. pesos ~ Computer Control of Machines and Processes John G. Bollinger Neil A. Duffie University of Wisconsin K M RR w ADDISON-WESLEY PUBLISHING COMPANY Reading, Massachusetts e Menlo Park, California e New York Don Mills, Ontario # Wokingham, England « Amsterdam ¢ Bonn Sydney ¢ Singapore # Tokyo # Madrid # San JuanThis book is in the Addison-Wesley Series in Electrical and Computer Engineering: Control Engineering The programs and applications presented in this book have been included for their instructional value. They have been tested with care, but are not guaranteed for any particular purpose. The publisher does not offer any warranties or representations, nor does it accept any liabilitics with respect: -~ to the programs or applications. Library of Congress Cataloging-in-Publication Data Bollinger, John G. Computer control of machines and processes. Includes index. 1. Automatic control—Data processing. 1. Duffie, Neil A. I. Title. J213.B5952 1988 629.8'95 87-14019 ISBN 0-201-10645-0 Reprinted with corrections February, 1989 Copyright © 1988 by Addison-Wesley Publishing Company, Inc, All rights reserved. No part of this publication may be teproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Printed in the United States of America. Published simultaneously in Canada. 7:89 10 11 12 ma 99989796To our families John and NeilPreface We are living in an analog world surrounded by a universe of digital technology. For many years control engineers integrated analog devices, machines, and processes wherein the variables that were input, the manipulations and control actions computed, and the variables that were output were continuously varying signals in time. Thus, analog devices were controlled by analog controllers. There was also the need to be concemed with discrete events and the logical intelligence for machines and processes, commonly referred to as sequence control. These two different aspects of automatic machine and process control were traditionally handled by two separate bodies of knowledge and reflected vastly different technologies and levels of theoretical treatment. On one hand, feedback control theory formed a significant body of knowledge on which much research has been and continues to be done. It was not uncommon for a control engineer to work on only this aspect of a control problem. On the other hand, sequence control formed another body of knowledge that was originally implemented with little theory (originally in the form of relay or pneumatic logic) and was generally implemented by the practitioner of logic systems. Technological developments in computers and associated input/output hardware have changed much of the way we look at automatic control problems. Digital computers are now used to control machines and processes in feedback control and logic control modes. In fact, a computer can perform both types of control concurrently, and is as adept at logical computations as it is at arithmetic. The design of appropriate software to provide effective feedback control and logic control actions, and the design of computer input/output devices to viiPreface interface machines and processes to computers to meet complete contro! requirements, therefore, are key issues to be considered along with appropriate analysis and design methodologies. The goal of this text is to make the process of creation of effective controller designs as comprehensive, yet as straightforward, as possible. This has been achieved throughout the text by starting with fundamental principles in all of the areas that are important in controlling machines and processes with digital computers, and then using those fundamental principles to quickly develop theories, methods, and techniques that are used by computer control system designers, TIME-DOMAIN APPROACH From inside a digital computer, the world jooks like a stream of discrete numbers and logic values. They are interpreted as data in any way that the computer is programmed to interpret them. Algebraic and logical computations are per- formed on the data, and the computer gives back to the world a similar set of discrete numbers and logic values. Mathematically, the computer is dealing in time series, or sequences of numbers. Using the digital computer stimulates a natural desire to view the dynamic behavior of machines and processes as a computer would view them. A discrete system model responds to this need by providing a defining relationship for the behavior between input and output time series. Because all information is in the time domain, it is logical to begin by working in the time domain. There are conceptual advantages in working in the time domain, and for incorporating discussions of logic design and computer hardware and software, in an introduction to computer control of machines and processes. Some of these advantages include: 1, Only algebra is needed as a mathematical foundation. 2. The entire topic can be approached without any prior knowledge of classical control theory. 3. A large part of what is called control of machines and processes is, in fact, sequential logic control. Logic is designed in the time domain, and the treatment of this subject in an integrated-textbook format provides the student with a very effective background to approach a wide range of computer control problems. 4. A significant portion of the solution of a total, control engineering problem involves the generation of command information such as that needed for machine motion control. Time-series modeling sets the stage for developingPreface ix computer-generated command series to satisfy a wide range of functional- system requirements. 5. Computer hardware and software function in the time domain. A time- domain approach reduces the level of abstraction between theory and implementation. There are a number of well known frequency-domain techniques for feedback control system analysis and design that the more advanced student should be aware of. These are discussed in the closing chapters of the text. We have found that it is not necessary to discuss these topics at the beginning of introductory courses on computer control systems for machines and processes, but have included them in courses where the students have already had a first course in feedback control theory. CHAPTER CONTENTS Each chapter in the text starts with basic principles and develops the analysis, design, and implementation methods required to solve problems in important areas of computer control. 1. Introduction to Computer Control Chapter 1 briefly describes the history of computer control and the explosion in the application of electronic technology since the 1960s. 2. Elements of Discrete Modeling Chapter 2 begins with a discussion of basic process types, and develops the idea of representing process dynamics using difference equations. The use of difference equations to represent the process leads to the topics of discrete controllers and the difference equations that represent them. Methods of algebraically combining and solving the difference equations that represent the components of closed-loop systems are introduced, and this leads to the concept of the system transfer function—the ability to find system output as a function of system input, and the ability to analyze and predict the stability of a system. 3. System Response Two fundamental concepts are introduced in Chapter 3 —the representation of a time-varying signal sampled at specific intervals of time as a sample sequence or time series, and the use of a time-shift operator to calculate the response of a system to an input sample sequence, The method of partial fraction expansion is introduced, and a final value theorem is developed that allows steady-state values of dynamic system variables to be determined.Preface 4. Discrete Controller Design Chapter 4 focuses on the design of control algorithms for desired closed-loop response to specified command and distur- bance inputs. Critical issues in the selection of a sample period are also discussed. Finally, the design of feedforward, cascade, and noninteracting control algorithms 1s described 5. Control Computers Chapter 5 deals with control computer hardware and software. Topics include binary logic, basic computer hardware, registers, the concepts of instructions and data, computer input/output (I/O), and the program- driven and interrupt-driven options for control algorithm implementation at the assembly language level. A set of high-level language procedures are defined for analog I/O, logic 1/O, and interrupt handling. These are then used to further illustrate important controller implementation options. 6. Computer Interfacing Chapter 6 describes how interfaces for devices such as push-buttons, limit switches, lamps, and numeric displays can be constructed. Interfaces for logic-level I/O are described, as are digital-to-analog conversion and analog-to-digital conversion. Computer bus architectures are discussed along with issues such as address decoding, device selection, and interrupt interfacing. 7. Sensors for Computer Control Chapter 7 describes a spectrum of sensors that are often found in computer control systems for machines and processes 8. Command Generation for Machine and Process Control Command generation often represents a major portion of the “intelligence” of a system, and several techniques for command generation for machine and process control are described in Chapter 8. 9. Sequential Control Using Programmable Logic Controllers _Implemen- tation of logic control and solution of logic equations on a control computer are discussed in Chapter 9. The implications of serial solution of logic on a digital computer are then discussed, together with the relationship between computer solution of logic and relay-logic systems defined using ladder diagrams. A number of design methods for logic control are described, including the use of flowcharts, switching tables, and state diagrams. 10. Process Modeling Chapter 10 reviews a number of approaches to process modeling. These include physical modeling, wherein mathematical definitions of the dynamic properties of all the components that make up a process are combined to create a complete process model. Two alternative approaches are also described: the step-response and stochastic methods of process model identification. It is shown that the transient and steady-statePreface xi characteristics of the response of the process to a step input can be used to identify and develop a straightforward discrete model for the process. A stochastic method is then presented in which linear least-squares estimation is used to find the values of coefficients in a discrete process model that “best fit” observed process inputs and outputs. 11. Analysis Using Transform Methods The Laplace and z transformations are defined in Chapter 11 and used in the analysis of continuous and discrete control systems and their components. The concepts of ideal samplers, hold elements, and representation of a sample sequence by a train of impulses are introduced. The frequency response of a closed-loop system and its elements is discussed, and it is shown that the delay implicit in a hold element (such as a digital-to-analog converter) has an adverse effect on the stability of the system under computer control. 12. Design Using Transform Methods A number of methods for designing discrete controllers for machines and processes are described in Chapter 12. The design techniques are developed for both continuous and discrete systems, illustrating their similarities and differences. Root-locus and frequency response methods are discussed first. Pole-zero cancellation and integral control compensa- tion techniques are then described, as are techniques for anticipating and compensating for the detrimental effects of hold elements in a system. PID controller approximation is described, as are the bilinear transformation, hold equivalence, and pole-zero mapping techniques for continuous controller approx- imation. Direct design of controllers to obtain specified closed-loop transfer functions is also reviewed. Appendix A Appendix A gives a brief review of state-variable methods for control system analysis and design. Both continuous and discrete systems are discussed. Appendix B_ Appendix B provides a review of binary arithmetic. TECHNOLOGY-BASED INFORMATION Throughout the text, technology-based information has been avoided wherever possible. The approach taken has been to describe computer architectures, interfaces, and programming languages at a generic level. The simplified computer architecture that is used in Chapters 5 and 6 allows fundamental concepts to be illustrated, and allows system-level issues to be discussed, withoutxii Preface describing complex computer architectures in commercial use. There are a great number of books, many of them inexpensive paperbacks, that can be used to supplement this text if additional information on a computer system's specific hardware and software is required. EXAMPLES AND PROBLEMS A variety of processes are used in examples and problems in the text to illustrate the analysis, design, and implementation techniques that are introduced. Many of the problems reiterate basic concepts illustrated in examples. Others challenge the reader to extend basic concepts to more complex situations. Still others require the integration of concepts presented in-one or more chapters; leading’ - toward broader concepts at the system level. Many of the examples and problems are manufacturing related, and most refer to practical application of concepts that are introduced PROGRAMS It is important to view the computer as a tool in the control system development process as well as the hardware basis for discrete controller implementation. Programs, therefore, are included at many locations in the text to illustrate the use of the computer for analysis, design, simulation, and modeling as well as for control. The programs given are usually greatly simplified to improve the illustration of important concepts. We have chosen Pascal as the language in which to present these software concepts. The reader will find little fundamental difference between the simple programs presented and programs written in other structured languages. It is expected that the reader will use the programs as prototypes for development of more elaborate computer software. An under- standing of the basic techniques illustrated will promote use and familiarity with future and existing computer-aided design and analysis software for computer control systems. COURSES BASED ON THE TEXT~ This text, and the courses that it supports at the University of Wisconsin—Mad- ison, have been in a continuous state of development, evolution, and refinementPreface xiii since the mid-1960s. The students who have enrolled in our courses have included nearly equal numbers of graduate and