Lecture 7 EXPERT CONTROL SYSTEMS Artificial intelligence, in particular expert system techniques, have been developing rapidly in control engineering.

Lecture 7

EXPERT CONTROL SYSTEMS

Artificial intelligence, in particular expert system techniques, have been developing rapidly in control engineering.

Applications of expert-system techniques in control engineering control-system design, fault diagnosis, simulation, modeling and identification, on-line performance monitoring, adaptation and auto-tuning and supervisory

control.

Branches of Computational Intelligence

7.1 Elements of an Expert System conventional computer software can be viewed as the synergy of:

In contrast, computer software used in Expert Systems can be described as the synergy of:

The most significant characteristic of this class of systems is that it draws on human knowledge and emulates human experts in the manner with which they arrive at decisions.

Definition of Expert System

A computing system capable of representing and reasoning about some knowledge rich domain, which usually requires a human expert, with a view toward solving problems and/or giving advice. Such systems are capable of explaining their reasoning.

Does not have a psychological model of how the expert thinks, but a model of the expert’s model of the domain.

7.1 Elements of an Expert System

Definition of Expert System

“An Expert System is the embodiment of knowledge elicited from human experts, suitably encoded so that the computa-tional system can offer intelligent advice and derive intelli-gent conclusions on the operation of a system”.

7.1 Elements of an Expert System

knowledge --two components: • facts, which constitute ephemeral information

subject to changes with time (e.g., plant variables) and

• procedural knowledge, which refers to the manner in which experts in the specific field of application arrive at their decisions.

7.1 Elements of an Expert System

7.1 Elements of an Expert System Expert System Structure

Inferenceengine

Explanationfacility

Knowledgebase

acquisitionfacility

Userinterface

Knowledgebase

Experts User

Knowledge Base

Stores all relevant information, data, rules, cases, and relationships used by the expert system

knowledge specific to the domain facts specific to the problem being solved Knowledge Representation is the key issue

Aim is usually to present the knowledge in as “declarative （陈述的）” a fashion （样式、方式） as possible

Inference Engine

Seeks information and relationships from the knowledge base and provides answers, predictions, and suggestions in the way a human expert would

Manipulates the knowledge base to solve the given problem

This is the "procedural knowledge", how to put the facts and domain knowledge together to reach a solution.

Basic ways inference engines work: forward chaining (forward reasoning)

FACTS = X

IF X, THEN Y

add Y to the blackboard which contains the facts start with the FACTS and work forward through the

rules to find a solution match FACTS to all possible RULES. A method of reasoning that starts with the facts and

works forward to the conclusions

Forward ChainingForward Chaining In this process the knowledge base is searched for rules that match

the known facts, and the action part of these rules is performed.The process continues until a goal is reached.

Puts the symptoms together to reach a conclusion ex. Doctor diagnosing a patient

Goal

Initial Knowledge/Facts

Forward Chaining

Basic ways inference engines work backward chaining (backward reasoning)

starts with the knowledge base - thinks of these as goals we are trying to obtain:

Y = result of rule (solution)

verify if FACTS (X) support the rule start with possible solution, and search facts to

see if rules can be supported A method of reasoning that starts with

conclusions and works backward to the supporting facts

Backward ChainingBackward Chaining Starts form a goal, the conclusion. All the rules that contain this conclusion are then

checked to determine whether the conditions of these rules have been satisfied Ex. Doctor has end idea of what is wrong with patient but know they must prove it

by going from the diagnosis and finding symptoms

Goal

Initial Knowledge/Facts

Backward Chaining

Explanation Facility

Explanation facility A part of the expert system that allows a user or

decision maker to understand how the expert system arrived at certain conclusions or results

Knowledge Acquisition Facility

Knowledge acquisition facility Provides a convenient and efficient means of capturing

and storing all components of the knowledge base

Knowledgebase

Knowledgeacquisition

facility

Joe Expert

User Interface Expert systems are interactive; a session

between the user and the KBS is necessary to generate a solution.

The interface is important since it provides the user with the ability to interact with the system.

A good user interface will increase users’ confidence in the system.

A poor interface will frustrate users and can cause a loss of confidence in the results of the system.

User Interface

The user interface also implements the explanation capability.

Essential is the ability to answer questions such as: Why? How? What?

Frequently: the ability to define terms

7.2 Stages in the Development of

an Expert System

Objectives ---Problem Definition Knowledge Acquisition and Knowledge

Representation Rapid Prototype Implementation Test and maintain

objectives The essential problem is selecting an

appropriate domain: the problem must require some type of specialized

knowledge, if there are human "experts" this criteria is probably satisfied

must not be overly large: define the problem fairly narrowly.

in business organizations, it should a problem that is handled often enough that an investment is expected to have some payoff: the once every 5 years sort of problem going to payoff.

Knowledge Acquisition

" the transfer and transformation of potential problem-solving expertise from some knowledge source to a program.”

- Buchanan 1983.

Knowledge Acquisition

machine learning - building capabilities into the system that allow it to learn from what it is doing. the problem of induction - how many instances

must be observed before it can be added to the knowledge base as "true"

knowledge elicitation - extract the knowledge from the human expert, through some means direct - interaction with the human expert

interviews, protocol analysis, direct observation, etc.

indirect - utilize statistical techniques to analyze of data and draw conclusions about the structure of the data.

Knowledge Acquisition (cont.)

Knowledge Representation A method to represent the knowledge about the domain

major methods: Decision tree Programming language logic

Although a shell contains a way to represent knowledge, shell selection should be influenced by the matching the representation to the knowledge in the domain.

Knowledge must be coordinated, so that the knowledge base is consistent.

Prototype Typically use an "incremental" development

approach to an expert system. Build an initial prototype and adjust and expand Allow the expert to interact with the prototype to

get feedback Reevaluate if the project should be

continued, if major redesign (knowledge representation) is necessary, or to go ahead.

Test and maintain

New rules can be continually added and old ones refined/ removed.

This is a tricky process, but there does not seem to be much literature on it.

One characteristic of an Expert system should be maintainability, so the ability to add/change/delete rules is essential.

Participants in Expert Systems Development and Use Domain expert The individual or group whose expertise and knowledge is

captured for use in an expert system Knowledge user

The individual or group who uses and benefits from the expert system

Knowledge engineer Someone trained or experienced in the design,

development, implementation, and maintenance of an expert system

Expertsystem

Domain expert

Knowledge engineer

Knowledge user

General Approaches to Building Expert Systems Purchase a developed system

Not that many exist, as packages are common for certain applications that are common to many businesses.

See expertise embedded in some applications, e.g., Turbo-Tax, network diagnostics.

General Approaches to Building Expert Systems

Build "in-house" using a shell A shell provides an inference engine, a user

interface, and a way to represent knowledge. Develop the knowledge base for the particular

problem domain. The focus of development is on knowledge

acquisition. Many shells are available for purchase.

General Approaches to Building Expert Systems

Build from scratch using an AI language Requires specialized training to effectively

program in these languages. Few people are trained in these approaches, and

these approaches are time consuming and expensive (shells are typically a much more economical approach).

Expert Systems Development Alternatives

low

high

low high

Developmentcosts

Time to develop expert system

Useexistingpackage

Developfromshell

Developfrom

scratch

When to Use an Expert System (1) Provide a high potential payoff or significantly

reduced downside risk Capture and preserve irreplaceable human

expertise Provide expertise needed at a number of

locations at the same time or in a hostile environment that is dangerous to human health

When to Use an Expert System (2) Provide expertise that is expensive or rare Develop a solution faster than human experts

can Provide expertise needed for training and

development to share the wisdom of human experts with a large number of people

Limitations Lack common sense: A KBS handles

problems in a very narrow range. Difficult to capture “deep knowledge” of a

problem domain. MYCIN, which diagnosis bacterial blood diseases,

does not know what blood does or the function of spinal cord. One story is that MYCIN asked if a patient was pregnant after being told the patient was a man.

Inability to provide deep explanation, i.e., why it applied certain rules.

Limitations Lack robustness: expertise is brittle. When a

human expert cannot solve a problem readily, they use their deep knowledge to come up with a strategy to attack a problem.

Difficult to verify. An important consideration as KBS approaches are applied to critical applications.

Little learning from experience. There are some inferential techniques, but they have their own limitations.

Categories of Expert SystemsCategory Problem Addressed

Interpretation Inferring situation descriptions from observations

Prediction Inferring likely consequences of given situations

Diagnosis Inferring system malfunctions from observations

Design Configuring objects under constraints Planning Developing plans to achieve goals

Monitoring Comparing observations to plans, flagging exceptions

Debugging Prescribing remedies for malfunctions

Repair Administer a prescribed remedy Instruction Diagnosing, debugging, and correcting student

performance Control Interpreting, predicting, repairing, and monitoring system

behavior

7.3 Concepts and Characteristics of Expert Control Systems DefinitionExpert control (or knowledge-based control) refers to methods that utilize expert-system techniques and control theory to design control systems that can auto-mate some of the tasks currently performed by human experts, and which cannot be carried out by traditional control systems

key point EC is the incorporation of heuristics and logic through knowledge-based structures, thus making the control systems more flexible and adaptive than conventional control systems.

7.3 Concepts and Characteristics of Expert Control Systems comparison of conventional expert systems and expert control system

7.3 Concepts and Characteristics of Expert Control Systems comparison of expert control and traditional advanced control

The fundamental functions of ECSs

(1) Take over the skilled operators' routine tasks and give effective

controls for processes which are time-varying, non-linear, and

subjective to various disturbances.

(2) Take advantage of all the available prior knowledge and on-line

information;

(3) perform fault diagnosis on the control system operation and

components, including the detection of actuator and sensor

problems;

(4) operate reliably and conveniently;

(5) Increase the amount of process knowledge, and accordingly improve

the control system's performance;

The fundamental functions of ECSs

(6) represent control knowledge in an effective way which easily allows

for modification and extension;

(7) Maintain dialogue with the user and give explanation of reasoning

results, and also obtain information from the user;

(8) require a minimal amount of prior knowledge;

(9) Have a capability for real-time reasoning and decision making.

suitable application areas for ECSs (1) ill-structured processes for which mathematical models do not exist

or are inadequate;

(2) Complex problems which require answers within a limited time

interval, such as fault diagnosis and emergency handling;

(3) Situations where expertise is required for problem-solving but

where there are not enough experts for the task;

(4) Situations where qualitative or uncertain information must be

processed, and symbolic logic is required for problem-solving;

(5) complicated problems where a heavy computing burden and high

cost would be involved when using conventional algorithmic

methods;

(6) Cases where operating conditions change frequently and/or

severely.

7.3 Concepts and Characteristics of Expert Control Systems DefinitionExpert control (or knowledge-based control) is one of the intelligent control methods, which combines control theory and expert-system techniques to design and realize in the autonomous operation of complex, uncertain or ill-defined physical processes.

An ECS is an intelligent control system which uses expert-system techniques on difficult control problems where analytic models do not exist or are inadequate, and require expert knowledge for their problem-solving.

7.4 Classification of Expert Control Systems

Rule-based expert tuning or adaptive controllers Expert supervisory control systems Hybrid expert control systems Real-time control expert system

7.4 Classification of Expert Control Systems Rule-based expert tuning or adaptive controllers

7.4 Classification of Expert Control Systems Expert supervisory control systems

7.4 Classification of Expert Control Systems

Hybrid expert control systems a composite intelligent control system which utilizes a multilayer hierarchical structure and the incorporation of various techniques, including expert systems, pattern recognition, fuzzy logic, neural networks, and computer process control.

7.4 Classification of Expert Control Systems Real-time control expert systema typical real-time expert system with all the characteristics of an expert system, such as modularity (flexibility), heuristics and transparency, as well as the features of a control system, e.g. real-time operation, reliability, and adaptation, etc

7.5 Design Principles of Expert

Control Systems knowledge representation in ECS can be grouped into two parts: system modeling (including the controlled process and controllers), and maintaining the relevant information and knowledge essential to

perform the intelligent control and supervision tasks.

Multiple representation forms should be used in modeling mainly because:

7.5.1 Modeling with multiple representation forms

7.5 Design Principles of Expert

Control Systems One of the important features of intelligent control is to classify and

extract on-line information in an effective way. In a complex system, a

large number of sensor data and noisy signals could enter the system

continuously. It is very important to collect, catalogue and dispense the

information in an organized way. Therefore, the emphasis of

information processing is on eliciting and recognizing characteristic

information that can reflect the system properties, and converting them

into the knowledge the decision-making requires.

7.5.2 Eliciting and recognizing characteristic information

7.5 Design Principles of Expert

Control Systems 7.5.3 Hierarchical structure of decision-making

Knowledge Refinement

Knowledge base & Inference Engine

Planning & Management

Supervisory control & Emergency Handing

Real-time Intelligent Control

7.5 Design Principles of Expert

Control Systems In ECSs, the inference engine should provide the Mechanism that

evaluates, interprets, and executes the data and knowledge to generate

inferences or sequences of actions to be executed under time constraints.

ECSs need to reason about a number of past, present and future events.

ECSs must be capable of being interrupted, to accept inputs from

unscheduled or asynchronous events, reasoning by a variety of means and

techniques.

Usually, different inference strategies should be used in different decision

levels or different tasks.

7.5.4 Real-time inference with multiple strategies

7.5 Design Principles of Expert

Control Systems concentrate only on the intelligence in the higher levels, such as

supervision, learning or adaptation, planning, etc., and adopt traditional

control techniques such as PID algorithms at their real-time level.

7.5.5 Introducing intelligent control into the real-time level

7.5 Design Principles of Expert

Control Systems

ECS is essentially non-linear, time-dependent, and also

unstructured. Thus, it is very difficult to analyze the stability of an

ECS by mathematical methods." Therefore, on-line monitoring of

the system behavior (e.g. acceptable behavior, malfunction

behavior and fault behavior,") and prediction of the possible states

to keep the system behavior within an acceptable area, is an

effective way to achieve guaranteed system stability.

7.5.6 On-line stability monitoring

7.6 Architecture of Expert Control

Systems

Figure 7.8 A generic architecture of expert control system

7.6 Architecture of Expert Control

Systems

Fig. 7.9 general basic structure of expert control

7.7 Development Methods of Expert Control

Systems The main tasks of developing an ECS can be grouped into three parts:

(1) Build the models of the process; including problem definition, model

selection, knowledge acquisition, etc.

(2) Construct an expert controller; involving building the knowledge base and

inference engine, constructing the system structure, determining knowledge

representation paradigms, selecting the control strategies and parameters,

etc.

(3) Establish a user-friendly interface; consisting of human-computer

interface design and management.

7.7 Development Methods of Expert Control

Systems Figure 7.10 Schema diagram of ECS development

seven stages