CN108182553B - Coal-fired boiler combustion efficiency online measurement method - Google Patents

Coal-fired boiler combustion efficiency online measurement method Download PDF

Info

Publication number
CN108182553B
CN108182553B CN201810107733.3A CN201810107733A CN108182553B CN 108182553 B CN108182553 B CN 108182553B CN 201810107733 A CN201810107733 A CN 201810107733A CN 108182553 B CN108182553 B CN 108182553B
Authority
CN
China
Prior art keywords
svm
working condition
boiler
combustion efficiency
individual
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
CN201810107733.3A
Other languages
Chinese (zh)
Other versions
CN108182553A (en
Inventor
唐振浩
吴笑妍
曹生现
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.)
Northeast Electric Power University
Original Assignee
Northeast Dianli University
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 Northeast Dianli University filed Critical Northeast Dianli University
Priority to CN201810107733.3A priority Critical patent/CN108182553B/en
Publication of CN108182553A publication Critical patent/CN108182553A/en
Application granted granted Critical
Publication of CN108182553B publication Critical patent/CN108182553B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • Health & Medical Sciences (AREA)
  • Educational Administration (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Theoretical Computer Science (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Regulation And Control Of Combustion (AREA)

Abstract

The invention provides an on-line measurement method for combustion efficiency of a coal-fired boiler, which comprises the steps of obtaining process parameters of a boiler production process, and calculating the combustion efficiency of the boiler at a corresponding moment; determining a standardized process parameter set after feature extraction by adopting a boost tree method; adopting a K-adjacent value classifier to classify the working condition of the standardized process parameter set after the feature extraction; aiming at various working condition data sets: establishing LS-SVM measurement models of various working conditions, and optimizing the LS-SVM measurement models of various working conditions by adopting a differential evolution algorithm to obtain the LS-SVM measurement models of the boiler combustion efficiency of various working conditions after final optimization; measuring the combustion efficiency of the coal-fired boiler by adopting LS-SVM measurement models of the boiler combustion efficiency under various working conditions; the method can realize high-precision data-driven modeling according to the relevant variables to obtain the real-time boiler combustion efficiency, and can meet the production requirements.

