CN116611347A - Continuous decision method for energy efficiency optimization of data center - Google Patents

Continuous decision method for energy efficiency optimization of data center Download PDF

Info

Publication number
CN116611347A
CN116611347A CN202310735629.XA CN202310735629A CN116611347A CN 116611347 A CN116611347 A CN 116611347A CN 202310735629 A CN202310735629 A CN 202310735629A CN 116611347 A CN116611347 A CN 116611347A
Authority
CN
China
Prior art keywords
model
encoder
ventilation system
heating
sequence
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
CN202310735629.XA
Other languages
Chinese (zh)
Inventor
张发恩
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.)
Ainnovation Nanjing Technology Co ltd
Original Assignee
Ainnovation Nanjing Technology 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 Ainnovation Nanjing Technology Co ltd filed Critical Ainnovation Nanjing Technology Co ltd
Priority to CN202310735629.XA priority Critical patent/CN116611347A/en
Publication of CN116611347A publication Critical patent/CN116611347A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • 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
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/08—Fluids
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/06—Power analysis or power optimisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/08—Thermal analysis or thermal optimisation
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Geometry (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Air Conditioning Control Device (AREA)

Abstract

The application discloses a continuous decision method for energy efficiency optimization of a data center, which comprises the following steps: step S1, establishing a time sequence model of water circulation of a heating ventilation system based on a seq2seq structure of a circulating neural network; s2, performing performance evaluation on the time sequence model; and S3, performing constrained continuous operation strategy optimization on the overall heating and ventilation system by using the time sequence model with the expected performance. Aiming at the characteristics of the heating and ventilation system, the self-correlation time sequence characteristic of the heating and ventilation system is used as an observable hidden variable to be added into an LSTM-Cell model training network, and a time sequence model formed by training can carry out constrained continuous operation strategy optimization on the whole heating and ventilation system, so that the technical problems that the future energy consumption condition of the heating and ventilation system cannot be predicted and the optimal decision on the whole heating and ventilation system cannot be made in continuous time steps in the conventional energy efficiency optimization method are solved, and the energy saving and emission reduction effects of a data center are greatly improved.

