CN104597208B - Multi-channel cyclic sampling gas analysis method - Google Patents
Multi-channel cyclic sampling gas analysis method Download PDFInfo
- Publication number
- CN104597208B CN104597208B CN201510027439.8A CN201510027439A CN104597208B CN 104597208 B CN104597208 B CN 104597208B CN 201510027439 A CN201510027439 A CN 201510027439A CN 104597208 B CN104597208 B CN 104597208B
- Authority
- CN
- China
- Prior art keywords
- sample gas
- gas passage
- data
- switching time
- passage
- 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
Links
Landscapes
- Sampling And Sample Adjustment (AREA)
Abstract
The invention provides a multi-channel cyclic sampling gas analysis method. In a self-learning stage, when analysis meter equipment is switched into a sample gas channel B from a sample gas channel A, the switching time is started to be timed; the analysis meter equipment is used for sampling input data and output data of the sample gas channel B; the theoretical output value is calculated according to the input data of the sampled sample gas channel B; the error value of the theoretical output value and the sampled output data is acquired through comparison of the theoretical output value and the sampled output data; when the calculated average error value is less than the preset threshold value, timing is stopped, the timing result is defined as the switching time tA-B; and in a subsequent practical working stage, when the sample gas channel A is switched into the sample gas channel B each time, after the switching time tA-B transition is finished, practical data can be taken on. The multi-channel cyclic sampling gas analysis method can provide the sample gas analysis accuracy of all channels.
Description
Technical field
The invention belongs to monitoring technology on-line field, more particularly, to a kind of multichannel circulating sampling analysis method for gases.
Background technology
In industrial furnace running, for ensureing body of heater safety and meeting technique needs, often for example micro- to furnace atmosphere
The gas contents such as amount oxygen, hydrogen carry out on-line analyses.For reducing investment outlay, maintain easily, often cover analyser equipment pair using single
Multiple sample gas passages are circulated sampling analyses.
But the circulating sampling process just because of analysis system, analyser equipment at a time can only be sampled one and be led to
Road, will result in analyser in this sampling instant and can not know the sample gas information in other passages not being sampled.
It is additionally, since sample air-flow to require time for through analysis meter, and analysis meter there is also response time, so in passage
Cyclic switching during will necessarily produce the change of data and stable process, and analyser data is serious in this course
Distortion is it is necessary to abandon so that the sample gas analytical data that during this, analyser provides provides mistake enlightenment.Above-mentioned two reasons make
Multichannel circulating sampling gas analysis system provide data message imperfect it is difficult to meet produce and safeguard needs.
Content of the invention
The purpose of the embodiment of the present invention is to provide a kind of multichannel circulating sampling analysis method for gases, to solve existing skill
Art is so will necessarily produce the change of data and stable process during the cyclic switching of passage, and divides in this course
Analyzer data serious distortion is it is necessary to abandon so that the sample gas analytical data that during this, analyser provides provides mistake enlightenment
Problem.
The embodiment of the present invention is achieved in that a kind of multichannel circulating sampling analysis method for gases, by single set analysis
Instrument equipment is circulated sampling analyses to multiple sample gas passages, is switched to sample gas in described analyser equipment by sample gas passage A and leads to
During road B, need through a switching time tA-B, accurate sampled data could be exported, methods described includes：
In the self study stage, when described analyser equipment is switched to sample gas passage B by sample gas passage A, start switching time
Timing；Described analyser equipment is sampled to the input data of sample gas passage B and output data；The sample being obtained according to sampling
The input data of gas passage B, calculates theoretical output valve；The output number being obtained by relatively more described theory output valve and sampling
According to obtaining both error amounts；When the average error value calculating is less than predetermined threshold value, stop timing, and timing result is fixed
Justice is switching time tA-B；Follow-up in the real work stage, when carrying out sample gas passage A every time and being switched to sample gas passage B operation,
Complete described switching time tA-BTransition after, just assume real data.
On the other hand, the embodiment of the present invention additionally provides a kind of multichannel circulating sampling analysis method for gases, including：
According to the input variable quantity of multichannel Automatic Cycle sampling analysis system judge using least square support to
Amount machine model prediction；Analysis system carries out least square method supporting vector machine forecast model training respectively to each passage, and to model
Carry out error evaluation；Analysis system carries out self study training to the time response in passage handoff procedure, finds out cutting of each passage
Change response time；After training terminates, system executes circulating sampling data handling procedure, becomes using to the study of switching time characteristic
Fruit is managed to the measured data flow direction in passage handoff procedure.
A kind of beneficial effect of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention includes：Improve
The accuracy of each passage sample gas analytical data, promotes the popularization and application in the industrial production of circulating sampling analysis system.
