Black Boxes & White Noise

Transcription

1 Black Boxes & White Noise The Evolution of Automatic Control K. J. Åström Lund Institute of Technology Lund, Sweden

2 1. Introduction 2. Early Ideas 3. A Discipline Emerges 4. The Second Wave 5. Conclusions

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17 Natural Science and Engineering Science Many similarities but also differencies Natural Phenomena Analysis Isolate phenomena Simplicity Basic laws Technical Systems Synthesis Interaction Complexity System principles

18

19 1. Introduction 2. Early Ideas 3. A Discipline Emerges 4. The Second Wave 5. Conclusions

20 Industrial Process Control Windmills Mead 1787 Steam engines Watt, Boulton 1788 Maxwell 1868 Routh 1875 Water turbines Stodola 1893 Hurwitz 1895

21 Accuracy & Stability PID Control u(t ) = k( e( t ) + I T i t o e (s) ds +T d de dt ) Honeywell Taylor Instrument Leeds & Northrup Foxboro

22 Flight Control The Wright Brothers 1903 Sperry 1912 Fully Automatic transatlantic flight 1947 Apollo 1969

23 Minorsky 1922 It is an old adage that a stable ship is difficult to steer.

24 Telecommunication SYSTEM DATE CHANNELS PER PAIR LOSS IN DB (3000 MI) REPEATERS (3000 MI) 1 st Transcontinental nd Transcontinental Open Wire Carrier Cable Carrier First Coaxial

25 The Feedback Amplifier Black's patent 1928 Granted 1937 Singing = Instability Nyquist 1932 Bode 1945 Network Analysis and Feedback Amplifier Design V 1 R 1 R 2 A V 2 = R 2 1 V 1 R A 1 + R 2 R 1 V 2

26 1. Introduction 2. Early Ideas 3. A Discipline Emerges 4. The Second Wave 5. Conclusions

27 A Discipline Emerges Industrial Process Control Telecommunications Flight Control Mathematics Principles Theory Design Methodology Applications

28 War Pressures National Defense Research Committee MIT Radiation Laboratory MIT Servomechanism Laboratory MIT Instrumentation Laboratory MIT Lincoln Laboratory

29 The Black Box Concept Input Output Abstraction Information hiding

30 The Black Box View of Dynamical System Input Output Linearity Superposition Sinusoids One pair sufficies

31 The Notion of Transfer Function u y 2 ϕ a G (s) = 0 e st g (t )dt = L (g) { } { } G (s) = L y L u

32 System Principles Feedforward Feedback FF Process Σ FB Process Combination 1 FF Process Σ FB Σ Process 1

33 Two Paradigms Feedback Open loop Acts only on deviations Market driven Unmeasurable disturbance Less accurate model Feedforward Closed loop Act before deviations occur Planning Measurable disturbance Accurate model

34 Servomechanism Theory Foundations Complex variables Laplace transforms System Concepts Feedback Feedforward Design methodology Frequency response Graphical methods Analog simulation Implementation

35 Consequences Education Application Industrialization Organisation Journals Conferences

36 1. Introduction 2. Early Ideas 3. A Discipline Emerges 4. The Second Wave 5. Conclusions

37 The Second Wave Feedback from applications Challenging problems New technology New ideas

38 Key Elements Reexamination of fundamentals Vital interaction with other disciplines Theory to match new technology

39 Two views of Dynamical Systems External Description Electrical engineering Input/Output Black Box Internal descriptions Mechanical Engineering The notion of state dx dt = f (x,u) y = g (x,u)

40 Modern Control Theory Optimal control Computer control Stochastic control Robust control CACE System identification Adaptive control Intelligent control

41 Optimal Control Euler Lagrange Pontryagin Hamilton Jacobi Bellman SII STAGE Five J-2 Engines 2, 500, 000N Thrust ATTITUDE AND THRUST CONTROL Swivel Outer Four Engines GUIDANCE SATURN Optimum to Desired End Conditions NAVIGATION Inertial Updating of Position and Velocity

42 Modeling Disturbances Power spectra White noise Innovations

43 Stochastic Control Theory Filtering and prediction Merger of calculus of variations and theory of random processes Decision making under uncertainty Industrial process control Probability density Test limit Set point for regulator with low variance Set point for regulator with high variance Process output

44 System Identification Thick stock dilution valve Thick stock flow valve White water Beta ray gauge Moisture meter Simplified diagram of a kraft paper machine MODEL OF PROCESS DYNAMICS AND DISTURBANCES

45 Control of Basis Weight Wet basis weight Set point of stuff gate level

46 Adaptive Control Design Estimator Regulator parameters r Regulator u Process y

47 Dual Control Control actions should be both directing and investigating Consequences for decision making decisions under uncertainty

48 Two Principles Certainty Equivalence (H. Simon 1956) Make the best estimate act as if it was true. Dual Control Control should be investigating as well as directing.

49 Computer Aided Control Engineering How to disseminate complicated technology? Conceptual simplicity computational sophistication Combine human intuition with computational power Nice way to package theory

50 Computer Control

51 Control Design & Process Design

52 The Internal Model Principle Controller y y^ C^ Process v e Σ K B^ Σ x^ L A^ u B Σ x C Σ A

53 Applications Energy generation Energy transmission Process control Discrete manufacturing Instrumentation Telecommunication Transportation Heating, ventilation, aircondition Entertainment Physics Biology Economics

54 Mission Critical Flight Control Space flight Automotive CD player Camcorder manufacturing DC-motor photo diodes radial arm Optical Pick-up Unit

55 The Mercedes A-classA Automatic control gives extra freedom to the designer ESP Unstable behavior improved by Electronic Stabilization Program (ESP)

56 Control and Economics Committee on Policy Optimisation HSO 1978 To consider the present state of development of optimal control techniques as applied to macroeconomic policy. To make recommendations concerning the feasibility and value of applying thes techniques within Her Majesty s Treasury.

57 1. Introduction 2. Early Ideas 3. A Discipline Emerges 4. The Second Wave 5. Conclusions

58 Some Challenges The gap between theory and practice Intelligent systems Man machine interfaces Technology / society interfaces Academic positioning

59 Real World Energy Transportation Communication Manufactoring Instrumentation Entertainment Biology Economics Mathematical Modeling Automatic Control Analysis Simulation Synthesis Implementation Commission Operation

60 Conclusions A glimpse of an emerging field Automatic control is pervasive Some system principles Many challenges ahead