Description

Continuous decision method for energy efficiency optimization of data center
Technical Field
The application relates to the field of data mining and machine learning, in particular to a continuous decision method for optimizing energy efficiency of a data center.
Background
In recent years, with the development of cloud services, big data, AI computing and other technologies, enterprises and governments build a large number of data centers, however, the energy consumption of the data centers is generally higher, and the average PUE value (index for evaluating the energy efficiency of the data centers) is between 2.2 and 3.0. According to statistics of related departments, the electricity consumption of the data center in China accounts for about 3% of the total electricity consumption of the whole society, and the proportion is rising year by year.
Currently, many researches on energy conservation and emission reduction of a data center are carried out, and existing or newly developed energy consumption simulation software is generally adopted to simulate the energy consumption condition of the data center so as to assist in design decision and energy efficiency optimization of the data center. However, most of these researches are biased towards researching the energy saving potential of the data center in the design stage, and the energy efficiency optimization problem of the data center after the data center is put into use is not considered. For example, a *** engineer in 2016 proposes a model predictive control method for deep learning, which helps a heat and ventilation engineer to better control and optimize a heat and ventilation system by searching for a relationship between a control point of the heat and ventilation system and a PUE of a data center. However, the existing energy efficiency optimization methods have the following two defects:
1. the time sequence relation of the water circulation of the heating and ventilation system is not considered, the current energy consumption condition of the heating and ventilation system can be predicted only according to the current running environment of the heating and ventilation system, and the future energy consumption condition of the heating and ventilation system cannot be predicted and decision optimized;
2. the heating and ventilation system can be locally optimized in a single time step, but cannot be integrally optimized in a continuous time step, and the energy saving and emission reduction effect of the data center is limited.
Disclosure of Invention
The application aims to provide an energy efficiency optimization continuous decision method for a data center so as to solve the technical problems.
To achieve the purpose, the application adopts the following technical scheme:
the energy efficiency optimization continuous decision method for the data center comprises the following steps:
step S1, establishing a time sequence model of water circulation of a heating ventilation system based on a seq2seq structure of a circulating neural network;
s2, performing performance evaluation on the time sequence model;
and S3, performing constrained continuous operation strategy optimization on the overall heating and ventilation system by using the time sequence model with the expected performance.
As a preferred embodiment of the present application, in the step S1, the construction of the timing model specifically includes the following steps:
s11, constructing a multi-layer Encoder model by using an OH-LSTM Cell neural network structure;
step S12, setting the input sequence of the Encoder model as (X+OH, M), and setting the output sequence of the Encoder model as (y1+OH, M);
x is used for representing historical time sequence information of the heating and ventilation system in a specified time period;
OH is used for representing an observable hidden variable of the heating ventilation system in the specified time period;
X+OH is a combination feature of the history timing information and feature information combination input to the Encoder model in the specified time period;
y1 is used for representing a prediction target of the Encoder model;
y1+OH represents that the output of the Encoder model contains the predicted target y1 and the observable hidden variable OH;
m is used for representing the number of steps of the time steps of the input sequence or the output sequence of the Encoder model;
step S13, setting a loss function of the Encoder model;
step S14, constructing training data of the Encoder model according to the sequence structure set in the step S12, and then training to form the Encoder model by taking the training data as a model training sample;
s15, cutting the input sequence of the Encoder model to be (X+OH, M), and cutting the output sequence of the Encoder model to be (y1+OH, 1);
"1" represents the single said time step that was last output by the Encoder model;
s16, constructing a multi-layer Decoder model by using the OH-LSTM Cell neural network structure;
step S17, setting the input sequence of the Decoder model as (X+OH, M), and setting the output sequence of the Decoder model as (y2+OH, N);
n is used for representing the step number of the time steps output by the Decoder model;
y2 is used for representing a prediction target of the Decoder model;
y2+oh represents that the output of the Decoder model contains the predicted target y2 and the observable hidden variable OH;
step S18, setting a loss function of the Decoder model;
step S19, constructing training data of an Encoder-Decoder model according to the sequence structures set in the step S12 and the step S17;
step S20, migrating the model parameters of the Encoder model to the Decode model, and training to form the Encoder-Decode model as the time sequence model by taking the training data constructed in the step S19 as a training sample;
step S21, clipping the input sequence of the Encoder-Decoder model to be (X+OH, M+N), clipping the output sequence of the Encoder-Decoder model to be (y 2, N),
M+N is used to indicate that the input sequence of the Encoder-Decode model contains M+N of the time steps.
As a preferred embodiment of the present application, the characteristic information includes at least the observable hidden variable.
In a preferred embodiment of the present application, in the step S13, a loss function of the Encoder model is set to perform a weighted average operation on an output sequence loss of the Encoder model.
In a preferred embodiment of the present application, in the step S18, a loss function of the Decoder model is set to average the output sequence loss of the Decoder model.
As a preferred embodiment of the present application, in the step S2, the method for performing performance evaluation on the time series model includes a continuous sensitivity curve analysis evaluation method, and the continuous sensitivity curve analysis evaluation method specifically includes the following steps:
