CN108985381B - Method, device and equipment for determining nitrogen oxide emission prediction model - Google Patents

Method, device and equipment for determining nitrogen oxide emission prediction model Download PDF

Info

Publication number
CN108985381B
CN108985381B CN201810825985.XA CN201810825985A CN108985381B CN 108985381 B CN108985381 B CN 108985381B CN 201810825985 A CN201810825985 A CN 201810825985A CN 108985381 B CN108985381 B CN 108985381B
Authority
CN
China
Prior art keywords
coal
determining
sample
fired boiler
sub
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.)
Active
Application number
CN201810825985.XA
Other languages
Chinese (zh)
Other versions
CN108985381A (en
Inventor
符基高
肖红
张荣跃
张骋
王丽苗
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN201810825985.XA priority Critical patent/CN108985381B/en
Publication of CN108985381A publication Critical patent/CN108985381A/en
Application granted granted Critical
Publication of CN108985381B publication Critical patent/CN108985381B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • 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"
    • 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
    • G06Q50/10—Services
    • G06Q50/26—Government or public services

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Tourism & Hospitality (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Development Economics (AREA)
  • Molecular Biology (AREA)
  • General Business, Economics & Management (AREA)
  • Biophysics (AREA)
  • Marketing (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Game Theory and Decision Science (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Primary Health Care (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Educational Administration (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Regulation And Control Of Combustion (AREA)
  • Feedback Control In General (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a method for determining a nitrogen oxide emission prediction model, which takes coal-fired boiler parameters obtained by multi-dimensional asynchronous time sampling and nitrogen oxide emission as initial samples, utilizes the characteristic that a significant offset convolutional neural network can process multi-dimensional asynchronous time sampling data, realizes the purpose of determining a subspace prediction model of each subspace by using the multi-dimensional asynchronous time sampling data, clusters the initial data according to working condition distribution and combines the initial data by using a partial least square method, and avoids the problem of over-low prediction precision caused by diversified working condition distribution. In addition, the invention also provides a device, equipment and a computer readable storage medium for determining the nitrogen oxide emission prediction model, and the function of the device and the equipment corresponds to the function of the method.

Description

Method, device and equipment for determining nitrogen oxide emission prediction model
Technical Field
The invention relates to the field of industrial emission prediction, in particular to a method, a device and equipment for determining a nitrogen oxide emission prediction model and a computer readable storage medium.
Background
Although the development of new energy is rapid in recent years, the coal-fired power generation still occupies more than 50% of the national power generation amount by 2030 years. Nitrogen oxide (NOx), one of the main pollutants of coal-fired power generation, is an important cause for optical pollution and acid rain, and seriously affects air quality, human health, plant growth and the like. With the rapid development of economy and the increasing improvement of living standard, people also put forward new requirements on living environment, which brings new challenges to power production. China is facing three problems of resource shortage, energy supply safety and environmental pollution, and energy conservation and emission reduction become important components of economic and scientific development and energy strategy.
The optimization of the coal-fired boiler of the active power station can be mainly divided into a hard scheme and a soft scheme. The 'hard' scheme mainly means that hardware equipment is technically improved and replaced, and original equipment is replaced by new equipment with higher efficiency. On one hand, the equipment modification requires a large amount of capital investment, the improvement margin of the equipment is limited, and meanwhile, the waste of the old equipment is caused; on the other hand, the problem cannot be solved fundamentally by simply replacing the equipment, and along with the abrasion and the aging of the equipment, the operation working condition can be seriously deviated from the optimal design value, so that the efficiency is reduced. The 'soft' scheme is that the production efficiency is improved through advanced detection, control and optimization methods under the condition that original equipment is not replaced. For example, the coal as fired information is utilized to optimize air distribution, and advanced control strategies are adopted to adjust operation parameters to realize optimization of the combustion process, so that the carbon content and emission of fly ash in flue gas are reduced. The 'soft' scheme does not need to increase too many hardware devices, has low implementation cost and higher cost performance, and therefore has wider application prospect.
The core of the 'soft' scheme is to construct a model for predicting the emission of nitrogen oxides, the existing implementation scheme is an integrated learning modeling method combining soft clustering, a least square support vector machine and partial least square regression, and the model divides original data into sub data sets which are overlapped and have difference by utilizing soft clustering (SFCM); then establishing a Least Square Support Vector Machine (LSSVM) individual model on each group of data sets, and taking the sample membership degree and the output of each sub-model together as the input variable of a synthesis strategy; and finally, eliminating the correlation between the submodels and the membership degree variable by using partial least squares regression as a synthesis function, and extracting the part with larger difference.
However, in addition to the thermal test data, the data samples under the historical operating conditions stored in the database are more used for constructing the soft measurement model at present. The historical operating data is different from the test data and does not meet the uniform characteristic, the data volume is large, the sampling frequency of the multivariate time sequence data is different, and the multivariate time sequence data is distributed under multiple working conditions, but the method cannot solve the problem that the sampling frequency of the multivariate time sequence data is different from the corresponding situation of the time sequence and does data analysis, so a new method needs to be found to solve the problem.
Therefore, the method has research significance on solving the problem that the traditional nitrogen oxide emission prediction model cannot be obtained according to asynchronous time sampling data training.
Disclosure of Invention
The invention aims to provide a method, a device and equipment for determining a nitrogen oxide emission prediction model and a computer readable storage medium, which are used for solving the problem that the nitrogen oxide emission prediction model cannot be obtained according to asynchronous time sampling data training.
In order to solve the technical problem, the invention provides a method for determining a prediction model of nitrogen oxide emission, which is applied to a coal-fired boiler and comprises the following steps:
using a multidimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure BDA0001742496000000021
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiCorresponding actual emission of nitrogen oxides;
for independent variable sample space
Figure BDA0001742496000000022
Clustering to determine multiple sub-sample spaces
Figure BDA0001742496000000023
Wherein T is a positive integer between 1 and T, T being the number of the sub-sample spaces;
determining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure BDA0001742496000000024
Degree of membership of;
processing each of the sub-sample spaces with a significance shift convolutional neural network;
determining a subspace prediction model for each of said subsample spaces
Figure BDA0001742496000000025
Wherein h istIs a subspace sample
Figure BDA0001742496000000026
The subspace prediction model of (1);
respectively utilizing each subspace prediction model to calculate coal-fired boiler parameter xiCorresponding predicted emission of nitrogen oxides to obtain
Figure BDA0001742496000000031
Wherein
Figure BDA0001742496000000032
According to membership space mu ═ mu1,...,μN]NSpace for predicting discharge amount
Figure BDA0001742496000000033
And dependent variable sample space
Figure BDA0001742496000000034
Determining reconstructed samples
Figure BDA0001742496000000035
Wherein,
Figure BDA0001742496000000036
from the reconstructed samples
Figure BDA0001742496000000037
Determination of final nitrogen using partial least squaresAnd F (x) an oxide emission prediction model, wherein x is a coal-fired boiler parameter to be predicted.
Wherein the processing each of the sub-sample spaces with the significance shift convolutional neural network comprises:
determining a significance observation matrix according to each sub-sample space by utilizing a preset significance convolutional neural network;
determining an offset prediction matrix according to each sub-sample space by utilizing a preset offset convolutional neural network;
and updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
Wherein the reconstructing the sample from the basis
Figure BDA0001742496000000038
The method for determining the final nitrogen oxide emission prediction model F (x) by using the partial least square method comprises the following steps:
from the reconstructed samples
Figure BDA0001742496000000039
Determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B through a univariate nonlinear iterative algorithm;
determining a final nox emission amount prediction model f (x) z (x) W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure BDA00017424960000000310
