CN113919559A - Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system - Google Patents
Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system Download PDFInfo
- Publication number
- CN113919559A CN113919559A CN202111147126.8A CN202111147126A CN113919559A CN 113919559 A CN113919559 A CN 113919559A CN 202111147126 A CN202111147126 A CN 202111147126A CN 113919559 A CN113919559 A CN 113919559A
- Authority
- CN
- China
- Prior art keywords
- parameters
- equipment
- sequence
- prediction
- 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.)
- Pending
Links
Images
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses a method and a device for ultra-short-term prediction of equipment parameters of an integrated energy system, which complete mechanism analysis and correlation analysis by building a historical database and provide a theoretical basis for selecting characteristic parameters in the process of building a prediction model; the method comprises the steps of carrying out data division on a historical database according to different equipment working conditions and calculating corresponding characteristic identification quantities of the historical database, and providing a judgment standard for pre-judging, searching and acquiring input data of a prediction model under subsequent working conditions; finally, ultra-short-term prediction of the prediction parameters is realized through similarity analysis and gray scale modeling, and prediction results of the prediction parameters in 0-4 hours in the future and with the time resolution of 5 minutes are provided. The method effectively improves the prediction precision and provides an effective reference basis for the optimal scheduling and economic operation of the regional comprehensive energy system.
Description
Technical Field
The invention relates to an ultra-short term prediction method and device for equipment parameters of an integrated energy system, and belongs to the technical field of integrated energy prediction.
Background
The comprehensive energy system is an important technical means for improving the energy utilization efficiency and reducing the energy utilization cost, and accurate supply and demand prediction is a key basis and an important precondition for economic planning, cooperative management and optimized scheduling of the comprehensive energy system. How to make accurate supply and demand prediction is the key point of ultra-short-term prediction of equipment parameters of the comprehensive energy system, and the ultra-short-term prediction of the equipment parameters needs to accurately predict the change condition of the equipment parameters within 0-4 hours in the future according to the internal change rule of the parameters and considering various influences of external factors on the parameter change.
At present, ultrashort-term prediction research aiming at equipment parameters of an integrated energy system is greatly developed in the aspect of basic theory, certain achievements are obtained in practical engineering application, the method can be divided into two types under the longitudinal view, one type is a prediction method based on a mechanism model, the method depends on mathematical description of equipment, the physical significance is clear, the functional relation is clear, the prediction process has stronger extrapolation performance, the difficulty of mathematical description is sharply increased along with the increase of the complexity of the mechanism of the equipment, and the feasibility of prediction implementation and the reliability of the prediction result are also sharply reduced. The other type is a prediction method based on a data model, and the method is characterized in that the incidence relation among variables is extracted on the basis of data samples, the development period is short, the calculation demand is small, the comprehensive cost of prediction solution is relatively low, but similarly, the method is difficult to explain the decision basis and the operation logic, and the uncertainty of the reliability of the prediction result and the hidden safety risk limit the popularization and the application of the prediction method in the industrial field.
Overall, the comprehensive energy system devices are of various types, and the operating rules of different devices have fluctuation in different time-space spans, and the uncertainty of the influence of various external factors makes the prediction algorithm based on a single model of a mechanism model or a data model poor in effect.
Based on the method, the invention provides the ultra-short term prediction method and the device of the equipment parameters of the comprehensive energy system, in order to improve the overall performance of the ultra-short term prediction method of the comprehensive energy system equipment and enhance the application effect of the method.
Disclosure of Invention
The purpose is as follows: in order to overcome the defects in the prior art, the invention provides the ultra-short-term prediction method and the device for the equipment parameters of the comprehensive energy system, which finish mechanism analysis and association degree analysis by building a historical database and provide a theoretical basis for selecting the characteristic parameters in the process of building a prediction model; the method comprises the steps of carrying out data division on a historical database according to different equipment working conditions and calculating corresponding characteristic identification quantities of the historical database, and providing a judgment standard for pre-judging, searching and acquiring input data of a prediction model under subsequent working conditions; finally, ultra-short-term prediction of the prediction parameters is realized through similarity analysis and gray scale modeling, and prediction results of the prediction parameters in 0-4 hours in the future and with the time resolution of 5 minutes are provided.
The technical scheme is as follows: in order to solve the technical problems, the technical scheme adopted by the invention is as follows:
the invention provides an ultra-short-term prediction method for equipment parameters of a comprehensive energy system, which comprises the following steps of:
and acquiring historical time sequence data of equipment parameters including prediction parameters of the comprehensive energy system equipment, and constructing a historical database.
And selecting characterization parameters related to the prediction parameters from the equipment parameters in the historical database according to the mechanism analysis of the equipment.
And performing correlation analysis on the characterization parameters, and extracting characteristic parameters from the characterization parameters.
And calculating the characteristic identification quantity under different operation condition types according to the characteristic parameters.
And constructing an online prediction model of the comprehensive energy system equipment.
Determining the type of the operating condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameter which is most similar to the original basic sequence of the characteristic parameter at the current moment from the corresponding historical database as the input of an online prediction model of the equipment of the comprehensive energy system, and solving the online prediction model of the equipment of the comprehensive energy system to obtain the predicted value of the equipment parameter of the equipment to be predicted.
An ultra-short term prediction device for equipment parameters of an integrated energy system comprises the following modules:
and the data acquisition module is used for acquiring historical time sequence data of equipment parameters including the prediction parameters of the comprehensive energy system equipment and constructing a historical database.
And the mechanism analysis module is used for selecting the characterization parameters related to the prediction parameters from the equipment parameters of the historical database according to the mechanism analysis of the equipment.
And the correlation analysis module is used for performing correlation analysis on the characterization parameters and extracting the characteristic parameters from the characterization parameters.
And the working condition analysis module is used for calculating the characteristic identification quantity under different operation working condition types according to the characteristic parameters.
And the online prediction model building module is used for building an online prediction model of the comprehensive energy system equipment.
And the real-time prediction module is used for determining the type of the operation condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameters which are most similar to the original basic sequence of the characteristic parameters at the current moment from the corresponding historical database as the input of the online prediction model of the comprehensive energy system equipment, and solving the online prediction model of the comprehensive energy system equipment to obtain the predicted value of the equipment parameters of the equipment to be predicted.
Preferably, the method further comprises the following steps: when a dead pixel exists in the historical database, the dead pixel is taken as a center, normal data before and after the dead pixel is selected according to the time sequence for curve fitting, the numerical value of the dead pixel is calculated according to the fitting function interpolation, and the numerical value is taken as the correction value of the dead pixel.
As a preferred scheme, the correlation analysis of the characterization parameters and the extraction of the characterization parameters from the characterization parameters include the following steps:
1) obtaining prediction parameter x from historical database0And a comparison sequence of p characterization parameters,
x0={x0(1),x0(2),…x0(i)…,x0(n)}
in the formula, x0As a reference sequence, x0(i) To predict the ith relevant parameter of the parameter,for the comparison sequence of the ith characterizing parameter,the j-th related parameter of the ith comparison sequence is shown, n is the number of elements contained in the sequence, and p is the number of the characterization parameters.
2) And performing dimensionless initialization processing on the reference sequence and the comparison sequence to obtain the dimensionless initialized reference sequence and comparison sequence.
x'0={1,x0(2)/x0(1),…,x0(n)/x0(1)}
3) Calculating a reference sequence x 'after dimension initialization'0And comparing the sequencesGrey correlation coefficient on the kth correlation parameter:
in the formula, eta is a resolution coefficient,the minimum difference of the two levels is represented,representing the two-step maximum difference.
4) Calculating gray correlation degree gamma of each characterization parameteri。
5) Setting a grey correlation threshold gamma0Extracting { gamma ] therefromi|γi>γ0The characterization parameters of i ═ 1,2, …, p } form a new characteristic parameter sequence x1,x2,…,xqWherein q is the number of characteristic parameters, and q is less than or equal to p.
As a preferred scheme, the calculating the feature identification quantity under different operation condition categories according to the feature parameters includes the following steps:
and dividing the working condition type according to the operating working condition of the equipment, and dividing the historical data according to the working condition type.
And calculating the characteristic identification quantity corresponding to the working condition category according to the divided historical data. The method comprises the following steps:
1) establishing a characteristic parameter x1,x2,…,xqFor the prediction parameter x0Is a regression model
In the formula (I), the compound is shown in the specification,in order to be the regression coefficient, the method,is an object estimate calculated based on a regression model.
2) Each regression coefficient was obtained from the following equation
In the formula (I), the compound is shown in the specification,are respectively x0,x1,x2,…,xqT, R are the dispersion matrix and regression matrix of the regression model, respectively; is provided with
In the formula, TijRepresenting the elements of the ith row and jth column of the matrix T, RiRepresenting the ith element in the matrix R.
3) Characterization of the prediction parameter x by λ0And a characteristic parameter x1,x2,…,xqBulk phase under this regimeThe degree of closeness is calculated by the formula:
and taking the lambda as the characteristic identification quantity corresponding to each group of working condition data.
As a preferred scheme, the building of the online prediction model of the integrated energy system device comprises the following steps:
1) when the prediction period is reached, the current moment is taken as a starting point, and the parameter x is simultaneously measured0,x1,x2,...,xqTracing m historical data as basic original sequences of online modeling:
in the formula, α is a background value.
3) The gray scale degree prediction model is established as
Wherein a is the coefficient of development, biAs a drive factor, h1(k-1) is a linearity correction amount; h is2The grey effect is indicated.
4) Gray scale prediction model can be converted to
5) performing one-time subtraction reduction operation on the above formula to obtain an online prediction model of the prediction parameters:
preferably, the parameters a and bi、h1(k-1)、h2The specific calculation expression of (2) is:
U=(BTB)-1BTY
in the formula, BTA transposed matrix representing the matrix B, (B)TB)-1Representation matrix BTInverse matrix of B
The method for determining the operation condition type of the equipment to be predicted comprises the following steps:
by means of the original base sequence, it is possible to,calculating a prediction parameter x0And a characteristic parameter x1,x2,...,xqThe real-time identification number lambda at the current time.
And determining the type of the operating condition of the equipment to be predicted according to the real-time identification quantity lambda and the judgment condition of the current operating condition of the equipment. The determination condition is thatAnd taking the value of the next L, wherein L is the total working condition classification number.
