CN109765331A - A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine - Google Patents
A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine Download PDFInfo
- Publication number
- CN109765331A CN109765331A CN201811168969.4A CN201811168969A CN109765331A CN 109765331 A CN109765331 A CN 109765331A CN 201811168969 A CN201811168969 A CN 201811168969A CN 109765331 A CN109765331 A CN 109765331A
- Authority
- CN
- China
- Prior art keywords
- oxygen content
- particle
- smoke gas
- data
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Feedback Control In General (AREA)
Abstract
A kind of hard measurement system of the alkali recovery furnace oxygen content of smoke gas based on least square method supporting vector machine, including alkali recovery furnace combustion process, field intelligent instrument and sensor, PLC control station, DCS database and oxygen content of smoke gas display screen, the hard measurement system of alkali recovery furnace oxygen content of smoke gas based on least square method supporting vector machine, including module update module, data preprocessing module, principal component analysis and gray scale relating module, Chaos-Particle Swarm Optimization optimizing module and least square method supporting vector machine model module.The present invention can be in the case where directly not using zirconium oxide oxygen amount sensor, according to fire box temperature, combustion chamber draft, black liquor flow, air-supply electric current, air inducing electric current, six auxiliary variables of total blast volume actual measured value, the predicted value of oxygen content of smoke gas is obtained by oxygen content of smoke gas soft-sensing model, it can preferably solve the problems, such as that oxygen content of smoke gas is difficult to accurately predict, improve the efficiency of combustion of alkali recovery furnace.
Description
Technical field
It is the present invention relates to oxygen content of smoke gas field of measuring technique, in particular to a kind of based on least square method supporting vector machine
Oxygen content of smoke gas hard measurement system.
Background technique
In black liquid alkali recovery process, oxygen content of smoke gas can reflect alkali recovery furnace efficiency of combustion and heat utilization efficiency.
Oxygen content of smoke gas by real-time high-precision is measured, the combustion case of alkali recovery furnace is can reflect, it is timely to facilitate operator
The ratio for regulating and controlling air quantity and black liquor, makes system high efficiency energy-saving run.If oxygen content value is excessively high, indicate to enter in burner hearth excessive
Air, extra gas discharge will take away amount of heat, so that the alkali recovery furnace thermal efficiency reduces, also result in more sulphur
The polluted gas such as compound are discharged into air, pollute environment.And if oxygen content value is too low, burner hearth is in oxygen debt state, black liquor
It can not be sufficiently burned, combustion thermal efficiency reduces, and causes to aggravate in flue gas containing gases such as carbon monoxide, hydrogen, hydrogen sulfide
Environmental pollution.Therefore it needs strictly to monitor oxygen content of smoke gas, keeps suitable oxygen content value.
Alkali recovery system is mainly by zirconia sensor and Thermomagnetic type sensor measurement oxygen content of smoke gas at present, this
A little devices can directly measure oxygen content of smoke gas, and measurement accuracy is higher, and reaction speed is fast, but with the variation of time, zirconium oxide
Popping one's head in, it is easy to aging to be blocked by dust, so that oxygen amount meter is difficult to obtain higher stability and accuracy, and repairs complicated, expense
Expensive, the regular effect of needs.In addition, the zirconium head of zirconia sensor is mounted on the node of smoke canal elbow, unstable flue gas
Change in flow will affect the measurement work of zirconium head, and by the metal pipe abrasion of zirconium head, its service life be caused to greatly shorten.
Summary of the invention
In order to overcome zirconia probe easy to aging, so that the oxygen amount meter is difficult to obtain asking for higher stability and accuracy
Topic, the oxygen content of smoke gas hard measurement system based on least square method supporting vector machine that the purpose of the present invention is to provide a kind of can be with
In the case where directly not using zirconium oxide oxygen amount sensor, by fire box temperature, combustion chamber draft, black liquor flow, air-supply electricity
The actual measured value of six stream, air inducing electric current and total blast volume auxiliary variables, the flue gas oxygen based on least square method supporting vector machine contain
Amount soft-sensing model obtains oxygen content predicted value, can preferably solve the problems, such as that oxygen content of smoke gas is difficult to accurately predict, improves
The efficiency of combustion of alkali recovery furnace.
