CN118014776A - Load dynamic prediction method for closed circulation water cooling system - Google Patents
Load dynamic prediction method for closed circulation water cooling system Download PDFInfo
- Publication number
- CN118014776A CN118014776A CN202410418995.7A CN202410418995A CN118014776A CN 118014776 A CN118014776 A CN 118014776A CN 202410418995 A CN202410418995 A CN 202410418995A CN 118014776 A CN118014776 A CN 118014776A
- Authority
- CN
- China
- Prior art keywords
- variable data
- equipment
- target
- independent
- parameter
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 30
- 238000001816 cooling Methods 0.000 title claims abstract description 26
- 230000009467 reduction Effects 0.000 claims abstract description 23
- 239000011159 matrix material Substances 0.000 claims description 24
- 230000008569 process Effects 0.000 claims description 12
- 230000008859 change Effects 0.000 claims description 9
- 230000004927 fusion Effects 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 6
- 238000012847 principal component analysis method Methods 0.000 claims description 5
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000012800 visualization Methods 0.000 claims description 3
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000012545 processing Methods 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 description 13
- 239000002699 waste material Substances 0.000 description 4
- 239000000498 cooling water Substances 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 239000000758 substrate Substances 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Development Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of data processing, in particular to a load dynamic prediction method of a closed circulation water cooling system, which comprises the following steps: acquiring variable data of a plurality of devices; according to the ecological level parameters of the equipment variable data, acquiring the equipment independent variable data, and further acquiring a plurality of main component directions after dimension reduction; and carrying out system load prediction according to the directions of the plurality of main components after dimension reduction. The invention makes the system load prediction loss smaller, is beneficial to the temperature precompensation of the closed circulating water system to the equipment, improves the running stability of the equipment and optimizes the resource utilization.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a load dynamic prediction method of a closed circulation water cooling system.
Background
The closed circulation cooling water system aims at coping with the heat loads, and the principle is that a closed circulation water path is formed inside the cooling equipment, a water pump is conveyed into the cooling equipment for cooling, and then the cooled water is recovered into a water tank, so that the cooling water system is a cooling water system which does not pollute the environment.
The closed circulation water cooling system is combined with the prediction system, so that the demand fluctuation in different time periods can be better dealt with through the prediction of the system load, the stability of the system is improved, the energy consumption is reduced, and the resource utilization is optimized; in the prior art, a device load condition is predicted by a mixed prediction model, but a plurality of variable relations are required to be input into the mixed model, which often has more complex error influence on a prediction result, and the mixed model also has a weight distribution problem when the variables are more, so that the main component variable dimension reduction is required to be carried out on the input variable relations, but the traditional PCA main component analysis algorithm is too focused on data characteristics to neglect variable characteristics, and most of the time, the dimension reduction result is not suitable for the prediction task of the mixed model.
Disclosure of Invention
In order to solve the problems, the invention provides a load dynamic prediction method of a closed circulation water cooling system, which comprises the following steps:
acquiring a plurality of equipment variable data, wherein the equipment variable data comprises a plurality of time data;
acquiring an ecological level parameter of each device variable data according to the difference of time data between each device variable data and other device variable data; screening all the equipment variable data according to the ecological level parameters and the independent uncorrelated conditions among the equipment variable data to obtain all the equipment independent variable data; acquiring a plurality of main component directions after dimension reduction according to the independent variable data of all the devices;
And carrying out system load prediction according to the directions of the plurality of main components after dimension reduction.
Preferably, the obtaining the ecological level parameter of each device variable data according to the difference of the time data between each device variable data and other device variable data includes the following specific methods:
recording any one device variable data as target device variable data, and obtaining a plurality of combinations of the target device variable data;
according to the difference of time data between the variable data of the target equipment and the variable data of other equipment, obtaining the average matching direction and the average amplitude difference value of each combination of the variable data of the target equipment;
the calculation method for obtaining the ecological level parameters of the variable data of the target equipment comprises the following steps:
In the method, in the process of the invention, An ecological level parameter representing target device variable data; /(I)Representing the total number of all combinations of target device variable data; /(I)First/>, representing target device variable dataCosine values of average matching directions of the combinations; /(I)First/>, representing target device variable dataAverage amplitude differences of the combinations.