Brief description
For the technical scheme being illustrated more clearly that in the embodiment of the present invention, below will be to embodiment or description of the prior art
In required use accompanying drawing be briefly described it should be apparent that, drawings in the following description be only the present invention some
Embodiment, for those of ordinary skill in the art, on the premise of not paying creative work, can also be attached according to these
Figure obtains other accompanying drawings.
Fig. 1 is a kind of configuration diagram of multichannel circulating sampling gas analysis system provided in an embodiment of the present invention；
Fig. 2 is a kind of schematic flow sheet of multichannel circulating sampling gas analysis system provided in an embodiment of the present invention；
Fig. 3 is the schematic flow sheet that a kind of passage provided in an embodiment of the present invention switches training method；
Fig. 4 is a kind of schematic flow sheet of Channel Prediction model training method provided in an embodiment of the present invention；
Fig. 5 is a kind of schematic flow sheet of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention；
Fig. 6 is a kind of schematic flow sheet of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention；
Fig. 7 is a kind of schematic flow sheet of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention；
Fig. 8 is a kind of schematic flow sheet of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, below in conjunction with drawings and Examples, right
The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only in order to explain the present invention, and
It is not used in the restriction present invention.
In order to technical solutions according to the invention are described, to illustrate below by specific embodiment.
Embodiment one
As shown in figure 1, being that the embodiment of the present invention provides a kind of multichannel circulating sampling gas analysis system configuration diagram,
Involved list set analyser equipment in the present invention is included (because analyser equipment is related to multiple dresses in shown analysis system
Put, including：Display device, sensing device, controller, due to being distributed in whole system different parts, therefore not in figure one
One mark is out).The system architecture that the present embodiment provides is it is adaptable to each method embodiment proposed by the present invention, but does not limit
The scope that each method embodiment proposed by the present invention is suitable for.
In Fig. 1, SV-0A is the three-way magnetic valve of sample gas passage A, and SV-0B is the three-way magnetic valve of sample gas passage B, SV-00C
For the purge solenoid valve of sample gas channel C (nitrogen), SV-00D is analysis channel three-way magnetic valve, and M1 is sampling pump, and M2 is discharge
Pump.
Wherein, SV-0A, SV-0B and SV-00D are two-bit triplet electromagnetic valve.From the figure, it can be seen that electromagnetic valve is divided into 3
Direction, respectively with 1,2,3 expressions.When electromagnetic valve obtains electric, 1-2 directional combination communicates, and other directional combinations are obstructed；Work as electromagnetic valve
During dead electricity, 1-3 directional combination communicates, and other directional combinations are obstructed.
In the present embodiment, related hardware system construction main thoroughfare has：Sample gas passage A (equipment before SV-0A and pipe
Road), sample gas passage B (equipment before SV-0B and pipeline), sampling channel is (between SV-0A, SV-0B, SV-00C and SV-00D
Equipment between four switches and pipeline), analysis channel (equipment between SV-00D and M1 and pipeline), discharge-channel
(between the equipment between SV-0A, SV-0B and M2 and pipeline), purge passage (is used for being passed through high pressure before SV-00D and SV-00C
The equipment of nitrogen and pipeline).
Embodiment two
The flow chart being illustrated in figure 2 a kind of multichannel circulating sampling analysis method for gases provided in an embodiment of the present invention,
As shown in Figure 2, a kind of embodiment of multichannel circulating sampling analysis method for gases that the present invention provides includes：
The software system that the data processing method of this patent is made up of with industrial PLC control system data processing software is real
Existing.The function of data processing software is data base administration, forecast model foundation and computing；The function of industrial PLC control system is
Man-machine interaction (the Human Machine of the hardware exchange, analytical data and system operatio of execution analysis system passage
Interface, writes a Chinese character in simplified form：HMI).
Because the data processing ultimate principle of multichannel Automatic Cycle sampling analysis system is unrelated with port number, for narration side
Just, the present embodiment to be carried out in detail taking the dual pathways Automatic Cycle sampling analysis system of hydrogen content analysis in continuous annealing furnace stove as a example
Thin introduction.
Data handling system and multichannel Automatic Cycle sampling analysis system start simultaneously at work.As shown in Fig. 2 leading to more
When road Automatic Cycle sampling analysis system work starts, data processing software also begins to be initialized.
In initialization procedure, data base is set up by data processing software；Obtain from the man machine interface of industrial PLC control system
Know input variable quantity, circulating sampling cycle T s, passage weight COEFFICIENT K q and the channel priorities of analysis system, and using industry
PLC control system is set up with the input variable of analysis system and is connected.
The circulating sampling cycle is each channel sample sum analysis time, that is, circulating sampling cycle T s=sample gas passage A point
Ts2 analysis time of analysis time ts1+ sample gas passage B.
Passage weight coefficient is the proportion that this channel sample accounts for the whole circulating sampling cycle analysis time.The dual pathways is adopted
Sample system, two passage weight coefficients meet following relation：
The weight coefficient Kq2=1 of the weight coefficient Kq1+ sample gas passage B of sample gas passage A
And, the pass, between channel sample analysis time, circulating sampling cycle and weight coefficient taking sample gas passage A as a example
System is as follows：
The sampling analyses time ts1=system of sample gas passage A executes the weight coefficient Kq1 of cycle T s × sample gas passage A
Channel priorities represent channel sample order, if taking triple channel circulating sampling system as a example it is assumed that passage is preferential
Level is " 132 ", then system is sampled according to the order of No. 1 passage, No. 3 passages, No. 2 passages.
The input variable quantity of analysis system is used for judging whether to adopt model prediction.
Because the data of analysis system is usually relevant with some variables, such as in the example of continuous annealing furnace hydrogen content analysis
In, the hydrogen content of the annealing furnace just stream with the hydrogen being passed through in stove and all gas the summation summation of nitrogen and hydrogen (mainly)
Amount compares Kf, in-furnace temperature TfWith furnace pressure PfDeng relating to parameters, these variables are exactly the input variable of analysis system.And analyze
The basis of system model predictions is exactly input variable and the sample of analyser institute gathered data.So whether having clear and definite input
Variable is the foundation whether analysis system adopts model prediction, and that is, in step 201, system according to input variable quantity will be
No is not 0 judging whether to adopt model prediction.In the example of annealing furnace hydrogen content analysis, analysis system is by Kf、TfAnd PfMake
To carry out model prediction for input variable.
Because in example, the input variable quantity in step 201 is more than 0, so system enters step 202, forecast model is instructed
Practice process.This patent adopts least square support vector machines, and (Least Squares Support Vector Machine, writes a Chinese character in simplified form：LS-
SVM) method carries out model prediction.Fig. 5 illustrates Channel Prediction model training flow process.Described forecast model training, mainly protects
Demonstrate,prove the software system of this industrial PLC control system composition monitor and export sample gas passage A data when, in addition not monitored
Sample gas passage B and C prediction data can be calculated in forecast model mode, and be presented on display device, for operator
Member observes.After having trained forecast model, the just passage switching training of progressive step 203.Described passage switching training is implemented
Method provided in example two, will not be described here.Refer to one of embodiment two feasible program in the present embodiment, increase
The setting of sample gas interchannel priority, therefore, after step 203, enters step 204, according to priority be set for follow
Ring sampling and analysis.
Because the input variable quantity in step 201 is equal to 0 in example, that is, operator are not had not have input variable, this
When will skip forecast model training process, but directly carry out the passage switching training of step 205.Described passage switches training
Method provided in embodiment two, will not be described here.One of embodiment two feasible program is refer in the present embodiment,
Increased the setting of sample gas interchannel priority, therefore, after step 205, enter step 206, the setting according to priority is entered
Row circulating sampling and analysis.
The present embodiment combines concrete implementation environment, by complete system environmentss, describes the present invention and implements
The scene that the method that example proposes is suitable for, and specifically that link in execution flow process to be realized.Real with the present invention
After applying the method for example proposition, can carry for the dual pathways Automatic Cycle sampling analysis system of hydrogen content analysis in continuous annealing furnace stove
For passage switching time, it is to avoid the data distortion that brings during passage switching in existing system, data loss problem, and based on pre-
Survey the display clear area that former passage after switching channel filled up by model, improve stability and the accuracy of software data.
Embodiment three
In the step 202 of embodiment two, after the forecast model training of each passage terminates, system enters passage switching instruction
Practice process.The step 203 that the present embodiment is directed in embodiment two or 205 offer passages switch implementing of training method.
As shown in figure 3, specifically including：
In the case of keeping annealing furnace hydrogen content basicly stable, system switches training by passage as shown in Figure 3 and obtains
Take each passage switching time.Taking dual pathways switching training as a example：
In step 301, timer T1 timing is just being started to sample gas passage B switching instant from sample gas passage A, supervising simultaneously
Survey sampled data rate of change.
In step 302, when data variation rate<5%, that is, when thinking that sampled data tends towards stability, enter step 303；No
Then return to step 301 and continue monitoring sampled data rate of change.
In step 303, timer T1 stops, and switches to sample gas passage A by sample gas passage B sampling simultaneously, starts timer
T2, monitors sampled data rate of change.
In step 304, when data variation rate<When 5%, that is, when thinking that sampled data tends towards stability, enter step 305；
Otherwise return to step 303 and continue monitoring sampled data rate of change.
In step 305, stop timer T2, cycle index counter, from Jia 1, records T1, T2 time, once switching instruction
White silk terminates.
Within step 306, judge to switch whether frequency of training reaches 5 times, if not, return to step 301 and proceed to cut
Change training, otherwise execution step 309.
In a step 309, after by the 5 bout switchings of this flow performing, system chooses the maximum of recorded T1, T2 time
Value tA-B、tB-AAs training result, whether true enter discriminatory analysis data after row of channels switching in normal work as system
Foundation that is real and providing system to show.
This gives a kind of feasible switching channel training method, wherein, the switching frequency of training of 5 times and 5%
Data variation rate be all that it does not limit the protection domain of the present embodiment, on this basis through putting into practice the preferred value that draws
The numerical value rationally deducing falls within the scope of the present invention.
Because analysis meter requires time for the analysis process of sample gas in passage handoff procedure, that is, there is passage switching
After need through after a while, analysis meter just can analyze accurate atmosphere data, and the switching of therefore analysis system there is
This feature of response time.And passage switching training seeks to know response time by training, and with setting sample later
Gas passage B switches to the response time t of sample gas passage AB-A, and, then should meet adopting of sample gas passage A taking this handoff procedure as a example
Sample ts1 analysis time is much larger than the response time t switching to sample gas passage AB-A, in system, it is defaulted as ts1 >=2 × tB-ARelation.
So after passage switching training process, system can execute cycle in system initialization process to system according to above-mentioned relation
TsWith each passage weight COEFFICIENT KqVerified and adjusted.
Example IV
As shown in figure 4, the present embodiment is in the way of deducing with reference to formula, to illustrate the prediction mould that the embodiment of the present invention proposes
The exploitativeness realized principle and method, further support forecast model in embodiment two and subsequent embodiment of type.
The present embodiment taking the analysis of continuous annealing furnace hydrogen content as a example it is contemplated that the acquisition condition of production process data, is chosen
It is passed through flow-rate ratio K of hydrogen in stove and all gas the summation summation of nitrogen and hydrogen (mainly)f, in-furnace temperature TfAnd stove
Interior pressure PfComposition input vector, hydrogen content Q in stoveHFor output vector.
In step 401, artificial setting penalty coefficient c and core width cs.
Set (Kf,Tf,Pf,QH)=(x1,x2,x3,y).
In step 402, pretreatment is normalized to data.
For collection initial data,WithIt is the maximum and minimum value in gathered data respectively.Return
After one changes,And the input vector as LS-SVM model.Output vector y is also carried out being similar to
Normalization, obtain yiOutput vector as LS-SVM model.
In step 403, training set and test set are manually selected.
Sample set is set up in data base by data processing software
And input and output vector exists in time in sample set
Continuous great amount of samples is chosen as training set, a small amount of sample is as test set during certain change.
In step 404, using least square method supporting vector machine, forecast model is trained.
Basic process is：Set up nonlinear solshingWherein ω is weight vector, b
For departure,It is the nuclear space mapping function that luv space is mapped to a high-dimensional feature space.
According to empirical risk minimization, set up forecast model
Wherein, εiFor error variance；C >=0 is penalty coefficient, for taking one between training error and model complexity
Compromise.
Construction Lagrange function is solved：
Wherein, ai(i=1,2 ..., N) is Lagrange multiplier, optimizes Lagrange function and obtains：
Variable ω and ε in subtractive (2)i, and according to Mercer condition, set：
And adopt RBFAs prediction machine kernel function, wherein σ is core width.Can get following system of linear equations：
Artificial setting penalty coefficient c and core width cs, and use solving equations aiAnd b.Finally obtain nonlinear solshing
As output vector yiPredictor formula：
In step 405, after forecast model is set up, test set is updated in model and carries out performance evaluation.
Model performance evaluation index adopts mean error computing formulaIn formula, Qi
For actual measured value, fiFor predictive value, m is checking number of times.
In a step 406, judge that channel pattern evaluates whether qualified, if qualified, terminate channel pattern training；If
Unqualified, return to the training that step 404 proceeds forecast model.
In the present invention, mean error ε is mainly used in weighing the penalty coefficient c of artificial setting and the accuracy of core width cs.
If mean error ε can not reach requirement, by manually adjusting penalty coefficient c and core width cs, software can be according to training set and survey
Examination collection recalculates, until mean error reaches requirement.
The present embodiment combines specific model and algorithmic formula, gives in detail for the forecast model how realizing passage
Thin elaboration, supports the realization about channel pattern training step in each embodiment further.
Embodiment five
The present embodiment is the specific implementation based on the step 204 in embodiment two, and has passed through embodiment and sentencing
Disconnected have input variable after execution flow process implement step.When in embodiment two step 203 passage switching training according to
After the completion of embodiment three specific implementation, system can have input by presetting system execution cycle, weight and priority execution
The multichannel circulating sampling analysis of variable.As shown in figure 5, specifically including step：
In step 501, judge whether to continue executing with circulating sampling.If it is, continuing executing with step 502；If not,
Then sampling analyses terminate.
Cycle and weight calculation two channel sample ts1 and ts2 analysis time are executed according to system, is assigned to timer respectively
T1, T2, as the judgement end time of two timers；Channel sample order is determined according to priority, assumed priority row in Fig. 6
Sequence is sample gas passage A, sample gas passage B.
In step 502, run each channel data forecast model.
Described each channel data forecast model, is by step 202, has being obtained by example IV computing of choosing.
In step 503, when switching to sample gas passage A sampling, timer T1 starts.When timer reaches tB-AMoment
Afterwards, that is, when thinking that sample gas passage A analytical data is true and reliable, sample gas passage A analytical data is delivered to HMI and shown in real time always
Out, and before sample gas passage A true samples data undelivered, the HMI video data of sample gas passage A is all by sample gas passage
The forecast model of A is calculated.
In step 504, while sample gas passage A true samples data delivers to HMI display, these data are also fed to count
According to storehouse, and for calculating the real-time mean error of forecast model.When error is more than 5%, execution step 505；Otherwise, execute step
Rapid 506.
In step 505, this channel pattern forecast error is pointed out to report to the police by data processing software, by artificial according to rear record
The data entering forecast model is modified qualified after, then forecast model is replaced.
In step 506, judge whether timer T1 reaches ts1, if do not reached, execution step 507；Otherwise execute step
Rapid 509.
In step 507, the forecast model data display of sampling channel B and the system of sampling channel A is kept to adopt in real time
Sample analytical data shows.
In step 508, after timer T1 reaches ts1 moment, timer T1 stops timing, and sample gas passage A switches to
Sample gas passage B, timer T2 start.When timer reaches tA-BAfter moment, sample gas passage B analytical data is delivered to HMI always
Show in real time.Equally, before sample gas passage B true samples data undelivered, the HMI video data of sample gas passage B all by
The forecast model of sample gas passage B is calculated.
In step 509, while sample gas passage B true samples data delivers to HMI display, these data are also fed to count
According to storehouse, and for calculating the real-time mean error of forecast model.When error is more than 5%, execution step 510；Otherwise, execute step
Rapid 511.
In step 510, this channel pattern forecast error is pointed out to report to the police by data processing software, by artificial according to rear record
The data entering forecast model is modified qualified after, then forecast model is replaced.
In step 511, after timer T2 reaches ts2 moment, timer T2 stops timing, return to step 501, judges
Whether execute the circulating sampling process in a cycle.
And while each passage true samples data delivers to HMI display, these data are also fed to data base, and are used for
Calculate the real-time mean error of forecast model.When error is more than 5%, data processing software points out this channel pattern forecast error
Report to the police, by artificial according to the data of rear typing forecast model is modified qualified after, then forecast model is replaced.
Analysis system can continue reciprocally to execute above-mentioned circulating sampling process in the case of no manual intervention, and by two passages
The measured data of gas analyses and prediction data are supplied to HMI and show.
Embodiment six
If analysis system run into no the situation of input variable when, as shown in Fig. 2 system then only execution passage switching instruction
Practice, then execute the multichannel circulating sampling analysis of no input variable again.Channel sample order is determined according to priority, in Fig. 6
Assumed priority is ordered as sample gas passage A, sample gas passage B.The multichannel circulating sampling of no input variable analyzes process such as Fig. 6 institute
Show, cycle and weight calculation two channel sample ts1 and ts2 analysis time executed according to system, be assigned to respectively timer T1,
T2；When switching to sample gas passage A sampling, timer T1 starts.When timer reaches tB-AAfter moment, that is, think sample gas passage A
When analytical data is true and reliable, sample gas passage A analytical data is delivered to HMI and shown in real time always, and in timer T1
Reach tB-ABefore moment, HMI does not show analysis numerical value.After timer T1 reaches ts1 moment, timer T1 stops timing, sample gas
Passage A switches to sample gas passage B, and timer T2 starts.When timer reaches tA-BAfter moment, sample gas passage B analytical data is sent
Show in real time to HMI and always, reach t in timer T2A-BBefore moment, HMI does not show.When timer T2 reaches the ts2 moment
Afterwards, timer T2 stops timing, and sample gas passage B switches to sample gas passage A, starts the circulating sampling process in next cycle.Analysis
System can continue reciprocally to execute above-mentioned circulating sampling process in the case of no manual intervention, and the gas analyses number by two passages
It is believed that breath is supplied to HMI at times and shows.Concrete execution step reference implementation example six, will not be described here.
Embodiment seven
The flow chart being illustrated in figure 7 a kind of multichannel circulating sampling analysis method for gases of present invention offer, is to this
It is related to, in invention embodiment, the refinement that passage switches training method, wherein, by single set analyser equipment to multiple sample gas
Passage is circulated sampling analyses, when described analyser equipment is switched to sample gas passage B by sample gas passage A, needs through one
Individual switching time tA-B, accurate sampled data could be exported.This gives specifically how to draw described switching time
tA-BMethod, in the present embodiment, according to whether calculating described switching time tA-B, and by real work process qualitatively
Self study stage and real work stage have been divided into it, so as to being understood more readily from this programme, methods described includes：
In the self study stage：
In a step 101, when analyser equipment is switched to sample gas passage B by sample gas passage A, start the meter of switching time
When.
In a step 102, analyser equipment is sampled to the input data of sample gas passage B and output data.
In step 103, the input data of the sample gas passage B being obtained according to sampling, calculates theoretical output valve；By than
More described theory output valve and the output data obtaining of sampling, obtain both error amounts.
At step 104, when the average error value calculating is less than predetermined threshold value, stop timing, and timing result is fixed
Justice is switching time tA-B.
Subsequently in the real work stage：
In step 105, carry out every time sample gas passage A be switched to sample gas passage B operation when, complete described switching time
tA-BTransition after, just assume real data.
The present embodiment, by improve the accurate of each passage sample gas analytical data using average error value during self study
Property, promote the popularization and application in the industrial production of circulating sampling analysis system.
There is a kind of feasible program in conjunction with the present embodiment, wherein, the described self study stage also includes：Described single set analyser
Equipment, according to the data collecting, is predicted the training of model using least squares support vector machine.
There is a kind of feasible program in conjunction with the present embodiment, wherein, the described sample gas passage A that carries out every time is switched to sample gas passage
During B operation, complete described switching time tA-BTransition after, just assume real data, also include：When carrying out described switching
Between tA-BTransition when, keep sample gas passage B forecast model predict the outcome present；Complete described switching time tA-B's
After transition, just by described present switch to real data.
There is a kind of feasible program in conjunction with the present embodiment, wherein, between the plurality of sample gas passage, there is priority relationship,
In self study stage, the calculating of described switching time, carry out according to the priority relationship between the plurality of sample gas passage.
There is a kind of feasible program in conjunction with the present embodiment, wherein, when described analyser equipment starts, methods described is also wrapped
Include：Described analyser equipment determines whether various kinds gas channel parameters；When judged result is sometimes, completes least square first and prop up
The training that support vector machine is predicted model obtains forecast model；Secondly, switch over passage training, obtain switching time tA-B；
When judged result is not have, directly switches over passage training, obtain switching time tA-B.
There is a kind of feasible program in conjunction with the present embodiment, wherein, wrap between described sample gas passage A and described sample gas passage B
Contain：It is switched to the switching time t of described sample gas passage B from described sample gas passage AA-BIt is switched to institute with from described sample gas passage B
State the switching time t of sample gas passage AB-A.
There is a kind of feasible program in conjunction with the present embodiment, wherein, the input data to sample gas passage B for the described analyser equipment
Sampled with output data, specifically included：Analyser equipment passes through the sensor of input port and output in sample gas passage B
The sensor of port, carries out the sampling of described input data and output data.
There is a kind of feasible program in conjunction with the present embodiment, wherein, described real data includes：The sampling of analyser equipment obtains
Input data and output data.
Embodiment eight
Embodiments provide a kind of multichannel circulating sampling analysis method for gases, as shown in figure 8, including：
In step 1001, judged using according to the input variable quantity of multichannel Automatic Cycle sampling analysis system
A young waiter in a wineshop or an inn takes advantage of supporting vector machine model to predict.
In step 1002, analysis system carries out least square method supporting vector machine forecast model training respectively to each passage,
And error evaluation is carried out to model.
In step 1003, analysis system carries out self study training to the time response in passage handoff procedure, finds out each
The handoff response time of passage.
In step 1004, after training terminates, system executes circulating sampling data handling procedure, using special to switching time
Property learning outcome in passage handoff procedure measured data flow direction be managed.
The present embodiment is predicted to the channel data not being sampled using least square method supporting vector machine model, by right
The self study of passage switching time characteristic solves the system response impact to data for the time lag.In the present embodiment, the process of self study can
With the method described in reference implementation example seven, the present embodiment is to set about for embodiment seven learning model from systems soft ware side
Concrete application.
Preferably, when input variable quantity is more than 0, judge to adopt least square method supporting vector machine model prediction.
There is a kind of feasible program in conjunction with the present embodiment, wherein, also include：According to each passage switching time characteristic
Practise achievement, the actual measurement number during the actual circulating sampling passage switching of execution, to the passage being just switched in sampling
Give certain time delay according to meeting and just give human-machine interaction picture and show and use.
Embodiment nine
The present embodiment is an example about the training of continuous annealing furnace hydrogen content analytical data forecast model, data processing
Software mainly completes under the MATLAB platform of The MathWorks company, and data base then sets up and puts down in the ACCESS of Microsoft
Under platform.Industrial PLC control system is Siemens's PCS7 system.
Before data processing, software prompt manually sets penalty coefficient c and core width cs, and system provides default value respectively
It is 5000 and 0.5.Data acquisition and normalized are carried out simultaneously, and sample data is given expression to curve mode on picture
Come.Choose continuous 300 groups of samples by artificial according to data and curves as training set, 60 groups of subsequent samples are as test set.
The part sample of test set and result such as table 1 below.
The part sample of table 1 test set and result
In this example, mean error requires to be less than 5%.As shown in table 1, mean error meets and requires it is believed that channel pattern
It is qualified to evaluate, and this Channel Prediction model training terminates.
The present invention provide for circulating sampling analysis system data processing method, using least square method supporting vector machine
Model prediction and the self study to passage switching time, substantially increase the integrity and accurately of each passage sample gas analytical data
Property, promote the popularization and application in the industrial production of circulating sampling analysis system.
Example discussed above is only that the basic embodiment to the present invention is described, and not the scope of the present invention is entered
Row limits, and on the premise of without departing from design spirit of the present invention, those skilled in the art make to technical scheme
Various conversion and improvement, all should fall in the protection domain of claims of the present invention determination.
Those of ordinary skill in the art are further appreciated that all or part of step realizing in above-described embodiment method is can
Completed with the hardware instructing correlation by program, described program can be stored in a computer read/write memory medium
In, described storage medium, including ROM/RAM, disk, CD etc..
The foregoing is only presently preferred embodiments of the present invention, not in order to limit the present invention, all essences in the present invention
Any modification, equivalent and improvement made within god and principle etc., should be included within the scope of the present invention.
Claims (7)
1. a kind of multichannel circulating sampling analysis method for gases based on self study, by single set analyser equipment to multiple sample gas
Passage is circulated sampling analyses it is characterised in that when described analyser equipment is switched to sample gas passage B by sample gas passage A,
Need through a switching time tA-B, accurate sampled data could be exported, methods described includes：
In the self study stage, when described analyser equipment is switched to sample gas passage B by sample gas passage A, start the meter of switching time
When；
Described analyser equipment is sampled to the input data of sample gas passage B and output data；
The input data of the sample gas passage B being obtained according to sampling, calculates theoretical output valve；By relatively more described theory output valve
The output data obtaining with sampling, obtains both error amounts；
When the average error value calculating is less than predetermined threshold value, stops timing, and timing result is defined as switching time tA-B；
Follow-up in the real work stage, when carrying out sample gas passage A every time and being switched to sample gas passage B operation, complete described switching
Time tA-BTransition after, just assume real data.
2. the method for claim 1 it is characterised in that the described self study stage also include：
Described single set analyser equipment according to the data collecting, using least squares support vector machine carry out sample gas passage A with
The training of the forecast model of sample gas passage B.
3. method as claimed in claim 2 is it is characterised in that the described sample gas passage A that carries out every time is switched to sample gas passage B behaviour
When making, complete described switching time tA-BTransition after, just assume real data, also include：
Carrying out described switching time tA-BTransition when, keep sample gas passage B forecast model predict the outcome present；
Complete described switching time tA-BTransition after, just by described present switch to real data.
4. described method as arbitrary in claim 1-3 is closed it is characterised in that there is priority between the plurality of sample gas passage
System, in self study stage, the calculating of described switching time, is carried out according to the priority relationship between the plurality of sample gas passage.
5. the method for claim 1 is it is characterised in that when described analyser equipment starts, methods described also includes：
Described analyser equipment determines whether various kinds gas channel parameters；
When judged result is sometimes, complete least squares support vector machine first and be predicted the training of model to obtain sample gas passage
A and the forecast model of sample gas passage B；Secondly, switch over passage training, obtain switching time tA-B；
When judged result is not have, directly switches over passage training, obtain switching time tA-B.
6. as the arbitrary described method of claim 1,2,3 or 5 it is characterised in that described sample gas passage A and described sample gas passage
Include between B：It is switched to the switching time t of described sample gas passage B from described sample gas passage AA-BWith from described sample gas passage B
It is switched to the switching time t of described sample gas passage AB-A.
7. if the arbitrary described method of claim 1,2,3 or 5 is it is characterised in that described analyser equipment is to sample gas passage B's
Input data and output data are sampled, and specifically include：
Analyser equipment, by the sensor of the sensor of input port in sample gas passage B and output port, carries out described defeated
Enter the sampling of data and output data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510027439.8A CN104597208B (en) | 2015-01-20 | 2015-01-20 | Multi-channel cyclic sampling gas analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510027439.8A CN104597208B (en) | 2015-01-20 | 2015-01-20 | Multi-channel cyclic sampling gas analysis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104597208A CN104597208A (en) | 2015-05-06 |
| CN104597208B true CN104597208B (en) | 2017-02-22 |
Family
ID=53123127
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510027439.8A Active CN104597208B (en) | 2015-01-20 | 2015-01-20 | Multi-channel cyclic sampling gas analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104597208B (en) |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4119003B2 (en) * | 1998-04-09 | 2008-07-16 | 大陽日酸株式会社 | Gas analyzer and method |
| CN102288737B (en) * | 2011-05-19 | 2012-11-21 | 浙江大学 | Circulating pipe network water quality integrative simulation test system with multichannel flow switching and calibrating devices |
| CN102662034B (en) * | 2012-05-28 | 2014-09-17 | 中国科学院地理科学与资源研究所 | Multichannel double-circulation soil carbon discharge delta13C observing system |
| CN104090077B (en) * | 2014-07-16 | 2015-09-02 | 中冶南方工程技术有限公司 | The hyperchannel circulating sampling gas analysis system of a kind of tape channel self-checking function and control method |
-
2015
- 2015-01-20 CN CN201510027439.8A patent/CN104597208B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN104597208A (en) | 2015-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106909705B (en) | Blast furnace molten iron quality forecasting method and system | |
| CN106352542B (en) | A kind of prediction technique of storage-type electric water heater residue bathing time | |
| US9646111B2 (en) | Process monitoring system and method | |
| CN108311544B (en) | A kind of roll-force Parameter Self-learning method and device | |
| CN103336433B (en) | Mixed self-adapting Predictive Control System based on Backstepping and forecast Control Algorithm thereof | |
| CN111307123A (en) | Real-time abnormity diagnosis and interpolation method of regimen monitoring data | |
| CN104700153A (en) | PH (potential of hydrogen) value predicting method of BP (back propagation) neutral network based on simulated annealing optimization | |
| CN106843172A (en) | Complex industrial process On-line quality prediction method based on JY KPLS | |
| CN202306196U (en) | Temperature field digital/analog simulation system for heating furnace | |
| CN105807632A (en) | Heating furnace temperature controller design method based on improved Dahlin algorithm | |
| CN107909204A (en) | A kind of processing parameter optimization method and device | |
| CN101765258A (en) | Three-phase electrode positioning device in smelting process of electro-fused magnesia furnace and control method thereof | |
| CN104732340B (en) | Target plans power grid aid decision-making method when one kind is based on year | |
| CN201476905U (en) | Neural network PID temperature controlled thermocouple automatic verification system | |
| CN205384740U (en) | Terminal Test system device is gathered to digital intelligent electric flux | |
| CN105400951B (en) | A kind of humidification hybrid control system and its control method for annealing process of silicon steel | |
| CN104597208B (en) | Multi-channel cyclic sampling gas analysis method | |
| CN105706010B (en) | Production line analogue means | |
| CN104578412A (en) | Big data technology-based state detection system and method | |
| CN107262533A (en) | A kind of control method for laminar cooling and device | |
| CN101833795B (en) | Dynamic recorder and recording method for heat treatment parameters | |
| CN116179840A (en) | Laser surface heat treatment temperature monitoring control system and control method | |
| CN114925569A (en) | Steel plate quenching temperature prediction method combining finite element and neural network | |
| CN103439964B (en) | A kind of online updating system and method for OCSVM monitoring model | |
| CN102419593A (en) | Sensor fault diagnosis method based on data mining |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |