through the process shown in fig. 7, a trained temperature prediction model can be obtained.
In the embodiment of the present invention, the trained temperature prediction model is used as a controller of a data center air conditioner end group control system, as shown in fig. 8, which is a schematic structural diagram of the controller provided in the embodiment of the present invention. By utilizing the controller, the return air temperature actually output by the tail end of the air conditioner is controlled to be closer to an expected value, the ideal group control of the temperature of the machine room control system is more effectively realized, and various problems of complex nonlinearity and the like in the control process of the air conditioning system are better solved.
Specifically, the cabinet return air temperature T' is taken as an air conditioner temperature controller input command u, the control system output is y, and the nonlinear functional relationship y between the input and the output is g (u). The ultimate goal of the system control is to be able to determine the optimum value of the input control quantity u so that the actual output y of the system is approximately equal to the desired output y of the systemd. In this system, the function of the neural network, which is referred to as a function transformation and can be represented by a functional relationship u ═ f (y), can be regarded as an input/output mapd) Descriptive expression is performed. To satisfy that the system output y can be equal to the desired output ydCombining the two formulas to obtain y ═ g [ f (y)d)]. Obviously, when f (·) is g-1When (·), y ═ ydThe requirements of (1).
As shown in fig. 9, which is a schematic flow chart of an implementation of a method for controlling a temperature at an end of an air conditioner in a data center according to an embodiment of the present invention, the method includes the following steps:
and S91, acquiring the air supply temperature and the IT equipment power consumption of each air conditioning equipment in the data center.
And S92, predicting the acquired air supply temperature of each air conditioner and the power consumption of the IT equipment by using the temperature prediction model to obtain the return air temperature prediction value of each air conditioner.
And S93, adjusting the return air temperature of each air conditioner according to the obtained return air temperature predicted value of each air conditioner.
For a better understanding of the present invention, the following description is given in conjunction with specific examples.
The data center is provided with 100 cabinets in total, 100 air conditioner tail ends are configured for refrigeration at the same time, the cabinets are distributed in sequence according to 8 rows, meanwhile, the cabinets are arranged in the machine room in an alternate mode, namely, the cabinets are arranged face to form a cold/hot channel, therefore, the heat radiation condition of IT equipment is reflected by the temperature of the hot channel (the return air temperature of the cabinets) of the data center, and the air supply temperature of an air conditioning system is reflected by the temperature of the cold channel of the machine room. When the method is specifically implemented, the air supply temperature of the air conditioner is effectively controlled within the range of 18-28 ℃.
First, a weight interval [ W ] is uniformly dividedmin,Wmax]Setting an pheromone table for each parameter for the part with equal spn, setting tau (0) as an initial pheromone value, recommending rho to be 0.7-0.95, taking rho to be 0.8, setting Q as pheromone increment intensity, taking Q to be 1, and exiting the condition E due to training errors0The optimal solution reserves the number σ ports and the number of other parameters. And inputting power consumption temperature sample data to obtain a corresponding output result, and then calculating an error E. Recording the weight of the error sigma group, and comparing the initial error E0And minimum error EminUpdating the pheromone tau (t +1) ═ rho (t) + delta tau (t), iterating for a plurality of times until the requirement is met, and outputting N groups of weight combinations.
Based on N groups of weight combinations output by the ant colony optimization algorithm, a three-layer BP neural network is adopted, the number of neurons in an implicit layer is preliminarily determined to be 5 according to an empirical formula, then the learning research of the mapping relation between input and output is completed, and the BP neural network capable of predicting the return air temperature of the cabinet is established.
Figure BDA0002346877640000151
Wherein n is1In one embodiment, the value of n is a constant between 1 and 10, and a is a predetermined constant.
And training the temperature prediction model by utilizing the established BP neural network according to the N groups of weight combinations output by the ant colony optimization algorithm, and selecting a group of weight combinations with the minimum difference value between the output value and the expected value as weight parameters of the temperature prediction model.