Degree of membership.
Wherein the coal-fired boiler parameter xiThe basis for partitioning into subsample spaces is:
determining coal-fired boiler parameter xiFor the maximum value of the membership degree of each subsample space, judging whether the maximum value isIs greater than
Figure BDA00017424960000000311
If yes, the parameter x of the coal-fired boiler is determinediDividing into a sub-sample space corresponding to the maximum value;
otherwise, determining the parameter x of the coal-fired boileriGreater than or equal to the degree of membership of each subsample space
Figure BDA0001742496000000041
And a plurality of membership degrees of (c) and the coal-fired boiler parameter xiAnd dividing into a sub-sample space corresponding to the plurality of membership degrees, wherein delta is a boundary margin parameter.
The coal-fired boiler parameters comprise any one or more of generator power, coal feeding quantity of a coal mill, air quantity of an inlet of the coal mill and oxygen content of smoke.
The invention also provides a device for determining the prediction model of the discharge amount of the nitrogen oxides, which is applied to a coal-fired boiler and comprises the following components:
an initial sample determination module: for taking a multi-dimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure BDA0001742496000000042
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiCorresponding actual emission of nitrogen oxides;
a sub-sample space determination module: for space of independent variable samples
Figure BDA0001742496000000043
Clustering to determine multiple sub-sample spaces
Figure BDA0001742496000000044
Wherein T is a positive integer between 1 and T, T being the number of the sub-sample spaces;
a membership degree determination module: for ensuringDetermining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure BDA0001742496000000045
Degree of membership of;
a significance offset convolutional neural network module: updating each of the subsample spaces with a significance shift convolutional neural network;
a subspace prediction model determination module: subspace prediction model for determining each of said subsample spaces
Figure BDA0001742496000000046
Wherein h istIs a subspace sample
Figure BDA0001742496000000047
The subspace prediction model of (1);
a predicted emission determination module: for calculating coal-fired boiler parameter x using each of the subspace prediction models, respectivelyiCorresponding predicted emission of nitrogen oxides to obtain
Figure BDA0001742496000000048
Wherein
Figure BDA0001742496000000049
A reconstructed sample determination module: for space mu ═ mu according to degree of membership1,...,μN]NSpace for predicting discharge amount
Figure BDA00017424960000000410
And dependent variable sample space
Figure BDA00017424960000000411
Determining reconstructed samples
Figure BDA00017424960000000412
Wherein,
Figure BDA00017424960000000413
the nitrogen oxide emission prediction model determination module: for reconstructing samples from said
Figure BDA00017424960000000414
And determining a final nitrogen oxide emission prediction model F (x) by using a partial least square method, wherein x is a coal-fired boiler parameter to be predicted.
Wherein the significant offset convolutional neural network module specifically comprises:
a significance observation matrix determination unit: the system comprises a preset significance convolutional neural network, a significance observation matrix and a significance estimation matrix, wherein the preset significance convolutional neural network is used for determining a significance observation matrix according to each sub-sample space;
an offset prediction matrix determination unit: the device comprises a preset offset convolutional neural network, a plurality of sub-sample spaces and an offset prediction matrix, wherein the preset offset convolutional neural network is used for determining the offset prediction matrix according to each sub-sample space;
an update unit: and the system is used for updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
The module for determining the prediction model of the emission of nitrogen oxides specifically comprises:
a parameter determination unit: for reconstructing samples from said
Figure BDA0001742496000000051
Determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B through a univariate nonlinear iterative algorithm;
nitrogen oxide emission prediction model determination unit: for determining the final nox emission amount prediction model f (x), z (x), W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure BDA0001742496000000052
Degree of membership.
In addition, the invention also provides a device for determining the prediction model of the emission of the nitrogen oxides, which is applied to a coal-fired boiler and comprises:
a memory: for storing a computer program;
a processor: for executing the computer program to implement the steps of a method for determining a model for predicting an amount of nox emission as described above.
Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when being executed by a processor, implements the steps of a method for determining a model for predicting nox emissions as described above.
In summary, the method for determining the nitrogen oxide emission prediction model provided by the invention includes that coal-fired boiler parameters obtained by multi-dimensional asynchronous time sampling and nitrogen oxide emission are used as initial samples, the initial samples are clustered to obtain a plurality of sub-sample spaces, then the sub-sample spaces are processed by using a significant offset convolutional neural network, then the sub-space prediction model of each sub-sample space is determined, the prediction emission corresponding to each coal-fired boiler parameter is predicted by using the sub-space prediction model, a reconstructed sample is determined according to the prediction emission and the membership degree of each coal-fired boiler parameter to each sub-sample space, and finally the final prediction model is determined according to the reconstructed sample by using a partial least square method. Therefore, the method utilizes the characteristic that the significant offset convolutional neural network can process multi-dimensional asynchronous time sampling data, achieves the purpose of determining a subspace prediction model of each subspace by utilizing the multi-dimensional asynchronous time sampling data, and also clusters initial data according to working condition distribution and combines the initial data by utilizing a partial least square method, so that the problem of low prediction precision caused by diversified working condition distribution is avoided.
The invention also provides a device, equipment and a computer readable storage medium for determining the nitrogen oxide emission prediction model, wherein the function of the device corresponds to that of the method, and the details are not repeated.
Drawings
In order to more clearly illustrate the embodiments or technical solutions of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without creative efforts.
FIG. 1 is a flowchart illustrating an embodiment of a method for determining a model for predicting NOx emissions according to the present invention;
FIG. 2 is a schematic diagram of a significant deviation convolutional neural network according to the present invention;