Description

Coal-fired boiler combustion efficiency online measurement method
Technical Field
The invention belongs to the technical field of thermal power generation control, and particularly relates to an on-line measurement method for combustion efficiency of a coal-fired boiler.
Background
Thermal power generation is a main mode of power production in China, and the generated energy of the thermal power generation is far superior to the power generation modes such as hydroelectric power generation, wind power generation and the like. In the thermal power generation production process, coal-fired power generation is dominant, and the generator set has the characteristics of high power, good stability and low cost. The coal-fired boiler is important production equipment of a coal-fired power plant and is used for converting energy released by combustion of coal in a hearth into heat energy of high-temperature and high-pressure steam. The combustion efficiency of the coal-fired boiler refers to the ratio of the heat actually released in the combustion process of coal to the heat released after the coal is completely combusted, and the combustion degree of the reaction fuel. The improvement of the combustion efficiency is an important index of the production of the coal-fired boiler, and the accurate measurement of the boiler combustion efficiency is a key for improving the boiler combustion efficiency and reducing the operation and maintenance cost.
The combustion process of the coal-fired boiler comprises complex physical and chemical changes, is influenced by various factors such as fuel temperature, fuel components, boiler load and the like, and has strong nonlinear relation among parameters, so that the combustion efficiency of the boiler is difficult to accurately measure on line due to the characteristics. In the existing boiler combustion efficiency measuring methods, a mechanism method for calculating the boiler combustion efficiency can only reflect the past boiler combustion efficiency and cannot predict the development trend of the boiler combustion efficiency; the other type of method can predict the combustion efficiency trend of the boiler based on data statistics, such as an autoregressive moving average model, an ash box model and the like, but has the problems of large required data volume, complex calculation process and sensitivity to noise.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an on-line measuring method for the combustion efficiency of a coal-fired boiler, which comprises the following steps:
step 1: acquiring process parameters of a boiler production process through a plant-level monitoring information system of a thermal power plant, calculating boiler combustion efficiency at a corresponding moment according to a mechanism model, and obtaining a DATA set DATA ═ X of the boiler production processk，Tk1., K }, where X is Xk∈RnIs the process parameter in the kth sample, TkThe boiler combustion efficiency of the kth sample is obtained, K is the number of samples in the data set, and n is the dimension of the process parameter;
step 2: standardizing process parameters of a data set of a boiler production process by adopting a conversion function to obtain a standardized data set SDATA ═ Sk，Tk1., K }, where Sk∈RnNormalized process parameters in the kth sample;
and step 3: determining the importance of the standardized process parameters by adopting a boost tree method, extracting the process parameters with the importance greater than an importance threshold gamma, and obtaining a standardized process parameter set CDATA ═ C after feature extractionk，Tk1,. K }, wherein,
Figure BDA0001568266550000011
normalized process parameter, n, in the k-th sample after feature selectioncFor feature extractionThe dimensions of the subsequent normalized process parameters;
and 4, step 4: adopting a K-adjacent value classifier to classify the working conditions of the standardized process parameter set CDATA after the feature extraction to obtain various working condition data sets LDATAi＝{Lp(i)，Tp(i)1, p (i), and collecting all kinds of working condition data sets LDATAiDivision into training data sets xLDATAi＝{Lh(i)，Th(i)1,. h, (i) and a test data set, ctldata }i＝{Le(i)，Te(i)|e(i)＝1，...，E(i)}；
Wherein, I is 1, 2, I is the number of working conditions,
Figure BDA0001568266550000021
is a process parameter in the p sample of the i-th working condition, Tp(i)The combustion efficiency of the boiler in the p sample under the i-type working condition is represented by P (i), the number of samples in the data set under the i-type working condition is represented by h (i), the number of the h sample in the training data set under the i-type working condition is represented by H (i), the number of samples in the training data set under the i-type working condition is represented by e (i), the number of the e sample in the testing data set under the i-type working condition is represented by E (i), and the number of samples in the testing data set under the i-type working condition is represented by E (i);
Figure BDA0001568266550000022
is a process parameter, T, in the h sample of the i-th class of operating conditions in the training dataseth(i)To train the boiler combustion efficiency in the h sample of the i-th condition in the data set,
Figure BDA0001568266550000023
for testing the process parameter, T, in the ith sample of the class i operating condition in the data sete(i)The combustion efficiency of the boiler in the ith sample of the ith working condition in the test data set is obtained;
and 5: aiming at various working condition data sets: training data set xLDATA of various working conditionsiAs input, LS-SVM measurement models of various working conditions are established, a differential evolution algorithm is adopted to optimize the LS-SVM measurement models of various working conditions, and a test data set cLDATA is adoptediFor optimizationThe LS-SVM measurement model is tested to obtain the LS-SVM measurement model of the boiler combustion efficiency of various working conditions after final optimization;
taking the i-th working condition as an example, the method is executed according to the following steps:
a: initializing parameters of a differential evolution algorithm;
the parameters of the differential evolution algorithm comprise: population size NαIndividual dimension D, maximum iteration algebra G, mutation factor F, cross probability CR ∈ [0, 1 ∈ ]]Initialized population L of differential evolution algorithmα，0Wherein α ═ 1, 2.. Nα；
B: obtaining LS-SVM parameter values according to the information of each individual in the population to obtain NαA set of LS-SVM parameter values;
c: taking the radial basis function as an LS-SVM kernel function, training an LS-SVM measurement model established by aiming at LS-SVM parameter values corresponding to each individual in the population and a training data set of the i-th working condition to obtain NαAn LS-SVM measurement model;
the LS-SVM model of the i-th working condition is as follows:
Figure BDA0001568266550000024
wherein h (i) is the number of training samples in the training data set of the i-th working condition, L (i) · 1, 2(i)New sample data for calculating the combustion efficiency of the boiler is needed for the i-th working condition,
Figure BDA0001568266550000031
inputting data L of i-th working condition correspondingly(i)Of the LS-SVM measurement model, ah(i)、b(i)Measuring model parameters, L, for LS-SVM under class i operating conditionsh(i)Is the process parameter of the h training sample in the training data set of the i-th working condition, K (L)(i)，Lh(i)) The kernel function of the LS-SVM measurement model under the i-th working condition is used;
kernel function K (L) of LS-SVM measurement model under i-th working condition(i)，Lh(i)) As shown below：
Figure BDA0001568266550000032
Wherein σ(i) 2The width of the kernel function of the LS-SVM under the i-th working condition is obtained;
d: inputting the test data of the i-th working condition into the LS-SVM model of the i-th working condition established by each individual in the population, and calculating the root mean square error value epsilon of the LS-SVM model of the i-th working condition established by each individual in the populationαI.e. the value of the fitness function f for each individual in the populationα＝εα；
The root mean square error value epsilon of the LS-SVM model of the i-th working condition established by each individual in the populationαThe calculation formula of (a) is as follows:
Figure BDA0001568266550000033
wherein,
Figure BDA0001568266550000034
calculating an LS-SVM measurement model output value of the process parameter in the e sample in the ith working condition test data set by adopting the alpha individual corresponding measurement model;
e: judging whether the current iteration times G reach the maximum iteration algebra G, if so, finishing the iteration, and acquiring the optimal LS-SVM parameter and the model parameter ah(i)、b(i)Obtaining an LS-SVM measurement model of the boiler combustion efficiency under the i-th working condition after final optimization, and executing the step 6, otherwise, executing the step F;
f: and updating the g-th iteration population, making the iteration algebra g equal to g +1, and returning to B.
F-1: alpha individual L for the current iterationα，gRandomly selecting three individuals in the current population
Figure BDA0001568266550000035
And
Figure BDA00015682665500000312
will be provided with
Figure BDA0001568266550000036
And
Figure BDA0001568266550000037
after variation of the difference with the individual
Figure BDA0001568266550000038
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gWherein z is1，z2，z3∈[1，Nα]；
The device is to
Figure BDA0001568266550000039
And
Figure BDA00015682665500000310
after variation of the difference with the individual
Figure BDA00015682665500000311
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gThe calculation formula of (a) is as follows:
Figure BDA0001568266550000041
f-2: variant individuals V of the g-th iterationα，gAnd the alpha individual L of the g iterationα，gPerforming cross operation to generate a new individual U of the g-th iterationα，g；
The j component of the new individual generating the g iteration
Figure BDA0001568266550000042
The formula of (a) is as follows:
Figure BDA0001568266550000043
wherein j is 1jIs [0, 1 ]]Random numbers in the interval, randnα∈[1，D]A random integer of (a);
f-3: determining new individual U for the g-th iterationα，gWith the alpha individual L of the g iterationα，gThe individual with small fitness function value is taken as the g +1 th iteration population individual Lα，g+1Namely, the next generation individual;
the g +1 th iteration population individual Lα，g+1The calculation formula of (a) is as follows:
Figure BDA0001568266550000044
wherein, f () is the corresponding individual fitness function value;
f-4: let iteration algebra g be g +1, return to B.
Step 6: saving the optimized LS-SVM measurement models of the boiler combustion efficiency under various working conditions;
and 7: reading the online measured value of the process parameter of the boiler production process in real time, and measuring the combustion efficiency of the coal-fired boiler by adopting LS-SVM (least squares-support vector machine) measurement models of the boiler combustion efficiency under various working conditions;
step 7-1: reading the on-line measured value of the process parameter of the boiler production process in real time, standardizing the process parameter of the boiler production process by adopting a transfer function to obtain the standardized process parameter
Figure BDA0001568266550000045
Step 7-2: according to the normalized process parameters
Figure BDA0001568266550000046
And classifying the clustering center distances of various working conditions, inputting the clustering center distances into an LS-SVM measurement model of the optimized boiler combustion efficiency for determining the working condition types, and acquiring an online measurement value of the boiler combustion efficiency.
The calculation formula for standardizing the process parameters of the data set of the boiler production process by adopting the conversion function is as follows:
Figure BDA0001568266550000047
the process parameters of the boiler production process comprise: the economizer water supply temperature, the superheater wall temperature, the furnace flue gas pressure, the main steam temperature, the reheater wall temperature, the reheater steam pressure, the reheater steam temperature, the furnace flue gas temperature, the high-pressure water supply pressure, the superheater desuperheater steam temperature, the generator active power, the primary air outlet flow, the primary air temperature, the secondary air inlet flow, the superheater desuperheater inlet flow, the water supply flow, the main steam flow, the total air volume, the oxygen supply volume, the coal supply volume and the secondary air door position feedback at unit moment.
The invention has the beneficial effects that:
the invention provides an on-line measurement method for combustion efficiency of a coal-fired boiler, which can realize high-precision data-driven modeling according to related variables to obtain real-time boiler combustion efficiency and can meet production requirements; the method has wide application range, is suitable for vector expression of different data sources, and has good adaptability; the method can conveniently share information with other links, and is convenient for providing reference information for the operation of other links.
Drawings
FIG. 1 is a flow chart of a method for on-line measurement of combustion efficiency of a coal-fired boiler in an embodiment of the present invention;
FIG. 2 is a flowchart of a process for optimizing LS-SVM measurement models for various operating conditions using a differential evolution algorithm in an embodiment of the present invention;
FIG. 3 is a graph of boiler combustion efficiency modeled using a DE-LSSVM model in accordance with an embodiment of the present invention;
FIG. 4 is a graph of boiler combustion efficiency measured using a DE-LSSVM model in accordance with an embodiment of the present invention.
Detailed Description
The technical solution in the embodiment of the present invention will be clearly and completely described below with reference to the drawings in the embodiment of the present invention, in which the embodiment of the present invention is implemented in a hardware environment: one PC, CPU: 2.50GHz, RAM: 2.00GB, software Environment: MATLAB R2014a, operating platform: windows XP.
The invention provides an on-line measuring method for combustion efficiency of a coal-fired boiler, which comprises the following steps as shown in figure 1:
step 1: acquiring process parameters of a boiler production process through a plant-level monitoring information system of a thermal power plant, calculating boiler combustion efficiency at a corresponding moment according to a mechanism model, and obtaining a DATA set DATA ═ X of the boiler production processk，Tk1., K }, where X is Xk∈RnIs the process parameter in the kth sample, TkAnd the boiler combustion efficiency of the kth sample, K is the number of samples in the data set, and n is the dimension of the process parameter.
In this embodiment, the 93 process parameters of the initial boiler production process include: economizer feedwater temperature (3 points of measurement) at unit time, superheater wall temperature (4 points of measurement), furnace flue gas pressure (17 points of measurement), main steam pressure (3 points of measurement), main steam temperature (4 points of measurement), reheater wall temperature, reheater steam pressure (5 points of measurement), reheater steam temperature (4 points of measurement), furnace flue gas temperature (6 points of measurement), high-pressure feedwater temperature, high-pressure feedwater pressure, superheater desuperheater steam temperature (8 points of measurement), generator active power, primary air outlet flow (4 points of measurement), primary air temperature (6 points of measurement), secondary air inlet flow (2 points of measurement), superheater desuperheater inlet flow (4 points of measurement), feedwater flow, main steam flow, total air volume, oxygen supply, coal supply volume, Secondary damper position feedback (8 measurement points) and data set details are shown in table 1.
In the present embodiment, the DATA set DATA of the boiler production process is { X ═ Xk，Tk1., 807}, the number of samples in the data set K807, and the process parameter dimension n 93.
TABLE 1 details of data sets for boiler production process
Input dimension Output dimension Sample spacing Number of training samples Number of samples tested
93 1 1min 486 321
Step 2: standardizing process parameters of a data set of a boiler production process by adopting a conversion function to obtain a standardized data set SDATA ═ Sk，Tk1., K }, where Sk∈RnIs the normalized process parameter in the kth sample.
In this embodiment, the calculation formula for performing the normalization processing on the process parameters of the data set in the boiler production process by using the transfer function is shown in formula (1):
Figure BDA0001568266550000061
and step 3: determining the importance of the standardized process parameters by using a boost tree method, wherein the extraction importance is greater than the weightObtaining the process parameter of the key threshold gamma to obtain a normalized process parameter set CDATA ═ C after feature extractionk，Tk1,. K }, wherein,
Figure BDA0001568266550000062
normalized process parameter, n, in the k-th sample after feature selectioncIs the dimension of the normalized process parameter after feature extraction.
In the present embodiment, the importance threshold γ is 0.1, the importance of the normalized process parameter is determined by using a Boosted tree method to obtain a normalized process parameter set after feature extraction, and as shown in table 2, the dimension n of the normalized process parameter after feature extraction isc＝16。
TABLE 2 normalized Process parameters after feature extraction
Figure BDA0001568266550000063
Figure BDA0001568266550000071
And 4, step 4: adopting a K-adjacent value classifier to classify the working conditions of the standardized process parameter set CDATA after the feature extraction to obtain various working condition data sets LDATAi＝{Lp(i)，Tp(i)1, p (i), and collecting all kinds of working condition data sets LDATAiDivision into training data sets xLDATAi＝{Lh(i)，Th(i)1,. h, (i) and a test data set, ctldata }i＝{Le(i)，Te(i)|e(i)＝1，...，E(i)}；
Wherein, i is 1, 2.., 6,
Figure BDA0001568266550000072
is a process parameter in the p sample of the i-th working condition, Tp(i)The combustion efficiency of the boiler in the p sample under the i type working condition, P (i) is the number of samples in the data set under the i type working condition, h (i) is the hThe number of h sample in the training data set of the i-type working condition, H (i) the number of samples in the training data set of the i-type working condition, e (i) the number of e sample in the testing data set of the i-type working condition, and E (i) the number of samples in the testing data set of the i-type working condition;
Figure BDA0001568266550000073
is a process parameter, T, in the h sample of the i-th class of operating conditions in the training dataseth(i)For the boiler combustion efficiency in the h sample of the i-th condition in the training set data set,
Figure BDA0001568266550000074
is a process parameter, T, in the e sample of the i-th class of operating conditions in the test set data sete(i)The combustion efficiency of the boiler in the e sample of the i type operating condition in the test set data set is tested.
In the present embodiment, the normalized process parameter set after feature extraction is classified into I-6 classes, the training data set and the test data set of each class of operating condition are shown in table 3,
TABLE 3 training data set and test data set for various operating conditions
Figure BDA0001568266550000075
Figure BDA0001568266550000081
And 5: aiming at various working condition data sets: training data set xLDATA of various working conditionsiAs input, LS-SVM measurement models of various working conditions are established, a differential evolution algorithm is adopted to optimize the LS-SVM measurement models of various working conditions, and a test data set cLDATA is adoptediAnd testing the optimized LS-SVM measurement model to obtain the finally optimized LS-SVM measurement model of the boiler combustion efficiency under various working conditions.
In this embodiment, taking the sixth type of operating condition as an example, a specific process is shown in fig. 2.
A: a differential evolution algorithm (DE) parameter is initialized.
The parameters of the differential evolution algorithm comprise: population size N α50, 2 individual dimension D, 500 maximum iteration algebra G, 0.8 mutation factor F and 0.7 cross probability CR, and initialized population L of differential evolution algorithmα，0Wherein α ═ 1, 2.., 50.
In this embodiment, the alpha-th individual of the population of the initialized differential evolution algorithm is
Figure BDA0001568266550000082
Initiating the jth component of the alpha individuals of the differential evolution algorithm population
Figure BDA0001568266550000088
The formula (2) is shown as follows:
Figure BDA0001568266550000083
wherein j is 1, 2,
Figure BDA0001568266550000084
in order to specify a search space lower threshold,
Figure BDA0001568266550000085
for a given search space ceiling threshold, rand is [0, 1 ]]Random numbers within the interval.
B: obtaining LS-SVM parameter values according to the information of each individual in the population to obtain NαSet LS-SVM parameter values.
In this embodiment, the LS-SVM parameter values specifically include a penalty factor C and a width σ of the radial basis function2。
C: and taking the radial basis function as an LS-SVM kernel function, and training to obtain 50 LS-SVM measurement models by aiming at LS-SVM parameter values corresponding to each individual in the population and LS-SVM measurement models established by the training data set of the sixth working condition.
The LS-SVM model of the sixth working condition is shown as formula (3):
Figure BDA0001568266550000086
wherein H (6) ═ 1, 2., H (6) is 141 for the number of training samples in the training data set of the sixth type of operating condition, L(6)New sample data for boiler efficiency needs to be calculated for the sixth class of operating conditions,
Figure BDA0001568266550000087
sample data L for calculating boiler combustion efficiency corresponding to input(6)Of the LS-SVM measurement model, ah(6)、b(6)LS-SVM measurement model parameter for the sixth class of operating conditions, Lh(6)Is the process parameter of the h sample in the training data set for the sixth class of operating conditions, K (L)(6)，Lh(6)) Is the kernel function of the LS-SVM measurement model.
In the present embodiment, the kernel function K (L) of the LS-SVM measurement model for the sixth class of operating conditions(6)，Lh(6)) As shown in formula (4):
Figure BDA0001568266550000091
wherein σ(6) 2The width of the LS-SVM kernel for the sixth class of operating conditions.
D: inputting the test data of the sixth working condition into the LS-SVM model of the sixth working condition established by each individual in the population, and calculating the root mean square error value epsilon of the LS-SVM model of the sixth working condition established by each individual in the populationαI.e. the value of the fitness function f for each individual in the populationα＝εα。
The root mean square error value epsilon of the LS-SVM model of the sixth working condition established by each individual in the populationαIs represented by equation (5):
Figure BDA0001568266550000092
wherein e (6) is the e sample number in the sixth working condition test data set; e (6) is that the sample number value of the sixth working condition test data set is 92;
Figure BDA0001568266550000093
the LS-SVM measurement model output value T of the process parameter in the e sample in the sixth working condition test data set calculated by adopting the alpha individual corresponding measurement modele(6)And correspondingly calculating the real value of the combustion efficiency of the boiler for the process parameters in the e sample in the test data set of the sixth working condition.
E: judging whether the current iteration times g reach the maximum iteration algebra 500, if so, finishing the iteration, and acquiring the optimal LS-SVM parameter and the model parameter ah(6)、b(6)Obtaining an LS-SVM measurement model of the boiler combustion efficiency under the finally optimized sixth working condition, and executing the step 6; otherwise, executing F;
f: and updating the g-th iteration population, making the iteration algebra g equal to g +1, and returning to B.
F-1: alpha individual L for the current iterationα，gRandomly selecting three individuals in the current population
Figure BDA0001568266550000094
And
Figure BDA0001568266550000095
will be provided with
Figure BDA0001568266550000096
And
Figure BDA0001568266550000097
after variation of the difference with the individual
Figure BDA0001568266550000098
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gWherein z is1，z2，z3∈[1，50]。
In this embodiment, the
Figure BDA0001568266550000101
And
Figure BDA0001568266550000102
after variation of the difference with the individual
Figure BDA0001568266550000103
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gIs represented by equation (6):
Figure BDA0001568266550000104
f-2: variant individuals V of the g-th iterationα，gAnd the alpha individual L of the g iterationα，gPerforming cross operation to generate a new individual U of the g-th iterationα，g。
In the present embodiment, the jth component of the new individual of the g-th iteration is generated
Figure BDA0001568266550000105
Is represented by formula (7):
Figure BDA0001568266550000106
wherein, randjIs [0, 1 ]]Random numbers in the interval, randnα∈[1，2]Is a random integer of (a).
F-3: determining new individual U for the g-th iterationα，gWith the alpha individual L of the g iterationα，gThe individual with small fitness function value is taken as the g +1 th iteration population individual Lα，g+1I.e. the next generation of individuals.
In this embodiment, the new individual U for the g +1 th iteration is determined by equation (5)α，g+1With the alpha individual L of the g iterationα，gThe fitness function value of (1).
G +1 th iteration population individual Lα，g+1Is represented by equation (8):
Figure BDA0001568266550000107
wherein, f () is the corresponding individual fitness function value.
F-4: let iteration algebra g be g +1, return to B.
Step 6: and storing the optimized LS-SVM measurement model of the boiler combustion efficiency under various working conditions.
And 7: and reading the on-line measured values of the process parameters in the boiler production process in real time, and measuring the combustion efficiency of the coal-fired boiler by adopting LS-SVM (least squares-support vector machine) measurement models of the boiler combustion efficiency under various working conditions.
In the present embodiment, a sample data example is shown in table 4.
TABLE 4 sample data example
Figure BDA0001568266550000108
Figure BDA0001568266550000111
Step 7-1: reading the on-line measured value of the process parameter of the boiler production process in real time, standardizing the process parameter of the boiler production process by adopting a transfer function to obtain a standardized process parameter CoThe sample data examples are shown in Table 4.
Step 7-2: according to the standardized process parameter CoAnd classifying the clustering center distances of the various working conditions, inputting the clustering center distances into an LS-SVM measurement model of the optimized boiler combustion efficiency for determining the working condition types as shown in Table 5, and obtaining an online measurement value of the boiler combustion efficiency.
In the present embodiment, sample data is used as an example. And classifying the standardized process parameter values and the clustering center distances of various working conditions, and finally determining that the on-line measured value of the boiler combustion efficiency is 85.8810% in an LS-SVM measurement model of the optimized boiler combustion efficiency input into a third working condition according to a majority decision method, wherein the boiler combustion efficiency at the corresponding moment is calculated by a mechanism model to be 86.0010% and the error value is 0.1196%.
TABLE 5 Cluster center distance between normalized Process parameters and various operating conditions
Figure BDA0001568266550000112
The results obtained by the method for on-line measurement of combustion efficiency of a coal-fired boiler (DE-LSSVM) and statistical analysis of measurement errors provided by the present invention are shown in tables 6 and 7, and the combustion efficiency of the boiler obtained by modeling and measurement of the DE-LSSVM model is shown in FIGS. 3 and 4.
TABLE 6 statistical analysis of boiler combustion efficiency modeling results
Figure BDA0001568266550000113
Figure BDA0001568266550000121
TABLE 7 statistical analysis of boiler combustion efficiency model measurements
Figure BDA0001568266550000122
From table 6, it is seen that the maximum root mean square error of modeling occurs 0.2968% when the DE-LSSVM model under the sixth working condition models, and the maximum average relative error of modeling occurs 0.0345% when the DE-LSSVM model under the sixth working condition models; from Table 7, the maximum root mean square error during measurement is 1.8525% when the DE-LSSVM model under the sixth working condition is used for measurement; the maximum average relative error is measured to be 0.0474% of that of the DE-LSSVM model under the sixth working condition. In the actual operation, the modeling process is performed off-line, the calculation time is the measurement time, and in the experiment, the average calculation time of each group of data is 1.719 ms. In both aspects of calculation time and measurement accuracy, the method provided by the invention can better meet the production requirement.

Claims (5)

1. An on-line measuring method for combustion efficiency of a coal-fired boiler is characterized by comprising the following steps:
step 1: acquiring process parameters of a boiler production process through a plant-level monitoring information system of a thermal power plant, calculating boiler combustion efficiency at a corresponding moment according to a mechanism model, and obtaining a DATA set DATA ═ X of the boiler production processk，Tk1., K }, where X is Xk∈RnIs the process parameter in the kth sample, TkThe boiler combustion efficiency of the kth sample is obtained, K is the number of samples in the data set, and n is the dimension of the process parameter;
step 2: standardizing process parameters of a data set of a boiler production process by adopting a conversion function to obtain a standardized data set SDATA ═ Sk，Tk1., K }, where Sk∈RnNormalized process parameters in the kth sample;
and step 3: determining the importance of the standardized process parameters by adopting a boost tree method, extracting the process parameters with the importance greater than an importance threshold gamma, and obtaining a standardized process parameter set CDATA ═ C after feature extractionk，Tk1,. K }, wherein,
Figure FDA0001568266540000011
normalized process parameter, n, in the k-th sample after feature selectioncDimension of the standardized process parameter after feature extraction;
and 4, step 4: adopting a K-adjacent value classifier to classify the working conditions of the standardized process parameter set CDATA after the feature extraction to obtain various working condition data sets LDATAi＝{Lp(i)，Tp(i)1, p (i), and collecting all kinds of working condition data sets LDATAiDivision into training data sets xLDATAi＝{Lh(i)，Th(i)1,. h, (i) and a test data set, ctldata }i＝{Le(i)，Te(i)|e(i)＝1，...，E(i)}；
Wherein, I is 1, 2, I is the number of working conditions,
Figure FDA0001568266540000012
is a process parameter in the p sample of the i-th working condition, Tp(i)The combustion efficiency of the boiler in the p sample under the i-type working condition is represented by P (i), the number of samples in the data set under the i-type working condition is represented by h (i), the number of the h sample in the training data set under the i-type working condition is represented by H (i), the number of samples in the training data set under the i-type working condition is represented by e (i), the number of the e sample in the testing data set under the i-type working condition is represented by E (i), and the number of samples in the testing data set under the i-type working condition is represented by E (i);
Figure FDA0001568266540000013
is a process parameter, T, in the h sample of the i-th class of operating conditions in the training dataseth(i)To train the boiler combustion efficiency in the h sample of the i-th condition in the data set,
Figure FDA0001568266540000014
for testing the process parameter, T, in the ith sample of the class i operating condition in the data sete(i)The combustion efficiency of the boiler in the ith sample of the ith working condition in the test data set is obtained;
and 5: aiming at various working condition data sets: training data set xLDATA of various working conditionsiAs input, LS-SVM measurement models of various working conditions are established, a differential evolution algorithm is adopted to optimize the LS-SVM measurement models of various working conditions, and a test data set cLDATA is adoptediTesting the optimized LS-SVM measurement model to obtain the finally optimized LS-SVM measurement model of the boiler combustion efficiency under various working conditions;
step 6: saving the optimized LS-SVM measurement models of the boiler combustion efficiency under various working conditions;
and 7: reading the online measured value of the process parameter of the boiler production process in real time, and measuring the combustion efficiency of the coal-fired boiler by adopting LS-SVM (least squares-support vector machine) measurement models of the boiler combustion efficiency under various working conditions;
step 7-1: reading the on-line measured value of the process parameter of the boiler production process in real time, standardizing the process parameter of the boiler production process by adopting a transfer function to obtain the standardized process parameter
Figure FDA0001568266540000021
Step 7-2: according to the normalized process parameters
Figure FDA0001568266540000022
And classifying the clustering center distances of various working conditions, inputting the clustering center distances into an LS-SVM measurement model of the optimized boiler combustion efficiency for determining the working condition types, and acquiring an online measurement value of the boiler combustion efficiency.
2. The method of claim 1, wherein the formula for normalizing the process parameters of the data set of the boiler production process using the transfer function is as follows:
Figure FDA0001568266540000023
3. the method of claim 1, wherein the process parameters of the boiler production process comprise: the economizer water supply temperature, the superheater wall temperature, the furnace flue gas pressure, the main steam temperature, the reheater wall temperature, the reheater steam pressure, the reheater steam temperature, the furnace flue gas temperature, the high-pressure water supply pressure, the superheater desuperheater steam temperature, the generator active power, the primary air outlet flow, the primary air temperature, the secondary air inlet flow, the superheater desuperheater inlet flow, the water supply flow, the main steam flow, the total air volume, the oxygen supply volume, the coal supply volume and the secondary air door position feedback at unit moment.
4. The method for measuring the combustion efficiency of the coal-fired boiler on line according to the claim 1, characterized in that the method comprises the following steps for various working condition data sets: training data set xLDATA of various working conditionsiAs input, LS-SVM measurement models of various working conditions are established, a differential evolution algorithm is adopted to optimize the LS-SVM measurement models of various working conditions, and a test data set cLDATA is adoptediAnd testing the optimized LS-SVM measurement model to obtain the finally optimized LS-SVM measurement model of the boiler combustion efficiency under various working conditions, taking the i-th working condition as an example, and executing the following steps:
a: initializing parameters of a differential evolution algorithm;
the parameters of the differential evolution algorithm comprise: population size NαIndividual dimension D, maximum iteration algebra G, mutation factor F, cross probability CR ∈ [0, 1 ∈ ]]Initialized population L of differential evolution algorithmα，0Wherein α ═ 1, 2.. Nα；
B: obtaining LS-SVM parameter values according to the information of each individual in the population to obtain NαA set of LS-SVM parameter values;
c: taking the radial basis function as an LS-SVM kernel function, training an LS-SVM measurement model established by aiming at LS-SVM parameter values corresponding to each individual in the population and a training data set of the i-th working condition to obtain NαAn LS-SVM measurement model;
the LS-SVM model of the i-th working condition is as follows:
Figure FDA0001568266540000031
wherein h (i) is the number of training samples in the training data set of the i-th working condition, L (i) · 1, 2(i)New sample data for calculating the combustion efficiency of the boiler is needed for the i-th working condition,
Figure FDA0001568266540000032
inputting data L of i-th working condition correspondingly(i)Of the LS-SVM measurement model, ah(i)、b(i)Measuring model parameters, L, for LS-SVM under class i operating conditionsh(i)Is the process parameter of the h training sample in the training data set of the i-th working condition, K (L)(i)，Lh(i)) The kernel function of the LS-SVM measurement model under the i-th working condition is used;
kernel function K (L) of LS-SVM measurement model under i-th working condition(i)，Lh(i)) As follows:
Figure FDA0001568266540000033
wherein σ(i) 2The width of the kernel function of the LS-SVM under the i-th working condition is obtained;
d: inputting the test data of the i-th working condition into the LS-SVM model of the i-th working condition established by each individual in the population, and calculating the root mean square error value epsilon of the LS-SVM model of the i-th working condition established by each individual in the populationαI.e. the value of the fitness function f for each individual in the populationα＝εα；
The root mean square error value epsilon of the LS-SVM model of the i-th working condition established by each individual in the populationαThe calculation formula of (a) is as follows:
Figure FDA0001568266540000034
wherein,
Figure FDA0001568266540000035
calculating an LS-SVM measurement model output value of the process parameter in the e sample in the ith working condition test data set by adopting the alpha individual corresponding measurement model;
e: judging whether the current iteration times G reach the maximum iteration algebra G, if so, finishing the iteration, and acquiring the optimal LS-SVM parameter and the model parameter ah(i)、b(i)To obtainFinally, executing the step 6 by the LS-SVM measurement model of the boiler combustion efficiency under the i-th working condition after optimization, otherwise, executing the step F;
f: and updating the g-th iteration population, making the iteration algebra g equal to g +1, and returning to B.
5. The method for measuring the combustion efficiency of the coal-fired boiler according to claim 4, wherein the F comprises the following steps:
f-1: alpha individual L for the current iterationα，gRandomly selecting three individuals in the current population
Figure FDA0001568266540000041
And
Figure FDA0001568266540000042
will be provided with
Figure FDA0001568266540000043
And
Figure FDA0001568266540000044
after variation of the difference with the individual
Figure FDA0001568266540000045
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gWherein z is1，z2，z3∈[1，Nα]；
The device is to
Figure FDA0001568266540000046
And
Figure FDA0001568266540000047
after variation of the difference with the individual
Figure FDA0001568266540000048
Carrying out vector synthesis to obtain the variant individual V of the g iterationα，gThe calculation formula of (a) is as follows:
Figure FDA0001568266540000049
f-2: variant individuals V of the g-th iterationα，gAnd the alpha individual L of the g iterationα，gPerforming cross operation to generate a new individual U of the g-th iterationα，g；
The j component of the new individual generating the g iteration
Figure FDA00015682665400000410
The formula of (a) is as follows:
Figure FDA00015682665400000411
wherein j is 1jIs [0, 1 ]]Random numbers in the interval, randnα∈[1，D]A random integer of (a);
f-3: determining new individual U for the g-th iterationα，gWith the alpha individual L of the g iterationα，gThe individual with small fitness function value is taken as the g +1 th iteration population individual Lα，g+1Namely, the next generation individual;
the g +1 th iteration population individual Lα，g+1The calculation formula of (a) is as follows:
Figure FDA00015682665400000412
wherein, f () is the corresponding individual fitness function value;
f-4: let iteration algebra g be g +1, return to B.
CN201810107733.3A 2018-02-02 2018-02-02 Coal-fired boiler combustion efficiency online measurement method Active CN108182553B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810107733.3A CN108182553B (en) 2018-02-02 2018-02-02 Coal-fired boiler combustion efficiency online measurement method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810107733.3A CN108182553B (en) 2018-02-02 2018-02-02 Coal-fired boiler combustion efficiency online measurement method

Publications (2)

Publication Number Publication Date
CN108182553A CN108182553A (en) 2018-06-19
CN108182553B true CN108182553B (en) 2022-04-01

Family

ID=62552103

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810107733.3A Active CN108182553B (en) 2018-02-02 2018-02-02 Coal-fired boiler combustion efficiency online measurement method

Country Status (1)

Country Link
CN (1) CN108182553B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109492319B (en) * 2018-11-23 2022-08-12 东北电力大学 Soft measurement method for oxygen content of boiler flue gas of thermal power plant
CN110017470A (en) * 2019-02-27 2019-07-16 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 Boiler combustion multiobjective optimization control method based on differential evolution algorithm
CN110532674B (en) * 2019-08-27 2022-08-12 东北电力大学 Method for measuring temperature of boiler furnace of coal-fired power plant
CN113420461A (en) * 2021-07-20 2021-09-21 山东恒拓科技发展有限公司 Cement decomposing furnace online simulation system and establishing method thereof

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7756591B2 (en) * 2006-04-25 2010-07-13 Pegasus Technologies, Inc. System for optimizing oxygen in a boiler
CN103400015B (en) * 2013-08-15 2016-05-18 华北电力大学 Based on the combustion system combining modeling method of numerical simulation and test run data
CN105605610A (en) * 2016-01-13 2016-05-25 上海交通大学 Boiler combustion optimization method based on particle swarm algorithm
CN106327021A (en) * 2016-08-31 2017-01-11 西安艾贝尔科技发展有限公司 Boiler combustion optimization air distribution method based on online model prediction
CN106527176A (en) * 2016-10-21 2017-03-22 哈尔滨理工大学 MFOA (modified fruit fly optimization algorithm)-SVM (support vector machine)-based boiler thermal efficiency and NOX modeling method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
基于DE-SVM非线性组合预测模型的研究;陈涛;《计算机工程与应用》;20110501;第47卷(第13期);33-36 *
基于最小二乘支持向量机的燃煤锅炉燃烧优化研究;陈南锟;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20170315(第3期);C042-1734 *
特征选择 GBDT 特征重要度;默一鸣;《https://blog.csdn.net/yimingsilence/article/details/71713751百度快照》;20170512;1 *
赵文杰 等.基于多LS-SVM集成模型的锅炉NO_x排放量建模.《电子测量与仪器学报 》.2016,第30卷(第7期), *

Also Published As

Publication number Publication date
CN108182553A (en) 2018-06-19

Similar Documents

Publication Publication Date Title
CN108182553B (en) Coal-fired boiler combustion efficiency online measurement method
CN106249728B (en) A kind of thermal power generation unit on-line performance monitoring method based on characteristics of components
CN108446529B (en) Organic Rankine cycle system fault detection method based on generalized mutual entropy-DPCA algorithm
CN111754093B (en) Fly ash carbon content prediction method based on coal quality prediction and PSO-SVM
CN109670629B (en) Coal-fired boiler thermal efficiency prediction method based on long-short term memory neural network
CN111159844B (en) Abnormity detection method for exhaust temperature of gas turbine of power station
CN112163380B (en) Neural network hearth oxygen concentration prediction system and method based on numerical simulation
CN110532674A (en) A kind of coal-fired power station boiler fire box temperature measurement method
CN110207094B (en) IQGA-SVR boiler heating surface contamination characteristic identification method based on principal component analysis
CN106649919A (en) Method and system for predicting carbon content of fly ash in coal-fired power plant boiler
CN111352408B (en) Multi-working-condition process industrial process fault detection method based on evidence K nearest neighbor
CN109471420B (en) CVA-SFA-based method for monitoring control performance of air preheater of large coal-fired power generator set of intelligent power plant
CN115510904B (en) Boiler heating area ash monitoring method based on time sequence prediction
CN111931436A (en) Burner nozzle air quantity prediction method based on numerical simulation and neural network
CN106289754B (en) A kind of fired power generating unit component capabilities variation determination method based on Statistical Control Technique
CN109270917B (en) Intelligent power plant steam turbine bearing-oriented closed-loop control system fault degradation state prediction method
CN116401948A (en) Online prediction method and system for generating amount of power station boiler ash based on LSTM
CN116029433A (en) Energy efficiency reference value judging method, system, equipment and medium based on grey prediction
CN115526433A (en) Power plant reheat flue gas baffle operation prediction method based on integrated hybrid model
CN113688895A (en) Method and system for detecting abnormal firing zone of ceramic roller kiln based on simplified KECA
Zheng et al. The bidirectional information fusion using an improved LSTM model
CN111242345A (en) Nuclear power unit electric power prediction method based on cluster analysis and random forest regression
CN113779706B (en) Impeller mechanical loss model construction method based on data credibility
Zhou et al. Prediction model of coal consumption based on random forest variable selection and random-grid hyperparametric optimization algorithm
CN117664576B (en) Bearing health assessment method based on improved optimizer deep learning model

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