The method comprises the following steps of selecting an expansion sequence of an original basic sequence of the characteristic parameter most similar to the current moment from a corresponding historical database as the input of an online prediction model of the comprehensive energy system equipment, solving the online prediction model of the comprehensive energy system equipment to obtain an equipment parameter prediction value of the equipment to be predicted, and comprises the following steps:
1) in the corresponding historical working condition data sequence, starting from the first historical data of the prediction parameters, selecting m data points to form a historical sequencePassing and predicting measured sequences of parametersCarrying out similarity judgment, and finding out an initial sequence number point s of a similar sequence meeting the following formula in the historical data sequence;
in the formula (I), the compound is shown in the specification,data of the (s + k-1) th point in the prediction parameter history sequence from the(s) th point to the (s + m-1) th point;
In the formula (I), the compound is shown in the specification,the method comprises the steps that data of (s + m + k-1) th points in a characteristic parameter historical sequence from (s + m) th points to (s + m + f-1) th points are obtained, k is the number of points to be predicted, and f is the total number of predicted points;
will be provided withIncorporating original base sequences of feature parametersIn the method, an extended sequence of the original basic sequence of the characteristic parameters is obtained
To pairPerforming one-time accumulation to generate (1-AGO) to obtain an input sequence of the prediction model
4) Will be provided withSubstituting the following formula to obtain the predicted value of the device prediction parameter at f points in the future as:
has the advantages that: compared with the prior art, the ultra-short-term prediction method and the ultra-short-term prediction device for the equipment parameters of the comprehensive energy system have the following beneficial effects that: the method integrates the theoretical accuracy of a mechanism model and the strong generalization deepening of a data driving model, and finally establishes an online gray scale degree ultra-short-term prediction model by means of correlation analysis, working condition data division and similarity principles, so that ultra-short-term prediction of specific parameters of the integrated energy system equipment under the multi-factor coupling condition is realized, the prediction precision is effectively improved, and an effective reference basis is provided for optimal scheduling and economic operation of the regional integrated energy system.
Drawings
FIG. 1 is a flow chart of the steps of the method of the present invention.
FIG. 2 is a schematic view of the structure of the apparatus of the present invention.
Detailed Description
The present invention will be further described with reference to the following examples.
As shown in fig. 1, in a first aspect, the present invention provides a method for ultra-short term prediction of equipment parameters of an integrated energy system, including the following steps:
and acquiring historical time sequence data of equipment parameters including prediction parameters of the comprehensive energy system equipment, and constructing a historical database.
And selecting characterization parameters related to the prediction parameters from the equipment parameters in the historical database according to the mechanism analysis of the equipment.
And performing correlation analysis on the characterization parameters, and extracting characteristic parameters from the characterization parameters.
And calculating the characteristic identification quantity under different operation condition types according to the characteristic parameters.
And constructing an online prediction model of the comprehensive energy system equipment.
Determining the type of the operating condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameter which is most similar to the original basic sequence of the characteristic parameter at the current moment from the corresponding historical database as the input of an online prediction model of the equipment of the comprehensive energy system, and solving the online prediction model of the equipment of the comprehensive energy system to obtain the predicted value of the equipment parameter of the equipment to be predicted.
When the dead pixel in the historical database comprises data missing and data gross error, data correction is carried out by adopting a curve fitting method, specifically, the dead pixel is taken as a center, normal data before and after the dead pixel is selected according to the time sequence for curve fitting, the numerical value of the dead pixel is interpolated according to a fitting function and is taken as the correction value of the dead pixel.
The mechanism analysis of the equipment is to utilize a mechanism model with definite physical or practical significance to search the change rule of the prediction parameters along with the characterization parameters, taking a generator set as an example, and the mechanism model related in the operation process of the equipment comprises the following steps:
1) equation of energy balance
2) Equation of mass balance
3) Equation of working medium properties
f(h,p,T)=0
4) Boundary constraint equation
f(λ1,λ2,λ3,…)=0
In the formula: ρ represents density, c represents specific heat capacity, V represents volume, T represents temperature, τ represents time, q represents specific heat capacityinRepresenting input energy, qoutRepresenting output energy, DinDenotes the inlet flow, DoutDenotes the outlet flow, h denotes the enthalpy, P denotes the pressure, lambda1、λ2、λ3Representing the constraint.
Therefore, when the unit output energy is taken as a prediction parameter, the related characterization parameters can be selected to be temperature, pressure and flow. When the characteristic parameters are determined, characteristic parameters are extracted from the characteristic parameters through correlation analysis.
The correlation analysis of the characterization parameters and the extraction of the characteristic parameters from the characterization parameters comprise the following steps:
1) obtaining prediction parameter x from historical database0And a comparison sequence of p characterization parameters,
x0={x0(1),x0(2),…x0(i)…,x0(n)}
in the formula, x0As a reference sequence, x0(i) To predict the ith relevant parameter of the parameter,for the comparison sequence of the ith characterizing parameter,the j-th related parameter of the ith comparison sequence is shown, n is the number of elements contained in the sequence, and p is the number of the characterization parameters.
2) Further, performing dimensionless initialization processing on the reference sequence and the comparison sequence to obtain the dimensionless initialized reference sequence and comparison sequence.
x′0={1,x0(2)/x0(1),…,x0(n)/x0(1)}
3) Calculating a reference sequence x 'after dimension initialization'0And comparing the sequencesGrey correlation coefficient on the kth correlation parameter:
wherein eta is a resolution coefficient, eta belongs to [0, 1], and when eta is less than or equal to 0.55, the resolution capability is optimal.
The minimum difference of the two levels is represented,representing the two-step maximum difference.
4) Further, the gray correlation degree gamma of each characterization parameter is calculatedi。
5) Setting a grey correlation threshold gamma00.6, { γ ] extracted therefromi|γi>γ0,i=1,2, …, p } to form a new characteristic parameter sequence x1,x2,...,xqWherein q is the number of characteristic parameters, and q is less than or equal to p.
The method for calculating the characteristic identification quantity under different operation condition categories according to the characteristic parameters comprises the following steps:
eight working condition categories of a full state, a 90% state, a 70% state, a 50% state, a 30% state, a 10% state, an ascending state and a descending state are divided according to the running working conditions of the equipment, and historical data are divided according to the working condition categories.
And calculating the characteristic identification quantity corresponding to the working condition category according to the divided historical data. The method comprises the following steps:
1) establishing a characteristic parameter x1,x2,...,xqFor the prediction parameter x0Is a regression model
In the formula (I), the compound is shown in the specification,in order to be the regression coefficient, the method,is an object estimate calculated based on a regression model.
2) Further, each regression coefficient can be obtained by the following formula
In the formula (I), the compound is shown in the specification,are respectively x0,x1,x2,...,xqT, R are the dispersion matrix and regression matrix of the regression model, respectively, having
In the formula, TijRepresenting the elements of the ith row and jth column of the matrix T, RiRepresenting the ith element in the matrix R.
3) Further, the prediction parameter x is characterized by λ0And a characteristic parameter x1,x2,...,xqThe overall correlation degree under the working condition is calculated by the formula:
therefore, lambda is used as a characteristic identification quantity corresponding to each group of working condition data, and when x is acquired0,x1,x2,...,xqWhen the data is actually measured, the real-time identification quantity of the data can be calculated and compared with the characteristic identification quantity, so that the current working condition state of the equipment is judged in advance, and a basis for predicting sample selection is further provided for implementation of real-time prediction.
The method for constructing the online prediction model of the comprehensive energy system equipment comprises the following steps:
when the prediction period is reached, the current moment is taken as a starting point, and the parameter x is simultaneously measured0,x1,x2,...,xqTracing m pieces of historical data as basic original sequences of online modeling, and establishing a prediction model:
1) basic original sequence
in the formula, α is a background value and is 0.5 as a default.
3) Further, a gray scale degree prediction model is established as
Wherein a is the coefficient of development, biAs a drive factor, h1(k-1) is a linearity correction amount; h is2The grey effect is indicated. The parameters a, bi、h1(k-1)、h2The specific calculation expression of (2) is:
U=(BTB)-1BTY
in the formula, BTA transposed matrix representing the matrix B, (B)TB)-1Representation matrix BTInverse matrix of B
4) Further, the gray scale prediction model may be converted to
5) performing one-time subtraction reduction operation on the above formula to obtain an online prediction model of the prediction parameters:
the method for determining the operation condition type of the equipment to be predicted comprises the following steps:
by means of the original base sequence, it is possible to,calculating a prediction parameter x0And a characteristic parameter x1,x2,...,xqThe real-time identification number lambda at the current time.
And determining the type of the operating condition of the equipment to be predicted according to the real-time identification quantity lambda and the judgment condition of the current operating condition of the equipment. The determination condition is thatL is 1,2, …, and L is the value of L, wherein L is the total number of the working condition classifications.
The method comprises the following steps of selecting an expansion sequence of an original basic sequence of the characteristic parameter most similar to the current moment from a corresponding historical database as the input of an online prediction model of the comprehensive energy system equipment, solving the online prediction model of the comprehensive energy system equipment to obtain an equipment parameter prediction value of the equipment to be predicted, and comprises the following steps:
1) in the corresponding historical working condition data sequence, starting from the first historical data of the prediction parameters, selecting m data points to form a historical sequencePassing and predicting measured sequences of parametersCarrying out similarity judgment, and finding out an initial sequence number point s of a similar sequence meeting the following formula in the historical data sequence;
in the formula (I), the compound is shown in the specification,data of the (s + k-1) th point in the prediction parameter history sequence from the(s) th point to the (s + m-1) th point;
In the formula (I), the compound is shown in the specification,the method comprises the steps that data of (s + m + k-1) th points in a characteristic parameter historical sequence from (s + m) th points to (s + m + f-1) th points are obtained, k is the number of points to be predicted, and f is the total number of predicted points;
will be provided withIncorporating original base sequences of feature parametersIn the method, an extended sequence of the original basic sequence of the characteristic parameters is obtained
To pairPerforming one-time accumulation to generate (1-AGO) to obtain an input sequence of the prediction model
4) Will be provided withSubstituting the following formula to obtain the predicted value of the device prediction parameter at f points in the future as:
as shown in fig. 2, in a second aspect, the present invention provides an ultra-short term prediction apparatus for equipment parameters of an integrated energy system, including the following modules:
and the data acquisition module is used for acquiring historical time sequence data of equipment parameters including the prediction parameters of the comprehensive energy system equipment and constructing a historical database.
And the mechanism analysis module is used for selecting the characterization parameters related to the prediction parameters from the equipment parameters of the historical database according to the mechanism analysis of the equipment.
And the correlation analysis module is used for performing correlation analysis on the characterization parameters and extracting the characteristic parameters from the characterization parameters.
And the working condition analysis module is used for calculating the characteristic identification quantity under different operation working condition types according to the characteristic parameters.
And the online prediction model building module is used for building an online prediction model of the comprehensive energy system equipment.
And the real-time prediction module is used for determining the type of the operation condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameters which are most similar to the original basic sequence of the characteristic parameters at the current moment from the corresponding historical database as the input of the online prediction model of the comprehensive energy system equipment, and solving the online prediction model of the comprehensive energy system equipment to obtain the predicted value of the equipment parameters of the equipment to be predicted.