Preferably, the method for obtaining the plurality of combinations of the variable data of the target device includes the following specific steps:
And combining the variable data of the target equipment with the variable data of all other equipment in pairs to obtain a plurality of combinations of the variable data of the target equipment.
Preferably, the obtaining the average matching direction and the average amplitude difference value of each combination of the target device variable data according to the difference of the time data between the target device variable data and other device variable data includes the following specific steps:
For each piece of equipment variable data, performing nonlinear fitting on time sequence by utilizing a visualization tool to obtain a data change curve of each piece of equipment variable data; for any one combination of target equipment variable data, aligning the data change curves of the two equipment variable data in the combination through a dynamic time warping algorithm, and obtaining a plurality of matching point pairs of the data change curves of the two equipment variable data in the combination, wherein each matching point pair has a matching direction and an amplitude difference value; taking the average value of the matching directions of all the matching point pairs as the average matching direction of the combination; and taking the average value of the amplitude differences of all the matching point pairs as the average amplitude difference of the combination.
Preferably, the method includes the steps of screening all the variable data of the equipment according to the ecological level parameters and the independent uncorrelated conditions among the variable data of the equipment to obtain the independent variable data of all the equipment, and comprises the following specific steps:
Acquiring a quantity parameter sequence;
any one of the quantity parameters in the quantity parameter sequence is recorded as Will be in all device variable data randomly selected/>The individual device variable data together form a data set, noted as a number parameter/>Target variable data sets of (a);
Acquiring quantity parameters according to independent uncorrelated conditions among equipment variable data in a target variable data set Independent uncorrelation degrees of the target variable data sets;
Acquiring a quantity parameter according to the ecological level parameter of the equipment variable data in the target variable data set The penalty degree of the target variable data set;
Obtaining quantity parameters according to punishment degree and independent uncorrelation degree An objective function output value of the objective variable data set;
And acquiring target function output values of all quantity parameters in the quantity parameter sequence, recording a target variable data set corresponding to the minimum value of the target function output values as an independent variable data set, and taking all device variable data in the independent variable data set as device independent variable data.
Preferably, the method for obtaining the number parameter sequence includes the following specific steps:
Presetting an initial quantity parameter For the initial quantity parameter/>And gradually increasing according to the step length of 1 until the number parameter is equal to the total number of all the device variable data, stopping increasing, and obtaining a sequence formed by a plurality of number parameters arranged from small to large, and recording the sequence as a number parameter sequence.
Preferably, the number parameter is obtained according to the independent uncorrelated condition between the device variable data in the target variable data setThe specific formula of the independent uncorrelation degree of the target variable data set is as follows:
In the method, in the process of the invention, Representing quantity parameters/>Independent uncorrelation degrees of the target variable data sets; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the inner product symbol; /(I)The representation takes absolute value.
Preferably, the number parameter is obtained according to the ecological level parameter of the device variable data in the target variable data setThe specific formula of the punishment degree of the target variable data set is as follows:
In the method, in the process of the invention, Representing quantity parameters/>The penalty degree of the target variable data set; /(I)Representing the average value of the ecological level parameters of all the equipment variable data; /(I)Representing the/>, in a set of target variable dataThe ecological level parameters of the individual device variable data.
Preferably, the number parameter is obtained according to the punishment degree and the independent uncorrelation degreeThe specific method for outputting the target function of the target variable data set comprises the following steps:
parameter of quantity Independent degree of independence and quantity parameter/>, of a target variable data set of (a)The difference in penalty degree of the target variable data set as a quantity parameter/>An objective function output value of the objective variable data set.
Preferably, the method for obtaining the directions of the plurality of main components after the dimension reduction according to the independent variable data of all the devices includes the following steps:
presetting a characteristic parameter Constructing a covariance matrix of the independent variable data of the equipment through the independent variable data of all the equipment, and constructing a covariance matrix of the variable data of the equipment through the variable data of all the equipment; taking a convolution result of the covariance matrix of the independent variable data of the equipment and the covariance matrix of the variable data of the equipment as a fusion matrix; decomposing the characteristic values of the fusion matrix in each characteristic direction by using a PCA principal component analysis method, acquiring the characteristic value of each characteristic direction, and marking a sequence formed by sequencing all the characteristic directions from large to small according to the characteristic values as a characteristic direction sequence; front in feature Direction sequence-The characteristic directions are all used as the main component directions after dimension reduction.
The technical scheme of the invention has the beneficial effects that: according to the ecological level parameter and the independent uncorrelated condition among the variable data of the equipment, screening the variable data of all the equipment to obtain the independent variable data of all the equipment; according to the independent variable data of all the equipment, the directions of the main components after dimension reduction are obtained, so that the linear loss of new main components and operation load data is minimized, the problem of fuzzy weight of a sub-model in a mixed model is avoided, the obtained directions of the main components after dimension reduction are input into the mixed model for regression, the system load prediction loss is smaller, the temperature precompensation of the closed circulating water system to the equipment is facilitated, the operation stability of the equipment is improved, and the resource utilization is optimized.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart showing the steps of a load dynamic prediction method of a closed circulation water cooling system;
FIG. 2 is a characteristic relation flow chart of a load dynamic prediction method of a closed circulation water cooling system.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of specific implementation, structure, characteristics and effects of a closed circulation water cooling system load dynamic prediction method according to the invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of the closed circulation water cooling system load dynamic prediction method provided by the invention with reference to the accompanying drawings.
Referring to fig. 1, a flow chart of steps of a method for dynamically predicting load of a closed circulation water cooling system according to an embodiment of the invention is shown, and the method includes the following steps:
step S001: and acquiring a plurality of equipment variable data, wherein the equipment variable data comprises a plurality of time data.
Specifically, firstly, a plurality of equipment variable data needs to be collected, and the specific process is as follows:
Acquiring all equipment operation monitoring time sequence data sequences within half a year through different types of sensors, wherein the equipment operation monitoring time sequence data sequences comprise a current monitoring time sequence data sequence, a vibration monitoring time sequence data sequence, a power monitoring time sequence data sequence, a raw material delivery monitoring time sequence data sequence, a production capacity monitoring time sequence data sequence, an operation environment temperature time sequence and a humidity monitoring time sequence data sequence; carrying out mean value normalization on all the equipment operation monitoring time sequence data, and recording each normalized equipment operation monitoring time sequence data as equipment variable data; wherein each device variable data contains a number of time data.
Acquiring temperature monitoring data of the equipment within half a year through a temperature sensor; subtracting a preset temperature threshold value under normal operation from the temperature monitoring output of the equipment, and carrying out nonlinear fitting again on time sequence to obtain the equipment operation load data.
It should be noted that, since each piece of equipment variable data may have an influence on the equipment operation load data, it is necessary to analyze each piece of equipment variable data with emphasis.
So far, a plurality of equipment variable data are obtained through the method.
Step S002: acquiring an ecological level parameter of each device variable data according to the difference of time data between each device variable data and other device variable data; screening all the equipment variable data according to the ecological level parameters and the independent uncorrelated conditions among the equipment variable data to obtain all the equipment independent variable data; and acquiring a plurality of main component directions after dimension reduction according to the independent variable data of all the devices.
In the conventional principal component analysis method of PCA, a covariance matrix is constructed through all device variable data, reflects the association degree between different device operation monitoring time sequence data sequences, then decomposes the covariance matrix to obtain characteristic values in a plurality of directions, sorts the characteristic values according to the size, and selects the characteristic direction (k represents the dimension after dimension reduction is expected) corresponding to the maximum k characteristic values; these feature directions constitute a new coordinate system and are called principal components; however, the principal components obtained by the dimension reduction logic are all device variable data with large variance characteristics, and the variable data with abundant variation characteristics is preserved, but the principal components have no practical prediction significance for the mixed model.
1. And acquiring the ecological level parameters of the variable data of each device.
The device operation system is regarded as an ecological circle, the ecological level represents the level of each device variable data in the ecological circle, all the device variable data in the device operation system are mutually influenced, and energy transfer exists among the device variable data; the higher the ecological level, the purer the device variable data has the self-variable characteristic, and the larger the transmission trend is.