step L1, randomly extracting a sample list with appointed continuous step length as input of the Encoder model;
step L2, selecting n sensitivity parameters to be analyzed and evaluated;
step L3, constructing a parameter integer list of sensitivity parameters;
step L4, inputting the parameter integer list and the t-th time step in the continuous step length designated in the step L1 into the input sequence of the Encoder model for Cartesian product combination so as to construct n groups of input sequences of the Encoder model;
step L5, carrying out Cartesian product combination on the input sequence of the Encoder model and n groups of input sequences of the Decoder model to construct n groups of input sequences of the Encoder-Decoder model;
step L6, taking the n groups of input sequences of the Encoder-Decoder model constructed in the step L5 as the input of the Encoder-Decoder model, and obtaining the output of the Encoder-Decoder model;
and step L7, drawing a time sequence distribution curve which can represent the energy consumption condition of the heating and ventilation system in the appointed continuous step time period according to the output of the Encoder-Decoder model.
As a preferred embodiment of the present application, in the step S2, the method for performing performance evaluation on the time sequence model includes a delay sensitivity curve analysis and evaluation method, and the delay sensitivity curve analysis and evaluation method specifically includes the following steps:
m1, randomly extracting M sections of sample list groups with appointed continuous step sizes as input of the Encoder model;
step M2, selecting n sensitivity parameters to be analyzed and evaluated;
m3, constructing a parameter integer list of sensitivity parameters;
m4, carrying out Cartesian product combination on the parameter integer list and M groups of input sequences of the Encoder model to construct n x M groups of input sequences of the Encoder model;
m5, carrying out Cartesian product combination on the sensitivity parameters of each group in the n-M groups and the input sequence of inputting the t-th time step in the appointed continuous step length into the Encoder model so as to construct an input sequence of the Encoder-Decode model in the n-M groups;
m6, taking the n.m groups of input sequences of the Encoder-Decoder model constructed in the step M5 as the input of the Encoder-Decoder model, and obtaining the output of the Encoder-Decoder model;
and M7, drawing a time sequence distribution curve which can represent the energy consumption condition of the heating and ventilation system in different time periods according to the output of the Encoder-Decoder model.
As a preferred embodiment of the present application, the number of steps of the continuous step size extracted randomly is 30 time steps.
As a preferred embodiment of the present application, t=30.
As a preferred solution of the present application, in the step S3, the process of performing constrained global continuous optimization on the operation policy of the hvac system according to the time sequence model includes the following steps:
step N1, extracting historical sample data of the heating and ventilation system at the current moment to construct a control parameter combination candidate set of the Decoder model in j continuous historical time steps, wherein j is a natural number more than or equal to 1;
step N2, constructing k groups of control parameter combination candidate sets;
step N3, combining k groups of the control parameter combination candidate sets with environmental parameters to construct m groups of input sequences of the Decoder models;
step N4, combining the constructed k groups of input sequences of the Decoder models with the input sequences of the Encoder models which are also k groups, and constructing k groups of input sequences of the Encoder-Decoder models;
step N5, inputting the k groups of input sequences of the Encoder-Decode model constructed in the step N4 into the Encoder-Decode model, and outputting a prediction result of the output power of the heating and ventilation system;
step N6, eliminating the control parameter combination candidate set which does not meet constraint conditions according to the prediction result;
step N7, selecting the control parameter combination candidate set with the lowest predicted output power of the heating and ventilation system from at least one group of control parameter combination candidate sets which are in accordance with the constraint conditions as a control parameter set of the heating and ventilation system at the next time step of the current moment and issuing the control parameter set to corresponding equipment in the heating and ventilation system, wherein each equipment in the heating and ventilation system operates according to the issued control parameter;
and step N8, repeating the steps N1 to N7 in the next time step of the current moment so as to carry out continuous optimization decision on the global running state of the heating and ventilation system.
Aiming at the characteristics of the heating and ventilation system, the self-correlation time sequence characteristic of the heating and ventilation system is used as an observable hidden variable to be added into an LSTM-Cell model training network, and a time sequence model formed by training can carry out constrained continuous operation strategy optimization on the whole heating and ventilation system, so that the technical problems that the future energy consumption condition of the heating and ventilation system cannot be predicted and the optimal decision on the whole heating and ventilation system cannot be made in continuous time steps in the conventional energy efficiency optimization method are solved, and the energy saving and emission reduction effects of a data center are greatly improved.
Drawings
In order to more clearly illustrate the technical solution of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below. It is evident that the drawings described below are only some embodiments of the present application and that other drawings may be obtained from these drawings without inventive effort for a person of ordinary skill in the art.
FIG. 1 is a method step diagram of a continuous decision method for optimizing data center energy efficiency according to an embodiment of the present application;
FIG. 2 is a diagram of method steps for constructing the timing model;
FIG. 3 is a diagram of method steps for evaluating the timing model using the continuous sensitivity curve analysis evaluation method;
FIG. 4 is a diagram of method steps for evaluating the timing model using the time delay sensitivity curve analysis evaluation method;
FIG. 5 is a diagram of method steps for constrained global continuous optimization of the operating strategy of the HVAC system according to the timing model;
FIG. 6 is a schematic representation of the structure of an improved OH-LSTM Cell neural network of the present application;
FIG. 7 is a schematic diagram of the construction of the Encoder-Decoder model;
FIG. 8 is a schematic diagram of a timing distribution curve made by the continuous sensitivity curve analysis evaluation method;
FIG. 9 is a second timing distribution diagram made by the continuous sensitivity curve analysis evaluation method;
FIG. 10 is a schematic diagram of a timing distribution curve made by the described method of analysis and evaluation of a delay sensitivity curve.
Detailed Description
The technical scheme of the application is further described below by the specific embodiments with reference to the accompanying drawings.
Wherein the drawings are for illustrative purposes only and are shown in schematic, non-physical, and not intended to be limiting of the present patent; for the purpose of better illustrating embodiments of the application, certain elements of the drawings may be omitted, enlarged or reduced and do not represent the size of the actual product; it will be appreciated by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The same or similar reference numbers in the drawings of embodiments of the application correspond to the same or similar components; in the description of the present application, it should be understood that, if the terms "upper", "lower", "left", "right", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, only for convenience in describing the present application and simplifying the description, rather than indicating or implying that the apparatus or elements being referred to must have a specific orientation, be constructed and operated in a specific orientation, so that the terms describing the positional relationships in the drawings are merely for exemplary illustration and should not be construed as limiting the present patent, and that the specific meaning of the terms described above may be understood by those of ordinary skill in the art according to specific circumstances.
In the description of the present application, unless explicitly stated and limited otherwise, the term "coupled" or the like should be interpreted broadly, as it may be fixedly coupled, detachably coupled, or integrally formed, as indicating the relationship of components; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between the two parts or interaction relationship between the two parts. The specific meaning of the above terms in the present application will be understood in specific cases by those of ordinary skill in the art.
Aiming at the characteristics of the heating and ventilation system, the self-correlation time sequence characteristic of the heating and ventilation system is used as an observable hidden variable to be added into an LSTM-Cell model training network, and a time sequence model formed by training can carry out constrained continuous operation strategy optimization on the whole heating and ventilation system, so that the technical problems that the future energy consumption condition of the heating and ventilation system cannot be predicted and the optimal decision on the whole heating and ventilation system cannot be made in continuous time steps in the conventional energy efficiency optimization method are solved, and the energy saving and emission reduction effects of a data center are greatly improved.
Referring to fig. 1, the method for continuously determining energy efficiency optimization of a data center provided by the embodiment of the application includes the following steps:
step S1, establishing a time sequence model of water circulation of a heating ventilation system based on a seq2seq structure of a circulating neural network;
s2, evaluating performance of the time sequence model;
and S3, performing constraint continuous operation strategy optimization on the overall situation of the heating and ventilation system by using a time sequence model with performance meeting expectations.
Referring to fig. 2, in step S1, the construction of the timing model specifically includes the following steps:
s11, constructing a multi-layer Encoder model by using an OH-LSTM Cell neural network structure, and specifically constructing the Encoder model by using a keras frame of the OH-LSTM Cell neural network; the OH-LSTM Cell neural network structure is a time-cycled neural network structure which is improved based on the existing LSTM (long-short-term memory network) neural network. The improvement points of the application are that: the pephole of OH (observed hidden variable) is added to LSTM so that the output y of the timing model can snoop the closed loop information through the pephole and feed back to the hidden layer state (hidden states). OH is an observable hidden variable of water circulation of the heating and ventilation system (such as a related variable of water inlet and outlet temperature, pressure and the like which have indirect influence on the output power of the heating and ventilation system). The peak refers to establishing a calculation channel for the autoregressive variables, that is, predicting variables (such as inlet and outlet water temperature, pressure and the like) related to the output power of the heating and ventilation system, and finally obtaining the power value to be predicted according to the variable values of the autoregressive variables. OH is an autoregressive variable.
FIG. 6 shows a schematic diagram of the OH-LSTM Cell neural network. In fig. 6, xt represents a parameter vector of the heating and ventilation system (vector formed by heating and ventilation system control parameters) at the current time point;
OHt represents the observable hidden variables of the heating and ventilation system at the current point in time;
ht represents a hidden variable for the current time;
OHHt is the hidden layer output after OHt and Ht are fused (spliced);
c is a memory neuron of the LSTM neural network.
Fig. 7 shows a schematic diagram of the construction of the Encoder-Decoder model according to the present application. With continued reference to fig. 2 and with reference to fig. 7, after completing the construction of the Encoder model of step S11, step S12 is entered,
step S12, setting the input sequence of the Encoder model as (X+OH, M), and setting the output sequence of the Encoder model as (y1+OH, M);
x is used for representing historical time sequence information of the heating and ventilation system in a certain period of time;
OH is used to represent the observable hidden variable of the heating ventilation system in the specified time period;
X+OH is a combination feature of the combination of the historical time sequence information and the feature information of the input Encoder model in the appointed time period; the characteristic information at least comprises an observable hidden quantity OH, and the characteristic information generally comprises the characteristics of working frequency, external temperature and humidity and the like of equipment such as a heating and ventilation system cooling tower, a cooling pump, a chiller and the like;
y1 is used for representing the prediction target of the Encoder model;
y1+OH represents that the output of the Encoder model contains a predicted target y1 and an observable hidden variable OH;
m is used to represent the sequence length of the input sequence or output sequence of the Encoder model, which in this embodiment is the number of steps of the time steps of the input or output of the Encoder model.
Step S13, setting a loss function of the Encoder model. Preferably, the loss function of the Encoder model is set to a weighted average of the output sequence loss of the Encoder model.
Step S14, constructing training data of the Encoder model according to the sequence structure set in the step S12, and then training to form the Encoder model by taking the constructed training data as a model training sample; the training data comprises historical time sequence information of the heating ventilation system and characteristic information of the heating ventilation system, wherein the characteristic information at least comprises an observable hidden variable OH.
S15, cutting an input sequence of the Encoder model to be (X+OH, M), and cutting an output sequence of the Encoder model to be (y1+OH, 1); the input and output sequences of the Encoder model are tailored to provide data preparation for Decoder model prediction. Since the hidden variables required for the prediction of the Decoder model are output step by step over time, the input of the Decoder model only requires the output of the last step of the Encoder model, so the number "1" in the output sequence (y1+OH, 1) of the Encoder model indicates a single time step of the final output of the Encoder model over a period of time.
Step S16, constructing a multi-layer Decoder model by using an OH-LSTM Cell neural network structure, wherein the construction process is consistent with the construction process of the Encoder model, and the details are not repeated here;
step S17, setting the input sequence of the Decoder model as (X+OH, M), and setting the output sequence of the Decoder model as (y2+OH, N);
n is used for representing the number of time steps output by the Decoder model;
y2 is used for representing the prediction target of the Decoder model;
y2+OH represents that the output of the Decoder model contains a predicted target y2 and an observable hidden variable OH;
step S18, setting a loss function of a Decoder model; preferably, the loss function of the Decoder model is set to average the output sequence loss of the Decoder model.
Step S19, constructing training data of an Encoder-Decoder model according to the sequence structures set in the step S12 and the step S17; the training data comprises historical time sequence information of the heating ventilation system and characteristic information of the heating ventilation system, wherein the characteristic information at least comprises an observable hidden variable OH.
Step S20, migrating model parameters of the Encoder model to the Decoder model, and training to form an Encoder-Decoder model as a time sequence model by taking the training data constructed in the step S19 as a training sample;
step S21, clipping the input sequence of the Encoder-Decoder model to (X+OH, M+N), clipping the output sequence of the Encoder-Decoder model to (y 2, N),
M+N is used to represent that the input sequence of the Encoder-Decode model contains M+N time steps.
In order to test performance of a time sequence model, the present application provides a continuous sensitivity curve analysis and evaluation method, please refer to fig. 3, which specifically includes the following steps:
step L1, randomly extracting a sample list with appointed continuous step length as input of an Encoder model, wherein data in the sample list is training data for constructing a time sequence model; it is also preferred that the number of steps in successive steps is 30, i.e. the random access heating system samples data over successive 30 time steps.
In step L2, n sensitivity parameters to be analyzed and evaluated are selected, where the sensitivity parameters refer to adjustable control parameters in an industrial scene, such as frequencies of various pumps in a heating and ventilation system, fan frequencies, and the like.
Step L3, constructing a parameter integer list of sensitivity parameters; the integral list of sensitivity parameters is made based on the actual control range of sensitivity parameters, such as the controllable frequency of the fan [30,50].
Step L4, the parameter integer list and the t time step in the continuous step length appointed by the step L1 are input into the input sequence of the Encoder model to be subjected to Cartesian product combination so as to construct n groups of input sequences of the Decoder model; if the designated continuous time step is 30, the value of t is preferably 30, that is, the parameter integer list is combined with the input sequence of the last time step of the designated continuous time step to the Encoder model by cartesian products.
Step L5, carrying out Cartesian product combination on the input sequence of the Encoder model and the input sequence of the n groups of the Decoders model to construct the input sequence of the n groups of the Encoders-Decoders model; it should be noted here that, instead of directly combining the input sequence of the Encoder model with the input sequence of the n groups of Decode models, the input sequence of the n groups of Encoder-Decode models is obtained by Cartesian product combination of the output sequence of the Encoder model output at the last time step within a specified continuous step length with the input sequence of the n groups of Decode models. Since the output sequence of the Encoder model is obtained by input sequence prediction, the process of constructing the input sequence of the Encoder-Decode model is summarized as a Cartesian product combination of the input sequence of the Encoder model with the input sequences of the n groups of Decode models for better clarity of presentation in step L5. In addition, the inputs of the n sets of Encoder-Decoder models are constructed instead of one set of inputs in order to increase the confidence in the model performance assessment results.
And step L6, taking the input sequence of the n groups of the Encoder-Decoder models constructed in the step L5 as the input of the Encoder-Decoder models, and obtaining the output of the Encoder-Decoder models.
And step L7, drawing a time sequence distribution curve which can represent the energy consumption condition of the heating and ventilation system in a specified continuous step time period according to the output of the Encoder-Decoder model.
Fig. 8 and 9 are schematic diagrams of the time-series distribution curve obtained by the continuous sensitivity curve analysis and evaluation method.
The application further provides a model performance evaluation method, which is a time delay sensitivity curve analysis evaluation method, referring to fig. 4, the specific process of performing performance evaluation on a time sequence model by the time delay sensitivity curve analysis evaluation method comprises the following steps:
m1, randomly extracting M sections of sample list groups with appointed continuous step sizes as input of an Encoder model; the sample list group comprises m groups of sample lists, and data in the sample list is training data for constructing a time sequence model; preferably, the time step (step size) of the specified consecutive step size is 30, so that the sufficiency of the data can be ensured.
Step M2, selecting n sensitivity parameters to be analyzed and evaluated.
And M3, constructing a parameter integer list of sensitivity parameters.
M4, carrying out Cartesian product combination on the parameter integer list and M groups of input sequences of the Encoder model to construct n.m groups of input sequences of the Encoder model; the construction process of the input sequence of the Decoder model is the same as step L4 in the continuous sensitivity curve analysis and evaluation method, and will not be described here again.
M5, inputting the sensitivity parameter of each group in the n-M groups and the t-th time step in the appointed continuous step length into the input sequence of the Encoder model to carry out Cartesian product combination so as to construct the input sequence of the n-M groups of Encoder-Decoder models; t is also preferably 30, i.e. the sensitivity parameters of each of the n x m groups are combined in cartesian products with the input sequence to the Encoder model at the last time step of the succession of steps of the current time period.
And M6, taking the n.m groups of input sequences constructed in the step M5 as the input of the Encoder-Decoder model, and obtaining the output of the Encoder-Decoder model.
And M7, drawing a time sequence distribution curve which can represent the energy consumption condition of the heating and ventilation system in different time periods according to the output of the Encoder-Decoder model.
Fig. 10 is a schematic diagram of a time sequence distribution curve obtained by the time delay sensitivity curve analysis and evaluation method.
Continuing with fig. 5, in step S3, the process of performing constrained global continuous optimization on the operation strategy of the hvac system according to the trained timing model specifically includes the following steps:
step N1, historical sample data of a heating and ventilation system are provided at the current moment to construct a control parameter combination candidate set of a Decoder model in j continuous historical time steps, wherein j is a natural number more than or equal to 1; the historical sample data comprises various control parameters of the heating ventilation system in historical time steps; the current time refers to the time when global continuous optimization of the heating and ventilation system is needed;
step N2, constructing k groups of control parameter combination candidate sets; a set of control parameter combination candidates includes control parameters for a specified time step, such as control parameters including j time steps, the control parameters for each time step may not be uniform;
step N3, combining the k groups of control parameter combination candidate sets with the environment parameters to construct an input sequence of m groups of Decoder models; the environmental parameters are temperature and humidity information and the like of the operation of the heating and ventilation system; the combination method of the control parameter combination candidate set and the environmental parameters is that the control parameter combination candidate set and the environmental parameters are spliced mathematically, and the existing data splicing modes are numerous, so that the specific splicing process is not described herein;
step N4, combining the constructed input sequence of the k groups of Decoder models with the input sequence of the k groups of the Encoder models to construct the input sequence of the k groups of the Encoder-Decoder models; the construction method of the input sequence of the Encoder-Decode model as described in the continuous sensitivity curve analysis and evaluation method or the time-lapse sensitivity curve analysis and evaluation method is not described in detail herein;
step N5, inputting the input sequence of the constructed k groups of the Encoder-Decoder models into the Encoder-Decoder models, and outputting a prediction result of the output power of the heating and ventilation system;
step N6, eliminating the control parameter combination candidate set which does not accord with the constraint condition according to the prediction result; the rejection rule is that each control parameter has a reasonable control range, and when the prediction result shows that the parameter value corresponding to the control parameter exceeds the reasonable control range, the candidate set of the control parameter combination where the control parameter is located is rejected;
step N7, selecting a control parameter combination candidate set with the lowest predicted output power of the heating and ventilation system from at least one group of control parameter combination candidate sets which are in accordance with constraint conditions as a control parameter set of the heating and ventilation system at the next time step of the current moment, and transmitting the control parameter set to corresponding equipment in the heating and ventilation system, wherein each equipment in the heating and ventilation system operates according to the transmitted control parameters;
and step N8, repeatedly executing N1 to N7 in the next time step of the current moment so as to carry out continuous optimization decision on the global running state of the heating ventilation system.
It should be understood that the above description is only illustrative of the preferred embodiments of the present application and the technical principles employed. It will be apparent to those skilled in the art that various modifications, equivalents, variations, and the like can be made to the present application. However, such modifications are intended to fall within the scope of the present application without departing from the spirit of the present application. In addition, some terms used in the description and claims of the present application are not limiting, but are merely for convenience of description.

Claims (3)

1. The continuous decision method for optimizing the energy efficiency of the data center is characterized by comprising the following steps of:
step S1, establishing a time sequence model of water circulation of a heating ventilation system based on a seq2seq structure of a circulating neural network;
s2, performing performance evaluation on the time sequence model;
s3, performing constraint continuous operation strategy optimization on the overall situation of the heating and ventilation system by using the time sequence model with expected performance;
in the step S1, the construction of the timing model specifically includes the following steps:
s11, building a multi-layer Encoder model;
step S12, setting the input sequence of the Encoder model as (X+OH, M), and setting the output sequence of the Encoder model as (y1+OH, M);
x is used for representing historical time sequence information of the heating and ventilation system in a specified time period;
OH is used for representing an observable hidden variable of the heating ventilation system in the specified time period;
X+OH is a combination feature of the history timing information and feature information combination input to the Encoder model in the specified time period;
y1 is used for representing a prediction target of the Encoder model;
y1+OH represents that the output of the Encoder model contains the predicted target y1 and the observable hidden variable OH;
m is used for representing the number of steps of the time steps of the input sequence or the output sequence of the Encoder model;
step S13, setting a loss function of the Encoder model;
step S14, constructing training data of the Encoder model according to the sequence structure set in the step S12, and then training to form the Encoder model by taking the training data as a model training sample;
s15, cutting the input sequence of the Encoder model to be (X+OH, M), and cutting the output sequence of the Encoder model to be (y1+OH, 1);
"1" represents the single said time step that was last output by the Encoder model;
s16, constructing a multi-layer Decoder model;
step S17, setting the input sequence of the Decoder model as (X+OH, M), and setting the output sequence of the Decoder model as (y2+OH, N);
n is used for representing the step number of the time steps output by the Decoder model;
y2 is used for representing a prediction target of the Decoder model;
y2+oh represents that the output of the Decoder model contains the predicted target y2 and the observable hidden variable OH;
step S18, setting a loss function of the Decoder model;
step S19, constructing training data of an Encoder-Decoder model according to the sequence structures set in the step S12 and the step S17;
step S20, migrating the model parameters of the Encoder model to the Decode model, and training to form the Encoder-Decode model as the time sequence model by taking the training data constructed in the step S19 as a training sample;
step S21, clipping the input sequence of the Encoder-Decoder model to be (X+OH, M+N), clipping the output sequence of the Encoder-Decoder model to be (y 2, N),
M+N is used for representing that the input sequence of the Encoder-Decode model comprises M+N time steps;
in the step S2, the method for performing performance evaluation on the time sequence model includes a delay sensitivity curve analysis and evaluation method, and the delay sensitivity curve analysis and evaluation method specifically includes the following steps:
m1, randomly extracting M sections of sample list groups with appointed continuous step sizes as input of the Encoder model;
step M2, selecting n sensitivity parameters to be analyzed and evaluated;
m3, constructing a parameter integer list of sensitivity parameters;
m4, carrying out Cartesian product combination on the parameter integer list and M groups of input sequences of the Encoder model to construct n x M groups of input sequences of the Encoder model;
m5, carrying out Cartesian product combination on the sensitivity parameters of each group in the n-M groups and the input sequence of inputting the t-th time step in the appointed continuous step length into the Encoder model so as to construct an input sequence of the Encoder-Decode model in the n-M groups;
m6, taking the n.m groups of input sequences of the Encoder-Decoder model constructed in the step M5 as the input of the Encoder-Decoder model, and obtaining the output of the Encoder-Decoder model;
m7, drawing a time sequence distribution curve which can represent the energy consumption condition of the heating and ventilation system in different time periods according to the output of the Encoder-Decoder model;
in the step S3, the process of performing constrained global continuous optimization on the operation strategy of the hvac system according to the time sequence model includes the following steps:
step N1, extracting historical sample data of the heating and ventilation system at the current moment to construct a control parameter combination candidate set of the Decoder model in j continuous historical time steps, wherein j is a natural number more than or equal to 1;
step N2, constructing k groups of control parameter combination candidate sets;
step N3, combining the k groups of control parameter combination candidate sets with environmental parameters to construct k groups of input sequences of the Decoder model;
step N4, combining the constructed k groups of input sequences of the Decoder models with the input sequences of the Encoder models which are also k groups of input sequences of the Encoder models, constructing k groups of input sequences of the Encoder-Decoder models, and randomly extracting m groups of sample list groups with appointed continuous step sizes as one group of input sequences of the Encoder models;
step N5, inputting the k groups of input sequences of the Encoder-Decode model constructed in the step N4 into the Encoder-Decode model, and outputting a prediction result of the output power of the heating and ventilation system;
step N6, eliminating the control parameter combination candidate set which does not meet constraint conditions according to the prediction result;
step N7, selecting the control parameter combination candidate set with the lowest predicted output power of the heating and ventilation system from at least one group of control parameter combination candidate sets which are in accordance with the constraint conditions as a control parameter set of the heating and ventilation system at the next time step of the current moment and issuing the control parameter set to corresponding equipment in the heating and ventilation system, wherein each equipment in the heating and ventilation system operates according to the issued control parameter;
and step N8, repeating the steps N1 to N7 in the next time step of the current moment so as to carry out continuous optimization decision on the global running state of the heating and ventilation system.
2. The data center energy efficiency optimization continuous decision method of claim 1, wherein the number of steps of the continuous step size randomly extracted is 30 of the time steps.
3. The data center energy efficiency optimization continuous decision method of claim 1, wherein t = 30.
CN202310735629.XA 2020-02-27 2020-02-27 Continuous decision method for energy efficiency optimization of data center Pending CN116611347A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310735629.XA CN116611347A (en) 2020-02-27 2020-02-27 Continuous decision method for energy efficiency optimization of data center

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010122279.6A CN111368408B (en) 2020-02-27 2020-02-27 Data center energy efficiency optimization continuous decision method based on seq2seq
CN202310735629.XA CN116611347A (en) 2020-02-27 2020-02-27 Continuous decision method for energy efficiency optimization of data center

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
CN202010122279.6A Division CN111368408B (en) 2020-02-27 2020-02-27 Data center energy efficiency optimization continuous decision method based on seq2seq

Publications (1)

Publication Number Publication Date
CN116611347A true CN116611347A (en) 2023-08-18

Family

ID=71210048

Family Applications (2)

Application Number Title Priority Date Filing Date
CN202310735629.XA Pending CN116611347A (en) 2020-02-27 2020-02-27 Continuous decision method for energy efficiency optimization of data center
CN202010122279.6A Active CN111368408B (en) 2020-02-27 2020-02-27 Data center energy efficiency optimization continuous decision method based on seq2seq

Family Applications After (1)

Application Number Title Priority Date Filing Date
CN202010122279.6A Active CN111368408B (en) 2020-02-27 2020-02-27 Data center energy efficiency optimization continuous decision method based on seq2seq

Country Status (1)

Country Link
CN (2) CN116611347A (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117404853B (en) * 2023-12-14 2024-03-08 山西省水利建筑工程局集团有限公司 External circulating water cooling system and method for tunnel boring machine

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109284576B (en) * 2018-10-29 2023-04-21 东北电力大学 Distributed electric heating load scheduling method based on measured data and modeling system thereof
CN110703899B (en) * 2019-09-09 2020-09-25 创新奇智(南京)科技有限公司 Data center energy efficiency optimization method based on transfer learning

Also Published As

Publication number Publication date
CN111368408B (en) 2023-06-23
CN111368408A (en) 2020-07-03

Similar Documents

Publication Publication Date Title
CN113835344B (en) Control optimization method of equipment, display platform, cloud server and storage medium
CN109634121B (en) Multi-parent genetic algorithm air source heat pump multi-objective optimization control method based on radial basis function neural network
CN107665230A (en) Training method and device for the users' behavior model of Intelligent housing
CN111401788B (en) Attribution method and device of service timing sequence index
Deng et al. Integrating feature optimization using a dynamic convolutional neural network for chemical process supervised fault classification
CN112415924A (en) Energy-saving optimization method and system for air conditioning system
CN104881735A (en) System and method of smart power grid big data mining for supporting smart city operation management
CN112990258A (en) Fault diagnosis method and system for water chilling unit
CN111368408B (en) Data center energy efficiency optimization continuous decision method based on seq2seq
CN115238599B (en) Energy-saving method and model reinforcement learning training method and device for refrigerating system
CN116579447A (en) Time sequence prediction method based on decomposition mechanism and attention mechanism
CN115370437A (en) Cold, heat and electricity tri-generation energy storage system based on compressed fluid and control method
Wang et al. A transformer-based multi-entity load forecasting method for integrated energy systems
CN114326987A (en) Refrigeration system control and model training method, device, equipment and storage medium
Wang et al. A data mining approach to discover critical events for event-driven optimization in building air conditioning systems
BLASI PERFORMANCE INCREMENT OF HIGH SCHOOL STUDENTS USING ANN MODEL AND SA ALGORITHM.
Huang et al. Operation optimization for centrifugal chiller plants using continuous piecewise linear programming
CN116795198A (en) Energy consumption optimization method and device for data center and storage medium
CN116578858A (en) Air compressor fault prediction and health degree evaluation method and system based on graphic neural network
CN115828769A (en) Method for predicting working condition of cooling tower and reducing consumption based on intelligent calculation
CN115081609A (en) Acceleration method in intelligent decision, terminal equipment and storage medium
CN114650190B (en) Energy-saving method, system, terminal equipment and storage medium for data center network
CN117634217B (en) Intelligent building management system and method based on digital twinning
CN117892073B (en) Irrigation area water metering system and water metering method
Ma et al. Monotonic neural network: combining deep learning with domain knowledge for chiller plants energy optimization

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