Example 1:
taking a blast furnace in energy equipment in an integrated energy system as an example, the method for predicting the parameters of the gas generation amount of the blast furnace comprises the following specific steps:
step 1, collecting historical time sequence data of parameters of a blast furnace of the comprehensive energy system, and constructing a historical database.
Acquiring historical time sequence data of all equipment parameters including gas generation amount in blast furnace equipment of the comprehensive energy system, and correcting data of possible dead spots in the historical time sequence data by adopting a curve fitting method, wherein the dead spots comprise data loss or data gross error, and the specific process comprises the following steps: and taking the dead pixel as a center, selecting normal data before and after the dead pixel according to the time sequence to perform curve fitting, interpolating and calculating the numerical value of the dead pixel according to a fitting function, and taking the numerical value as a correction value of the dead pixel.
And 2, determining characterization parameters related to the blast furnace gas generation amount based on the mechanism analysis of the blast furnace.
Reasonable characterization parameters are selected according to the analysis result of the blast furnace gas generation mechanism, in the blast furnace smelting process, materials such as iron ore, coke, limestone and the like are loaded from the top of the blast furnace, and are subjected to reduction reaction with preheated air (mixed rich oxygen and coal powder) in the blast furnace, and finally blast furnace gas is discharged from the top of the blast furnace. Therefore, when the parameters of the gas generation amount need to be predicted, the related characteristic parameters can be selected from coke ratio, coal ratio, oxygen enrichment, air quantity, air temperature, air pressure and pressure difference in the furnace.
And 3, performing correlation analysis on the characteristic parameters, and extracting the characteristic parameters related to the blast furnace gas generation amount.
After the characteristic parameters are determined, extracting characteristic parameters from the characteristic parameters through correlation analysis to serve as a basis for calculating the characteristic identification quantity of the subsequent working conditions and establishing a prediction model, and the specific implementation steps are as follows:
1) and obtaining a reference sequence of gas generation amount, a comparison sequence of coke ratio, coal ratio, oxygen enrichment, air quantity, air temperature, air pressure and pressure difference in the furnace from a historical database.
x0={x0(1),x0(2),…x0(i)…,x0(n)}
In the formula, x0As a reference sequence, x0(i) Is the ith relevant parameter of the gas generation amount,for the comparison sequence of the ith characterizing parameter,is the j-th related parameter of the ith comparison sequence, n is the number of elements contained in the sequence, and p is a characterization parameterAnd (4) the number.
2) Further, performing dimensionless initialization processing on the reference sequence and the comparison sequence to obtain the dimensionless initialized reference sequence and comparison sequence.
x′0={1,x0(2)/x0(1),…,x0(n)/x0(1)}
3) Calculating a reference sequence x 'after dimension initialization'0And comparing the sequencesGrey correlation coefficient on the kth correlation parameter:
wherein eta is a resolution coefficient, eta belongs to [0, 1], and when eta is less than or equal to 0.55, the resolution capability is optimal.
The minimum difference of the two levels is represented,representing the two-step maximum difference.
4) Further, the gray correlation degree gamma of each characterization parameter is calculatedi。
5) Setting a grey correlation threshold gamma00.6, { γ ] extracted therefromi|γi>γ0The characterization parameters of i ═ 1,2, …, p } form a new characteristic parameter sequence x1,x2,...,xqWherein, in the step (A),q is the number of characteristic parameters, and q is less than or equal to p.
In the gray correlation degree obtained by calculation, the air quantity is greater than the coke ratio, greater than the air pressure, greater than the coal ratio, greater than the air temperature, greater than 0.6, greater than the oxygen enrichment and greater than the pressure difference in the blast furnace, so that the air quantity, the coke ratio, the air pressure, the coal ratio and the air temperature are taken as characteristic parameters for predicting the blast furnace gas generation amount.
And 4, calculating the characteristic identification quantity of the blast furnace gas generation quantity under different working condition types.
After the characteristic parameters are determined, according to the operation conditions of the blast furnace equipment, dividing the operation conditions into eight operation condition categories of full load, 90% load, 70% load, 50% load, 30% load, 10% load, load increase and load decrease, dividing the historical data of the blast furnace gas generation amount and the characteristic parameters according to the operation condition categories, and calculating corresponding characteristic identification amounts, wherein the implementation steps comprise:
1) establishing the air quantity, coke ratio, air pressure, coal ratio and air temperature to the coal gas generation quantity x0Is a regression model
In the formula (I), the compound is shown in the specification,in order to be the regression coefficient, the method,for prediction parameter estimation based on regression model calculation, x1,x2,...,xqThe characteristic parameters of the air quantity, the coke ratio, the air pressure, the coal ratio and the air temperature are q-5.
2) Further, each regression coefficient can be obtained by the following formula
In the formula (I), the compound is shown in the specification,are respectively x0,x1,x2,...,xqT, R are the dispersion matrix and regression matrix of the regression model, respectively, having
In the formula, TijRepresenting the elements of the ith row and jth column of the matrix T, RiRepresenting the ith element in the matrix R.
3) Further, lambda is used for representing the overall correlation degree of the gas generation quantity and the air quantity, the coke ratio, the air pressure, the coal ratio and the air temperature under the working condition, and the calculation formula is as follows:
therefore, the step is repeated to calculate the lambda value of each group of working condition data, the lambda value is used as the characteristic identification quantity corresponding to each working condition, when the actual measurement data of the gas generation quantity, the air quantity, the coke ratio, the air pressure, the coal ratio and the air temperature is obtained, the actual characteristic identification quantity can be calculated and compared with the characteristic identification quantity, so that the current working condition state of the blast furnace is judged, and the basis for selecting the prediction sample is further provided for the implementation of the actual prediction.
Step 5, establishing an online prediction model
When the prediction period is reached, the current moment is taken as a starting point, and m historical data of gas generation amount, air volume, coke ratio, air pressure, coal ratio and air temperature are traced back to serve as basic original sequences of online modeling, wherein the process of establishing the prediction model comprises the following steps:
1) obtaining basic original sequences of parameters of gas generation quantity, coke ratio, coal ratio, air quantity, air pressure and air temperature
in the formula, α is a background value and is 0.5 as a default.
3) Further, a gray scale degree prediction model is established as
Wherein a is the coefficient of development, biAs a drive factor, h1(k-1) is a linearity correction amount; h is2The grey effect is indicated. The parameters a, bi、h1(k-1)、h2The specific calculation expression of (2) is:
U=(BTB)-1BTY
in the formula, BTA transposed matrix representing the matrix B, (B)TB)-1Representing momentsArray BTInverse matrix of B
4) Further, the gray scale prediction model may be converted to
5) performing a subtraction reduction operation on the above formula to obtain a gray predicted value of the blast furnace gas generation amount as follows:
step 6, input prediction and output prediction
With the original basic sequence in step 5,calculating the real-time identification quantity lambda of the gas generation quantity, the air quantity, the coke ratio, the air pressure, the coal ratio and the air temperature at the current moment based on the step 4, and further judging the current working condition of the equipment to meet the requirementAnd taking the value of the next L, wherein L is the total working condition classification number.
After the current working condition state of the equipment is determined, an expansion sequence of the original basic sequence of the characteristic parameters most similar to the current moment is required to be searched from the corresponding historical working condition data sequence to be used as the input of a prediction model, and the specific implementation process is as follows:
1) in the corresponding historical working condition data sequence, starting from the first historical data of the prediction parameters, selecting m data point groupsInto a historical sequencePassing and predicting measured sequences of parametersCarrying out similarity judgment, and finding out an initial sequence number point s of a similar sequence meeting the following formula in the historical data sequence;
in the formula (I), the compound is shown in the specification,data of the (s + k-1) th point in the prediction parameter history sequence from the(s) th point to the (s + m-1) th point;
In the formula (I), the compound is shown in the specification,the method comprises the steps that data of (s + m + k-1) th points in a characteristic parameter historical sequence from (s + m) th points to (s + m + f-1) th points are obtained, k is the number of points to be predicted, and f is the total number of predicted points;
will be provided withIncorporating original base sequences of feature parametersIn the method, an extended sequence of the original basic sequence of the characteristic parameters is obtained
To pairPerforming one-time accumulation to generate (1-AGO) to obtain an input sequence of the prediction model
4) Will be provided withSubstituting the following formula to finally obtain the predicted value of the total f points of the predicted parameters at 0-4 hours in the future and with the time resolution of 5 minutes as follows:
example 2:
the invention provides an implementation method and system for ultra-short-term prediction of comprehensive energy system equipment, wherein a historical database is constructed, characterization parameters are obtained based on mechanism analysis for correlation analysis and characteristic parameters are extracted, working condition data are divided by combining with the operating conditions of the energy equipment, a sample space is built, then an online gray scale ultra-short-term prediction model is built by using the actual measurement data of the prediction parameters and the characteristic parameters, the input of the online prediction model is obtained by pre-judging the real-time working conditions, and finally the ultra-short-term prediction of the prediction parameters is realized. The model gives consideration to the theoretical accuracy of the mechanism model and the strong generalization and deepening of the data driving model, can realize the ultra-short-term prediction of the comprehensive energy equipment under the multi-factor coupling condition, effectively improves the prediction precision, and provides an effective reference basis for the optimal scheduling and the economic operation of the regional comprehensive energy system.
The above description is only of the preferred embodiments of the present invention, and it should be noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the invention and these are intended to be within the scope of the invention.
Claims (9)
1. An ultra-short-term prediction method for equipment parameters of an integrated energy system is characterized by comprising the following steps: the method comprises the following steps:
acquiring historical time sequence data of equipment parameters including prediction parameters of the comprehensive energy system equipment, and constructing a historical database;
selecting a characterization parameter related to a prediction parameter from equipment parameters of a historical database according to mechanism analysis on the equipment;
performing correlation analysis on the characterization parameters, and extracting characteristic parameters from the characterization parameters;
calculating characteristic identification quantities under different operation condition categories according to the characteristic parameters;
constructing an online prediction model of the comprehensive energy system equipment;
determining the type of the operating condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameter which is most similar to the original basic sequence of the characteristic parameter at the current moment from the corresponding historical database as the input of an online prediction model of the equipment of the comprehensive energy system, and solving the online prediction model of the equipment of the comprehensive energy system to obtain the predicted value of the equipment parameter of the equipment to be predicted.
2. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 1, characterized in that: further comprising: when a dead pixel exists in the historical database, the dead pixel is taken as a center, normal data before and after the dead pixel is selected according to the time sequence for curve fitting, the numerical value of the dead pixel is calculated according to the fitting function interpolation, and the numerical value is taken as the correction value of the dead pixel.
3. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 1, characterized in that: the correlation analysis of the characterization parameters and the extraction of the characteristic parameters from the characterization parameters comprise the following steps:
1) obtaining prediction parameter x from historical database0And a comparison sequence of p characterization parameters,
x0={x0(1),x0(2),…x0(i)…,x0(n)}
in the formula, x0As a reference sequence, x0(i) To predict the ith relevant parameter of the parameter,for the comparison sequence of the ith characterizing parameter,for the jth correlation parameter of the ith comparison sequence,n is the number of elements contained in the sequence, and p is the number of characterization parameters;
2) carrying out dimensionless initialization processing on the reference sequence and the comparison sequence to obtain a dimensionless initialized reference sequence and a comparison sequence;
x'0={1,x0(2)/x0(1),…,x0(n)/x0(1)}
3) calculating a reference sequence x 'after dimension initialization'0And comparing the sequencesGrey correlation coefficient on the kth correlation parameter:
in the formula, eta is a resolution coefficient,the minimum difference of the two levels is represented,represents the two-step maximum difference;
4) calculating gray correlation degree gamma of each characterization parameteri;
5) Setting a grey correlation threshold gamma0Extracting { gamma ] therefromi|γi>γ0The characterization parameters of i ═ 1,2, …, p } form a new characteristic parameter sequence x1,x2,…,xqWherein q is the number of characteristic parameters,q≤p。
4. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 3, characterized in that: the method for calculating the characteristic identification quantity under different operation condition categories according to the characteristic parameters comprises the following steps:
dividing working condition types according to the operating working conditions of the equipment, and dividing historical data according to the working condition types;
calculating the characteristic identification quantity corresponding to the working condition category according to the divided historical data; the method comprises the following steps:
1) establishing a characteristic parameter x1,x2,…,xqFor the prediction parameter x0Is a regression model
In the formula (I), the compound is shown in the specification,in order to be the regression coefficient, the method,is an object estimation value calculated based on a regression model;
2) each regression coefficient was obtained from the following equation
In the formula (I), the compound is shown in the specification,are respectively x0,x1,x2,…,xqT, R are the dispersion matrix and regression matrix of the regression model, respectively; is provided with
In the formula, TijRepresenting the elements of the ith row and jth column of the matrix T, RiRepresents the ith element in the matrix R;
3) characterization of the prediction parameter x by λ0And a characteristic parameter x1,x2,…,xqThe overall correlation degree under the working condition is calculated by the formula:
and taking the lambda as the characteristic identification quantity corresponding to each group of working condition data.
5. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 1, characterized in that: the method for constructing the online prediction model of the comprehensive energy system equipment comprises the following steps:
1) when the prediction period is reached, the current moment is taken as a starting point, and the parameter x is simultaneously measured0,x1,x2,…,xqTracing m historical data as basic original sequences of online modeling:
wherein, alpha is a background value;
3) the gray scale degree prediction model is established as
Wherein a is the coefficient of development, biAs a drive factor, h1(k-1) is a linearity correction amount; h is2The amount of gray effect;
4) gray scale prediction model can be converted to
5) performing one-time subtraction reduction operation on the above formula to obtain an online prediction model of the prediction parameters:
6. the ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 5, characterized in that: the parameters a, bi、h1(k-1)、h2The specific calculation expression of (2) is:
U=(BTB)-1BTY
in the formula, BTA transposed matrix representing the matrix B, (B)TB)-1Representation matrix BTInverse matrix of B
7. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 4, characterized in that: the method for determining the operation condition type of the equipment to be predicted comprises the following steps:
by means of the original base sequence, it is possible to,calculating a prediction parameter x0And a characteristic parameter x1,x2,…,xqReal-time identification quantity lambda at the current moment;
determining the type of the operating condition of the equipment to be predicted according to the real-time identification quantity lambda and the judgment condition of the current operating condition of the equipment; the determination condition is thatAnd taking the value of the next L, wherein L is the total working condition classification number.
