Muutke küpsiste eelistusi

E-raamat: Practical Grey-box Process Identification: Theory and Applications

Teised raamatud teemal:
  • Formaat - PDF+DRM
  • Hind: 159,93 €*
  • * hind on lõplik, st. muud allahindlused enam ei rakendu
  • Lisa ostukorvi
  • Lisa soovinimekirja
  • See e-raamat on mõeldud ainult isiklikuks kasutamiseks. E-raamatuid ei saa tagastada.
Practical Grey-box Process Identification: Theory and ApplicationsSuurem pilt
Teised raamatud teemal:

DRM piirangud

  • Kopeerimine (copy/paste):

    ei ole lubatud

  • Printimine:

    ei ole lubatud

  • Kasutamine:

    Digitaalõiguste kaitse (DRM)
    Kirjastus on väljastanud selle e-raamatu krüpteeritud kujul, mis tähendab, et selle lugemiseks peate installeerima spetsiaalse tarkvara. Samuti peate looma endale  Adobe ID Rohkem infot siin. E-raamatut saab lugeda 1 kasutaja ning alla laadida kuni 6'de seadmesse (kõik autoriseeritud sama Adobe ID-ga).

    Vajalik tarkvara
    Mobiilsetes seadmetes (telefon või tahvelarvuti) lugemiseks peate installeerima selle tasuta rakenduse: PocketBook Reader (iOS / Android)

    PC või Mac seadmes lugemiseks peate installima Adobe Digital Editionsi (Seeon tasuta rakendus spetsiaalselt e-raamatute lugemiseks. Seda ei tohi segamini ajada Adober Reader'iga, mis tõenäoliselt on juba teie arvutisse installeeritud )

    Seda e-raamatut ei saa lugeda Amazon Kindle's. 

In process modelling, knowledge of the process under consideration is typically partial with significant disturbances to the model. Disturbances militate against the desirable trait of model reproducibility. "Grey-box" identification takes advantage of two sources of process information that may be available: any invariant prior knowledge and response data from experiments.



"Practical Grey-box Process Identification" is in three parts: The first part is a short review of the theoretical fundamentals of grey-box identification, focussing particularly on the theory necessary for the software presented in the second part. Part II puts the spotlight on MoCaVa, a MATLAB®-compatible software tool, downloadable from springeronline.com, for facilitating the procedure of effective grey-box identification. Part III demonstrates the application of MoCaVa using two case studies drawn from the paper and steel industries. More advanced theory is laid out in an appendix and the MoCaVa source code enables readers to expand on its capabilities to their own ends.
Part I Theory of Grey—box Process Identification
1 Prospects and Problems
3(20)
1.1 Introduction
3(1)
1.2 White, Black, and Grey Boxes
4(9)
1.2.1 White—box Identification
5(1)
1.2.2 Black—box Identification
6(4)
1.2.3 Grey—box Identification
10(3)
1.3 Basic Questions
13(4)
1.3.1 Calibration
14(1)
1.3.2 How to Specify a Model Set
15(2)
1.4 ...and a Way to Get Answers
17(1)
1.5 Tools for Grey—box Identification
18(5)
1.5.1 Available Tools
18(3)
1.5.2 Tools that Need to Be Developed
21(2)
2 The MoCaVa Solution
23(54)
2.1 The Model Set
23(8)
2.1.1 Time Variables and Sampling
24(1)
2.1.2 Process, Environment, and Data Interfaces
25(2)
2.1.3 Multi—component Models
27(2)
2.1.4 Expanding a Model Class
29(2)
2.2 The Modelling Shell
31(10)
2.2.1 Argument Relations and Attributes
34(3)
2.2.2 Graphic Representations
37(4)
2.3 Prior Knowledge
41(10)
2.3.1 Hypotheses
42(1)
2.3.2 Credibility Ranking
43(1)
2.3.3 Model Classes with Inherent Conservation Law
43(1)
2.3.4 Modelling 'Actuators'
44(2)
2.3.5 Modelling 'Input Noise'
46(3)
2.3.6 Standard I/O Interface Models
49(2)
2.4 Fitting and Falsification
51(6)
2.4.1 The Loss Function
52(2)
2.4.2 Nesting and Fair Tests
54(1)
2.4.3 Evaluating Loss and its Derivatives
55(1)
2.4.4 Predictor
56(1)
2.4.5 Equivalent Discrete—time Model
56(1)
2.5 Performance Optimization
57(5)
2.5.1 Controlling the Updating of Sensitivity Matrices
58(1)
2.5.2 Exploiting the Sparsity of Sensitivity Matrices
59(1)
2.5.3 Using Performance Optimization
60(2)
2.6 Search Routine
62(3)
2.7 Applicability
65(12)
2.7.1 Applications
65(2)
2.7.2 A Method for Grey—box Model Design
67(1)
2.7.3 What is Expected from the User?
68(1)
2.7.4 Limitations of MoCaVa
69(1)
2.7.5 Diagnostic Tools
69(2)
2.7.6 What Can Go Wrong?
71(6)
Part II Tutorial on MoCaVa
3 Preparations
77(6)
3.1 Getting Started
77(1)
3.1.1 System Requirements
77(1)
3.1.2 Downloading
77(1)
3.1.3 Installation
77(1)
3.1.4 Starting MoCaVa
78(1)
3.1.5 The HTML User's Manual
78(1)
3.2 The 'Raw' Data File
78(1)
3.3 Making a Data File for MoCaVa
78(5)
4 Calibration
83(64)
4.1 Creating a New Project
83(2)
4.2 The User's Guide and the Pilot Window
85(1)
4.3 Specifying the Data Sample
86(2)
4.3.1 The Time Range Window
86(2)
4.4 Creating a Model Component
88(13)
4.4.1 Handling the Component Library
89(1)
4.4.2 Entering Component Statements
90(2)
4.4.3 Classifying Arguments
92(3)
4.4.4 Specifying I/O Interfaces
95(3)
4.4.5 Specifying Argument Attributes
98(2)
4.4.6 Specifying Implicit Attributes
100(1)
4.4.7 Assigning Data
100(1)
4.5 Specifying Model Class
101(2)
4.6 Simulating
103(3)
4.6.1 Setting the Origin of the Free Parameter Space
103(1)
4.6.2 Selecting Variables to be Plotted
104(1)
4.6.3 Appraising Model Class
105(1)
4.7 Handling Data Input
106(1)
4.8 Fitting a Tentative Model Structure
107(6)
4.8.1 Search Parameters
108(3)
4.8.2 Appraising the Search Result
111(2)
4.9 Testing a Tentative Model Structure
113(8)
4.9.1 Appraising a Tentative Model
116(2)
4.9.2 Nesting
118(1)
4.9.3 Interpreting the Test Results
119(2)
4.10 Refining a Tentative Model Structure
121(1)
4.11 Multiple Alternative Structures
122(2)
4.12 Augmenting a Disturbance Model
124(8)
4.13 Checking the Final Model
132(2)
4.14 Terminals and 'Stubs'
134(1)
4.15 Copying Components
135(3)
4.16 Effects of Incorrect Disturbance Structure
138(2)
4.17 Exporting/Importing Parameters
140(1)
4.18 Suspending and Exiting
141(2)
4.18.1 The Score Table
142(1)
4.19 Resuming a Suspended Session
143(1)
4.20 Checking Integration Accuracy
143(4)
5 Some Modelling Support
147(38)
5.1 Modelling Feedback
147(7)
5.1.1 The Model Class
148(5)
5.1.2 User's Functions and Library
153(1)
5.2 Rescaling
154(5)
5.3 Importing External Models
159(26)
5.3.1 Using Dymola™ as Modelling Tool for MoCaVa
160(6)
5.3.2 Detecting Over—parametrization
166(4)
5.3.3 Assigning Variable Input to Imported Models
170(3)
5.3.4 Selective Connection of Arguments to Dymola™ Models
173(12)
Part III Case Studies
6 Case 1: Rinsing of the Steel Strip in a Rolling Mill
185(50)
6.1 Background
185(1)
6.2 Step 1: A Phenomenological Description
185(4)
6.2.1 The Process Proper
185(3)
6.2.2 The Measurement Gauges
188(1)
6.2.3 The Input
189(1)
6.3 Step 2: Variables and Causality
189(5)
6.3.1 The variables
189(1)
6.3.2 Cause and effect
190(1)
6.3.3 Data Preparation
191(1)
6.3.4 Relations to Measured Variables
192(2)
6.4 Step 3: Modelling
194(9)
6.4.1 Basic Mass Balances
194(7)
6.4.2 Strip Input
201(2)
6.5 Step 4: Calibration
203(3)
6.6 Refining the Model Class
206(7)
6.6.1 The Squeezer Rolls
206(5)
6.6.2 The Entry Rolls
211(2)
6.7 Continuing Calibration
213(2)
6.8 Refining the Model Class Again
215(2)
6.8.1 Ventilation
215(2)
6.9 More Hypothetical Improvements
217(5)
6.9.1 Effective Mixing Volumes
217(2)
6.9.2 Avoiding the pitfall of 'Data Description'
219(3)
6.10 Modelling Disturbances
222(3)
6.10.1 Pickling
222(1)
6.10.2 State Noise
223(2)
6.11 Determining the Simplest Environment Model
225(8)
6.11.1 Variable Input Acid Concentration
225(1)
6.11.2 Unexplained Variation in Residual Acid Concentration
225(4)
6.11.3 Checking for Possible Over—fitting
229(4)
6.11.4 Appraising Roller Conditions
233(1)
6.12 Conclusions from the Calibration Session
233(2)
7 Case 2: Quality Prediction in a Cardboard Making Process
235(78)
7.1 Background
235(1)
7.2 Step 1: A Phenomenological Description
235(2)
7.3 Data Preparation
237(7)
7.4 Step 2: Variables and Causality
244(4)
7.4.1 Relations to Measured Variables
247(1)
7.5 Step 3: Modelling
248(23)
7.5.1 The Bending Stiffness
248(5)
7.5.2 The Paper Machine
253(7)
7.5.3 The Pulp Feed
260(2)
7.5.4 Control Input
262(3)
7.5.5 The Pulp Mixing
265(2)
7.5.6 Pulp Input
267(2)
7.5.7 The Pulp Constituents
269(2)
7.6 Step 4: Calibration
271(8)
7.7 Expanding the Tentative Model Class
279(11)
7.7.1 The Pulp Refining
279(5)
7.7.2 The Mixing—tank Dynamics
284(3)
7.7.3 The Machine Chests
287(2)
7.7.4 Filtering the "Kappa" Input
289(1)
7.8 Checking for Over—fitting: The SBE Rule
290(3)
7.9 Ending a Calibration Session
293(2)
7.9.1 'Black—box' vs 'White—box' Extensions
293(1)
7.9.2 Determination vs Randomness
294(1)
7.10 Modelling Disturbances
295(1)
7.11 Calibrating Models with Stochastic Input
296(10)
7.11.1 Determination vs Randomness Revisited
299(5)
7.11.2 A Local Minimum
304(2)
7.12 Conclusions from the Calibration Session
306(7)
Appendices
A Mathematics and Algorithms
313(28)
A.1 The Model Classes
313(3)
A.2 The Loss Derivatives
316(1)
A.3 The ODE Solver
317(4)
A.3.1 The Reference Trajectory
317(1)
A.3.2 The State Deviation
318(1)
A.3.3 The Equivalent Discrete—time Sensitivity Matrices
318(3)
A.4 The Predictor
321(1)
A.4.1 The Equivalent Discrete—time Model
322(1)
A.5 Mixed Algebraic and Differential Equations
322(4)
A.6 Performance Optimization
326(4)
A.6.1 The SensitivityUpdateControl Function
327(3)
A.6.2 Memoization
330(1)
A.7 The Search Routine
330(1)
A.8 Library Routines
331(6)
A.8.1 Output Conversion
331(1)
A.8.2 Input Interpolators
331(3)
A.8.3 Input Filters
334(1)
A.8.3 Disturbance Models
335(2)
A.9 The Advanced Specification Window
337(4)
A.9.1 Optimization for Speed
337(1)
A.9.2 User's Checkpoints
338(1)
A.9.3 Internal Integration Interval
338(1)
A.9.4 Debugging
339(2)
Glossary 341(4)
References 345(4)
Index 349


Professor (emeritus) Torsten Bohlin has been employed in the following capacities:



1963 - 1971 at the IBM Nordic Laboratories as Research Engeneer working with computerized industrial process ontrol. 1971 appointed (by the king) Professor of the chair of Automatic Control at Linkping Technical Institute. 1972 - 1996 Professor in Automatic Control at the Royal Institute of Tecknology (KTH) in Stockholm. 1972 - 1988 Head of the Department of Automatic Control, Member of the board of the school of Technical Physics, and Member of the faculty of KTH. Member of the Swedish IFAC comittee, TFF (national), and IEEE Reviewer 66 times
Lisainfo e-raamatute kohta Lisainfo e-raamatute kohta
Püsilink: https://www.kriso.ee/db/97818462840382e.html
Märksõnad: