CN108428915A - A kind of fuel cell exhaust process anode pressure control method based on iterative learning - Google Patents
A kind of fuel cell exhaust process anode pressure control method based on iterative learning Download PDFInfo
- Publication number
- CN108428915A CN108428915A CN201810249546.9A CN201810249546A CN108428915A CN 108428915 A CN108428915 A CN 108428915A CN 201810249546 A CN201810249546 A CN 201810249546A CN 108428915 A CN108428915 A CN 108428915A
- Authority
- CN
- China
- Prior art keywords
- sequence
- exhaust process
- iteration
- anode
- anode pressure
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M8/00—Fuel cells; Manufacture thereof
- H01M8/04—Auxiliary arrangements, e.g. for control of pressure or for circulation of fluids
- H01M8/04082—Arrangements for control of reactant parameters, e.g. pressure or concentration
- H01M8/04089—Arrangements for control of reactant parameters, e.g. pressure or concentration of gaseous reactants
- H01M8/04104—Regulation of differential pressures
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/30—Hydrogen technology
- Y02E60/50—Fuel cells
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Sustainable Development (AREA)
- Sustainable Energy (AREA)
- Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- General Chemical & Material Sciences (AREA)
- Fuel Cell (AREA)
- Feedback Control In General (AREA)
Abstract
The fuel cell exhaust process anode pressure control method based on iterative learning that the invention discloses a kind of, including：Iteration control parameter is initialized to memory module；Obtain output of the controlled device under controlled quentity controlled variable sequence；Error criterion is calculated, control effect is assessed；Pass through control information and iterative learning matrix update controlled quentity controlled variable sequence；Into exhaust process next time, return to step 2 repeats step 2 to the process of step 5.The method of the present invention is measured in order to control with anode incoming gas mass flow, iteration updates controlled quentity controlled variable sequence, control exhaust process anode pressure accurately tracks setting value, since this method is data-driven, Model free control, eliminate the modeling work to controlled device, have certain learning compensation ability for the Parameter Perturbation of controlled device, improves system Control platform, extend the service life of Proton Exchange Membrane Fuel Cells.
Description
Technical field
The invention belongs to new energy automatic control technology field more particularly to a kind of fuel cell rows based on iterative learning
Gas process anode pressure control method.
Background technology
Energy demand with growth and environmental protection standard, it is becoming for world today's development to develop clean new energy technology
Gesture.Fuel cell technology directly converts the chemical energy in fuel to electric energy by electrochemical reaction, is before one kind has extensively
The energy utilization technology of scape.Wherein Proton Exchange Membrane Fuel Cells has high-energy source transfer efficiency, high-energy density, pollution-free row
It puts, the small feature low with running temperature of noise, thus has obtained extensive commercial application.To save design and operating cost, greatly
Most Proton Exchange Membrane Fuel Cells are eliminated additional fuel re-cycling arrangement, are improved using the method for operation of closing anode
Fuel availability.The component that this method of operation can cause vapor and nitrogen etc. to be unfavorable for electrochemical reaction accumulates in anode stream
Road influences fuel battery performance, it is necessary to these foreign gas components be discharged by periodical exhaust process.However it periodically arranges
The fluctuation that gas process can cause anode pressure larger again, and cathode pressure is basically unchanged when load stabilization, cathode and anode both sides are larger
It is therefore desirable to the anode-side pressures to exhaust process to apply high level control to proton exchange membrane application for pressure difference.
Domestic and foreign scholars are concentrated mainly on the release for the research of exhaust process Anodic pressure control problems at present
With the static optimization of duration, still there is deficiency for the optimal control in dynamic of exhaust process.Currently, there is scholar's use ratio-
Integrated Derivative (PID, Proportion Integration Differentiation) controller or Model Predictive Control
(MPC, Model Predictive Control) controls anode pressure, achieves certain effect.However since PID is controlled
Device processed belongs to subsequent control, and anode pressure unavoidably will produce larger fluctuation；MPC is the forecast Control Algorithm based on model,
Its control effect depends on the accuracy of Controlling model, in controlled device dynamic characteristic complexity, the foundation of Controlling model or distinguishes
Know and need a large amount of work, and difficulty is larger.Meanwhile the controller parameter of PID and MPC control strategies often adjust in advance and
It is no longer adjusted after coming into operation, when Parameter Perturbation, when dynamic characteristic changes therewith, preset parameter occur for controlled device
The control effect of controller will deteriorate, or even influence system stability.
Invention content
The problem of for existing control program and deficiency, it is contemplated that Proton Exchange Membrane Fuel Cells exhaust process has
There is periodicity and dynamic characteristic is complicated, the present invention provides a kind of, and the fuel cell exhaust process based on iterative learning is positive
Extreme pressure force control method saves the priori and identification modeling of controlled device, has the energy of compensation object parameters perturbation
Power realizes that anode of proton exchange membrane fuel cell pressure accurately tracks setting value in exhaust process, reduces anode pressure value
Fluctuation, promotes the service life of Proton Exchange Membrane Fuel Cells.
For achieving the above object, the present invention uses following technical scheme：A kind of fuel cell based on iterative learning
Exhaust process anode pressure control method, using the anode pressure in periodical exhaust process as controlled volume, anode incoming gas matter
Amount flow is measured in order to control, and design iteration learns control flow, and the feature with data-driven and model-free realizes proton exchange membrane
Fuel cell exhaust process anode pressure accurately tracks setting value, is as follows：
Step 1, select anode pressure for exhaust process controlled volume, anode incoming gas mass flow controls for exhaust process
Amount, initialization iterative learning controls relevant parameter, and to memory module, (memory module can be hard disc of computer or other storages
Equipment), the initialization iterative learning control relevant parameter includes initialization exhaust process Anodic pressure set points sequence
yref, the initial controlled quentity controlled variable sequence u of anode incoming gas mass flow0, iterative learning matrix ΓPAnd ΓD, juxtaposition iteration variable k is
Initial value；
Step 2, the controlled quentity controlled variable sequence in memory module is applied to Proton Exchange Membrane Fuel Cells exhaust process, obtains the
The output sequence y of k iteration final vacuum process anode pressurek；
Step 3, corresponding error criterion is calculated, controlled quentity controlled variable sequence u after kth time iteration is assessedkControl to anode pressure
Effect, including calculate anode pressure absolute error sequence ek, error differential sequenceWith root-mean-square error RMSEkIf root mean square misses
Poor RMSEkLess than the root-mean-square error threshold value RMSE of settingc, then controlled quentity controlled variable sequence is not updated, is taken in currently stored module
Controlled quentity controlled variable sequence be subsequent exhaust process use controlled quentity controlled variable sequence, go to step 5, otherwise go to step 4；
Step 4, according to absolute error sequence and error differential sequence, the controlled quentity controlled variable for calculating newly using iterative learning matrix
Sequence uk+1, mould controlled quentity controlled variable sequence in the block is updated storage, while updating iteration variable k=k+1；
Step 5, into exhaust process next time, return to step 2 repeats step 2 to process (process of step 5
It is Infinite Cyclic in fuel battery service life).
In step 1 of the present invention, the exhaust process anode pressure setting value sequences yref, anode incoming gas mass flow
Initial controlled quentity controlled variable sequence u0It is respectively provided with following form：
yref=[yref(T0)yref(T1)…yref(TN-1)]1×N (1)
u0=[u0(T0)u0(T1)…u0(TN-1)]1×N (2)
Wherein, an exhaust process is from T0Moment continues to TN-1Moment, N are number of samples, are generally taken as exhaust process and hold
Continuous time TN-1-T0With the ratio of sampling time T, yref(Ti), i=0,1 ..., N-1 indicates i+1 in an exhaust process
The anode pressure setting value of sampling instant, u0(Ti), i=0,1 ..., when N-1 indicates that i+1 samples in an exhaust process
The anode incoming gas mass flow initial value at quarter, and u0=a [11 ... 1]1×N,Or it is true by other reference controllers
It is fixed.
In step 1 of the present invention, the initialization iterative learning matrix ΓP、ΓDIt is initial value with iteration variable k is set, respectively
As following formula is realized：
K=0 (5)
Wherein,I=0,1 ..., N-1 is ratio iterative learning matrix ΓPI-th of diagonal entry,I=0,1 ..., N-1 is Differential iteration learning matrix ΓDI-th of diagonal entry, ΓPAnd ΓDIt can ensure to change
It withholds and is arbitrarily chosen under the premise of holding back, iteration variable k has recorded the newer number of controlled quentity controlled variable sequence.
In step 2 of the present invention, the output sequence y of the kth time iteration final vacuum process anode pressurekWith following shape
Formula：
yk=[yk(T0)yk(T1)…yk(TN-1)]1×N (6)
Wherein, yk(Ti), i=0,1 ..., N-1 indicates anode pressure i+1 in an exhaust process after kth time iteration
The sampled value at a moment, and the value of the 0th iteration, that is, k be 0 when indicate initial controlled quentity controlled variable sequence is not updated.
In step 3 of the present invention, controlled quentity controlled variable sequence u after the kth time iterationkWith following form：
uk=[uk(T0)uk(T1)…uk(TN-1)]1×N (7)
Wherein, uk(Ti), i=0,1 ..., anode incoming gas mass flow is once being arranged after N-1 indicates kth time iteration
The value of i+1 sampling instant during gas, and have u when k=0k=u0。
In step 3 of the present invention, anode pressure absolute error sequence e is calculated by following formulak, error differential sequenceWith
Root-mean-square error RMSEk：
ek=[ek(T0)ek(T1)…ek(TN-1)]1×N=yref-yk (8)
Wherein, ek(Ti), i=0,1 ..., anode pressure absolute error was vented once after N-1 indicates kth time iteration
The value of i+1 sampling instant in journey,I=0,1 ..., anode pressure error differential exists after N-1 indicates kth time iteration
The value of i+1 sampling instant in exhaust process.
Root-mean-square error threshold value RMSE described in step 3 of the present inventioncWith anode pressure dimension having the same, determine repeatedly
The degree of agreement of anode pressure output and setting value at the end of generation.
It is described according to absolute error sequence and error differential sequence in step 4 of the present invention, it is calculated using iterative learning matrix
Obtain new controlled quentity controlled variable sequence uk+1Formula it is as follows：
As a preferred embodiment of the present invention, step 1 Anodic pressure set points sequences yrefGenerally it is taken as proton exchange
Membrane fuel battery cathod pressure, with the control targe ensured in exhaust process be by iterative learning reduce cathode, anode it
Between pressure difference extend fuel cell to reduce in exhaust process the stress acted on because of pressure difference in proton exchange membrane
Service life.
As a preferred embodiment of the present invention, initially controlled in step 1 for obtaining anode incoming gas mass flow
Measure sequence u0Other reference controllers can be PID or MPC etc., then u0It is that the reference controller is electric in pem fuel
In the exhaust process of pond one time controlled quentity controlled variable sequence is generated by anode incoming gas mass flow controls anode pressure.
As a preferred embodiment of the present invention, initially controlled in step 1 for obtaining anode incoming gas mass flow
Measure sequence u0Other reference controllers can also be iterative learning controller of the present invention, then u0Exist for the reference controller
The more newly-generated controlled quentity controlled variable sequence of certain iteration in Proton Exchange Membrane Fuel Cells exhaust process anode pressure iteration control.
As a preferred embodiment of the present invention, logic is redirected in step 3 and ensure that in controlled device generation Parameter Perturbation
When, it, can also even if existing controlled quentity controlled variable sequence can not obtain satisfied exhaust process anode pressure control effect in memory module
By being again started up iterative learning, according to exhaust process, periodically Correction and Control amount sequence, the control up to obtaining satisfaction are imitated
Fruit.
Using above-mentioned technical proposal, the method for the present invention has advantageous effect below：
The method of the present invention is directed to the feature of Proton Exchange Membrane Fuel Cells periodicity exhaust process characteristic complexity, is entered with anode
Stream gas mass flow is measured in order to control, and iteration updates controlled quentity controlled variable sequence, and control exhaust process anode pressure accurately tracks setting value,
The fluctuation for reducing anode pressure extends Proton Exchange Membrane Fuel Cells, especially the wherein service life of proton exchange membrane,
Improve system overall control quality；
The method of the present invention is data-driven, Model free control, does not need the priori in relation to controlled device characteristic, saves
Modeling and identification work to controlled device especially in the case of controlled device dynamic characteristic complexity simplify control
The design process of device；
It is root-mean-square error threshold value that the method for the present invention, which judges whether to the modified condition of controlled quentity controlled variable sequence, is based on tradition
The control method of model is compared, and control deterioration situation has certain study caused by Parameter Perturbation occurs for controlled device
Compensation ability.
Description of the drawings
The present invention is done with reference to the accompanying drawings and detailed description and is further illustrated, it is of the invention above-mentioned or
Otherwise advantage will become apparent.
Fig. 1 is that the present invention is based on the Proton Exchange Membrane Fuel Cells exhaust process anode pressure controlling parties that iterative learning controls
Method structure diagram；
Fig. 2 is that the present invention is based on the fuel cell exhaust process anode pressure control method flow charts of iterative learning；
Fig. 3 is that the Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of the present invention tracks examination in setting value
Test Anodic pressure output curve；
Fig. 4 is that the Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of the present invention tracks examination in setting value
Test Anodic incoming gas mass flow curve；
The Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of Fig. 5 present invention is in setting value tracking test
Anodic pressure root-mean-square error with iterations change curve；
Fig. 6 is that the Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of the present invention is compensated in Parameter Perturbation
Test Anodic pressure output curve；
Fig. 7 is that the Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of the present invention is compensated in Parameter Perturbation
Test Anodic incoming gas mass flow curve；
The Proton Exchange Membrane Fuel Cells exhaust process anode pressure control method of Fig. 8 present invention compensates real in Parameter Perturbation
Test change curve of the Anodic pressure root-mean-square error with iterations.
Specific implementation mode
The present invention will be further described with reference to the accompanying drawings and embodiments.
In order to keep the technical problem to be solved in the present invention, technical solution and advantageous effect clearer, below in conjunction with the accompanying drawings
And specific embodiment carries out carrying out detailed description to the specific implementation mode of the method for the present invention.Below with reference to attached drawing description
Embodiment is exemplary, and is only used for explaining the present invention, and is not construed as limiting the claims.
The hydrogen-air proton exchange established herein with the famous scholar Anna G.Stefanopoulou of fuel cell field
Membrane cell model illustrates the design and embodiment of the method for the present invention as controlled device.The hydrogen-air proton
The periodical exhaust process of exchange film fuel battery model determines by the aperture of air bleeding valve, and air bleeding valve is fully closed when normal operation (opens
It is then opened by the rate of 100%/s when degree is, needs exhaust 0%), until air bleeding valve standard-sized sheet (aperture 100%), is not required to be vented
Shi Ze is closed by the rate of -100%/s.
Fig. 1 illustrates the control method structure diagram of the present invention, including links such as data processing, controlled quentity controlled variable updates.
In an exhaust process, it is k that might as well enable iterations at this time, and storage link is responsible for controlled quentity controlled variable sequence uk, namely exhaust
Process anode incoming gas mass flow, effect to controlled device namely hydrogen-air Proton Exchange Membrane Fuel Cells；It is controlled
The influence of object not only controlled amount, is also influenced by Parameter Perturbation, output quantity of the controlled device under the effect of controlled quentity controlled variable sequence
Sequences ykNamely anode pressure is sent in the output consecutive sample values of exhaust process to data processing link, is calculated output tracking and is set
Definite value yrefEach error criterion namely absolute error sequence ek, error differential sequenceWith root-mean-square error RMSEk, judgement is
No termination iteration updates link Correction and Control amount sequence if continuing iteration by controlled quentity controlled variable, and by revised controlled quentity controlled variable sequence
Row are sent to memory module, the controlled quentity controlled variable sequence in update module, and controlled quentity controlled variable sequence is kept in memory module if final value iteration not
Become, so far completes the iteration control of an exhaust process, enable k=k+1, into being vented iteration next time.Fig. 2 illustrates the present invention
The iteration control flow chart of method.
Embodiment 1：Setting value tracking test.With the controlled quentity controlled variable sequence generated with reference to discrete ratios-differential (PI) controller
Based on, using Iterative Learning Control Algorithm, controlled quentity controlled variable sequence is updated, promotes exhaust process Anodic pressure tracking setting value
Effect.
Embodiment 2：Parameter Perturbation compensation experiment.When Parameter Perturbation occurs for controlled device, at the end of 1 iteration of embodiment
Controlled quentity controlled variable sequence based on, using Iterative Learning Control Algorithm, update controlled quentity controlled variable sequence, target compensation Parameter Perturbation is to control
The influence of effect.The Parameter Perturbation of controlled device is grate flow channel flow area from 0.0005m in the present embodiment2Step reduces
1% to 0.000495m2。
In 2 embodiments of the method for the present invention, sampling time T is taken as 0.1s, studied Proton Exchange Membrane Fuel Cells
An exhaust process under 40kW constant load continuous operation operating modes the period be 10s, wherein：The fuel cell starts to arrange when 2s
Gas, air bleeding valve start to open at, but due to valve rate limit, until valve is just opened completely when 3s；Stop to fuel cell when 4s
It is only vented, air bleeding valve is begun to shut off, until just stop exhaust when 5s completely, but is needed to the just row of being restored to of fuel cell when 10s
Operating mode before gas, therefore number of samples N=100.
Step 1. initializes iterative learning and controls relevant parameter, exhaust process Anodic pressure set points sequences yrefIt is as follows
Shown in formula：
yref=pca·[1 1…1]1×100 (12)
Wherein, anode pressure setting value sequences yrefUse length for 100 cathode pressure pcaConstant value vector, due to the matter
Proton exchange film fuel cell is in 40kW constant load continuous operation operating modes, and constant cathode pressure is pca=2.007 ×
105Pa, therefore yref=2.007 × 105·[1 1…1]1×100。
To ensure the stabilization of controlled device, in embodiment 1, the initial controlled quentity controlled variable sequence u of anode incoming gas mass flow0
It is generated in an exhaust process with reference to Discrete PI controller by one, the controller form is as follows：
Wherein, KP=3.056 × 10-8For proportionality coefficient, KI=4.086 × 10-8For integral coefficient, emulation moment t corresponds to
Controlled quentity controlled variable uPI(t) by the absolute error e (t) of moment t, up to the error accumulation amount of moment t since emulationAnd
Controller parameter calculates；And in example 2, initial controlled quentity controlled variable sequence u0For 1 iteration ends of embodiment when memory module in
Controlled quentity controlled variable sequence.Therefore in the 2 of the method for the present invention embodiments, the initial controlled quentity controlled variable sequence u of anode incoming gas mass flow0It can
It is expressed as：
Wherein, uPIFor the controlled quentity controlled variable sequence generated in an exhaust process with reference to Discrete PI controller, uemb1To implement
Controlled quentity controlled variable sequence in memory module when 1 iteration ends of example.
Initialization ratio iterative learning matrix ΓPWith Differential iteration learning matrix ΓD, iteration variable k, respectively such as following formula institute
Show：
K=0 (17)
Wherein, ΓPAnd ΓDIt can guarantee controlled device iteration convergence in 2 embodiments, respectively dimension is 100 and diagonal element
The identical diagonal matrix of element；The controlled quentity controlled variable sequence set in the memory module of iteration variable k=0 expressions at this time is initial controlled quentity controlled variable sequence
Arrange u0, while indicating not to initial controlled quentity controlled variable sequence u0It is iterated update.
Controlled quentity controlled variable sequence in memory module is applied to Proton Exchange Membrane Fuel Cells exhaust process by step 2., obtains quilt
Control the continuous sampling y of object anode pressure output sequence after kth time iterationk=[yk(0)yk(1)…yk(99)]1×100, wherein
yk(i), i=0,1 ..., 99 indicate the sampling at anode pressure i+1 moment in an exhaust process after kth time iteration
Value.
Step 3. calculates anode pressure absolute error sequence ek, error differential sequenceWith root-mean-square error RMSEk, calculate
Mode is distinguished as follows：
ek=[ek(0)ek(1)…ek(99)]1×100=yref-yk (18)
Wherein, ek(i), i=0,1 ..., anode pressure absolute error is in an exhaust process after 99 expression kth time iteration
The value of middle i+1 sampling instant,Anode pressure error differential is once being arranged after indicating kth time iteration
The value of i+1 sampling instant during gas.Controlled quentity controlled variable sequence of the controlled device after kth time iteration is assessed by error criterion
uk=[uk(0)uk(1)…uk(99)]1×100The control effect to anode pressure under effect, if root-mean-square error RMSEkIt is less than
The root-mean-square error threshold value RMSE of settingc, then controlled quentity controlled variable sequence is not updated, gos to step 5, take currently stored module
In controlled quentity controlled variable sequence be controlled quentity controlled variable sequence that subsequent exhaust process uses, otherwise go to step 4, wherein uk(i), i=0,
1 ..., 99 indicate anode incoming gas mass flow i+1 sampling instant in an exhaust process after kth time iteration
Value；Root-mean-square error threshold value RMSEcWith anode pressure dimension having the same, determine at the end of iteration anode pressure output with
The degree of agreement of track setting value takes RMSE in 2 embodiments of the method for the present inventionc=50Pa.
Step 4. is according to absolute error and error differential sequence and ratio, Differential iteration learning matrix, the control for calculating newly
Amount sequence u processedk+1It is as follows：
Mould controlled quentity controlled variable sequence in the block is updated storage, and updates iteration variable k=k+1.
Step 5. enters exhaust process, return to step 2 next time, repeats step 2 to the process of step 5.
Using the fuel cell exhaust process anode pressure control method based on iterative learning in the present invention, in embodiment 1
Proton Exchange Membrane Fuel Cells exhaust process anode pressure is controlled in the case of 2 descriptions, control effect such as Fig. 3~8
It is shown.
Fig. 3 is the curve graph that exhaust process anode pressure changes with iterations in embodiment 1 (setting value tracking test),
It can be seen from the figure that with the increase of iterations, based on the control effect (k=0) with reference to Discrete PI controller, sun
The effect of pole pressure output value tracking fixed valure sequence is constantly promoted.Fig. 4 be 1 Anodic incoming gas mass flow of embodiment with
Iterations variation curve graph, as iterations increase, anode incoming gas mass flow is also being continuously increased, with row
Throughput reaches balance.From root-mean-square error shown in fig. 5 it can also be seen that anode pressure setting value tracking effect is constantly being promoted,
Until when iteration variable k=572, RMSE572Less than threshold value 50Pa, iteration terminates.Fig. 6 is that (Parameter Perturbation compensation is real for embodiment 2
Test) in the curve graph that changes with iterations of exhaust process anode pressure, it can be seen from the figure that since object parameters are taken the photograph
Dynamic, (k=0), which is adjusted, with the controlled quentity controlled variable sequence at the end of 1 iteration of embodiment has relatively large deviation, but as iterations increase
Add, anode pressure setting value tracking effect constantly improves.Fig. 7 is 2 Anodic incoming gas mass flow of embodiment with iteration time
The curve graph of number variation, as iterations increase, anode incoming gas mass flow constantly adjusts, to adapt to new exhaust
Journey flox condition.As can be seen from Figure 8, root-mean-square error is gradually reduced with iterations, as iteration variable k=71,
RMSE71Less than threshold value 50Pa, iteration terminates.
The fuel cell exhaust process anode pressure control method based on iterative learning that the present invention provides a kind of, it is specific real
Now there are many method of the technical solution and approach, the above is only a preferred embodiment of the present invention, it is noted that for this
For the those of ordinary skill of technical field, without departing from the principle of the present invention, several improvement and profit can also be made
Decorations, these improvements and modifications also should be regarded as protection scope of the present invention.Each component part being not known in the present embodiment is available
The prior art is realized.
Claims (8)
1. a kind of fuel cell exhaust process anode pressure control method based on iterative learning, which is characterized in that including following
Step：
Step 1, select anode pressure for exhaust process controlled volume, anode incoming gas mass flow is exhaust process controlled quentity controlled variable,
It initializes iterative learning and controls relevant parameter to memory module, the initialization iterative learning control relevant parameter includes initialization
Exhaust process Anodic pressure set points sequences yref, the initial controlled quentity controlled variable sequence u of anode incoming gas mass flow0, iterative learning
Matrix ΓPAnd ΓD, juxtaposition iteration variable k is initial value；
Step 2, the controlled quentity controlled variable sequence in memory module is applied to Proton Exchange Membrane Fuel Cells exhaust process, obtains kth time
The output sequence y of iteration final vacuum process anode pressurek；
Step 3, corresponding error criterion is calculated, controlled quentity controlled variable sequence u after kth time iteration is assessedkTo the control effect of anode pressure,
Including calculating anode pressure absolute error sequence ek, error differential sequenceWith root-mean-square error RMSEkIf root-mean-square error
RMSEkLess than the root-mean-square error threshold value RMSE of settingc, then controlled quentity controlled variable sequence is not updated, takes currently stored mould in the block
Controlled quentity controlled variable sequence is the controlled quentity controlled variable sequence that subsequent exhaust process uses, and goes to step 5, otherwise goes to step 4；
Step 4, according to absolute error sequence and error differential sequence, the controlled quentity controlled variable sequence for calculating newly using iterative learning matrix
uk+1, mould controlled quentity controlled variable sequence in the block is updated storage, while updating iteration variable k=k+1；
Step 5, into exhaust process next time, return to step 2 repeats step 2 to the process of step 5.
2. according to the method described in claim 1, it is characterized in that, in step 1, the exhaust process anode pressure setting value sequence
Arrange yref, the initial controlled quentity controlled variable sequence u of anode incoming gas mass flow0It is respectively provided with following form：
yref=[yref(T0) yref(T1) … yref(TN-1)]1×N (1)
u0=[u0(T0) u0(T1) … u0(TN-1)]1×N (2)
Wherein, an exhaust process is from T0Moment continues to TN-1Moment, N are number of samples, yref(Ti), i=0,1 ..., N-1
Indicate the anode pressure setting value of i+1 sampling instant in an exhaust process, u0(Ti), i=0,1 ..., N-1 indicates one
The anode incoming gas mass flow initial value of i+1 sampling instant in secondary exhaust process, and u0=a [1 1 ... 1
]1×N,Or it is determined by other reference controllers.
3. according to the method described in claim 2, it is characterized in that, in step 1, the initialization iterative learning matrix ΓP、ΓD
It is initial value with iteration variable k is set, respectively as following formula is realized：
K=0 (5)
Wherein,For ratio iterative learning matrix ΓPI-th of diagonal entry,For Differential iteration learning matrix ΓDI-th of diagonal entry, iteration variable k has recorded control
Measure the newer number of sequence.
4. according to the method described in claim 3, it is characterized in that, in step 2, the kth time iteration final vacuum process anode
The output sequence y of pressurekWith following form：
yk=[yk(T0) yk(T1) … yk(TN-1)]1×N (6)
Wherein, yk(Ti), i=0,1 ..., N-1 indicates after kth time iteration anode pressure in an exhaust process when i+1
The sampled value at quarter, and the value of the 0th iteration, that is, k be 0 when indicate initial controlled quentity controlled variable sequence is not updated.
5. according to the method described in claim 4, it is characterized in that, in step 3, controlled quentity controlled variable sequence u after the kth time iterationkTool
There is following form：
uk=[uk(T0) uk(T1) … uk(TN-1)]1×N (7)
Wherein, uk(Ti), i=0,1 ..., anode incoming gas mass flow was vented once after N-1 indicates kth time iteration
The value of i+1 sampling instant in journey, and have u when k=0k=u0。
6. according to the method described in claim 5, it is characterized in that, in step 3, it is absolute that anode pressure is calculated by following formula
Error sequence ek, error differential sequenceWith root-mean-square error RMSEk：
ek=[ek(T0) ek(T1) … ek(TN-1)]1×N=yref-yk (8)
Wherein, ek(Ti), i=0,1 ..., anode pressure absolute error is in an exhaust process after N-1 indicates kth time iteration
The value of i+1 sampling instant,Anode pressure error differential is once being arranged after indicating kth time iteration
The value of i+1 sampling instant during gas.
7. according to the method described in claim 6, it is characterized in that, the RMSE of root-mean-square error threshold value described in step 3cWith anode
Pressure dimension having the same determines the degree of agreement of anode pressure output and setting value at the end of iteration.
8. described micro- according to absolute error sequence and error the method according to the description of claim 7 is characterized in that in step 4
Sub-sequence calculates to obtain new controlled quentity controlled variable sequence u using iterative learning matrixk+1Formula it is as follows：
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810249546.9A CN108428915B (en) | 2018-03-26 | 2018-03-26 | Fuel cell exhaust process anode pressure control method based on iterative learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810249546.9A CN108428915B (en) | 2018-03-26 | 2018-03-26 | Fuel cell exhaust process anode pressure control method based on iterative learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108428915A true CN108428915A (en) | 2018-08-21 |
| CN108428915B CN108428915B (en) | 2021-04-27 |
Family
ID=63159120
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810249546.9A Active CN108428915B (en) | 2018-03-26 | 2018-03-26 | Fuel cell exhaust process anode pressure control method based on iterative learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108428915B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112563541A (en) * | 2020-12-11 | 2021-03-26 | 昆明理工大学 | Fuel cell cathode pressure control method for improving particle swarm PID |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104407642A (en) * | 2014-12-01 | 2015-03-11 | 杭州电子科技大学 | Temperature control method for continuous casting billet induction heating process, based on iterative learning control |
| CN104701839A (en) * | 2014-09-03 | 2015-06-10 | 国家电网公司 | Air conditioner load modeling method based on least squares parameter identification |
| CN104850679A (en) * | 2015-04-03 | 2015-08-19 | 浙江工业大学 | Static pressure control method of variable air volume (VAV) air-conditioning system fan on basis of iterative learning |
| CN106443585A (en) * | 2016-09-09 | 2017-02-22 | 中国地质大学（武汉） | Accelerometer combined LED indoor 3D positioning method |
| CN106654319A (en) * | 2016-12-27 | 2017-05-10 | 东南大学 | Temperature modeling method for proton exchange membrane fuel cell (PEMFC) system based on variation particle swarm and differential evolution hybrid algorithm |
-
2018
- 2018-03-26 CN CN201810249546.9A patent/CN108428915B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104701839A (en) * | 2014-09-03 | 2015-06-10 | 国家电网公司 | Air conditioner load modeling method based on least squares parameter identification |
| CN104407642A (en) * | 2014-12-01 | 2015-03-11 | 杭州电子科技大学 | Temperature control method for continuous casting billet induction heating process, based on iterative learning control |
| CN104850679A (en) * | 2015-04-03 | 2015-08-19 | 浙江工业大学 | Static pressure control method of variable air volume (VAV) air-conditioning system fan on basis of iterative learning |
| CN106443585A (en) * | 2016-09-09 | 2017-02-22 | 中国地质大学（武汉） | Accelerometer combined LED indoor 3D positioning method |
| CN106654319A (en) * | 2016-12-27 | 2017-05-10 | 东南大学 | Temperature modeling method for proton exchange membrane fuel cell (PEMFC) system based on variation particle swarm and differential evolution hybrid algorithm |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112563541A (en) * | 2020-12-11 | 2021-03-26 | 昆明理工大学 | Fuel cell cathode pressure control method for improving particle swarm PID |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108428915B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2017156888A1 (en) | Sofc system thermo-electric coordinated control method on the basis of fractional order sliding mode variable structure | |
| CN111725546B (en) | Test system and test method for fuel cell stack working condition | |
| CN111029625B (en) | Output power and temperature control method for solid oxide fuel cell | |
| CN112713288B (en) | Control system and control method for fuel cell bubbling humidifier | |
| CN110838590B (en) | Gas supply control system and method for proton exchange membrane fuel cell | |
| CN109444749A (en) | Temperature compensation system and method, storage medium for lithium battery capacity measurement | |
| CN108428915A (en) | A kind of fuel cell exhaust process anode pressure control method based on iterative learning | |
| CN106300365A (en) | A kind of static voltage stability control method based on air conditioner load | |
| CN112563541A (en) | Fuel cell cathode pressure control method for improving particle swarm PID | |
| CN112508333A (en) | Estimation system and method for greenhouse effect of biomass energy utilization | |
| CN104167915A (en) | Method and circuit for controlling Buck DC/DC converter | |
| WO2023019536A1 (en) | Deep reinforcement learning-based photovoltaic module intelligent sun tracking method | |
| CN115621508A (en) | Method for controlling pressure and flow of cathode gas of fuel cell system | |
| CN111682759B (en) | Fuel cell DC-DC converter operation control method, computer readable storage medium and fuel cell | |
| CN115347218A (en) | Cascade control method of air supply system of proton exchange membrane fuel cell | |
| CN111106371B (en) | Fuel cell air management system and method for acquiring related control signals thereof | |
| CN114134017A (en) | Anaerobic fermentation temperature control system and method based on BP neural network prediction | |
| CN113555592A (en) | Nitrogen concentration self-adaptive control method of fuel cell system | |
| CN112711187A (en) | Multi-electric field cooperative control method for dry type electric dust remover of coal-fired unit | |
| CN117878972B (en) | Quick frequency adjustment method and device for power distribution network with participation of multiple photovoltaic sources | |
| CN110970936A (en) | Method for calculating primary frequency modulation performance of deep peak shaving unit | |
| CN105199973B (en) | A kind of Yeast Cultivation online adaptive control method based on differential evolution algorithm | |
| CN114678567B (en) | Fuel cell system power optimization method aiming at control parameters | |
| Hähnel et al. | Online identification of an electric PEMFC model for power control by NMPC | |
| CN117096384A (en) | Humidification system of proton exchange membrane fuel cell and control method |
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 |