Specifically, for each piece of equipment variable data, nonlinear fitting is performed on time sequence by using a visualization tool, and a data change curve of each piece of equipment variable data is obtained; recording any one device variable data as target device variable data, combining the target device variable data with all other device variable data in pairs to obtain a plurality of combinations of the target device variable data; for any one combination of target equipment variable data, aligning the data change curves of the two equipment variable data in the combination through a dynamic time warping algorithm, and obtaining a plurality of matching point pairs of the data change curves of the two equipment variable data in the combination, wherein each matching point pair has a matching direction and an amplitude difference value; taking the average value of the matching directions of all the matching point pairs as the average matching direction of the combination; and taking the average value of the amplitude differences of all the matching point pairs as the average amplitude difference of the combination.
And acquiring the ecological level parameter of each device variable data according to the difference of the moment data between each device variable data and other device variable data.
As an example, the calculation method for obtaining the ecological level parameter of the target device variable data is as follows:
In the method, in the process of the invention, An ecological level parameter representing target device variable data; /(I)Representing the total number of all combinations of target device variable data; /(I)First/>, representing target device variable dataCosine values of average matching directions of the combinations; /(I)First/>, representing target device variable dataAverage amplitude differences of the combinations.
It should be noted that the number of the substrates,The greater the range between 0-180 degrees, the greater the interpretation of the target device variable dataThe timing of the matching point pairs in each combination is earlier, i.e. the target device variable data and the/>The earlier the target device variable data changes and is transferred to the other device variable data in the combination compared with the other device variable data, and the other device variable data has compliance for the target device variable data; /(I)Will/>The larger the value is, the more the value is converted into 0-2, which indicates that the variable data of the target equipment is in the/>The higher the self-varying features in the individual combinations; /(I)The value of the target equipment variable data is positive or negative, and the larger the value is, the larger the amplitude energy in the target equipment variable data is, and the trend of the amplitude energy is that the amplitude energy is transmitted to the lower-order equipment variable data; thus/>Description of target device variable data at/>The self-variable characteristics and the average value of the transmission trend in the individual combinations are further obtained to obtain the ecological level parameters of the variable data of the target equipment; the larger the ecological level parameter is, the higher the ecological level of the variable data of the target equipment is.
Optionally, in other embodiments, according to a correlation situation between each device variable data and other device variable data, the method for obtaining the ecological level parameter of each device variable data includes:
Recording any one device variable data as target device variable data, and recording a data sequence consisting of all time data in the target device variable data as a time data sequence of the target device variable data; the calculation method for obtaining the ecological level parameters of the variable data of the target equipment comprises the following steps:
In the method, in the process of the invention, An ecological level parameter representing target device variable data; /(I)Representing the total number of all device variable data; /(I)Represents the/>A time of day data sequence of individual device variable data; /(I)A time data sequence representing target device variable data; representation/> And/>DTW distance between; /(I)Representing an exponential function based on natural constants, the examples employ/>Model to present inverse proportional relationship,/>For model input, the practitioner may choose the inverse proportion function according to the actual situation.
So far, the ecological level parameter of each device variable data is obtained.
2. And acquiring the independent variable data of the equipment.
It should be noted that, since the device variable data with similar linear relationships has redundancy problem in the mixed model, the weight distribution of the plurality of input device variable data is blurred, and thus all the device independent variable data need to be obtained.
Presetting an initial quantity parameterWherein the present embodiment is described as/>To describe the example, the present embodiment is not particularly limited, wherein/>Depending on the particular implementation.
Specifically, for the initial quantity parameterAnd gradually increasing according to the step length of 1 until the number parameter is equal to the total number of all the device variable data, stopping increasing, and obtaining a sequence formed by a plurality of number parameters arranged from small to large, and recording the sequence as a number parameter sequence.
Any one of the quantity parameters in the quantity parameter sequence is recorded asWill be in all device variable data randomly selected/>The individual device variable data together form a data set, noted as a number parameter/>Is a target variable data set of (1).