FIG. 3 is a schematic diagram illustrating an embodiment of a method for determining a model for predicting an amount of nitrogen oxide emissions according to the present invention;
fig. 4 is a block diagram illustrating an embodiment of a device for determining a model for predicting an amount of nox discharged according to the present invention.
Detailed Description
The core of the invention is to provide a method, a device and equipment for determining a nitrogen oxide emission prediction model and a computer readable storage medium, which solve the problem that the nitrogen oxide emission prediction model cannot be determined according to multi-dimensional asynchronous time sampling data and improve the prediction accuracy.
In order that those skilled in the art will better understand the disclosure, the invention will be described in further detail with reference to the accompanying drawings and specific embodiments. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
An embodiment of a method for determining a model for predicting nox emissions according to the present invention is described below, and referring to fig. 1, the embodiment of the method includes:
step S101: using a multidimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure BDA0001742496000000071
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiCorresponding to the actual emission of nitrogen oxides.
According to the invention, multi-dimensional parameters of the coal-fired boiler are mainly researched, the sampling frequency of the parameters of the coal-fired boiler in each dimension is different, the parameters of the coal-fired boiler specifically refer to parameters influencing the discharge amount of nitrogen oxides of the coal-fired boiler, and specifically can comprise any one or more of the power of a generator, the coal feeding amount of a coal mill, the air quantity of an inlet of the coal mill and the oxygen content of smoke.
Step S102: for independent variable sample space
Figure BDA0001742496000000072
Clustering to determine multiple sub-sample spaces
Figure BDA0001742496000000073
Wherein T is a positive integer between 1 and T, and T is the number of the sub-sample spaces.
Specifically, clustering is performed through distribution of parameters of the coal-fired boiler, and it should be particularly noted that, in a conventional clustering method, for example, fuzzy mean clustering (FCM), the original samples are divided into several mutually disjoint sub-classes based on a maximum membership principle, and thus the obtained sub-classes are mutually isolated. In practical problems, when the membership degree of a sample belonging to a certain class is far greater than the membership degree of the sample belonging to other classes, the sample can be classified into the class by using the principle of the maximum membership degree; however, for samples belonging to classes with close membership, if they are still simply divided into a single subclass, their properties cannot be described accurately to a large extent.
In order to better process the boundary samples of the adjacent subspaces, the present embodiment may adopt an SFCM clustering algorithm, which divides the space by membership degree cut sets, and allows the boundary samples between classes to belong to multiple subclasses simultaneously, and the subsample sets may intersect with each other. The specific algorithm is as follows:
firstly, clustering initial samples by using an FCM clustering method to obtain a coal-fired boiler parameter xiMembership to each subsample space, for coal fired boiler parameter xiThe division into which sub-sample space is to be divided is determined taking into account two principles:
determining coal-fired boiler parameter xiFor the maximum value of the membership degree of each subsample space, judging whether the maximum value is greater than the maximum value
Figure BDA0001742496000000081
If yes, the parameter x of the coal-fired boiler is determinediDividing into a sub-sample space corresponding to the maximum value; otherwise, determining the parameter x of the coal-fired boileriGreater than or equal to the degree of membership of each subsample space
Figure BDA0001742496000000082
And a plurality of membership degrees of (c) and the coal-fired boiler parameter xiAnd dividing into a sub-sample space corresponding to the plurality of membership degrees, wherein delta is a boundary margin parameter.
Thus, the coal-fired boiler parameter x is determined when the boundary between subclasses can be clearly determined, for example, when the membership degree is close to 1iBelonging to only a single subsample space; coal-fired boiler parameter x when the boundary between subclasses is not clear, e.g. when membership is slightly greater than 1/TiPossibly belonging to multiple sub-sample spaces simultaneously.
Step S103: determining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure BDA0001742496000000083
Degree of membership.
It should be noted that, in the process of determining the sub-sample space, the division needs to be performed according to the membership degree. Specifically, step S102 and step S103 may be determined simultaneously, and the order of the two is not limited in this embodiment.
Step S104: processing each of the sub-sample spaces with a significance shift convolutional neural network.
In summary, step S104 includes: determining a significance observation matrix according to each sub-sample space by utilizing a preset significance convolutional neural network; determining an offset prediction matrix according to each sub-sample space by utilizing a preset offset convolutional neural network; and finally updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
Specifically, as shown in fig. 2, the mixing time sequence, which includes a sequence value (i.e., a value indicating each input parameter), an indication line (indicator) (indicating a data parameter of mixing) and a duration (duration), enters two convolutional neural networks, namely a significance network (significance network) and an Offset network (Offset network). The significance network includes a plurality of convolutional layers, while the offset network uses only 1 convolutional layer, with the size of the 1-dimensional convolutional kernel being 1. The 1-dimensional convolution kernels are used in each convolution layer of the two networks, the sizes of the convolution kernels are fixed, but the number of the convolution kernels is variable, and the number of the convolution kernels used in the last convolution layer is the same as the dimension of the output variable to be predicted (namely, the multidimensional coal-fired boiler parameter in the model). The results of these two convolutional networks can be represented by S and off, respectively, and the variable to be predicted by xIRepresentation (refers to values of multidimensional coal-fired boiler parameters).
xIThe value of (A) is obtained by calculating the weight according to a gating mechanism sigma (S) after the off correction of the offset network:
Figure BDA0001742496000000091
and finally, obtaining final output by a full connection layer:Hn＝WHn-1+b。
step S105: determining a subspace prediction model for each of said subsample spaces
Figure BDA0001742496000000092
Wherein h istIs a subspace sample
Figure BDA0001742496000000093
The subspace prediction model of (1).
Step S106: respectively utilizing each subspace prediction model to calculate coal-fired boiler parameter xiCorresponding predicted emission of nitrogen oxides to obtain
Figure BDA0001742496000000094
Wherein
Figure BDA0001742496000000095
Step S107: according to membership space mu ═ mu1,...,μN]NSpace for predicting discharge amount
Figure BDA0001742496000000096
And dependent variable sample space
Figure BDA0001742496000000097
Determining reconstructed samples
Figure BDA0001742496000000098
Wherein,
Figure BDA0001742496000000099
step S108: from the reconstructed samples
Figure BDA00017424960000000910
And determining a final nitrogen oxide emission prediction model F (x) by using a partial least square method, wherein x is a coal-fired boiler parameter to be predicted.
The partial least squares method can effectively eliminate the collinearity between variables and suppress noise interference in the samples by extracting the maximum variance of the input samples, that is, the reconstructed samples in the present embodiment. The partial least square method is used as a synthesis strategy to extract components with large differences in each sub-model, and redundant and collinearity information is eliminated.
For the partial least square method synthesis function, in addition, because the density and radius of each sub-sample set space are different, for any sample, the fuzzy membership vector also influences the final integrated output result, so the input sample Z ═ Z is considered1,...,zN]TAnd single output sample
Figure BDA00017424960000000911
Wherein N is the number of samples,
Figure BDA00017424960000000912
a univariate Nonlinear Iterative (NIPALS) algorithm may be employed in extracting the principal components. And solving and extracting the PLS principal components to obtain a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B. Thus, the synthesis function can be expressed as F (x) ═ Z (x) · W (P)TW)-1BqT。
In summary, step S108 specifically includes: from the reconstructed samples
Figure BDA0001742496000000101
Determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B through a univariate nonlinear iterative algorithm; determining a final nox emission amount prediction model f (x) z (x) W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure BDA0001742496000000102
Degree of membership.
In summary, a soft measurement model of nitrogen oxide emission is established by using a convolutional neural network ensemble learning model modeling method based on significance shift based on historical operating data of the power station, and a model structure diagram is shown in fig. 3. The model firstly utilizes an SFCM clustering algorithm to carry out fuzzy partition on initial samples according to input variables to obtain T sub-sample spaces, utilizes a significant convolutional neural network to process each sub-sample, determines sub-models, and then utilizes a synthesis strategy to combine the sub-models to obtain final output. Specifically, the whole model building process may be divided into a training stage and a testing stage, and after the prediction model is obtained in step S108, the prediction model may be trained by using the test sample. It may be that when the initial samples are determined in step S101, the samples for training and the samples for testing are determined separately for the purpose of optimizing and verifying the final prediction model.
Step S105 is described in detail below, and based on the characteristics of the operating condition, the SFCM algorithm is used to divide the initial sample to obtain T different sub-sample spaces. The following describes a process of determining a subspace prediction model according to any one of the subsample spaces, and the process of determining the corresponding subspace prediction model by using the other subsample spaces may refer to the following processes:
assuming that a certain subsample space contains n samples, a significant shift convolutional neural network is selected as an individual regression model, and our aim is to predict the element subset xnRelative future value of
Figure BDA0001742496000000103
Wherein
Figure BDA0001742496000000104
Is xnA subset of features of (1). Is provided with
Figure BDA0001742496000000105
i∈1,2,...,dI. Wherein, S, F:
Figure BDA0001742496000000106
is displayA neural network and an offset neural network, σ being a normalized activation function independent on each row, i.e.
Figure BDA0001742496000000111
For any one
Figure BDA0001742496000000112
And σ, such that σ (a)T1M＝1，
Figure BDA0001742496000000113
Is a Hadamard (element) matrix multiplication.
Fang Cheng
Figure BDA0001742496000000114
Exceeds the columns of the matrix in brackets, thus outputting a vector
Figure BDA00017424960000001112
Is a matrix
Figure BDA0001742496000000115
For the linear combination of line i, we will consider that S is the form of a fully convolved network (consisting entirely of convolutional layers) and F
Figure BDA0001742496000000116
Wherein
Figure BDA0001742496000000117
And off:
Figure BDA0001742496000000118
is a multi-layered perceptron, in which case F can be viewed as the sum of the projections (x → xI) and the convolutional network of all kernels of length 1 can be rewritten as the equation:
Figure BDA0001742496000000119
wherein S (·)mIs the mth column of matrix S (-). We refer to this network as the Significant Offset Convolutional Neural Network (SOCNN), while S and offIt is referred to as a saliency and offset network. Note that when off is 0 and σ is 1, the model simplifies the set of separate ar (m) models for each dimension.
The network maintains the time dimension to the top level, while the feature count for each time step (filter) in the hidden level is custom. The last convolutional layer, however, has a filter number corresponding to the output dimension, and the weighted frame shows how the outputs from the significance and offset networks are related to the equations
Figure BDA00017424960000001110
In combination.
Please note that equation
Figure BDA00017424960000001111
In a form that enhances the separation of the temporal dependence (in units of weight Wm), the observation Sm (the local meaning of S as a convolutional network is determined by its filters, which capture the temporal dependence and are independent of the relative position in time) and the temporally completely independent predictor off. This provides some interpretability of the appropriate functions and weights. For example, each of the past observations provided an adjusted single regressor for the target variable through the offset network. Note that due to the asynchronous sampling process, the successive values of x may be heterogeneous, so on the other hand, the significance network provides data-dependent weights for all regressors and accumulates them.
In summary, according to the method for determining a model for predicting nitrogen oxide emissions provided by this embodiment, coal-fired boiler parameters and nitrogen oxide emissions obtained by multidimensional asynchronous time sampling are used as initial samples, the initial samples are clustered to obtain a plurality of sub-sample spaces, a significant offset convolutional neural network is used to process each sub-sample space, so as to determine a sub-space prediction model of each sub-sample space, the sub-space prediction model is used to predict predicted emissions corresponding to each coal-fired boiler parameter, a reconstructed sample is determined according to the predicted emissions and the membership of each coal-fired boiler parameter to each sub-sample space, and finally, a partial least square method is used to determine a final prediction model according to the reconstructed sample. Therefore, the method utilizes the characteristic that the significant offset convolutional neural network can process multi-dimensional asynchronous time sampling data, achieves the purpose of determining the subspace prediction model of each subspace by utilizing the multi-dimensional asynchronous time sampling data, and avoids the problem of low prediction precision caused by diversified working condition distribution by clustering initial data according to the working condition distribution and utilizing a partial least square method fusion mode.