To achieve the goals above, the technical solution adopted by the present invention is that:
A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine, including field intelligent instrument and biography
Sensor 2, field intelligent instrument and sensor 2 are connect with alkali recovery furnace combustion process 1, while being also connected with DCS database 3, PLC
Control station 5 is connected respectively at alkali recovery furnace combustion process 1 with DCS database 3, DCS database 3 and oxygen content of smoke gas hard measurement
Model 4 connects, and variable data is transferred to DCS database 3 by the auxiliary variable of field instrument and the measurement alkali recovery furnace of sensor 2,
Input data of the data as oxygen content of smoke gas model 4 in DSC data library 3, PLC control station 5 is for controlling performance variable
Alkali recovery furnace combustion process 1 is controlled, oxygen content of smoke gas hard measurement value display screen 6 is used to show the data of oxygen content of smoke gas model 4,
The hard measurement system of the oxygen content of smoke gas model 4 includes；
(1) data preprocessing module: data scrubbing is carried out to the data acquired from DCS database, is rejected using 3 σ methods
Fault data are on this basis smoothed data using five-spot triple smoothing, reject random error；It uses again
Min-max standardized method carries out dimension to data and unitizes, and sample point is projected to [0,1] section；
(2) principal component analysis and gray scale relating module: number and type for selecting auxiliary variable choose cumulative variance
Contribution rate method determines principal component number, utilizes variance contribution ratio δiWith the accumulative variance contribution ratio η of preceding k pivotkTo measure
Information covers ratio, replaces original m variable with 85% or more information k (k < m) a pivot of covering of selection；Utilize grey
The degree of association analyzes the curvilinear trend relationship of each variable and oxygen content of smoke gas, determines the variable strong with oxygen content of smoke gas correlation；
(3) least square method supporting vector machine model module: use RBF gaussian kernel function as oxygen content of smoke gas hard measurement mould
The kernel function of type, and least square method supporting vector machine can overcome the correlation between variable；
(4) Chaos-Particle Swarm Optimization optimizing module: right with the Chaos particle swarm optimization algorithm (CPSO) with Stochastic inertia weight
The regularization parameter of LSSVM and nuclear parameter optimizing.
In the step (4), steps are as follows for the parameter optimization of the CPSO algorithm with Stochastic inertia weight:
Step 1: initialization CPSO parameter: population scale d, maximum number of iterations Gmax, Studying factors c1,c2, nuclear parameter σ and
The value range of regular parameter γ；
Step 2: d particle being randomly generated in determined range, each particle initial position is (σ0,γ0)；
Step 3: establishing LSSVM model using the position and training data of each particle, and using test data according to formulaSolve the fitness value x of each particleid, y in formulaiIt is true value,It is predicted value, then carries out
Sequence, sets p for each particle history adaptive optimal control angle valuebest, population is set by the smallest particle position of fitness value
History adaptive optimal control degree position gbest；
Step 4: the fitness value of solution and more each particle, if the current fitness value of each particle is optimal less than its history
Fitness value, i.e. xid< pbest, then current pbestIt is set as xidIf similarly current particle adaptive optimal control angle value is suitable less than global optimum
Answer angle value, i.e. xid< gbest, then g is enabledbestEqual to xid；
Step 5: according to formulaGenerate Stochastic inertia weight, it is assumed that optimizing Spatial Dimension is D, population
Number is N, wherein i-th of particle is expressed as Xi=(xi1,xi2,L,xiD), the desired positions of particle experience are denoted as Pi=(pi1,
pi2,L,piD), it is also possible to pbestIt indicates, the desired positions of whole particle experience are denoted as P in populationg=(pg1,pg2,L,pgD),
G can also be usedbestIt indicates.The speed V of particlei=(vi1,vi2,L,viD) indicate.For every generation particle, its d ties up (1≤d
≤ D) speed and position update according to the following formula:With
xid(k+1)=xid(k)+vid(k+1), so that the speed of more new particle and position, generate particle of new generation, wherein w is inertia
Weight, rand () are the random number between (0,1), c1It is local optimum specific gravity, c2It is global optimum's specific gravity；
Step 6: chaos processing is carried out to population, retains the smallest preceding N × P particle of fitness function value in population,
Wherein P indicates substitution probability, chaos masking search is carried out to this N × P particle, according to formulaBy chaos vector
[0, the 1] solution space of linear projection to Logistic system, xiFor optimal solution, xmax,xminFor optimal solution bound, further according to formula
zn+1=μ zn(1-zn), 0 < znThe grey iterative generation chaos sequence iteration initial value of < 1, μ=4 z0It cannot be 0, by the chaos sequence of generation
Column pass through formula xi=xmin+(xmax-xmin)*ziInverse mapping replaces original N to obtain real optimal solution to former solution space
× P population particle；
Step 7: judging whether that satisfaction reaches maximum number of iterations or precision of prediction, if satisfied, then LSSVM parameter optimization is temporary
Stop；Conversely, return step 4 continues searching LSSVM parameter；
Step 8: the optimal hyper parameter found being input in LSSVM algorithm, to oxygen content of smoke gas training sample
It practises, establishes soft-sensing model, oxygen content of smoke gas hard measurement is carried out to test data.
Temperature sensor, pressure sensor, electromagnetic flowmeter in the present invention, PLC control station, DCS database are existing
There is equipment, details are not described herein.Particle swarm optimization algorithm and support vector machines are also well-known technique, and details are not described herein.
Beneficial effects of the present invention:
The invention has the advantages that 1) can be in the case where directly not using zirconium oxide oxygen amount sensor, by right
Fire box temperature, combustion chamber draft, black liquor flow, air-supply electric current, six auxiliary variables of air inducing electric current and total blast volume actual measured value,
Oxygen content predicted value is obtained with the oxygen content of smoke gas soft-sensing model based on least square method supporting vector machine, is to solve zirconium oxide
Sensor is easy to wear and improves the efficiency of combustion effective measures of alkali recovery furnace；2) on-line measurement.
Detailed description of the invention
Fig. 1 is the hard measurement system structure signal of the alkali recovery furnace oxygen content of smoke gas based on least square method supporting vector machine
Figure.
Fig. 2 is least square method supporting vector machine of the present invention (LSSVM) soft-sensing model structure chart.
Fig. 3 is the CPSO algorithm parameter searching process figure that the present invention has Stochastic inertia weight.
Fig. 4 is oxygen content of smoke gas hard measurement value and measured value absolute error figure.
Fig. 5 is alkali recovery furnace technique flow chart.
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
1, referring to attached drawing 1, field intelligent instrument and sensor 2 are connect with alkali recovery furnace combustion process 1, at the same also with DCS
Database 3 is connected, and PLC control station 5 is connected respectively at alkali recovery furnace combustion process 1 with DCS database 3, DCS database 3 and cigarette
Gas oxygen content soft-sensing model 4 connects.Field instrument and sensor 2 measure the auxiliary variable of alkali recovery furnace, and variable data is passed
Defeated to arrive DCS database 3, input data of the data as oxygen content of smoke gas model 4 in DSC data library 3, PLC control station 5 is used for
Performance variable is controlled to control alkali recovery furnace combustion process 1, oxygen content of smoke gas hard measurement value display screen 6 is for showing that flue gas oxygen contains
The data of model 4 are measured, a kind of hard measurement system of the alkali recovery furnace oxygen content of smoke gas based on least square method supporting vector machine includes:
(1) data preprocessing module: data scrubbing is carried out to the data acquired from DCS database, is rejected using 3 σ methods
Fault data are on this basis smoothed data using five-spot triple smoothing, reject random error；It uses again
Min-max standardized method carries out dimension to data and unitizes, and sample point is projected to [0,1] section；
(2) principal component analysis and gray scale relating module: number and type for selecting auxiliary variable.Choose cumulative variance
Contribution rate method determines principal component number, utilizes variance contribution ratio δiWith the accumulative variance contribution ratio η of preceding k pivotkTo measure
Information covers ratio, replaces original m variable with 85% or more information k (k < m) a pivot of covering of selection；Utilize grey
The degree of association analyzes the curvilinear trend relationship of each variable and oxygen content of smoke gas, determines the variable strong with oxygen content of smoke gas correlation；
(3) least square method supporting vector machine model module: use RBF gaussian kernel function as oxygen content of smoke gas hard measurement mould
The kernel function of type, and least square method supporting vector machine can overcome the correlation between variable.
(4) Chaos-Particle Swarm Optimization optimizing module: right with the Chaos particle swarm optimization algorithm (CPSO) with Stochastic inertia weight
The regularization parameter of LSSVM and nuclear parameter optimizing；
2, least square method supporting vector machine (LSSVM) modeling procedure is as follows:
Assuming that data set is (xi,yi), i=1,2, L, m, xi∈Rn, x hereiniRefer to oxygen content of smoke gas hard measurement mould
Each auxiliary variable in type, n indicate that the dimension of input sample, m indicate number of samples.yiIndicate i-th of output, i.e. flue gas oxygen contains
The measured value of amount.Feature space LSSVM model is represented by
In formula:
--- sample data is mapped to high-dimensional feature space from lower dimensional space by mapping function；
W --- weight vector；
B --- amount of bias.
The objective function of LSSVM can be described as
In formula, ζiIt is the error variance of i-th of sample, γ is the compromise parameter between model generalization ability and precision, is claimed
For regularization parameter, generally higher than 0.
Lagrangian is constructed to solve, as follows:
In formula, α=[α1,α2,L,αm]T--- Lagrange multiplier.
According to KKT (Karush-Kuhn-Tucker) condition, partial derivative is asked to each variable of Lagrangian, can be obtained:
It is converted into matrix equation, as shown in formula:
In formula,--- meet the kernel function of Mercer condition.
After determining RBF gaussian kernel function, according toα, b are acquired, LSSVM is obtained
Nonlinear model:
Wherein, gaussian kernel functionσ is nuclear parameter, determines the diameter of RBF gaussian kernel function
To sphere of action and width, attached drawing 2 is LSSVM soft-sensing model structure chart.
3, referring to attached drawing 3, that steps are as follows is described for the parameter optimization of the CPSO algorithm with Stochastic inertia weight:
Step 1: initialization CPSO parameter: population scale d, maximum number of iterations Gmax, Studying factors c1,c2, nuclear parameter σ and
The value range of regular parameter γ；
Step 2: d particle being randomly generated in determined range, each particle initial position is (σ0,γ0)；
Step 3: establishing LSSVM model using the position and training data of each particle, and using test data according to formulaSolve the fitness value x of each particleid, it is ranked up, each particle history adaptive optimal control angle value is set
It is set to pbest, set the smallest particle position of fitness value to the history adaptive optimal control degree position g of populationbest；
Step 4: the fitness value of solution and more each particle, if the current fitness value of each particle is optimal less than its history
Fitness value, i.e. xid< pbest, then current pbestIt is set as xid.If similarly current particle adaptive optimal control angle value is suitable less than global optimum
Answer angle value, i.e. xid< gbest, then g is enabledbestEqual to xid；
Step 5: according to formulaStochastic inertia weight is generated, according toAnd xid(k+1)=xid(k)+vid(k+1) more new particle
Speed and position, generate particle of new generation；
Step 6: chaos processing is carried out to population, retains the smallest preceding N × P particle of fitness function value in population,
Wherein P indicates substitution probability, chaos masking search is carried out to this N × P particle, according to formulaBy chaos vector
Linear projection to Logistic system [0,1] solution space, further according to formula zn+1=μ zn(1-zn), 0 < znThe iteration of < 1, μ=4
Chaos sequence is generated, the chaos sequence of generation is passed through into formula xi=xmin+(xmax-xmin)*ziInverse mapping is to former solution space, to obtain
Real optimal solution out replaces N × P original population particle；
Step 7: judging whether that satisfaction reaches maximum number of iterations or precision of prediction, if satisfied, then LSSVM parameter optimization is temporary
Stop；Conversely, return step 4 continues searching LSSVM parameter；
Step 8: the optimal hyper parameter found being input in LSSVM algorithm, to oxygen content of smoke gas training sample
It practises, establishes soft-sensing model, oxygen content of smoke gas hard measurement is carried out to test data.
4, the oxygen content of smoke gas value gone out according to hard measurement system prediction, total blast volume in furnace should be passed through in real time by calculating
Value, always works at alkali recovery furnace in high efficiency range.Table 1 is different moments oxygen content of smoke gas measured value and hard measurement value.
Table 1 different moments oxygen content of smoke gas measured value and hard measurement value
As shown in Figure 4: curve is oxygen content of smoke gas hard measurement value and measured value absolute error, and absolute error is in 0.01-
It is fluctuated between 0.025, the oxygen amount value for illustrating that built oxygen content of smoke gas soft-sensing model obtains can be used as reference, facilitate in time
Existing oxygen amount meter is corrected, boiler combustion efficiency is improved.
As shown in Figure 5: what is provided in figure is alkali recovery furnace technique flow chart.The main material of alkali recovery furnace burning is to come from
The concentrated black liquid of evaporation section.Concentrated black liquid is through cascade evaporator, black liquor slot, the processing of concentrated black liquid heater, by black liquor spray gun high pressure spray
Enter burner hearth.In alkali recovery furnace, black liquor drop is by atomization, dry after-combustion.Grey black after part is burnt falls on alkali recovery furnace
Bed course on, grey black burns away, and inorganic matter constantly melts, and partial organic substances charcoal turns to elemental carbon, for burning and restoring sulfuric acid
Sodium is used.After the reaction was completed, saltcake is reduced into vulcanized sodium, and the fusant of generation is flowed out from chute.Air needed for black-liquor combustion
It is sent into air preheater heating through pressure fan, after air is heated to 150 DEG C, after distributing according to a certain percentage, is passed through respectively
Primary and secondary air tuyere is sent into combustion furnace；Tertiary air is directly sent into combustion furnace by air blower without preheating.What black-liquor combustion generated
High-temperature flue gas is taken out after the equipment such as economizer, cascade evaporator absorb waste heat, then after electrostatic precipitator dedusting by air-introduced machine
It is discharged out through chimney.The fusant that black-liquor combustion generates is expelled to dissolving tank through smelt spout, send after green liquor strainer
Caustic room carries out causticization, to recycle industrial soda.
Claims (2)
1. a kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine, which is characterized in that including live intelligence
Can instrument and sensor (2), field intelligent instrument and sensor (2) connect with alkali recovery furnace combustion process (1), while also with
DCS database (3) is connected, and PLC control station (5) is connected with alkali recovery furnace combustion process (1) and DCS database (3) respectively, DCS
Database (3) is connect with oxygen content of smoke gas soft-sensing model (4), the auxiliary of field instrument and sensor (2) measurement alkali recovery furnace
Variable data is transferred to DCS database (3) by variable, and the data in DSC data library (3) are as oxygen content of smoke gas model (4)
Input data, PLC control station (5) are used to control performance variable the soft survey of oxygen content of smoke gas to control alkali recovery furnace combustion process (1)
Magnitude display screen (6) is used to show the data of oxygen content of smoke gas model (4)；
The hard measurement system of the oxygen content of smoke gas model (4) includes；
(1) data preprocessing module: carrying out data scrubbing to the data acquired from DCS database, rejects fault using 3 σ methods
Data are on this basis smoothed data using five-spot triple smoothing, reject random error；Again using minimum-
Maximum standardization method carries out dimension to data and unitizes, and sample point is projected to [0,1] section；
(2) principal component analysis and gray scale relating module: number and type for selecting auxiliary variable choose cumulative variance contribution
Rate method determines principal component number, utilizes variance contribution ratio δiWith the accumulative variance contribution ratio η of preceding k pivotkCarry out scaling information
Cover ratio, replaces original m variable with 85% or more information k (k < m) a pivot of covering of selection；Utilize grey correlation
Degree analyzes the curvilinear trend relationship of each variable and oxygen content of smoke gas, determines the variable strong with oxygen content of smoke gas correlation；
(3) least square method supporting vector machine model module: use RBF gaussian kernel function as oxygen content of smoke gas soft-sensing model
Kernel function, and least square method supporting vector machine can overcome the correlation between variable；
(4) Chaos-Particle Swarm Optimization optimizing module: with the Chaos particle swarm optimization algorithm (CPSO) with Stochastic inertia weight to LSSVM just
Then change parameter and nuclear parameter optimizing.
2. a kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine according to claim 1,
It is characterized in that, in the step 4, steps are as follows for the parameter optimization of the CPSO algorithm with Stochastic inertia weight:
Step 1: initialization CPSO parameter: population scale d, maximum number of iterations Gmax, Studying factors c1,c2, nuclear parameter σ and canonical
The value range of parameter γ；
Step 2: d particle being randomly generated in determined range, each particle initial position is (σ0,γ0)；
Step 3: establishing LSSVM model using the position and training data of each particle, and using test data according to formulaSolve the fitness value x of each particleid, y in formulaiIt is true value,It is predicted value, then carries out
Sequence, sets p for each particle history adaptive optimal control angle valuebest, population is set by the smallest particle position of fitness value
History adaptive optimal control degree position gbest；
Step 4: the fitness value of solution and more each particle, if the current fitness value of each particle is less than its history adaptive optimal control
Angle value, i.e. xid< pbest, then current pbestIt is set as xidIf similarly current particle adaptive optimal control angle value is less than global optimum's fitness
Value, i.e. xid< gbest, then g is enabledbestEqual to xid；
Step 5: according to formulaGenerate Stochastic inertia weight, it is assumed that optimizing Spatial Dimension is D, and population number is
N, wherein i-th of particle is expressed as Xi=(xi1,xi2,L,xiD), the desired positions of particle experience are denoted as Pi=(pi1,pi2,L,
piD), it is also possible to pbestIt indicates, the desired positions of whole particle experience are denoted as P in populationg=(pg1,pg2,L,pgD), it is also possible to
gbestIt indicates, the speed V of particlei=(vi1,vi2,L,viD) indicate, for every generation particle, its d dimension (1≤d≤D) speed
Degree and position update according to the following formula:And xid(k+1)
=xid(k)+vid(k+1), so that the speed of more new particle and position, generate particle of new generation, wherein w is inertia weight,
Rand () is the random number between (0,1), c1It is local optimum specific gravity, c2It is global optimum's specific gravity；
Step 6: chaos processing being carried out to population, retains the smallest preceding N × P particle of fitness function value in population, wherein P
It indicates substitution probability, chaos masking search is carried out to this N × P particle, according to formulaBy chaos SYSTEM OF LINEAR VECTOR
Project to [0,1] solution space of Logistic system, xiFor optimal solution, xmax,xminFor optimal solution bound, further according to formula zn+1
=μ zn(1-zn), 0 < znThe grey iterative generation chaos sequence iteration initial value of < 1, μ=4 z0It cannot be 0, by the chaos sequence of generation
Pass through formula xi=xmin+(xmax-xmin)*ziInverse mapping replaces original N × P to obtain real optimal solution to former solution space
A population particle；
Step 7: judging whether that satisfaction reaches maximum number of iterations or precision of prediction, if satisfied, then LSSVM parameter optimization suspends；
Conversely, return step 4 continues searching LSSVM parameter；
Step 8: the optimal hyper parameter found being input in LSSVM algorithm, oxygen content of smoke gas training sample is learnt, is built
Vertical soft-sensing model carries out oxygen content of smoke gas hard measurement to test data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811168969.4A CN109765331A (en) | 2018-10-08 | 2018-10-08 | A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811168969.4A CN109765331A (en) | 2018-10-08 | 2018-10-08 | A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109765331A true CN109765331A (en) | 2019-05-17 |
Family
ID=66449864
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811168969.4A Pending CN109765331A (en) | 2018-10-08 | 2018-10-08 | A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109765331A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109506248A (en) * | 2018-12-04 | 2019-03-22 | 华北电力大学 | It is a kind of based on can online optimizing reasoning by cases query formulation Boiler combustion optimization |
| CN113450880A (en) * | 2021-08-31 | 2021-09-28 | 大唐环境产业集团股份有限公司 | Desulfurization system inlet SO2Intelligent concentration prediction method |
| CN113724796A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Chemical device tail gas oxygen content calculation method and device, storage medium and processor |
| CN113724795A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Method and device for calculating concentration of raffinate hydrogen peroxide, storage medium and processor |
| CN114637212A (en) * | 2022-05-17 | 2022-06-17 | 浙江浩普智能科技有限公司 | Oxygen content prediction control method and system |
| CN116679008A (en) * | 2023-06-09 | 2023-09-01 | 中国城市建设研究院有限公司 | Big data-based waste incineration flue gas monitoring method, device and storage medium |
Citations (3)
| 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 |
| CN103902813A (en) * | 2014-03-06 | 2014-07-02 | 东南大学 | Steam-driven draught fan full working condition online monitoring model modeling method based on CPSO-LSSVM |
| CN106127330A (en) * | 2016-06-18 | 2016-11-16 | 上海大学 | Fluctuating wind speed Forecasting Methodology based on least square method supporting vector machine |
-
2018
- 2018-10-08 CN CN201811168969.4A patent/CN109765331A/en active Pending
Patent Citations (3)
| 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 |
| CN103902813A (en) * | 2014-03-06 | 2014-07-02 | 东南大学 | Steam-driven draught fan full working condition online monitoring model modeling method based on CPSO-LSSVM |
| CN106127330A (en) * | 2016-06-18 | 2016-11-16 | 上海大学 | Fluctuating wind speed Forecasting Methodology based on least square method supporting vector machine |
Non-Patent Citations (4)
| Title |
|---|
| 刘彩虹: "《制造业供应链质量风险管理研究》", 31 December 2016, 广西师范大学出版社 * |
| 刘晓春: "《雷达天线罩电性能设计技术》", 31 January 2017, 航空工业出版社 * |
| 李艳等: "火电厂锅炉烟气氧含量精确测量仿真研究", 《计算机仿真》 * |
| 蓝金辉等: "《冶金计量》", 30 November 2006, 中国计量出版社 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109506248A (en) * | 2018-12-04 | 2019-03-22 | 华北电力大学 | It is a kind of based on can online optimizing reasoning by cases query formulation Boiler combustion optimization |
| CN113724796A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Chemical device tail gas oxygen content calculation method and device, storage medium and processor |
| CN113724795A (en) * | 2020-05-25 | 2021-11-30 | 中国石油化工股份有限公司 | Method and device for calculating concentration of raffinate hydrogen peroxide, storage medium and processor |
| CN113450880A (en) * | 2021-08-31 | 2021-09-28 | 大唐环境产业集团股份有限公司 | Desulfurization system inlet SO2Intelligent concentration prediction method |
| CN114637212A (en) * | 2022-05-17 | 2022-06-17 | 浙江浩普智能科技有限公司 | Oxygen content prediction control method and system |
| CN116679008A (en) * | 2023-06-09 | 2023-09-01 | 中国城市建设研究院有限公司 | Big data-based waste incineration flue gas monitoring method, device and storage medium |
| CN116679008B (en) * | 2023-06-09 | 2024-03-19 | 中国城市建设研究院有限公司 | Big data-based waste incineration flue gas monitoring method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109765331A (en) | A kind of oxygen content of smoke gas hard measurement system based on least square method supporting vector machine | |
| JP4876057B2 (en) | Plant control device and thermal power plant control device | |
| US8135653B2 (en) | Power plant control device which uses a model, a learning signal, a correction signal, and a manipulation signal | |
| CN107016455B (en) | Prediction system and method for oxygen content of flue gas at hearth outlet of circulating fluidized bed household garbage incineration boiler | |
| CN105864797B (en) | Real-time prediction system and method for boiler entering heat value of circulating fluidized bed household garbage incineration boiler | |
| CN106224939B (en) | Circulating fluid bed domestic garbage burning boiler bed temperature Forecasting Methodology and system | |
| EP1921280B1 (en) | Systems and methods for multi-level optimizing control systems for boilers | |
| CN109960873B (en) | Soft measurement method for dioxin emission concentration in urban solid waste incineration process | |
| CN103759277B (en) | Coal-fired power station boiler intelligent ash blowing closed loop control method, device and system | |
| Si et al. | Optimization of coal-fired boiler SCRs based on modified support vector machine models and genetic algorithms | |
| CN1312296C (en) | Optimizing control method for hybrid type blast furnace hot blast stove | |
| CN109508818B (en) | Online NOx prediction method based on LSSVM | |
| TWI705316B (en) | Boiler operation support device, boiler operation support method, and boiler learning model creation method | |
| CN105907906B (en) | Pebble stove burns stove process model building and energy consumption optimization method and system | |
| CN102460018B (en) | Especially the control method of fossil fuel boiler furnace combustion process and combustion system | |
| CN106594794B (en) | A kind of hybrid intelligent boiler efficiency burning optimization model update method | |
| CN109992844A (en) | A kind of boiler flyash carbon content prediction technique based on ADQPSO-SVR model | |
| CN110207094A (en) | IQGA-SVR boiler heating surface fouling characteristics discrimination method based on principal component analysis | |
| CN107024861B (en) | Online modeling method of converter dry dedusting system | |
| CN114492727A (en) | Intelligent prediction and control method for inlet smoke concentration of dust removal system | |
| CN113074459A (en) | Boiler control optimization method and system | |
| CN109140474A (en) | A kind of coal-fired power station boiler manually intelligent sootblowing | |
| JP2000222005A (en) | Controller for process | |
| JP2009222332A (en) | Control device for plant equipped with boiler and control method for plant equipped with boiler | |
| Wang et al. | A two-tier approach to the data-driven modeling on thermal efficiency of a BFG/coal co-firing boiler |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190517 |
|
| RJ01 | Rejection of invention patent application after publication |