0 2 4 6 8 10 12 14 16 18
−0.2
−0.1 0 0.1 0.2 0.3
Real and estimated model outputs - Fit = 62.95 % - Vaf = 86.29 %
System output Model output
0 2 4 6 8 10 12 14 16 18
−2
−1 0 1 2
Time (s)
Real and estimated model outputs - Fit = 94.89 % - Vaf = 99.74 %
System output Model output
Figure7.17: Global validationdata.
7.4.2.2 Blak-box ase
As a next step, let us present the validation results of the estimated blak-box
LPV models. The obtained averageloalBFTmeasurements aregiven inTable 7.3
while theglobalvalidationstepyieldsthe followingaveragetmeasurements
73.8%
and
97.2%
forφ ˙ 1
andφ ˙ 2
, respetively. Notie that, in the blak-box framework, thereare moreparameterstotune duringtheoptimization. This isthe mainreasonwhy we ould obtain slightlybetter BFTs in this ase than in the gray-box one.
i 2 4 6 8 10
φ ˙ 1 [BF T ]
84.64 82.34 82.33 85.03 84.80φ ˙ 2 [BF T ]
79.26 79.19 82.43 80.40 75.24Table7.3: Performane metris(BFT (%))forthe estimated frozen blak-box LPV
models on validationdata
φ (i) 2,0 ∈ [π/8 : π/16 : 6π/8]
.TheobtainedaverageBFTsequalto
[73.8% 97.2%]
and[59.9% 94.22%]
intheblak-andgray-boxase,respetively. Notiethatwhentheverysimplemodel,depitedin
Figure7.15, isused, the obtained average BFTmeasurements are
[53.02% 84.31%]
respetivelyon
φ ˙ 1
andφ ˙ 2
whihislowerthanwhatisobtainedbyusingtheestimatedLPV models. Therefore, the proposed modelreallyallows toapture the dynamis
of the system. All these results prove that, adding up prior information through
the knowledge of the LPV model struture, is an eient solution resulting in an
aurate LPV modelby applying loal experimental data and the
H ∞
-norm-based method developed in Chapter 5.Chapter 8
Summary of the obtained results and
future researh objetives
8.1 Thesis Points
In this Setion, the developed new sienti results are summarized as thesis
points with the referenes to the orresponding Chapters in this manusript. The
rst thesis deals with the re-struturing of blak-box state-spae LTI models into
gray-box ones. The seond and third thesis form a separate thesis group beause
they takle the identiation problem of state-spae LPV models by involving a
lassial interpolation step. Even though, the fourth and fth thesis deal alsowith
the identiationof state-spae LPV models, they ompose adistint thesis group,
beause here, anew behavioral approahand the
H ∞
-norm isused toestimatetheLPV model fromloalexperiments. In the following, the developed tehniques are
enumerated aording tothe abovepresented grouping.
Thesis 1 A new tehnique being able to restruture blak-box linear
time-invariant (LTI) state-spae models into gray-box ones, by traing bak the
identi-ation problem to a strutured
H ∞
synthesis problem, has been developed. Theblak- and gray-box LTI models are ompared in the frequeny domain. Then, by
minimizing the
H ∞
-norm-based ost funtion dened by Eq. (4.16), the unknown parameters found inthe gray-box LTI modelare determined.Thisthesispointisdeveloped andpresentedinChapter4. Asimulationexample
isusedtodemonstratethe eetivenessofthe proposed solutioninSetion6.2. The
obtained results have been published in[155, 153℄.
Thesis 2.1 Anew tehniqueperformingthe identiationof interpolatedlinear
parameter-varying (LPV) models from loally restrutured models by using
stru-tured
H ∞
synthesishasbeendeveloped. Inthisase,theloallyidentiedblak-boxLTImodels aretransformedintotheorrespondingloalfrozengray-boxLPV
mod-els, inevery working point,by usingloallythe
H ∞
-norm-based LTI re-struturing tehnique developed during the rst thesis point. This step is then followed by alassialleast-squares-based interpolation inordertoderive the nalgray-box LPV
model.
is used to demonstrate the eetiveness of the proposed solution in Chapter 6.3.3.
The obtained resultshave been published in [154℄.
Thesis2.2 Anewtehniqueperformingthe identiationof interpolatedlinear
parameter-varyingmodels fromloallyrestrutured models by usingthe
null-spae-based tehnique has been developed. Here, the loally identied blak-box LTI
models are again transformed into the orresponding loally frozen gray-box LPV
models by using anull-spae-based tehnique, developed in[115℄. This step is then
followed by a lassial least-squares-based interpolation in order to derive the nal
gray-box LPV model.
ThisthesispointisdevelopedandpresentedinSetion5.4. Asimulationexample
is used to demonstrate the eetiveness of the proposed solution in Chapter 6.3.3.
The obtained resultshave been published in [157℄.
Thesis 3.1. A new tehnique being able to identify blak and gray-box linear
parameter-varyingmodelsfromloalexperiments,bytraingbaktheidentiation
problem to a strutured
H ∞
-norm optimization problem, has been developed.The loallyestimatedblak-boxLTIandthefrozengray-boxLPVmodelsareplaedintothe
H ∞
-norm-basedglobalostfuntion denedby Eq.(5.12). Then,the nalLPV model is estimated by optimizing one single ost funtion without the appliationof the lassialinterpolation.
ThisthesispointisdevelopedandpresentedinSetion5.5. Asimulationexample
is used to demonstrate the eetiveness of the proposed solution in Chapter 6.3.3.
The obtained resultshave been published in [149, 151, 152, 156, 148, 150℄.
Thesis 3.2. A new
H ∞
-norm-based approah whih determines a set of loal models for linear parameter-varying model identiationhas been developed. Thedeveloped algorithmis ableto determineiteratively areliableset of loaloperating
points. Then,the determined set of working pointsan beappliedduring anyloal
model-based LPV modelidentiationtehnique.
ThisthesispointisdevelopedandpresentedinSetion5.3. Asimulationexample
is used to demonstrate the eetiveness of the proposed solution in Chapter 6.3.2.
The obtained resultshave been published in [149, 148℄.