In the following, embodiments of a device for determining a model for predicting an emission amount of nitrogen oxide according to embodiments of the present invention are described, and a device for determining a model for predicting an emission amount of nitrogen oxide described below and a method for determining a model for predicting an emission amount of nitrogen oxide described above may be referred to correspondingly.
Referring to fig. 4, the apparatus embodiment comprises:
the initial sample determination module 401: for taking a multi-dimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure BDA0001742496000000121
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiCorresponding to the actual emission of nitrogen oxides.
The sub-sample space determination module 402: for space of independent variable samples
Figure BDA0001742496000000122
Clustering to determine multiple sub-sample spaces
Figure BDA0001742496000000123
Wherein T is a positive integer between 1 and T, and T is the number of the sub-sample spaces.
Membership determination module 403: for determining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure BDA0001742496000000124
Degree of membership.
Significance offset convolutional neural network module 404: processing each of the subsample spaces with a significance shift convolutional neural network;
subspace prediction model determination module 405: subspace prediction model for determining each of said subsample spaces
Figure BDA0001742496000000131
Wherein h istIs a subspace sample
Figure BDA0001742496000000132
The subspace prediction model of (1).
The predicted emissions determination module 406: for calculating coal-fired boiler parameter x using each of the subspace prediction models, respectivelyiCorresponding predicted emission of nitrogen oxides to obtain
Figure BDA0001742496000000133
Wherein
Figure BDA0001742496000000134
Reconstructed sample determination module 407: for space mu ═ mu according to degree of membership1,...,μN]NSpace for predicting discharge amount
Figure BDA0001742496000000135
And dependent variable sample space
Figure BDA0001742496000000136
Determining reconstructed samples
Figure BDA0001742496000000137
Wherein,
Figure BDA0001742496000000138
the nox emission prediction model determination module 408: for reconstructing samples from said
Figure BDA0001742496000000139
And determining a final nitrogen oxide emission prediction model F (x) by using a partial least square method, wherein x is a coal-fired boiler parameter to be predicted.
The significant offset convolutional neural network module 404 specifically includes:
the significance observation matrix determination unit 4041: and the system is used for determining a significance observation matrix according to each sub-sample space by utilizing a preset significance convolutional neural network.
Offset prediction matrix determination unit 4042: and the method is used for determining an offset prediction matrix according to each sub-sample space by utilizing a preset offset convolutional neural network.
Update unit 4043: and the system is used for updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
The module 408 for determining a model for predicting the amount of nitrogen oxide emission specifically includes:
parameter determination unit 4081: for reconstructing samples from said
Figure BDA00017424960000001310
And determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B by a univariate nonlinear iterative algorithm.
Nitrogen oxide emission amount prediction model determination unit 4082: for determining the final nox emission amount prediction model f (x), z (x), W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure BDA00017424960000001311
Degree of membership of。
A device for determining a nox emission prediction model according to this embodiment is used to implement the foregoing method for determining a nox emission prediction model, and thus a specific implementation manner of the device may be found in the foregoing embodiment of a method for determining a nox emission prediction model, for example, the initial sample determining module 401, the sub-sample space determining module 402, the membership degree determining module 403, the significant shift convolutional neural network module 404, the sub-space prediction model determining module 405, the predicted emission determining module 406, the reconstructed sample determining module 407, and the nox emission prediction model determining module 408 are respectively used to implement steps S101, S102, S103, S104, S105, S106, S107, and S108 in the above method for determining a nox emission prediction model. Therefore, specific embodiments thereof may be referred to in the description of the corresponding respective partial embodiments, and will not be described herein.
In addition, since the determination device of the nox emission prediction model provided in this embodiment is used to implement the determination method of the nox emission prediction model, the function corresponds to the function of the method, and details are not described here.
In addition, the invention also provides a device for determining the prediction model of the emission of the nitrogen oxides, which is applied to a coal-fired boiler and comprises:
a memory: for storing a computer program;
a processor: for executing the computer program to implement the steps of a method for determining a model for predicting an amount of nox emission as described above.
Finally, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when being executed by a processor, implements the steps of a method for determining a model for predicting nox emissions as described above.
In addition, since the determining device of the nox emission amount prediction model and the computer readable storage medium provided by the present invention are used for implementing the determining method of the nox emission amount prediction model, the functions thereof correspond to the functions of the above method, and further, the detailed description thereof can also refer to the description of the above method embodiment, which will not be described herein.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other. The device disclosed by the embodiment corresponds to the method disclosed by the embodiment, so that the description is simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The present invention provides a method, an apparatus, a device and a computer readable storage medium for determining a model for predicting nox emission. The principles and embodiments of the present invention are explained herein using specific examples, which are presented only to assist in understanding the method and its core concepts. It should be noted that, for those skilled in the art, it is possible to make various improvements and modifications to the present invention without departing from the principle of the present invention, and those improvements and modifications also fall within the scope of the claims of the present invention.

Claims (7)

1. A method for determining a nitrogen oxide emission prediction model is applied to a coal-fired boiler, and is characterized by comprising the following steps:
using a multidimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure FDA0003331494670000011
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiThe corresponding actual discharge amount of the nitrogen oxides, wherein the parameters of the coal-fired boiler comprise any one or more of the power of a generator, the coal feeding amount of a coal mill, the air quantity at the inlet of the coal mill and the oxygen content of flue gas;
for independent variable sample space
Figure FDA0003331494670000012
Clustering to determine multiple sub-sample spaces
Figure FDA0003331494670000013
Wherein T is a positive integer between 1 and T, T being the number of the sub-sample spaces;
determining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure FDA0003331494670000014
Degree of membership of;
processing each of the sub-sample spaces with a significance shift convolutional neural network; determining a subspace prediction model for each of said subsample spaces
Figure FDA0003331494670000015
Wherein h istIs a subspace sample
Figure FDA0003331494670000016
The subspace prediction model of (1);
respectively utilizing each subspace prediction model to calculate coal-fired boiler parameter xiCorresponding predicted emission of nitrogen oxides to obtain
Figure FDA0003331494670000017
Wherein
Figure FDA0003331494670000018
According to membership space mu ═ mu1,...,μN]NSpace for predicting discharge amount
Figure FDA0003331494670000019
And dependent variable sample space
Figure FDA00033314946700000110
Determining reconstructed samples
Figure FDA00033314946700000111
Wherein,
Figure FDA00033314946700000112
from the reconstructed samples
Figure FDA00033314946700000113
Determining a final nitrogen oxide emission prediction model F (x) by using a partial least square method, wherein x is a coal-fired boiler parameter to be predicted;
the processing each of the sub-sample spaces with a significance shift convolutional neural network comprises:
determining a significance observation matrix according to each sub-sample space by utilizing a preset significance convolutional neural network;
determining an offset prediction matrix according to each sub-sample space by utilizing a preset offset convolutional neural network;
and updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
2. The method of claim 1, wherein the reconstructing the sample from the basis of the data is performed using a single-pass filter
Figure FDA00033314946700000114
The method for determining the final nitrogen oxide emission prediction model F (x) by using the partial least square method comprises the following steps:
from the reconstructed samples
Figure FDA00033314946700000115
Determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B through a univariate nonlinear iterative algorithm;
determining a final nox emission amount prediction model f (x) z (x) W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure FDA0003331494670000021
Degree of membership.
3. The method of claim 1 or 2, wherein the coal-fired boiler parameter xiThe basis for partitioning into subsample spaces is:
determining coal-fired boiler parameter xiFor the maximum value of the membership degree of each subsample space, judging whether the maximum value is greater than the maximum value
Figure FDA0003331494670000022
If yes, the parameter x of the coal-fired boiler is determinediDividing into a sub-sample space corresponding to the maximum value;
otherwise, determining the parameter x of the coal-fired boileriGreater than or equal to the degree of membership of each subsample space
Figure FDA0003331494670000023
And a plurality of membership degrees of (c) and the coal-fired boiler parameter xiAnd dividing into a sub-sample space corresponding to the plurality of membership degrees, wherein delta is a boundary margin parameter.
4. A device for determining a prediction model of nitrogen oxide emission is applied to a coal-fired boiler, and is characterized by comprising:
an initial sample determination module: for taking a multi-dimensional asynchronous time sampling sequence of the coal-fired boiler as an initial sample
Figure FDA0003331494670000024
Wherein i is a positive integer between 1 and N, N is the sample capacity, xiIs a multidimensional coal-fired boiler parameter, yiIs related to the coal-fired boiler parameter xiThe corresponding actual discharge amount of the nitrogen oxides, wherein the parameters of the coal-fired boiler comprise any one or more of the power of a generator, the coal feeding amount of a coal mill, the air quantity at the inlet of the coal mill and the oxygen content of flue gas;
a sub-sample space determination module: for space of independent variable samples
Figure FDA0003331494670000025
Clustering to determine multiple sub-sample spaces
Figure FDA0003331494670000026
Wherein T is a positive integer between 1 and T, T being the number of the sub-sample spaces;
a membership degree determination module: for determining coal-fired boiler parameter xiDegree of membership mu to each of said subsample spacesi＝[μ1i,...,μTi]TWherein, mutiIs xiSub-sample space
Figure FDA0003331494670000027
Degree of membership of;
a significance offset convolutional neural network module: processing each of the subsample spaces with a significance shift convolutional neural network;
a subspace prediction model determination module: subspace prediction model for determining each of said subsample spaces
Figure FDA0003331494670000028
Wherein h istIs a subspace sample
Figure FDA0003331494670000029
The subspace prediction model of (1);
a predicted emission determination module: for calculating coal-fired boiler parameter x using each of the subspace prediction models, respectivelyiCorresponding predicted emission of nitrogen oxides to obtain
Figure FDA0003331494670000031
Wherein
Figure FDA0003331494670000032
A reconstructed sample determination module: for space mu ═ mu according to degree of membership1,...,μN]NSpace for predicting discharge amount
Figure FDA0003331494670000033
And dependent variable sample space
Figure FDA0003331494670000034
Determining reconstructed samples
Figure FDA0003331494670000035
Wherein,
Figure FDA0003331494670000036
the nitrogen oxide emission prediction model determination module: for reconstructing samples from said
Figure FDA0003331494670000037
Determining a final nitrogen oxide emission prediction model F (x) by using a partial least square method, wherein x is a coal-fired boiler parameter to be predicted;
the significant offset convolutional neural network module specifically includes:
a significance observation matrix determination unit: the system comprises a preset significance convolutional neural network, a significance observation matrix and a significance estimation matrix, wherein the preset significance convolutional neural network is used for determining a significance observation matrix according to each sub-sample space;
an offset prediction matrix determination unit: the device comprises a preset offset convolutional neural network, a plurality of sub-sample spaces and an offset prediction matrix, wherein the preset offset convolutional neural network is used for determining the offset prediction matrix according to each sub-sample space;
an update unit: and the system is used for updating each sub-sample space according to the significance observation matrix and the offset prediction matrix.
5. The apparatus of claim 4, wherein the model for predicting the amount of nitrogen oxide emissions determining module specifically comprises:
a parameter determination unit: for reconstructing samples from said
Figure FDA0003331494670000038
Determining a weight matrix W, an input load matrix P, an output load vector q and a coefficient matrix B through a univariate nonlinear iterative algorithm;
nitrogen oxide emission prediction model determination unit: for determining the final nox emission amount prediction model f (x), z (x), W (P)TW)-1BqTWherein Z (x) is [. z ]1(x),...zT(x)]T，zt(x)＝[h1(x),...,hT(x),μ1(x),...,μT(x)]，μt(x) For coal-fired boiler parameter x pair sub-sample space
Figure FDA0003331494670000039
Degree of membership.
6. A device for determining a prediction model of nitrogen oxide emission, which is applied to a coal-fired boiler, is characterized by comprising:
a memory: for storing a computer program;
a processor: steps for executing said computer program for implementing a method for determining a model for predicting emissions of nitrogen oxides as claimed in any one of claims 1 to 3.
7. A computer-readable storage medium, wherein the computer-readable storage medium has stored thereon a computer program, which when executed by a processor, implements the steps of a method for determining a model for predicting nox emissions as claimed in any one of claims 1 to 3.
CN201810825985.XA 2018-07-25 2018-07-25 Method, device and equipment for determining nitrogen oxide emission prediction model Active CN108985381B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810825985.XA CN108985381B (en) 2018-07-25 2018-07-25 Method, device and equipment for determining nitrogen oxide emission prediction model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810825985.XA CN108985381B (en) 2018-07-25 2018-07-25 Method, device and equipment for determining nitrogen oxide emission prediction model

Publications (2)

Publication Number Publication Date
CN108985381A CN108985381A (en) 2018-12-11
CN108985381B true CN108985381B (en) 2022-02-15

Family

ID=64550968

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810825985.XA Active CN108985381B (en) 2018-07-25 2018-07-25 Method, device and equipment for determining nitrogen oxide emission prediction model

Country Status (1)

Country Link
CN (1) CN108985381B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109711453B (en) * 2018-12-21 2022-05-13 广东工业大学 Multivariable-based equipment dynamic health state evaluation method
CN111581581B (en) * 2020-04-23 2021-02-02 大唐环境产业集团股份有限公司 Method and system for detecting NOx concentration at SCR inlet under multi-boundary condition
CN111701444B (en) * 2020-07-01 2022-02-08 盐城工学院 Method for treating organic waste gas by using biotrickling filter based on convolutional neural network
CN112782976B (en) * 2020-12-14 2023-03-07 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 CNN algorithm-based intelligent combustion optimization closed-loop control method
CN113065765A (en) * 2021-04-01 2021-07-02 瀚蓝绿电固废处理(佛山)有限公司 Construction method and application of prediction model for CO emission in flue gas of waste incinerator
CN114065100B (en) * 2021-11-11 2024-06-14 国网河北能源技术服务有限公司 Boiler nitrogen oxide emission control method, terminal and storage medium
CN114707778A (en) * 2022-06-08 2022-07-05 中汽研汽车检验中心（天津）有限公司 Method for rapidly predicting automobile PN emission based on oil product aromatic hydrocarbon composition
CN118039026A (en) * 2024-01-19 2024-05-14 淮阴工学院 Nitrogen oxide prediction method and system based on integrated learning gas turbine

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103455635A (en) * 2013-09-24 2013-12-18 华北电力大学 Thermal process soft sensor modeling method based on least squares and support vector machine ensemble

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2016064125A (en) * 2014-09-19 2016-04-28 シナノケンシ株式会社 Onset risk forecasting system for cerebrovascular disease

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103455635A (en) * 2013-09-24 2013-12-18 华北电力大学 Thermal process soft sensor modeling method based on least squares and support vector machine ensemble

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
AUTOREGRESSIVE CONVOLUTIONAL NEURAL NETWORKS FOR ASYNCHRONOUS TIME SERIES;BINKOWSKI M et al.;《ICLR 2018 Conference》;20180216;第1-16页 *

Also Published As

Publication number Publication date
CN108985381A (en) 2018-12-11

Similar Documents

Publication Publication Date Title
CN108985381B (en) Method, device and equipment for determining nitrogen oxide emission prediction model
CN111148118B (en) Flow prediction and carrier wave turn-off method and system based on time sequence
CN111738477B (en) Power grid new energy consumption capability prediction method based on deep feature combination
CN111144609A (en) Boiler exhaust emission prediction model establishing method, prediction method and device
CN110263395A (en) The power plant's denitration running optimizatin method and system analyzed based on numerical simulation and data
CN113469449B (en) Optimization control method and system for desulfurization system
CN110020712B (en) Optimized particle swarm BP network prediction method and system based on clustering
CN112488145A (en) NO based on intelligent methodxOnline prediction method and system
CN111860701B (en) Denitration system working condition discrimination preprocessing method based on clustering method
CN111240282A (en) Process optimization method, device, equipment and computer readable storage medium
CN112434848A (en) Nonlinear weighted combination wind power prediction method based on deep belief network
CN116307212B (en) Air quality prediction method and system
Zhou et al. From neural network to neuro-fuzzy modeling: applications to the carbon dioxide capture process
CN113449919B (en) Power consumption prediction method and system based on feature and trend perception
CN112232373A (en) Method and system for predicting carbon content of boiler fly ash
CN114970353A (en) MSWI process dioxin emission soft measurement method based on missing data filling
CN112884213A (en) Coal-fired boiler NOx prediction method based on wavelet decomposition and dynamic hybrid deep learning
CN117150677A (en) Method and device for determining type selection parameters of electric dust collector
Hao et al. Prediction of nitrogen oxide emission concentration in cement production process: a method of deep belief network with clustering and time series
CN116204779A (en) Method, system and readable storage medium for judging operation state of energy storage salt cavern
CN111882006A (en) Data preprocessing method of desulfurization system based on PCA and correlation function
CN105426959A (en) Aluminium electrolysis energy conservation and emission reduction method based on BP neural network and self-adaptive MBFO algorithm
CN116187512A (en) BP neural network-based carbon emission capacity prediction method and device
CN113869359B (en) Method for predicting nitrogen oxides in urban solid waste incineration process based on modularized neural network
CN113111588B (en) NO of gas turbine X Emission concentration prediction method and device

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