As an example, the number parameters in the number parameter sequence are acquiredThe calculation method of the target function output value of the target variable data set comprises the following steps:
In the method, in the process of the invention, Representing the number parameter/>, in the number parameter sequenceIs set according to the target function output value of the target function; /(I)Representing a number parameter in a number parameter sequence; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the average value of the ecological level parameters of all the equipment variable data; /(I)Representing the/>, in a set of target variable dataEcological level parameters of the individual device variable data; /(I)Representing the inner product symbol; /(I)The representation takes absolute value.
It should be noted that the number of the substrates,Representing quantity parameters/>Independent degree of independence of target variable data sets,/>Represents the/>Personal device variable data and No. >Inner product of the device variable data, the closer the inner product is to 0, the description of the/>Personal device variable data and No. >The more orthogonal the device variable data are, namely the two device variable data are independent and uncorrelated; the smaller the ecological level difference, the device variable data can be derived from the same upper level variable, thus generating larger resource waste rate, requiring the resource waste rate as a penalty,Representing the variance of the ecological level parameters of all the equipment variable data, wherein the smaller the variance is, the higher the resource waste rate of the equipment variable data is; /(I)The smaller the value is, the lower the ecological level of the equipment variable data is; quantity parameter/>/>, In a target variable data setThe higher the resource wave rate of the individual equipment variable data is and the lower the ecological level is, the larger the penalty term of the objective function is; conversely, the larger the penalty term, the description/>The lower the resource waste rate of the variable data quantity of each device is and the higher the ecological level is, the/>The more likely the individual device variable data amount is the device independent variable data in the system, the smaller the penalty term; Representing quantity parameters/> Punishment degree of the target variable data set.
In other embodiments, the number parameters in the number parameter sequence are obtainedThe optional calculation method of the target function output value of the target variable data set is as follows:
In the method, in the process of the invention, Representing the number parameter/>, in the number parameter sequenceIs set according to the target function output value of the target function; /(I)Representing a number parameter in a number parameter sequence; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the divide/>, in a set of target variable data/>, Other than individual device variable dataDevice variable data; /(I)Representation/>And/>Pearson correlation coefficient between; /(I)The representation takes absolute value.
Further, the objective function output values of all the quantity parameters in the quantity parameter sequence are obtained, the objective variable data set corresponding to the minimum value of the objective function output values is recorded as an independent variable data set, and all the equipment variable data in the independent variable data set are used as equipment independent variable data.
So far, all the independent variable data of the equipment are obtained.
3. And acquiring a plurality of main component directions after dimension reduction.
Presetting a characteristic parameterWherein the present embodiment is described as/>To describe the example, the present embodiment is not particularly limited, wherein/>Depending on the particular implementation.
Specifically, a covariance matrix of the independent variable data of the equipment is built through the independent variable data of all the equipment, and a covariance matrix of the variable data of the equipment is built through the variable data of all the equipment; taking a convolution result of the covariance matrix of the independent variable data of the equipment and the covariance matrix of the variable data of the equipment as a fusion matrix; decomposing the characteristic values of the fusion matrix in each characteristic direction by using a PCA principal component analysis method, acquiring the characteristic value of each characteristic direction, and marking a sequence formed by sequencing all the characteristic directions from large to small according to the characteristic values as a characteristic direction sequence; front in the characteristic direction sequenceThe characteristic directions are all used as the main component directions after dimension reduction.
It should be noted that, the covariance matrix of the independent variable data of the device is utilized to perform convolution calculation on the covariance matrix of the variable data of the device to obtain a fusion matrix, because the convolution can fuse the characteristics of the independent variable data of the device with the characteristics of the variable data of all the devices, all the characteristic values decomposed in the fusion matrix not only have variance characteristics, but also contain independent characteristics and ecological level characteristics, and further the dimension-reducing variable obtained by the principal component analysis method of PCA is more beneficial to the prediction process, and the higher and more independent the dimension-reducing variable level is, the smaller the prediction loss obtained in the mixed model as the input variable is.
So far, the directions of a plurality of main components after dimension reduction are obtained through the method.
Step S003: and carrying out system load prediction according to the directions of the plurality of main components after dimension reduction.
Specifically, mapping all operation load data in half years into a coordinate space with a plurality of main component directions after dimension reduction as a transverse axis, carrying out nonlinear fitting on load values in a two-dimensional coordinate space with the main component directions after dimension reduction, taking a fitting curve as a predictive sub-model of each main component and the operation load data, then taking the weight of each sub-model as a normalization value of a corresponding characteristic value of the main component directions after dimension reduction, obtaining all main components according to all independent variable data of equipment at the current moment, respectively inputting the main components into a plurality of predictive sub-models for nonlinear regression, obtaining a plurality of load regression values, and carrying out weighted summation on all the load regression values to obtain the load value at the next moment of the system, thereby completing dynamic prediction of the load value of the closed circulation water cooling system.
The closed circulating water system can reversely compensate the same temperature value according to the equipment operation load value, so that the equipment operation state is more stable, and the resource utilization is optimized.
Thus, the present embodiment is completed; referring to fig. 2, a characteristic relation flow chart of a load dynamic prediction method of a closed circulation water cooling system is shown.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the invention, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present invention should be included in the scope of the present invention.
Claims (10)
1. The method for dynamically predicting the load of the closed circulation water cooling system is characterized by comprising the following steps of:
acquiring a plurality of equipment variable data, wherein the equipment variable data comprises a plurality of time data;
acquiring an ecological level parameter of each device variable data according to the difference of time data between each device variable data and other device variable data; screening all the equipment variable data according to the ecological level parameters and the independent uncorrelated conditions among the equipment variable data to obtain all the equipment independent variable data; acquiring a plurality of main component directions after dimension reduction according to the independent variable data of all the devices;
And carrying out system load prediction according to the directions of the plurality of main components after dimension reduction.
2. The method for dynamically predicting the load of a closed circulation water cooling system according to claim 1, wherein the obtaining the ecological level parameter of each equipment variable data according to the difference of time data between each equipment variable data and other equipment variable data comprises the following specific steps:
recording any one device variable data as target device variable data, and obtaining a plurality of combinations of the target device variable data;
according to the difference of time data between the variable data of the target equipment and the variable data of other equipment, obtaining the average matching direction and the average amplitude difference value of each combination of the variable data of the target equipment;
the calculation method for obtaining the ecological level parameters of the variable data of the target equipment comprises the following steps:
In the method, in the process of the invention, An ecological level parameter representing target device variable data; /(I)Representing the total number of all combinations of target device variable data; /(I)First/>, representing target device variable dataCosine values of average matching directions of the combinations; /(I)First/>, representing target device variable dataAverage amplitude differences of the combinations.
3. The method for dynamically predicting the load of a closed circulation water cooling system according to claim 2, wherein the obtaining the plurality of combinations of the variable data of the target equipment comprises the following specific steps:
And combining the variable data of the target equipment with the variable data of all other equipment in pairs to obtain a plurality of combinations of the variable data of the target equipment.
4. The method for dynamically predicting load of closed cycle water cooling system according to claim 2, wherein the obtaining the average matching direction and average amplitude difference of each combination of the target equipment variable data according to the difference of time data between the target equipment variable data and other equipment variable data comprises the following specific steps:
For each piece of equipment variable data, performing nonlinear fitting on time sequence by utilizing a visualization tool to obtain a data change curve of each piece of equipment variable data; for any one combination of target equipment variable data, aligning the data change curves of the two equipment variable data in the combination through a dynamic time warping algorithm, and obtaining a plurality of matching point pairs of the data change curves of the two equipment variable data in the combination, wherein each matching point pair has a matching direction and an amplitude difference value; taking the average value of the matching directions of all the matching point pairs as the average matching direction of the combination; and taking the average value of the amplitude differences of all the matching point pairs as the average amplitude difference of the combination.
5. The method for dynamically predicting the load of a closed circulation water cooling system according to claim 1, wherein the method for screening all the variable data of the equipment according to the independent uncorrelated conditions among the ecological level parameters and the variable data of the equipment to obtain the independent variable data of all the equipment comprises the following specific steps:
Acquiring a quantity parameter sequence;
any one of the quantity parameters in the quantity parameter sequence is recorded as Will be in all device variable data randomly selected/>The individual device variable data together form a data set, noted as a number parameter/>Target variable data sets of (a);
Acquiring quantity parameters according to independent uncorrelated conditions among equipment variable data in a target variable data set Independent uncorrelation degrees of the target variable data sets;
Acquiring a quantity parameter according to the ecological level parameter of the equipment variable data in the target variable data set The penalty degree of the target variable data set;
Obtaining quantity parameters according to punishment degree and independent uncorrelation degree An objective function output value of the objective variable data set;
And acquiring target function output values of all quantity parameters in the quantity parameter sequence, recording a target variable data set corresponding to the minimum value of the target function output values as an independent variable data set, and taking all device variable data in the independent variable data set as device independent variable data.
6. The method for dynamically predicting the load of a closed-loop water cooling system according to claim 5, wherein the obtaining the number parameter sequence comprises the following specific steps:
Presetting an initial quantity parameter For the initial quantity parameter/>And gradually increasing according to the step length of 1 until the number parameter is equal to the total number of all the device variable data, stopping increasing, and obtaining a sequence formed by a plurality of number parameters arranged from small to large, and recording the sequence as a number parameter sequence.
7. The method for dynamically predicting load of closed-loop water cooling system according to claim 5, wherein the quantitative parameters are obtained according to independent uncorrelated conditions among the variable data of the equipment in the target variable data setThe specific formula of the independent uncorrelation degree of the target variable data set is as follows:
In the method, in the process of the invention, Representing quantity parameters/>Independent uncorrelation degrees of the target variable data sets; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the/>, in a set of target variable dataDevice variable data; /(I)Representing the inner product symbol; /(I)The representation takes absolute value.
8. The method for dynamically predicting load of closed-loop water cooling system according to claim 5, wherein the quantity parameter is obtained according to the ecological level parameter of the equipment variable data in the target variable data setThe specific formula of the punishment degree of the target variable data set is as follows:
In the method, in the process of the invention, Representing quantity parameters/>The penalty degree of the target variable data set; /(I)Representing the average value of the ecological level parameters of all the equipment variable data; /(I)Representing the/>, in a set of target variable dataThe ecological level parameters of the individual device variable data.
9. The method for dynamically predicting load of closed-loop water cooling system according to claim 5, wherein the quantitative parameters are obtained according to punishment degree and independent uncorrelation degreeThe specific method for outputting the target function of the target variable data set comprises the following steps:
parameter of quantity Independent degree of independence and quantity parameter/>, of a target variable data set of (a)The difference in penalty degree of the target variable data set as a quantity parameter/>An objective function output value of the objective variable data set.
10. The method for dynamically predicting the load of the closed circulation water cooling system according to claim 1, wherein the method for obtaining the reduced-dimension directions of the plurality of main components according to the independent variable data of all the devices comprises the following specific steps:
presetting a characteristic parameter Constructing a covariance matrix of the independent variable data of the equipment through the independent variable data of all the equipment, and constructing a covariance matrix of the variable data of the equipment through the variable data of all the equipment; taking a convolution result of the covariance matrix of the independent variable data of the equipment and the covariance matrix of the variable data of the equipment as a fusion matrix; decomposing the characteristic values of the fusion matrix in each characteristic direction by using a PCA principal component analysis method, acquiring the characteristic value of each characteristic direction, and marking a sequence formed by sequencing all the characteristic directions from large to small according to the characteristic values as a characteristic direction sequence; front in feature Direction sequence-The characteristic directions are all used as the main component directions after dimension reduction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410418995.7A CN118014776B (en) | 2024-04-09 | 2024-04-09 | Load dynamic prediction method for closed circulation water cooling system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410418995.7A CN118014776B (en) | 2024-04-09 | 2024-04-09 | Load dynamic prediction method for closed circulation water cooling system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118014776A true CN118014776A (en) | 2024-05-10 |
| CN118014776B CN118014776B (en) | 2024-07-05 |
Family
ID=90958078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410418995.7A Active CN118014776B (en) | 2024-04-09 | 2024-04-09 | Load dynamic prediction method for closed circulation water cooling system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118014776B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102282710A (en) * | 2009-01-13 | 2011-12-14 | 丰田自动车株式会社 | Fuel cell system |
| CN107022733A (en) * | 2016-02-02 | 2017-08-08 | 中国科学院上海应用物理研究所 | A kind of fused salt heat diffusion treatment equipment and its application |
| CN111178602A (en) * | 2019-12-18 | 2020-05-19 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Circulating water loss prediction method based on support vector machine and neural network |
| CN112703457A (en) * | 2018-05-07 | 2021-04-23 | 强力物联网投资组合2016有限公司 | Method and system for data collection, learning and machine signal streaming for analysis and maintenance using industrial internet of things |
| US20240085061A1 (en) * | 2021-02-07 | 2024-03-14 | Octopus Energy Heating Limited | Methods and systems for modulating energy usage |
-
2024
- 2024-04-09 CN CN202410418995.7A patent/CN118014776B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102282710A (en) * | 2009-01-13 | 2011-12-14 | 丰田自动车株式会社 | Fuel cell system |
| CN107022733A (en) * | 2016-02-02 | 2017-08-08 | 中国科学院上海应用物理研究所 | A kind of fused salt heat diffusion treatment equipment and its application |
| CN112703457A (en) * | 2018-05-07 | 2021-04-23 | 强力物联网投资组合2016有限公司 | Method and system for data collection, learning and machine signal streaming for analysis and maintenance using industrial internet of things |
| CN111178602A (en) * | 2019-12-18 | 2020-05-19 | 中国大唐集团科学技术研究院有限公司火力发电技术研究院 | Circulating water loss prediction method based on support vector machine and neural network |
| US20240085061A1 (en) * | 2021-02-07 | 2024-03-14 | Octopus Energy Heating Limited | Methods and systems for modulating energy usage |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118014776B (en) | 2024-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Remaining useful life prediction using a novel feature-attention-based end-to-end approach | |
| CN110414788B (en) | Electric energy quality prediction method based on similar days and improved LSTM | |
| CN108009667A (en) | A kind of energy demand total amount and structure prediction system | |
| Kebriaei et al. | Short-term load forecasting with a new nonsymmetric penalty function | |
| CN113128113A (en) | Poor information building load prediction method based on deep learning and transfer learning | |
| CN112257928A (en) | Short-term power load probability prediction method based on CNN and quantile regression | |
| CN113536670B (en) | Federal migration learning method for predicting residual life of group product | |
| CN111079989A (en) | Water supply company water supply amount prediction device based on DWT-PCA-LSTM | |
| CN112232575A (en) | Comprehensive energy system regulation and control method and device based on multivariate load prediction | |
| CN111178602A (en) | Circulating water loss prediction method based on support vector machine and neural network | |
| CN113326966A (en) | CEEMD-LSTM-based multi-load prediction method for comprehensive energy system | |
| CN114118592B (en) | Smart power grids power consumption end short-term energy consumption prediction system | |
| CN111652422A (en) | Heat supply system load prediction method, device and system based on building classification | |
| CN118014776B (en) | Load dynamic prediction method for closed circulation water cooling system | |
| CN116933216B (en) | Management system and method based on flexible load resource aggregation feature analysis | |
| CN113393119A (en) | Stepped hydropower short-term scheduling decision method based on scene reduction-deep learning | |
| CN117436920A (en) | Hybrid neural network day-ahead electricity price prediction method embedded with multi-head attention mechanism | |
| Kan et al. | Stock price prediction based on artificial neural network | |
| CN116957262A (en) | Optimization scheduling method of park comprehensive energy system based on three-process coupling | |
| CN113344410B (en) | Multi-dimensional evaluation method, system, terminal and storage medium for energy conservation transformation of power distribution network | |
| Wang et al. | A new method for short-term electricity load forecasting | |
| Zhao et al. | Energy System Planning and Intelligent Regulation and Optimization based on Artificial Intelligence | |
| Duong et al. | Application of Seasonal Trend Decomposition using Loess and Long Short-Term Memory in Peak Load Forecasting Model in Tien Giang | |
| Li et al. | Smoothed deep neural networks for marine sensor data prediction | |
| Nazarov et al. | Machine Learning Price Prediction on Green Building Prices |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |