FIRST THOUGHT- CONTROLLED PROSTHETIC DEVICE 2019-04-11 Chalmers - - PowerPoint PPT Presentation
DEEP LEARNING IN REFINING PROCESSES A PRE-STUDY OF INTERNAL AND EXTERNAL VARIABLES IMPACT ON PULP PROPERTIES FREDRIK BENGTSSON*, ANDERS KARLSTRM* AND JAN HILL**, *CHALMERS UNIVERSITY OF TECHNOLOGY , GTEBORG, SWEDEN, **QUAL TECHAB,
A PRE-STUDY OF INTERNAL AND EXTERNAL VARIABLES IMPACT ON PULP PROPERTIES
FREDRIK BENGTSSON*, ANDERS KARLSTRÖM* AND JAN HILL**, *CHALMERS UNIVERSITY OF TECHNOLOGY , GÖTEBORG, SWEDEN, **QUAL TECHAB, TYRINGE, SWEDEN
2019-04-11 Chalmers University of Technology 2
2019-04-11 Chalmers University of Technology 3
Example used in this presentation: CD- refiners.
2019-04-11 Chalmers University of Technology 4
Temp.
rmax r5 r4 T5 T4 Tmax
Radial position
2 2
max 11 22 32 33
: Relates to the production screw, not Plate gap. : Corresponds to dilution wa
FZ FZ FZ FZ FZ CD CD CD
screw est H O est H O screw
T g P Y C GU g D C g g D P D
max
ter to each ref. zone. : Consistency in the periphery of each ref. zone. : Temperture maximum in the flat zone. Dependent on plate pattern C T
2019-04-11 Chalmers University of Technology 5
Vision: By controlling the (C)TMP-refiners, specific energy can be reduced significantly at the same time as pulp property variations are maintained at an acceptable level. To reach the vision: Introduce Tmax - control in the flat zone(FZ) and Consistency control in both FZ and the conical zone (CD). Keystones: 1. Manipulate the work load between the zones in an unortodox way to increase production and stabilize pulp quality. 2. Select proper operating points defined by requested pulp quality.
2019-04-11 Chalmers University of Technology 6
2019-04-11 Chalmers University of Technology 7
Flat zone Energy and Material balances 1,…,n1 Inlet mixing zone Outlet mixing zone Inlet mixing point Mixing point between FZ & CD CD zone Energy and Material balances 1,…,n2
Mass flow (Water) Mass flow (Steam) Mass flow (Chips and/or fibers) Work related to the motor load distribution
2019-04-11 Chalmers University of Technology 8
A new set of process variables, internal states , available:
→ TCtrl – Temperature control CCtrl – Consistency control → Increased amount of process data to handle. → Possibility to systematically analyze pulp properties, process stability as well as different estimation tools.
2019-04-11 Chalmers University of Technology 9
Between zones
2019-04-11 Chalmers University of Technology 10
Future challenge
TCtrl – in Automatic mode ; Major features
2019-04-11 Chalmers University of Technology 11
Two data sets: 0 - 600 hr
2019-04-11 Chalmers University of Technology 12
Two data sets 0-600 hr Future Challenge – minimize variations even more by using overall MPC for specific pulp properties
2019-04-11 Chalmers University of Technology 13
time
Oct Nov Dec Jan Feb Mar
Pulp property
Comment: The zero-elements in the vector xm(t) are possible to predict using piece-wise linear functions where residence time and/or Sp.Energy and consistency are used as independent variables. Pulp samples from lab. tests Pulp samples in time domain
t xm
s
x
s
t
t xm ˆ
2019-04-11 Chalmers University of Technology 14
Karlström, A. and Hill J. ”CTMP Process Optimization Part III: On the Modeling of Scott-Bond, Z-strength and Tensile index ”. Submitted for publication in NPPRJ 2017.
Pulp & handsheet models
(non-linear model based on first principals)
Consistency in FZ Consistency in CD Residence time in CD Residence time in FZ External variables Machine parameters Refiner segment parameters Measured internal Variables (Temp) Quality vector
Non-linear Linear
MPC based on pulp & handsheet estimations
Specific energy
2019-04-11 Chalmers University of Technology 15
2 1 ) ( ˆ
2 2 1 1
,q , , m b x θ x θ x θ f
m mk mk m m m m m m
x
{θ ,…, θ
Properties Intercept Cons.(FZ) Cons.(CD) R.time(FZ) R.time(CD) Spec.E CSF Freeness 767.4522 6.5706
1430.6061 Sheet density 311.3555
7.6204 100.8509 55.8056 Tensile strength
0.0831
62.7926 Tensile index 40.2999
1.0245
200.9831 Elongation to rupture 1.6649
0.0143
6.0259 Tensile energy absorption 80.0754
2.2176
314.1703 Tensile energy absorption index 0.4283
0.0046 0.7856
Tensile stiffness
1.3169 1757.0678
Tensile stiffness index 5.2630
0.5462
Tear strength 6435.5391
0.2833 811.0374 -25020.3831 Tear index 53.1587
0.0703
Short-span compressive test index 24.1904
0.3068
10.6965 ISO brightness 88.6031
0.0572
331.3528 Scott-Bond 155.0146
3.5924 120.7834
Z-strength 144.0442
3.1333
890.5317 Shives(>=0.3mm)
94.7427 -1459.5470 10070.7017 Long fibers 8.0583
47.3025 Fines 34.3848 0.2466
70.0445
Model parameters - Case I
Karlström, A. and Hill J. ”CTMP Process Optimization Part III: On the Modeling of Scott-Bond, Z-strength and Tensile index ”. Submitted for publication in NPPRJ 2017.
Dynamic modeling using a System identification approach Multivariate modeling approach
2019-04-11 Chalmers University of Technology 16
Machine learning is learning from data, system identification can be described as learning dynamic models from data.
Predictors used: Specific Energy, Consistency in FZ and CD, inlet consistency
2019-04-11 Chalmers University of Technology 17
1) Build models based on DC-gains and refiner dynamics! Introduce learning algorithms. 2) Use delay/filter to foresee future changes after latency!
The models seem to be useful!
In the same way as the multivariate regr. models
2019-04-11 Chalmers University of Technology 18
Control concepts provides a possibility to deside where in the operating window to run the refiners. This gives an opportunity to:
enough pulp (and handsheet) property control;
CCtrl and TCtrl – in Automatic mode
2019-04-11 Chalmers University of Technology 19
Control concepts provides a possibility to deside where in the operating window to run the refiners. This gives an opportunity to:
enough pulp (and handsheet) property control;
2019-04-11 Chalmers University of Technology 20
Next step! Introduce machine learning algorithms