undergraduate students, nearly equal numbers of electrical and mechanical engineering students, and students from all other engineering disciplines including aerospace, agricultural, chemical, civil, industrial, and nuclear. In addition, we have presented the material in the text in our industry outreach programs and in many short courses given for industry both in the United States and abroad, The spectrum of topics that are important in computer control is exceedingly broad, and instructors using this text in formal courses will want to tailor the material they present to the background of their students and the time available. A previous course in control theory is not required in the use of this text, and Chapters 1 through 9 can be covered in one semester at this introductory level. Instructors at institutions using the quarter system may wish to cover Chapters 1 through 4 together with Chapters 11 and 12 in one quarter at the introductory level. Chapters 5 through 10 can then be covered in a second quarter. Instructors of courses in which students have already had an introductory course in control theory may wish to cover the material in Chapters 11 and 12 before covering Chapters 5 through 10, de-emphasizing Chapters 2 through 4 if desired. Similarly, if students have already had experience with computer hardware and programming at the assembly language level, an instructor may wish to de-emphasize Chapters 5 and 6, and spend more time on Chapters 8 through 12. We suggest that instructors use inexpensive hardware and software documentation to supplement the text if a specific computer system is to be used by the student. Throughout the text, the succession of topics and ideas presented in each area follows a logical progression based on the premise that the student has little prior knowledge in the area. This has proved advantageous in our experience because of the great differences in backgrounds of students who enroll in computer control courses. We have found that even with a very heterogeneous student population with wide disparities in backgrounds and experiences, all of the students can leave courses in which this text is used with the confidence that they have been exposed to most of the major issues in the analysis, design, and implementation of computer control systems for machines and processes. ACKNOWLEDGEMENTS, We are sincerely thankful for the contributions of many people during the years in which this text has been in development. Professors David D. Bedworth, A. H. Haddad, David E. Hardt, Stephen P. Krosner, Robert D. Lorenz, and Kim A. Stelson have reviewed the manuscript, and we have tried to incorporate theirPreface valuable ideas and suggestions wherever possible. Special thanks go to Professors Howard L. Harrison and Thomas J. Higgins for their contributions, criticisms, and encouragement. Many students at the University of Wisconsin—Madison made significant contributions. These include Bruce R. Beadle, Dongik Joo, Richard A. Lund, J. Kelly Lee, Aun-Noew Poo, Mahesh K. Seth, Rafael Rangel-Sostmann, Jose Rodriguez-Ortiz, and the students of ME 547 who made so many helpful suggestions for improvements in the manuscript, Madison, Wisconsin J. G. B. and N. A. D.Contents Chapter 1 INTRODUCTION TO COMPUTER CONTROL 1 1.1. Introduction 1 1.2. Concepts of Computer Control 5 Summary oi Bibliography 10 Chapter 2 ELEMENTS OF DISCRETE MODELING 12 2.1 Introduction 12 2.2 Discrete Process Models 43 Other Process Types 24 2.3. The Discrete Controller 24 Proportional Control 25 Integral Control 26 Derivative Control 27 PID Contro} 27 xvxvi Contents 2.4 Transfer Functions and Block Diagrams 29 Discrete Transfer Functions 29 Block Diagrams 31 The Closed-Loop Transfer Function 31 Table of Transfer Functions 33 2.5 The Solution of Difference Equations 42 First-Order Equations 42 Second-Order Equations 44 Higher-Order Equations 47 Consideration of Initial Conditions 47 Nonhomogeneous Equations—Particular Solution 49 Nonhomogeneous Equations—Complete Solution 51 2.6 Stability Analysis 53 2.7 Summary 57 Bibliography 58 Problems 58 Chapter 3 SYSTEM RESPONSE 69 3.1. Introduction 69 3.2 System Inputs 70 Table of Generating Functions 75 3.3. System Response 77 Partial Fraction Expansion 82 Final-Value Theorem. 84 Response to a Unit Pulse Input 85 3.4 Summary 98 Bibliography 98 Problems 99. Chapter 4 DISCRETE CONTROLLER DESIGN mW 4.1. Introduction 777 4.2 Control Algorithms for Desired Closed-Loop Response 112 Control Algorithms for Desired Output-Input Relationships 114Contents 4.3 44 45 46 47 xvii Control Algorithms for Desired Error-Input Relationships 122 Control Algorithms for Desired Output-Disturbance Relationships 125 Controller Realizability 131 Controller Stability 133 Sample-Period Selection 137 Speed of Response to Inputs 138 Rejection of Disturbances 139 Stability 140 Feedforward Control 142 Disturbance Feedforward Control 142 Command Feedforward Control 145 Cascade Control 150 Noninteracting Control of Interacting Systems 155 Summary 762 Bibliography 762 Problems 163 Chapter 5 CONTROL COMPUTERS 173 5.1 Introduction —- 173 5.2 Basic Computer Organization 174 Central Processing Unit (CPU) 175 Memory = 175 Input/Output Devices 176 5.3. Binary Logic 176 AND Operation 176 OR Operation 177 NOT Operation 177 Other Operations 178 R-5 Flip-Flop 179 DFlip-Flop 180 Register or Buffer 187 Adder 187 Ripple-Carry Adder 182xviii 54 55 5.6 57 5.9 Chapter 6 Contents Instructions and Data 183 Load and Store Instructions 185 Arithmetic and Logical Instructions 136 Branching Instructions 188 Shifting and Rotating Instructions 189 Input/Output Instructions 197 Program-Driven Input/Output 197 Interrupt-Driven Input/Output 193 High-Level Languages for Computer Control 195 DAC (igital-to-Analog Conversion) 197 ADC (Analog-to-Digital Conversion) 197 DLC (Logic Output) 197 LCD (Logic Input) 198 CONNECT (Connect Control Procedure to Interrupt Input) 198 DISCONNECT (Disconnect Control Procedure from Interrupt Input) 198 Closed-Loop Computer Control 199 Programmed Synchronization to an External Clock Signal 200 Synchronization with an External Clock Using Interrupts 201 Control Routine Timing 204 Control Algorithms with One Sample Delay 205 Contro] Algorithms with Precalculation 206 Reduction of Computation Time 207 Summary 272 Bibliography 273 Problems 273 COMPUTER INTERFACING 221 64 6.2 introduction, 2217 Computer Architecture 222 Memory Address Bus (MAB) 222 Memory Data Bus (MDB) 222 READ and WRITE Control Signals 223 Input/Output Address Bus (IOAB) 223Contents Input/Output Data Bus (IODB) 223 IN and OUT Control Signals 224 CPU Interrupt Signal (INT) 224 Programmed-Driven vs. Interrupt-Driven I/O 6.3 Logic-Level Input/Output 231 64 Digital-to-Analog Conversion 234 Range 236 Resolution 236 Accuracy 237 65 Analog-to-Digital Conversion 238 Filtering 240 Resolution 243 Amplitude Uncertainty 244 Sample-and-Hold 245 Multiplexing 246 Conversion Time and Stability 248 6.6 Alpha-Numeric Input/Output 248 Output to Alpha-Numeric Displays 249 Keyboard Input 250 6.7 Numeric I/O 251 Binary-to-Decimal Conversion 252 Decimal-to-Binary Conversion 254 Seven-Segment Display Output 255 Thumbwheel Switch Input 256 68 Interrupt Interfacing 257 69 Computer Control Example 259 6.10 Summary 262 Bibliography 262 Problems 263 Chapter 7 228 SENSORS FOR COMPUTER CONTROL 7.1 Introduction 270 Accuracy 271 Resolution 271 270xx 7.2 73 7A 75 7h 77 78 7.9 Contents Repeatability 277 Range 277 Dynamic Response 271 Position Sensors 272 Potentiometer 272 Linear Variable Differential Transformer 273 Rotary Resolver 275 Phase Modulation 277 Amplitude Modulation 278 Linear Resolvers 280 Optical Angle Encoders 280 Absolute Optical Angle Encoders 287 Incremental Optical Angle Encoders 282 Capacitive Sensors 284 Other Position Sensors 285 Velocity Sensors 286 d.c. Tachometers 286 Linear Velocity Sensors 286 Differentiation of Position Feedback 288 Analog Differentiation 288 Digital Differentiation 289 Acceleration Sensors 290 Force Sensors 297 Temperature Sensors 292 Resistance Thermometer 293 Thermocouples 294 Thermistors 296 Strain Sensors 298 Strain Measurement 298 Torque Measurement 300 Optical Sensors 300 Four-Quadrant Photodiode 300 Linear Photadiode Arrays 302 Two-Dimensional Photodiode Arrays 304 Summary 305 Bibliography 305 Problems 306Contents xxi Chapter 8 COMMAND GENERATION IN MACHINE AND PROCESS CONTROL 314 8.1 Introduction 314 8.2 Command Generation Using Linear Interpolation 316 82 Command Generation Using Cubic Polynomials = 379 Caleulation of Cubis Qysthsivnh Mabisss 220 Simplification of Cubic Spline Calculations 325 8.4 Point-to-Point Motion Generation Using Spline Functions 337 Single Axis, Single Spline 337 Single Axis, Three Splines, Constant Acceleration 333 Single Axis, Three Splines, Decreasing Acceleration 334 Multiple Axes, Single Spline 335 Multiple Axes, Three Splines, Constant Acceleration 336 Multiple Axes, Three Splines, Decreasing Acceleration 338 8.5 Controlled Rate of Change of Acceleration 339 8.6 Feedforward Control Using Cubic Spline Functions 339 87 Open-Loop Position Control Using Stepping Motors 343 Constant-Velocity Profile 346 Constant-Acceleration Profile 349 Generation of Straight-Line Motion 353 Generation of Circular Motion 355 88 Summary 358 Bibliography 359 Problems 359 Chapter 9 SEQUENTIAL CONTROL USING PROGRAMMABLE LOGIC CONTROLLERS 369 9.1 Introduction 369 9.2 Boolean Algebra 377 Theorems of Boolean Algebra 37/ Boolean Equations for Switching Logic 373 9.3. Digital Solution of Boolean Equations 375 Boolean Assignment Statements 375 Scan Time 377xxii 94 95 9.6 ere Chapter 10 Contents Ladder Diagrams 387 Components of Ladder Diagrams 387 Control Relays 387 Solenoids 381 Lamp 383 Push-Button Switches 383 Limit Switches 383 Electrical Circuit Diagrams 383 Logic Functions 384 AND 384 OR = 385 EXCLUSIVE OR 385 NOT 385 Obtaining Boolean Equations from Ladder Diagrams 385 Converting Boolean Equations into Ladder Diagrams 386 Sequential Logic Design 388 Off-Dominant Design of Holding Functions 388 Timing Functions 389 Specification of Control Requirements 390 Switching Tables 392 Flowcharts 393 State Diagrams 394 Programming, Monitoring, and Diagnosis 405 Programming Languages 406 Programming Devices 406 Monitoring and Diagnosis 406 Simulation 407 Summary 408 Bibliography 409 Problems 409 PROCESS MODELING 421 10.1 10.2 Introduction 421 Physical Modeling 422 Steps in Physical Modeling 422Contents xxiii Linearization of Models 423 Simplification of Process Models 427 Verification of the Process Model 428 10.3 Step-Response Modeling 428 Delay 428 Integration 429 Integration with Dead Time 430 First-Order Systems 437 First-Order Systems with Dead Time 432 Second-Order Systems 432 Highly Underdamped Processes 433 Highly Overdamped Processes 435 Critically Damped Processes 435 Limitations 435 10.4 Stochastic Modeling 439 First-Order Process with Dead Time 439 Estimating Higher-Order Process Parameters 443 Residual Errors 445 Autocorrelation 445 Normality 448 Confidence Regions for the Parameters 448 10.5 Summary 452 Bibliography 453 Problems 454 Chapter 11 ANALYSIS USING TRANSFORM METHODS 467 11.1 Introduction 467 11.2. The Laplace Transformation 468 Laplace Transform of Function Multiplied by a Constant 468 Laplace Transform of the Sum of Two Functions 468 - . .--Laplace Transform of a Time-Delayed Function 469 Laplace Transform of Derivatives 470 11.3. Frequency Response 477 Obtaining Frequency Response from Laplace Transforms 473 Approximate Frequency Response 474xxiv 11.4 115 11.6 Chapter 12 Contents The z Transformation 482 Ideal Sampler 482 Laplace Transform of a Unit Impulse 485 Laplace Transform of a Modulated Train of Impulses 486 Definition of the z Transformation 486 Obtaining z Transforms from Laplace Transforms 487 Inverse 2 Transformation 490 Transfer Functions and Block Diagrams 490 Hold Elements 491 Stability 496 Frequency Response of the Sample-and-Hold 498 Summary 503 Bibliography .. 504 Problems 504 DESIGN USING TRANSFORM METHODS 508 12.4 12.2 12.4 12.5 126 introduction 508 Controller Design Using Root Locus 509 Root Locations in the s Plane 509 Root Locations in the z Plane 511 Setting Controller Gain 512 Controller Design Using Frequency Response 576 Rulel 518 Rule2 518 Rule3 518 Gain Margin 519 Phase Margin 519 Rule 4 519 Compensation for Process Dynamics 525 Addition of Integral Control 525 Pole-Zero Cancellation 530 Approximation of Zero-Order Hold Dynamics 535 Discrete Approximation of Continuous Controllers 537° PID Controller Approximation 537 The Bilinear Transformation 539Contents xxv Hold Equivalence 541 Pole-Zero Mapping 543 Results of Controller Approximation 544 127 The Direct Design Method 546 128 Summary 547 Bibliography 548 Problems 549 Appendix A ANALYSIS AND DESIGN USING STATE VARIABLES 551 CG A.1_ Introduction 557 A2_ State Equations for Continuous Systems 551 Process Transfer Function Matrix 554 Continuous Closed-Loop Control 555 A3_ State Equations for Discrete Systems 559 Discrete Process Transfer Function Matrix 563 Discrete Closed-Loop Control 564 Closed-Loop Characteristic Equation 565 State Estimator Feedback 569 A4 Summary 576 Bibliography 576 Problems 577 Appendix B BINARY NUMBERS AND ARITHMETIC 379 Binary-to-Decimal Conversion 580 Decimal-to-Binary Conversion 587 Octal and Hexadecimal Notation 582 Addition and Subtraction 585 2'S-Complement Arithmetic 586 Multiple-Precision Addition and Subtraction 590 Multiplication of Binary Numbers 592 Division of Binary Numbers 593 Problems 594xxvi Contents Appendix C 7-BIT ASCIL CHARACTER CODES 598 Appendix D STUDENT'S t DISTRIBUTION 599 Appendix E TABLE OF z TRANSFORMS 600 Index 603 |14 Chapter 1 Introduction to Computer Control INTRODUCTION Everywhere we look today we see evidence of the impact of computers and other electronic devices on our lives. In our homes, our automobiles, and our offices and factories, these devices are changing the way we work, entertain ourselves, and communicate with others. The impact of this rapidly developing electronics and computer technology is also accelerating the trend toward automation and the use of computer control in the manufacturing industry. A landmark of early manufacturing automation was the A. O. Smith Corporation’s fully automated automobile frame production facility built in 1921. Over 550 operations in the facility were synchronized via line shafts. There was, of course, no computer. Electronic sensor and actuator technologies had not yet progressed beyond their infancy as components of automatic systems, and control functions were performed by mechanical mechan- isms and pneumatics. Although electronic measurement, control, and actuation technology contin- ued to develop and improve during succeeding decades, it had reached only a marginal level of “credibility” in the manufacturing industry by 1950. Most electronic devices at that time were basically analog in nature, functioning with electronic:inputs or outputs proportional to desired physical quantities. In the mid-1950s digital devices began to appear, including digital volt- meters and N/C (numerically controlled) machine tools. The latter represented a significant step forward, because they not only enabled machining operations to be preprogrammed rather than performed manually, but also allowed for rapid 1Introduction to Computer Control reprogramming when it was necessitated by changes in parts to be manufactured. This flexibility (previously obtainable only in manual operations) demonstrated the future potential of automation in the manufacturing industry. The electronic logic in these systems, although “hard-wired” rather than computer-based, proved that “intelligence” could be designed into production processes, resulting in greater sophistication, higher rates, and greater precision than could be obtained manually. Since 1960 there has been an explosion in the application of electronic technology. Electronic transducers are found in virtually all manufacturing industries, including the metalworking, plastics processing, chemical processing, paper, and textiles industries. The advent of the digital computer has provided the core about which complex data-acquisition, monitoring, communication, and control systems can be constructed. One key element that emerged in the eighties was the development of “smart sensors’—that is, sensors with microcomputer-based calibration, computation, and decision-making power built into the unit. This trend has greatly enhanced the use of many physical measurement techniques, particularly those that are inherently nonlinear. The service industry has also become automated. The banking industry was one of the first to undergo computer automation of accounting, bookkeeping, and data-entry functions. These innovations were followed by the trend toward electronic money. In another domain, computers access a worldwide commercial air travel data-base to assist airlines and travel agents in making flight reservations. New aircraft have more than 100 on-board computers and can be flown and landed automatically. Where will this lead us? Clearly, automation of the service industry will continue, spurred by the desire to provide improved services at lower cost. In the factory of the future, computers will run machines using complex algorithms that achieve a level of machine intelligence necessary to operate, monitor, and diagnose the condition of both the manufacturing process and the product. This will require sophisticated electronic devices such as tool-failure sensors, operating parameter sensors (torque, speed, flow,.and.so on).and such inspection devices.as tactile and optical sensors. Much of the technology for the factory of the future already exists, and electronic devices play leading roles in it. Digital computers, sensors, and actuators are distributed throughout complex systems. Sensors of all types provide the computer with the information it needs to perform monitoring and control tasks. Developments in the use of laser and microelectronic sensors make possible the rapid measurement of physical properties that were previously difficult to measure. Actuator technology also continues to improve through developments in servo-drives, torque and force motors, and piezoelectric actuators for high-precision positioning systems,introduction 3 Camera control contro «= tive ag 1516 | 98 18 Robot Machine Spindle drive Tool changer RR ara J | OC. a Re se and X-axis drive 5 Machine tool Robot Figure 1.1 Automated machining cell with loading/unloading robot. ‘As an example of the numerous electronic devices that may be found in an advanced manufacturing system, consider the machining cell shown in Fig. 1.1. The cell includes a machine tool and a robot for loading and unloading parts. The various electronic devices typically found in such a cell, and the purposes they serve, are as follows: Machine Tool . Axis Drive Servo-Motor—machine motion . Tachometer—motor velocity sensor Resolver—machine position sensor Spindle Drive Motor—tool rotation . Limit Switch—overtravel protection . Stepping Mator—tool changer positioning . Tactile Probe—measurement of workpiece . Servo-Amplifiers—machine drives . Control Computer—machine control . Display—machine status Robot . Servo-Motor—arm motion . Optical Encoder—arm position sensorIntroduction to Computer Control 13. Control Valve—pneumatic gripper 14, Tactile Sensor—gripper force 15. Servo-Amplifiers—robot arms 16. Control Computer—robot control 17. Display—robot status 18. Camera—part identification, guidance As you can see, these devices include control computers, position sensors, velocity sensors, limit switches, and actuators such as servo-motors and stepping motors. Also shown are a vision system for part identification and guidance of the robot and a tactile sensor on the machine tool for measurement of workpiece dimensions and automatic tool adjustment. In this regatd, we should note that control computers are used to implement system communication, sequencing, and decision-making functions as well as feedback control functions. Numerous ideas about the application of digital computers to the manipula- tion of elements in automated manufacturing systems have been advanced, published, discussed, and marketed. It is interesting that many early computer systems, designed as either tatal-automation systems or process control systems, failed to gain immediate respect or failed to be commercially successful for a number of important reasons, First and foremost, many early systems were designed, sold, and installed before essential computer programing, control theory, and maintenance capabilities were developed. Second, it was not uncommon to see latge computer installations originally dedicated only to process monitoring and data collection rather than incorporating control. An anxious management, expecting to be rewarded by financial return on the computer investment, became discouraged by the inability of the computer staff to show prompt and tangible progress. Third, computer reliability, including the massive interface systems required for on-line monitoring and control, had not yet developed to the point where they generated firm confidence. Fourth, not a great deal was known about the overall dynamics of systems on. which control was to be implemented, and there were relatively few persons knowledgeable in the art and science of discrete systems modeling and control theory. The analytical treatment of computer-based feedback control systems is a culmination of a historical development that began with consideration of continuous analog control concepts. The mathematical foundations of differential equations and complex variable theory, including the Laplace transform, afforded designers of continuous linear control systems a means of model formulation, equation manipulation, analysis, design, and controller synthesis. As interest in discrete systems grew during the late 1950s, the z-transform and certain associated concepts such as the ideal sampler came into use. As a result, it became generally accepted that one should be well versed in continuous-Concepts of Computer Control 5 system control theory and in digital control theory before one proceeded to the design of discrete control systems. This situation can be most discouraging to both practicing engineers and students who wish to accomplish feedback control tasks with a digital computer. Fortunately, it need not be so. The authors propose that we reflect on the basic nature of the computer- —namely, its handling of dieerote information—and bogin with the basic mathematics that describes its nature. This text is based on the idea that a time- domain approach utilizing difference equations, sequences, and series provides an efficient and productive introduction to the implementation of computer control. The backward shift operator of classical finite-difference equation theory is used here as the basic manipulation tool. It provides a direct, efficient approach to leaming how to design computer feedback control systems for practical applications and allows rapid utilization of computers in the solution of a broad spectrum of practical problems. The material that follows first develops an approach to computer control design and synthesis using the backward shift operator. The concepts of process modeling, system stability, controller algorithm synthesis, and analytical manipu- lation of control dynamical equations are discussed. This presentation is divided into chapters on the elements of discrete modeling, system response, and discrete controller design. Important topics in the implementation of computer control are then presented in chapters on control computers, computer interacting, sensors, command generation, sequential control using programmable logic controllers and process modeling. The final chapters and Appendix A present approaches to the analysis and design of discrete systems and controllers via transform and state-variable techniques. CONCEPTS OF COMPUTER CONTROL The application of a digital computer for control may take many forms and may entail a wide range of ideas and principles. The simplest approach is to use the computer as a large and efficient data collector and calculator. This concept is illustrated schematically in Fig. 1.2, wherein the system shown is described as a computer monitor. The terms process, plant, and machine may be used inter- changeably, depending on that which is to be controlled. In control literature, the terms plant and process are commonly employed. The plant or process itself may involve a great number of internal, individual, or interrelated feedback control loops, or it may simply be a specific physical process. For example, a machine tool or robot has a number of axes, each of which can have control loops for position, velocity, current, and so on. These axes are combined with other machineIntroduction to Computer Control Operator Computer input = [|-———>}_— Computer interface Process Measured Manipulated process outputs process inputs Figure 1.2 Schematic diagram of a computer monitoring system. components to perform processing functions that are to be monitored and manipulated. In a computer monitoring system, human operators manipulate individual process controllers directly. The computer simply observes the process via an interface that is capable of detecting physical signals such as contact closures and analog voltages, converting analog signals to digital words, and converting digital words of various formats into meaningful binary information for computer processsing. Operators make decisions and take actions on the basis of analyses that the computer can be programmed to prepare. Many early computer installations were undertaken with this approach in the hope that the logging and processing of large amounts of data would lead to a better understanding of the overall behavior of the process and that, ultimately, a more sophisticated level of computer control could be developed. Figure 1.3 illustrates the concept of open-loop computer control. in this scheme the computer is used as an on-line processor of programs and data to generate specific commands for the manipulation of machine and process actuators. Open-loop principles are also employed in the use of computers to manipulate electric or electro-hydraulic stepping motors for machine tool control. In order to automate a process or plant completely, it is necessary to develop computer systems that are capable of comparing a desired set of results with actual results and of taking corrective action. This is the essence of closed-loop computer control or feedback control. Figure 1.4 depicts a computer control system wherein the computer is placed in a control loop by collecting operating data from sensors in the process, accepting desired reference states (the set point) from an operator, and closing the control loop by establishing manipulationConcepts of Computer Control 7 Operator Computer Computer -———» output interface Process Manipulated process inputs Figure 1.3 Schematic diagram of computer open-loop control. commands that cause specific changes in the process outputs. Within the computer, two distinct operations must usually be performed: The difference or ettor between the operating point and the set point or error is calculated from the equation Error = Set Point — Actual Output and a corrective action is determined by solving an equation or set of equations that have been established as characterizing a desirable way to control the system. The law governing determination of this manipulation can be referred to as the control algorithm. These operations are illustrated in block diagram form in Fig. 1.5. Figure 1,4 Schematic diagram of computer closed-loop control. Operator ‘Computer ‘Computer input =}———»| Computer. -K———+) output interface interface Process #—______J Measured Manipulated process outputs process inputs\ntroduction to Computer Control From operator e [ BT Eto | Controt law ¢ To output Desired |_oralgorithin’ | “Process intertace setpoint a manipulation Process output Computer From input interface Figure 1.5 Block diagram of computer functions for closed-loop control. Computer closed-loop feedback control is similar in concept to closed-loop feedback control of continuous analog systems. The difference lies primarily in the fact that the digital computer operates with specific sets of numbers at discrete points in time. The computer is programmed to sample values of the process output variables at certain instants in time. Calculation then proceeds in such a way as to establish the error and the desired manipulation corresponding to that error, using an appropriate control law or algorithm. The manipulation is then applied to the process. The process outputs are again sampled after a predetermined interval called the sample period, and the procedure is repeated. Clearly, the speed at which the computer can perform many operations relative to the speed at which the process can respond is a significant factor in how well the process can be kept under control. Techniques for designing control algorithms and establishing sampling rates are discussed in considerable detail in subsequent chapters. Digital control techniques have been utilized extensively in machine tools, chemical process plants, rolling mills, paper machines, and many other processes. The paper machine serves as a good example. Two of the primary output variables in the Fourdrinier papermaking process are basis weight (the weight of a certain area of paper) and moisture content. These two parameters can be related to the amount of fiber stock, which is controlled by the position of a stock valve, and to the amount of moisture being taken out in the dryer section, which may be considered in a simple analysis as proportional to the setting of the dryer steam valve. The computer can sample the output transducers (namely, a weight gauge and moisture indicator), compute the necessary valve actuations, and directly manipulate the stock and steam valves. : Another scheme that has been widely i is feedforward control. Processes are often subjected to disturbance inputs that affect the output of theSummary 9 process but can be neither predicted in advance nor controlled. Where a process is subjected to external disturbances that can be readily measured and monitored, it is possible to compute process manipulations and implement them directly without waiting for information to return via a feedback path. For example, feedforward control has proved to be a powerful technique in the digestion of wood chips in a pressure vessel under controlled conditions of temperature, pressure, time, and liquor composition. (Because of long process times at various stages in the digester, feedback control is incapable of producing alterations in process inputs in time to prevent significant loss of product when disturbances such as changes in the properties of incoming wood chips occur.) The computer can monitor and compute the optimal “cooking conditions” for material entering the digester and can alter control functions as these constituents proceed through the process. Quality and productivity are important elements in any modern manufactur- ing system. Consider the machining of a cylindrical object. One aspect that may be important to quality is diameter. The diameter of the cylinder may need to be within a certain tolerance in order to fit into a hole in another part. If cylinders are produced with a diameter outside of this tolerance, the condition must be detected and the parts scraped. Material and processing time are wasted and productivity suffers. An even more serious situation arises when it is not detected that out-of-tolerance parts are being manufactured, and an attempt is made to produce assemblies with parts that do not fit. Productivity losses are magnified whenever errors are allowed to persist while further value is being added. Statistical quality control based on checking samples of parts was developed because it was often too expensive to check every machined part. Today, quality can be built into manufacturing processes by using appropriate sensors, computer decision making, and actuators. For the cylinder machining process, this can be accomplished using opto-electronic gauging and robotic manipulators. Not only is it possible to provide 100 % inspection, but the information gleaned can also be used in the control of variables such as tool position. And by studying the successive corrections required to sustain part accuracy, engineers can make further inferences about tool wear and breakage. Remember, as you study this text, that information collected and not used to its fullest extent is information partially wasted. SUMMARY This general introduction outlines only a few of the many important concepts in computer control of machines and processes. In the material to follow, these and other significant concepts are developed in detail, with emphasis on design10 Introduction to Computer Control techniques and examples. The current “state of the art” has evolved over several decades of improvement in computer speed and reliability, reduction in computer size and cost, and increased availability of interface hardware. Modern computers can accomplish a great deal in real-time control, as the concepts and techniques developed im subsequent: chapters wil itustrate. First we must develop a mathematical basis for analyzing systems as viewed by a computer. Then we can investigate techniques for synthesizing control algorithms. Chapter 2 provides the concepts necessary to create mathematical models of real elements and systems that are compatible with the discrete nature of the computer and provide the analyst and designer with a view of the process to be controlled as “seen” by the computer. The more information that can economically be gathered about a process or a product, and the more we understand about how that information reflects performance, quality, and productivity, the better our opportunity to achieve improvement through computer control. This improvement can be the difference between success and failure of a system. In industry, achieving improvement over the competition is a key strategic objective. BIBLIOGRAPHY Astrom, K. J. and B. Wittenmark. Computer Controlled Systems: Theory and Design Englewood Cliffs, NJ: Prentice-Hall, 1984. Beeby, W., and P. Collier. New Directions through CAD/CAM, Dearborn, MI: Society of Manufacturing Engineers, 1986. Besant, C. B. Computer Aided Design and Manufacture, 2/e. New York, NY: Wiley, 1983. Bollinger, J. G. "Computer Control of Machine Tools,” Annals of the CIRP.Vol.. 21. No. 2, 1972. Cadzow, J. A., and H. R. Martens. Discrete Time and Computer Control Sustems. Englewood Cliffs, NJ: Prentice-Hall, 1970. Costa, A., and M. Garetti. “Design of a Control System for a Flexible Manufacturing Cell," Journal of Manufacturing Systems, Vol. 4, No. 1, 1985, pp. 65-84. Desrochers, A. A., "A Comparison of Adaptive Control Strategies for Hot Steel Rolling Mills,” Journal of Manufacturing Systems, Vol.1, No.2, November 1982, pp- 183-194. Dorf, R. C. Modern Control Systems, 4/e. Reading, MA: Addison-Wesley, 1986. Domfeld, D. A, D. M. Auslander, and P. M. Sagues. “Microprocessor-Controlled Manufacturing Processes,” Mechanical Engineering, Vol. 102, No. 13, December 1980, pp. 34-41.Bibliography "1 a Duffie, N. "An Approach to the Design of Distributed Machinery Control Systems,” IEEE Transactions on Industry Applications, Vol. IA-18, No. 4, July/August 1982, pp. 435-442. ElMaraghy, H. “Automated Tool Management in Flexible Manufacturing,” Journal of Manufacturing Systems, Vol. 4, No. 1, 1985, pp. I-13. Fotsch, R. “Machine Tool Justification Policies: Their Effect on Productivity and Profitability,” Journal of Manufacturing Systems, Vol. 3, No. 2, 1984, pp. 169-195. Groover, M. P. Automation, Production Systems and Computer-Integrated Manufacturing. Englewood Cliffs, NJ: Prentice-Hall, 1987. Harrison, H. L., and J. G. Bollinger. Introduction to Automatic Controls. New York, NY: Harper & Row, 1969. Hutchinson, G. K., and J. R. Holland. “The Economic Value of Flexible Automation,” Journal of Manufacturing Systems, Vol. 1, No. 2, 1982, pp. 215-228. International Trends in Manufacturing Technology. Bedford, UK: IFS (Publications) Ltd, 1983-87 (series). Jones, A. and C. McLean. “A Proposed Hierarchical Control Model for Automated Manufacturing Systems,” Journal of Manufacturing Systems, Vol. 5, No. 1, 1986, pp. 15-25. Kalpakjian, S. Manufacturing Processes for Engineering Materials. Reading, MA: Addison- Wesley, 1984. Koren, Yoram. Computer Control of Manufacturing Systems. New York, NY: McGraw-Hill, 1983. Maimon, O. “Real-Time Operational Control of Flexible Manufacturing Systems,” Journal of Manufacturing Systems, Vol. 6, No. 2, 1987, pp. 125-136. Ranky, P. G. Computer-Integrated Manufacturing. Englewood Cliffs, NJ: Prentice-Hall, 1986. Rembold, U., C. Blume, and R. Dillmann. Computer-Integrated Manufacturing Technology and Systems. New York, NY: Marcel Dekker, 1985. Riley, F. J. Assembly Automation. New York, NY: Industrial Press Inc., 1983 Roger, A. “The Microprocessor Invades the Production Line,” IEEE Spectrum, January 1979, pp. 53-59. Spur, G., G. Seliger, and B. Viehweger. “Cell Concepts for Flexible Automated Manufacturing,” Journal of Manufacturing Systems, Vol. 5, No. 3, 1986, pp. 171-179. Weck, M. “Machine Diagnostics in Automated Production,” Journal of Manufacturing Systems, Vol. 2, No. 2, 1983, pp. 101-106.al Chapter 2 Elements of Discrete Modeling INTRODUCTION 12 The purpose of this chapter is to explore some of the basic concepts and techniques applicable to the analysis and design of digital computer control systems. As a starting point, let us consider the way in which a computer can be utilized in controlling a process with many variables to be controlled. The nature of the computer is such that it can work with only one controlled variable at a time. Because of this, having sampled a particular variable (obtained its current value), the computer usually must initiate an appropriate action or correction quickly and then move on to the next controlled variable. Thus, although a given variable is continuous with time, the computer has knowledge of its value only at discrete points in time. Much of control theory is concerned with continuous systems: characterized by analog signals, and much of it uses differential equations as a basis for modeling. However, the nature of the digital computer makes it convenient to think in terms of discrete systems for which difference equations provide the appropriate mathematical basis. Thus, this chapter introduces the concepts of discrete modeling. Examples of how difference equations can be derived for specific first-order, second-order, and delay processes are presented first, followed by a discussion of block diagrams for discrete systems and of how a table of transfer functions can be used to obtain the transfer function of a given system. The concept of proportional, integral, and derivative control actions performed by a computer is introduced, and we show how closed-loop system models are obtained. Finally, a discussion of the solution of difference equations leads to the analysis of stability for discrete systems.Discrete Process Models 13 2.2 | DISCRETE PROCESS MODELS Only processes with a single manipulated variable (input) and a single controlled variable (output) will be considered here, but the results are applicable to processes with many controlled variables, provided that there are no interactions between variables. The interaction problem is discussed in Chapter 4, and the more generalized state-variable modeling approach is developed in Appendix A. The immediate objective is to develop a suitable mathematical relationship between some output variable and a corresponding input variable on the basis that the dynamics of the process are known. The essential ideas are illustrated in Fig, 2.1. A process with a single input and a single output is shown schematically in Fig. 2.1(a). For a large class of computer control applications, the input can be considered a series of steps, because it is the nature of computer control hardware to establish a level of input and to have this level held fixed until one sample interval later, at which time a new level can be produced. Normally, the process output is sampled via computer control hardware that nearly instantaneously “captures” the value and makes it available for subsequent control calculations. Figure 2.1(b) shows a motor/amplifier process for which the input is a voltage v(F) and the output is the motor shaft position 6(H). A hypothetical input sequence is plotted against time in Fig. 2.1(0). T is the time period between samples of the output; it is assumed that the time required to sample the process output and produce a new process input is small compared to the sample period T. The process input results in an output, as illustrated in Fig. 2.1(d). The mathematical problem can be stated as follows: Given the defining equation for the process, the value of the input at some arbitrary time # = 0, and the value of the output at time f = 0, together with any necessary previous values, determine the value of the output one sample period later—that is, at time t = T. The relationship can also be thought of in terms of normalized time n, where n is an integer index and n = #/T. We will now illustrate concepts of discrete modeling by considering several basic models that can be used either to define certain simple processes or to approximate the behavior of more complicated processes. Example 2.1 An integration process To provide an example of what is commonly known as an integration in control terminology, assume that the process illustrated in Fig. 2.1(b) can be defined approximately by the equation d6(t) oe = Koll dt oo) where 0(#) is the input, 6(#) is the output, and K is a constant that depends on motor/amplifier characteristics. Assume that 6(0) = 9, at time t = O and that a14 Elements of Discrete Modeling INPUT OUTPUT —— + PROCESS. }-——> (a) INPUT amplifier Motor OUTPUT v(t) Shaft Motor shi aft > E>) "position Amplifier input voltage _ Voltage and current to motor (b) 0, and the initial condition is h(0) =: 0 meters. The corresponding differential equation is dhit) —— +h) =2 a + h(t) Its solution is and is plotted in Fig. 2.3. From this solution, Program 2.1 yields the following values for h(#) in meters, obtained at 1-minute increments in time. h(0) = 0 A(1) = 1.264 A(2) = 1.729 h(3) = 1.900 h(a) = 1.963 If the sample period is chosen as T= 1 minute, the constants in the difference equation are <0 ets * = 0.368 K(1 — e~ *) = 0.2(0.632) = 0.1264 meters/degree Thus the difference equation for the process is h, = 0.368h,_, + 0.12640, _, i Figure 2.3 Response in height of liquid in tank for constant valve position of 10°. Ait) {meters} ~ Calculated from solution of differential equation X Calculated from difference equation (with a 1-minute sample period)18 Elements of Discrete Modeling program Continuous_tank_model; {Height of liquid in a tank is calculated for a constant input valve position.} const {0.1-minute time increment} {t-minute process time constant} {0.2-meters/degree process gain} {height [m}, position [deg], time [min}} begin {initial values} {10-degree valve position} {increment time} {solution of differential equation} writeln(t, h) until t >= 10.0 {stop after 10 minutes} end. Program 2.1 Continuous tank model with constant input for Example 2.2. With the initial condition h, = 0 meters, and with 0, = 10° forn = 0, 1,2,3,..., values at several sample times can be calculated, by using the difference equation. Some of these values are hy = 0 h, =O + 1.264 = 1.264 = 0.368(1.264) + 1.264 = 1.729 hy hy = 0.368(4.729) + 1.264 = 1.900 h, = 0.368(1,900) +: 1.264 = 1.963 This technique can be referred to as the recursion method for calculating the process output from the input. These values, which can be obtained via Program 2.2, are plotted in Fig. 2.3. Note that they correspond exactly to those evaluated previously from the solution of the differential equation. However, unlike the solution of the differential equation, the recursion method does not yield information about the response of the process at times other than the sample times ¢ = 0, 1, 2, 3,... minutes. 0 Example 2.3 Varying process input It must not be inferred that the process input is required to be the same constant value from sample period to sample period, In fact, the opposite is usually the case. The necessary assumption is that the input is constant during any sample period. To illustrate this point, the