8. The ultra-short term prediction method for the equipment parameters of the integrated energy system according to claim 5, characterized in that: the method comprises the following steps of selecting an expansion sequence of an original basic sequence of the characteristic parameter most similar to the current moment from a corresponding historical database as the input of an online prediction model of the comprehensive energy system equipment, solving the online prediction model of the comprehensive energy system equipment to obtain an equipment parameter prediction value of the equipment to be predicted, and comprises the following steps:
1) in the corresponding historical working condition data sequence, starting from the first historical data of the prediction parameters, selecting m data points to form a historical sequencePassing and predicting measured sequences of parametersCarrying out similarity judgment, and finding out an initial sequence number point s of a similar sequence meeting the following formula in the historical data sequence;
in the formula (I), the compound is shown in the specification,data of the (s + k-1) th point in the prediction parameter history sequence from the(s) th point to the (s + m-1) th point;
In the formula (I), the compound is shown in the specification,starting from the s + m th point to the s + m th pointData of an s + m + k-1 point in the characteristic parameter historical sequence ending at the s + m + f-1 point, wherein k is the number of points to be predicted, and f is the total number of predicted points;
will be provided withIncorporating original base sequences of feature parametersIn the method, an extended sequence of the original basic sequence of the characteristic parameters is obtained
To pairPerforming one-time accumulation to generate (1-AGO) to obtain an input sequence of the prediction model
4) Will be provided withSubstituting the following formula to obtain the predicted value of the device prediction parameter at f points in the future as:
9. an ultra-short term prediction device for equipment parameters of an integrated energy system is characterized in that: the system comprises the following modules:
the data acquisition module is used for acquiring historical time sequence data of equipment parameters including prediction parameters of the comprehensive energy system equipment and constructing a historical database;
the mechanism analysis module is used for selecting a characterization parameter related to the prediction parameter from the equipment parameters of the historical database according to the mechanism analysis of the equipment;
the correlation analysis module is used for performing correlation analysis on the characterization parameters and extracting characteristic parameters from the characterization parameters;
the working condition analysis module is used for calculating characteristic identification quantities under different operation working condition types according to the characteristic parameters;
the online prediction model building module is used for building an online prediction model of the comprehensive energy system equipment;
and the real-time prediction module is used for determining the type of the operation condition of the equipment to be predicted, selecting an extended sequence of the original basic sequence of the characteristic parameters which are most similar to the original basic sequence of the characteristic parameters at the current moment from the corresponding historical database as the input of the online prediction model of the comprehensive energy system equipment, and solving the online prediction model of the comprehensive energy system equipment to obtain the predicted value of the equipment parameters of the equipment to be predicted.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111147126.8A CN113919559A (en) | 2021-09-28 | 2021-09-28 | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111147126.8A CN113919559A (en) | 2021-09-28 | 2021-09-28 | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113919559A true CN113919559A (en) | 2022-01-11 |
Family
ID=79236872
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111147126.8A Pending CN113919559A (en) | 2021-09-28 | 2021-09-28 | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113919559A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115374572A (en) * | 2022-10-21 | 2022-11-22 | 南京安全无忧网络科技有限公司 | Process stability analysis system and method |
CN116737510A (en) * | 2023-08-08 | 2023-09-12 | 深圳阿比特科技有限公司 | Data analysis-based intelligent keyboard monitoring method and system |
CN116957367A (en) * | 2023-09-21 | 2023-10-27 | 南昌大学 | Parameter multi-time scale prediction method and system for comprehensive energy system operation strategy |
-
2021
- 2021-09-28 CN CN202111147126.8A patent/CN113919559A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115374572A (en) * | 2022-10-21 | 2022-11-22 | 南京安全无忧网络科技有限公司 | Process stability analysis system and method |
CN116737510A (en) * | 2023-08-08 | 2023-09-12 | 深圳阿比特科技有限公司 | Data analysis-based intelligent keyboard monitoring method and system |
CN116737510B (en) * | 2023-08-08 | 2023-11-28 | 深圳阿比特科技有限公司 | Data analysis-based intelligent keyboard monitoring method and system |
CN116957367A (en) * | 2023-09-21 | 2023-10-27 | 南昌大学 | Parameter multi-time scale prediction method and system for comprehensive energy system operation strategy |
CN116957367B (en) * | 2023-09-21 | 2024-01-09 | 南昌大学 | Parameter multi-time scale prediction method and system for comprehensive energy system operation strategy |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113919559A (en) | Ultra-short-term prediction method and device for equipment parameters of comprehensive energy system | |
CN104651559B (en) | Blast furnace liquid iron quality online forecasting system and method based on multivariable online sequential extreme learning machine | |
CN109935280B (en) | Blast furnace molten iron quality prediction system and method based on ensemble learning | |
CN110066895B (en) | Stacking-based blast furnace molten iron quality interval prediction method | |
CN108009667A (en) | A kind of energy demand total amount and structure prediction system | |
CN104657788B (en) | Key industry electricity demand forecasting method based on Industrial Cycle index | |
CN101211383A (en) | Blast furnace molten iron silicon content feature analysis and prediction method | |
CN110378064A (en) | A method of the prediction dead stock column temperature of blast furnace furnace core | |
CN109670625A (en) | NOx emission concentration prediction method based on Unscented kalman filtering least square method supporting vector machine | |
CN105574297B (en) | Self adaptation blast furnace molten iron silicon content trend prediction method | |
Li et al. | Dynamic time features expanding and extracting method for prediction model of sintering process quality index | |
CN111080481A (en) | Electric energy substitution potential gray analysis method based on Markov chain correction | |
CN102117365A (en) | On-line modeling and optimizing method suitable for recovering coking coarse benzene | |
CN113506121A (en) | Analysis method and device for price influence factors | |
CN113268883A (en) | Method for predicting corrosion rate of submarine crude oil pipeline based on PCA-ABC-SVM model | |
CN110533247B (en) | Monthly power consumption prediction method adopting temperature data anomaly point compensation | |
CN111915080A (en) | Raw fuel cost optimal proportioning method based on molten iron quality constraint | |
Shi et al. | Key issues and progress of industrial big data-based intelligent blast furnace ironmaking technology | |
CN113517037B (en) | Method and system for predicting sintering ore FeO by fusing data and knowledge | |
CN112989711B (en) | Aureomycin fermentation process soft measurement modeling method based on semi-supervised ensemble learning | |
Ji et al. | Application of the improved the ELM algorithm for prediction of blast furnace gas utilization rate | |
CN111798023B (en) | Comprehensive coke ratio prediction method in steelmaking sintering production | |
CN116738866B (en) | Instant learning soft measurement modeling method based on time sequence feature extraction | |
CN116757078A (en) | Method and system for measuring flow velocity of pulverized coal based on acting force | |
CN115631804A (en) | Method for predicting outlet concentration of sodium aluminate solution in evaporation process based on data coordination |
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 |