Collecting IT load power consumption P corresponding to 100 cabinets in machine roomi(i ═ 1, 2.., 100.) air conditioner supply air temperature Ti(i ═ 1, 2., 100) is input into the trained temperature prediction model as an input variable of the temperature prediction model, and the corresponding cabinet return air temperature T 'is output'j(j ═ 1, 2.., 100). Therefore, the data center air conditioning system is combined with the power consumption of the IT equipment, the return air temperature of the air conditioner is adjusted according to the air supply temperature of the cabinet, and the indoor temperature is reasonably and effectively controlled.
Based on the same technical concept, an embodiment of the present invention further provides a data center air conditioner terminal temperature control device, as shown in fig. 10, including:
the system comprises an acquisition unit 101, a processing unit and a control unit, wherein the acquisition unit is used for acquiring the air supply temperature and the IT equipment power consumption of each air conditioning equipment of a data center;
the prediction unit 102 is configured to predict the acquired supply air temperature of each air conditioner and the power consumption of the IT equipment by using a temperature prediction model to obtain a return air temperature prediction value of each air conditioner, where a weight parameter of the temperature prediction model is selected from N sets of weight combinations, the N sets of weight combinations are determined by using an ant colony optimization algorithm according to a weight definition domain, and N is an integer greater than 1, and the weight parameter is trained by using a neural network according to power consumption temperature sample data;
and the adjusting unit 103 is configured to adjust the return air temperature of each air conditioner according to the obtained return air temperature predicted value of each air conditioner.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
the dividing unit is used for uniformly dividing the weight definition domain into a plurality of sub-domains, and the boundary point of each sub-domain is a candidate weight;
the control unit is used for releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once and recording corresponding subregion identification;
the first determining unit is used for determining M sub-regions which are penetrated by each ant to form M groups of reference weights for training the temperature prediction model according to the recorded sub-region identifications, wherein M is an integer greater than 1;
the first training unit is used for training by using M groups of reference weights determined by the ants as candidate weight parameters of the temperature prediction model according to the power consumption temperature sample data and outputting a first return air temperature candidate prediction value of the air conditioning equipment;
and the selection unit is used for determining a first error between the first return air temperature candidate predicted value and a first return air temperature expected value, and selecting N groups of reference weights according to the sequence of the corresponding first errors from small to large to obtain N groups of weight combinations.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
and the updating unit is used for updating the pheromone corresponding to each candidate weight after each ant finishes traversing.
In one embodiment, the updating unit may update the pheromone corresponding to each candidate weight value according to the following formula:
τ(t+1)＝ρ(t)+Δτ(t)
wherein:
τ (t +1) represents pheromone corresponding to the candidate weight at the time of t + 1;
ρ represents a preset pheromone residual coefficient;
Δ τ (t) represents the total amount of pheromone increment on the candidate weight by all ants during traversal at time t.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
a second determining unit, configured to train the temperature prediction model by using each set of weight values included in the N sets of weight value combinations as an initial value of a weight value parameter;
the second training unit is used for respectively inputting the power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by the temperature prediction model corresponding to each group of weights;
and the third determining unit is used for determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to the second error between the second return air temperature prediction candidate value and the second return air temperature expected value.
In an implementation manner, the data center air conditioner terminal temperature control apparatus provided in an embodiment of the present invention further includes:
a construction unit configured to construct the neural network according to the following formula:
Figure BDA0002346877640000171
wherein n is1The number of neurons in the hidden layer;
n is the number of neurons in the input layer;
m is the number of neurons in the output layer;
a is a preset constant.
In addition, the method for controlling the temperature at the end of the air conditioner in the data center according to the embodiment of the invention described in conjunction with fig. 9 may be implemented by a computing device. Fig. 11 is a schematic diagram illustrating a hardware structure of a computing apparatus according to an embodiment of the present invention.
The computing device provided by the embodiments of the present invention may include a processor 111 and a memory 112 storing computer program instructions.
In particular, the processor 111 may include a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or may be configured as one or more Integrated circuits implementing embodiments of the present invention.
Memory 112 may include mass storage for data or instructions. By way of example, and not limitation, memory 112 may include a Hard Disk Drive (HDD), a floppy Disk Drive, flash memory, an optical Disk, a magneto-optical Disk, tape, or a Universal Serial Bus (USB) Drive or a combination of two or more of these. Memory 112 may include removable or non-removable (or fixed) media, where appropriate. The memory 112 may be internal or external to the data processing apparatus, where appropriate. In a particular embodiment, the memory 112 is a non-volatile solid-state memory. In a particular embodiment, the memory 112 includes Read Only Memory (ROM). Where appropriate, the ROM may be mask-programmed ROM, Programmable ROM (PROM), Erasable PROM (EPROM), Electrically Erasable PROM (EEPROM), electrically rewritable ROM (EAROM), or flash memory or a combination of two or more of these.
The processor 111 implements any of the traffic protection methods in the above embodiments by reading and executing computer program instructions stored in the memory 112.
In one example, the OTN device may also include a communication interface 113 and a bus 110. As shown in fig. 7, the processor 111, the memory 112, and the communication interface 113 are connected via the bus 110 to complete communication therebetween.
The communication interface 113 is mainly used for implementing communication between modules, apparatuses, units and/or devices in the embodiments of the present invention.
The bus 110 includes hardware, software, or both to couple the components of the OTN device to one another. By way of example, and not limitation, a bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hypertransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an infiniband interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a video electronics standards association local (VLB) bus, or other suitable bus or a combination of two or more of these. Bus 110 may include one or more buses, where appropriate. Although specific buses have been described and shown in the embodiments of the invention, any suitable buses or interconnects are contemplated by the invention.
In addition, in combination with the method for controlling the temperature of the air conditioner terminal in the data center in the above embodiments, embodiments of the present invention may provide a computer-readable storage medium to implement. The computer readable storage medium having stored thereon computer program instructions; the computer program instructions, when executed by a processor, implement any of the data center air conditioner terminal temperature control methods in the above embodiments.
It is to be understood that the invention is not limited to the specific arrangements and instrumentality described above and shown in the drawings. A detailed description of known methods is omitted herein for the sake of brevity. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present invention are not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications and additions or change the order between the steps after comprehending the spirit of the present invention.
The functional blocks shown in the above-described structural block diagrams may be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it may be, for example, an electronic circuit, an Application Specific Integrated Circuit (ASIC), suitable firmware, plug-in, function card, or the like. When implemented in software, the elements of the invention are the programs or code segments used to perform the required tasks. The program or code segments may be stored in a machine-readable medium or transmitted by a data signal carried in a carrier wave over a transmission medium or a communication link. A "machine-readable medium" may include any medium that can store or transfer information. Examples of a machine-readable medium include electronic circuits, semiconductor memory devices, ROM, flash memory, Erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, Radio Frequency (RF) links, and so forth. The code segments may be downloaded via computer networks such as the internet, intranet, etc.
It should also be noted that the exemplary embodiments mentioned in this patent describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above-described steps, that is, the steps may be performed in the order mentioned in the embodiments, may be performed in an order different from the order in the embodiments, or may be performed simultaneously.
As described above, only the specific embodiments of the present invention are provided, and it can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system, the module and the unit described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. It should be understood that the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive various equivalent modifications or substitutions within the technical scope of the present invention, and these modifications or substitutions should be covered within the scope of the present invention.

Claims (13)

1. A method for controlling the temperature of the tail end of an air conditioner of a data center is characterized by comprising the following steps:
acquiring air supply temperature and IT equipment power consumption of each air conditioning equipment of a data center;
predicting the acquired air supply temperature of each air-conditioning device and the power consumption of the IT device by using a temperature prediction model to obtain a return air temperature predicted value of each air-conditioning device, wherein weight parameters of the temperature prediction model are trained by using a neural network according to power consumption temperature sample data and are selected from N groups of weight combinations, the N groups of weight combinations are determined by using an ant colony optimization algorithm according to weight definition domains, and N is an integer greater than 1;
and adjusting and respectively adjusting the return air temperature of each air conditioning equipment according to the obtained return air temperature predicted value of each air conditioning equipment.
2. The method according to claim 1, wherein the N sets of weight combinations are determined according to the following procedure by using an ant colony optimization algorithm according to a weight definition domain:
uniformly dividing the weight definition domain into a plurality of sub-regions, wherein the boundary point of each sub-region is a candidate weight;
releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once, and recording corresponding subregion identification;
determining M sub-regions passed by each ant to form M groups of reference weights for training the temperature prediction model according to the recorded sub-region identifications, wherein M is an integer greater than 1;
according to the power consumption temperature sample data, training by using M groups of reference weights determined by the ants as candidate weight parameters of the temperature prediction model respectively, and outputting a first return air temperature candidate predicted value of the air-conditioning equipment;
and determining a first error between the first return air temperature candidate predicted value and a first return air temperature expected value, and selecting N groups of reference weights according to the sequence of the corresponding first errors from small to large to obtain N groups of weight combinations.
3. The method of claim 2, further comprising, after each ant traversal is complete:
and updating the pheromone corresponding to each candidate weight.
4. The method of claim 3, wherein the pheromone corresponding to each candidate weight is updated according to the following formula:
τ(t+1)＝ρ(t)+Δτ(t)
wherein:
τ (t +1) represents pheromone corresponding to the candidate weight at the time of t + 1;
ρ represents a preset pheromone residual coefficient;
Δ τ (t) represents the total amount of pheromone increment on the candidate weight by all ants during traversal at time t.
5. The method of claim 2, 3 or 4, wherein the temperature prediction model is obtained by training with a neural network according to the following procedures according to power consumption temperature sample data:
aiming at each group of weight values contained in the N groups of weight value combinations, training the temperature prediction model by using the group of weight values as initial values of weight value parameters;
respectively inputting the power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by a temperature prediction model corresponding to each group of weights;
and determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to a second error between the second return air temperature prediction candidate value and a second return air temperature expected value.
6. The method of claim 5, wherein the neural network is constructed according to the following formula:
Figure FDA0002346877630000021
wherein n is1The number of neurons in the hidden layer;
n is the number of neurons in the input layer;
m is the number of neurons in the output layer;
a is a preset constant.
7. A data center air conditioner terminal temperature control device, characterized by, includes:
the system comprises an acquisition unit, a processing unit and a control unit, wherein the acquisition unit is used for acquiring the air supply temperature and the IT equipment power consumption of each air conditioning equipment of a data center;
the prediction unit is used for predicting the acquired air supply temperature of each air-conditioning device and the power consumption of the IT device by using a temperature prediction model to obtain a return air temperature prediction value of each air-conditioning device, wherein weight parameters of the temperature prediction model are trained by using a neural network according to power consumption temperature sample data and are selected from N groups of weight combinations, the N groups of weight combinations are determined by using an ant colony optimization algorithm according to weight definition domains, and N is an integer greater than 1;
and the adjusting unit is used for adjusting and respectively adjusting the return air temperature of each air conditioning equipment according to the obtained return air temperature predicted value of each air conditioning equipment.
8. The apparatus of claim 7, further comprising:
the dividing unit is used for uniformly dividing the weight definition domain into a plurality of sub-domains, and the boundary point of each sub-domain is a candidate weight;
the control unit is used for releasing a plurality of ants, controlling each ant to pass through each candidate weight subregion once and recording corresponding subregion identification;
the first determining unit is used for determining M sub-regions which are penetrated by each ant to form M groups of reference weights for training the temperature prediction model according to the recorded sub-region identifications, wherein M is an integer greater than 1;
the first training unit is used for training by using M groups of reference weights determined by the ants as candidate weight parameters of the temperature prediction model according to the power consumption temperature sample data and outputting a first return air temperature candidate prediction value of the air conditioning equipment;
and the selection unit is used for determining a first error between the first return air temperature candidate predicted value and a first return air temperature expected value, and selecting N groups of reference weights according to the sequence of the corresponding first errors from small to large to obtain N groups of weight combinations.
9. The apparatus of claim 8, further comprising:
and the updating unit is used for updating the pheromone corresponding to each candidate weight after each ant finishes traversing.
10. The apparatus of claim 7, 8 or 9, further comprising:
a second determining unit, configured to train the temperature prediction model by using each set of weight values included in the N sets of weight value combinations as an initial value of a weight value parameter;
the second training unit is used for respectively inputting the power consumption temperature sample data to obtain a second return air temperature prediction candidate value output by the temperature prediction model corresponding to each group of weights;
and the third determining unit is used for determining a group of weights with the minimum second error as weight parameters of the temperature prediction model according to the second error between the second return air temperature prediction candidate value and the second return air temperature expected value.
11. The apparatus of claim 10, further comprising:
a construction unit configured to construct the neural network according to the following formula:
Figure FDA0002346877630000041
wherein n is1The number of neurons in the hidden layer;
n is the number of neurons in the input layer;
m is the number of neurons in the output layer;
a is a preset constant.
12. A computing device, comprising: at least one processor, at least one memory, and computer program instructions stored in the memory that, when executed by the processor, implement the method of any of claims 1-6.
13. A computer-readable storage medium having computer program instructions stored thereon, which when executed by a processor implement the method of any one of claims 1-6.
CN201911398281.XA 2019-12-30 2019-12-30 Data center air conditioner tail end temperature control method, device and medium Pending CN111144543A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911398281.XA CN111144543A (en) 2019-12-30 2019-12-30 Data center air conditioner tail end temperature control method, device and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911398281.XA CN111144543A (en) 2019-12-30 2019-12-30 Data center air conditioner tail end temperature control method, device and medium

Publications (1)

Publication Number Publication Date
CN111144543A true CN111144543A (en) 2020-05-12

Family

ID=70521944

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911398281.XA Pending CN111144543A (en) 2019-12-30 2019-12-30 Data center air conditioner tail end temperature control method, device and medium

Country Status (1)

Country Link
CN (1) CN111144543A (en)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112329338A (en) * 2020-10-26 2021-02-05 ***通信集团内蒙古有限公司 Cold source system control method and device based on fish swarm optimization BP neural network
CN112616292A (en) * 2020-11-27 2021-04-06 湖南大学 Data center energy efficiency optimization control method based on neural network model
CN112612305A (en) * 2020-12-04 2021-04-06 格力电器（武汉）有限公司 Temperature adjusting method, device, equipment, storage medium and air conditioning system
CN112834079A (en) * 2020-12-25 2021-05-25 山东朗进科技股份有限公司 Method for judging parameter drift of temperature sensor of air conditioning unit of railway vehicle
CN113757922A (en) * 2020-06-03 2021-12-07 ***通信集团浙江有限公司 Deep learning-based air conditioning system energy-saving control method, device and equipment and computer medium
CN114322260A (en) * 2021-12-21 2022-04-12 上海美控智慧建筑有限公司 Air conditioner automatic driving and model training and predicting method, device and equipment
CN114325141A (en) * 2020-09-29 2022-04-12 ***通信集团设计院有限公司 Method and device for monitoring abnormal electric quantity of base station and readable storage medium
CN115344069A (en) * 2022-07-28 2022-11-15 上海蓝色帛缔智能工程有限公司 Liquid cooling control method and system for data center
CN117241561A (en) * 2023-10-13 2023-12-15 北京科技大学 Data center distributed precise air conditioning unit optimized operation parameter estimation method

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102213475A (en) * 2011-03-22 2011-10-12 曙光信息产业(北京)有限公司 Adaptive management method for power consumption of data center
CN102353119A (en) * 2011-08-09 2012-02-15 北京建筑工程学院 Control method of VAV (variable air volume) air-conditioning system
CN103955743A (en) * 2014-04-22 2014-07-30 江苏大学 Ultrahigh-pressure water jet road mark line cleaning effect forecasting method and device
CN106482248A (en) * 2015-08-25 2017-03-08 中南大学 Control system of machine room air conditioner based on multi-connected air conditioner and method
CN106885326A (en) * 2017-04-17 2017-06-23 南京佳力图机房环境技术股份有限公司 A kind of dynamic heat pipe backboard air-conditioning system and its control method
CN107655160A (en) * 2017-11-27 2018-02-02 中山路得斯空调有限公司 A kind of central air-conditioning regulating system based on neural network prediction
US20180348250A1 (en) * 2014-08-14 2018-12-06 University Of North Dakota Flight parameter prediction using neural networks
CN108981097A (en) * 2018-06-19 2018-12-11 广东美的暖通设备有限公司 Air-conditioning system and its method of adjustment and device can be needed
CN108990383A (en) * 2018-08-15 2018-12-11 北京建筑大学 A kind of data center's air-conditioning system forecast Control Algorithm
CN109118020A (en) * 2018-09-06 2019-01-01 南京工业职业技术学院 A kind of subway station energy consumption short term prediction method and its forecasting system
CN109324679A (en) * 2018-09-21 2019-02-12 郑州云海信息技术有限公司 A kind of server energy consumption control method and device
CN109695944A (en) * 2018-11-29 2019-04-30 中国汽车工业工程有限公司 A kind of control method of the coating fresh air conditioner based on multi-model deep learning
CN110188956A (en) * 2019-05-31 2019-08-30 广州水沐青华科技有限公司 Load forecasting method based on ant colony neural network and the system comprising this method, memory

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102213475A (en) * 2011-03-22 2011-10-12 曙光信息产业(北京)有限公司 Adaptive management method for power consumption of data center
CN102353119A (en) * 2011-08-09 2012-02-15 北京建筑工程学院 Control method of VAV (variable air volume) air-conditioning system
CN103955743A (en) * 2014-04-22 2014-07-30 江苏大学 Ultrahigh-pressure water jet road mark line cleaning effect forecasting method and device
US20180348250A1 (en) * 2014-08-14 2018-12-06 University Of North Dakota Flight parameter prediction using neural networks
CN106482248A (en) * 2015-08-25 2017-03-08 中南大学 Control system of machine room air conditioner based on multi-connected air conditioner and method
CN106885326A (en) * 2017-04-17 2017-06-23 南京佳力图机房环境技术股份有限公司 A kind of dynamic heat pipe backboard air-conditioning system and its control method
CN107655160A (en) * 2017-11-27 2018-02-02 中山路得斯空调有限公司 A kind of central air-conditioning regulating system based on neural network prediction
CN108981097A (en) * 2018-06-19 2018-12-11 广东美的暖通设备有限公司 Air-conditioning system and its method of adjustment and device can be needed
CN108990383A (en) * 2018-08-15 2018-12-11 北京建筑大学 A kind of data center's air-conditioning system forecast Control Algorithm
CN109118020A (en) * 2018-09-06 2019-01-01 南京工业职业技术学院 A kind of subway station energy consumption short term prediction method and its forecasting system
CN109324679A (en) * 2018-09-21 2019-02-12 郑州云海信息技术有限公司 A kind of server energy consumption control method and device
CN109695944A (en) * 2018-11-29 2019-04-30 中国汽车工业工程有限公司 A kind of control method of the coating fresh air conditioner based on multi-model deep learning
CN110188956A (en) * 2019-05-31 2019-08-30 广州水沐青华科技有限公司 Load forecasting method based on ant colony neural network and the system comprising this method, memory

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
GWO-CHING LIAO: "Hybrid improved differential evolution and wavelet neural network with load forecasting problem of air conditioning", vol. 61, no. 2014, pages 673 - 682 *
刘静纨等: "神经网络模糊预测优化控制在VAV***中的应用", vol. 22, no. 12, pages 2909 - 2914 *
孙丽萍等: "中央制冷空调冷冻水***模糊RBF控制研究", vol. 21, no. 05, pages 110 - 116 *

Cited By (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113757922A (en) * 2020-06-03 2021-12-07 ***通信集团浙江有限公司 Deep learning-based air conditioning system energy-saving control method, device and equipment and computer medium
CN114325141B (en) * 2020-09-29 2023-09-05 ***通信集团设计院有限公司 Base station electric quantity abnormality monitoring method and device and readable storage medium
CN114325141A (en) * 2020-09-29 2022-04-12 ***通信集团设计院有限公司 Method and device for monitoring abnormal electric quantity of base station and readable storage medium
CN112329338A (en) * 2020-10-26 2021-02-05 ***通信集团内蒙古有限公司 Cold source system control method and device based on fish swarm optimization BP neural network
CN112616292A (en) * 2020-11-27 2021-04-06 湖南大学 Data center energy efficiency optimization control method based on neural network model
CN112616292B (en) * 2020-11-27 2022-04-22 湖南大学 Data center energy efficiency optimization control method based on neural network model
CN112612305A (en) * 2020-12-04 2021-04-06 格力电器（武汉）有限公司 Temperature adjusting method, device, equipment, storage medium and air conditioning system
CN112612305B (en) * 2020-12-04 2022-04-08 格力电器（武汉）有限公司 Temperature adjusting method, device, equipment, storage medium and air conditioning system
CN112834079A (en) * 2020-12-25 2021-05-25 山东朗进科技股份有限公司 Method for judging parameter drift of temperature sensor of air conditioning unit of railway vehicle
CN112834079B (en) * 2020-12-25 2023-10-24 山东朗进科技股份有限公司 Parameter drift judging method for temperature sensor of air conditioning unit of railway vehicle
CN114322260A (en) * 2021-12-21 2022-04-12 上海美控智慧建筑有限公司 Air conditioner automatic driving and model training and predicting method, device and equipment
CN114322260B (en) * 2021-12-21 2023-09-08 上海美控智慧建筑有限公司 Air conditioner automatic driving, model training and predicting method, device and equipment
CN115344069A (en) * 2022-07-28 2022-11-15 上海蓝色帛缔智能工程有限公司 Liquid cooling control method and system for data center
CN117241561A (en) * 2023-10-13 2023-12-15 北京科技大学 Data center distributed precise air conditioning unit optimized operation parameter estimation method
CN117241561B (en) * 2023-10-13 2024-02-09 北京科技大学 Data center distributed precise air conditioning unit optimized operation parameter estimation method

Similar Documents

Publication Publication Date Title
CN111144543A (en) Data center air conditioner tail end temperature control method, device and medium
US20200279149A1 (en) Method for reinforcement learning using virtual environment generated by deep learning
US20210018205A1 (en) Systems and methods for automated system identification
US11085663B2 (en) Building management system with triggered feedback set-point signal for persistent excitation
US10544951B2 (en) Optimized precooling of structures
CN112329338B (en) Cold source system control method and device based on fish swarm optimization BP neural network
CN105652677A (en) Intelligent home control method, device and system based on user behavior analysis
CN105243348B (en) Reader optimal deployment method based on passive ultrahigh frequency RFID positioning system
CN110826784B (en) Method and device for predicting energy use efficiency, storage medium and terminal equipment
CN113240184B (en) Building space unit cold load prediction method and system based on federal learning
KR101727434B1 (en) Method for estimating efficiency of refrigerator using random forest model
CN112272074A (en) Information transmission rate control method and system based on neural network
CN114353872A (en) Prediction method and device for machine room temperature
CN114326987B (en) Refrigerating system control and model training method, device, equipment and storage medium
CN113028610B (en) Method and device for global optimization and energy-saving control of dynamic load of central air conditioner
CN103745275A (en) Photovoltaic system electricity generation power prediction method and device
US20220154961A1 (en) Control method, computer-readable recording medium storing control program, and air conditioning control device
KR20200105365A (en) Method for reinforcement learning using virtual environment generated by deep learning
Setiawan et al. Sequence to sequence deep learning architecture for forecasting temperature and humidity inside closed space
Avila-Miranda et al. An optimal and intelligent control strategy to ventilate a greenhouse
KR101657137B1 (en) Intelligent Pigsty Air Vent Method Of Control
CN114636212A (en) GRNN-based multi-chiller system operation control method
CN112944599A (en) Multi-parameter coupling control method and device of air conditioning system
CN114707684A (en) Improved LSTM-based raw tobacco stack internal temperature prediction algorithm
CN113177675A (en) Air conditioner cold load prediction method based on optimization neural network of longicorn group algorithm

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination