ARTIFICIAL INTELLIGENCE

Artificial intelligence (Modelling air

fuel ratio control)
ABSTRACT

AI is a filed of study that encompasses computational techniques for performing

tasks that apparently require intelligence when performed by humans. Such problems

include diagnosing problems in automobiles, computers and people, designing new

computers writing stores and symphonies, finding mathematical theorems, assembling

and inspecting products in factories. It is a technology of information processing

concerned with process of reasoning, learning and perception.

AI is both an art and a science. The activity of developing intelligent computer

systems employs both proved mathematical principles, empirical results of studying

previous systems.

The ability of systems to sense its environment, make decisions and control

actions is known as intelligence. Intelligent action when carried out on a machine are

known as artificial intelligence which can be formally defined as- :It is a branch of

computer science, which deals with a growing set of computer problem solving

techniques being developed to imitated the human thought or decision making process on

to produce the same results as those processes.

pravinjadhav111@indiatimes.com narendranrc@rediffmail.com

G.C.O.E.AMRAVATI 0
 ARTIFICIAL INTELLIGENCE

CONTENTS
SR.NO TOPIC PAGE NO.

1 INTRODUCTION 1

2 AI TECHNIQUES 1

2.1 KBs 1
2.2 NNs 2
2.3 FUZZY LOGIC 2
2.4 GAs 2
2.5 CBR 2
3 COMPONENT OF AI 3

3.1 INTELELLIGENT DESIGN 3

3.2 FUZZY LOGIC 3
3.3 CBR 4
3.4 HYBRID SYSTEM 4

4 FUZZY MODELLING 5

5 CONCLUSION 7

6 REFRENCES 7

G.C.O.E.AMRAVATI 1
 ARTIFICIAL INTELLIGENCE
1. INTRODUCTION:
AI in Design research is concerned with the use of AI techniques to study design.

This is done by making hypotheses about designing and producing modes of design,

knowledge and activity. Using these hypotheses and modes, computer programs (i.e.,

design systems) are built that can aid designers and carry out some portion of the design

process. Thus hypotheses about design are tested and refined using design systems.

Design, as a complex, goal-directed, problem-solving activity, has long been

studied as a topic in AI. Apart from the intellectual challenge, it is also an important

contributor to economic success, and is a fundamental precursor to manufacturing.

AI techniques and components of an intelligent manufacturing system. Since its

emergence in the 1950s AI has provided several techniques with applications in

manufacturing. In the early years, knowledge-based systems, (KBS) attracted great

attention. Recently neural networks (NN), case-based reasoning (CBR), genetic

algorithms (GAs) and fuzzy logic have attracted more attention and have been

successfully employed in manufacturing.

This section contains a very brief and much simplified outline of the main. AI

techniques and the components of simplified model of an intelligent. Manufacturing

system.

2. AI TECHNIQUES:

2.1 KBs: The first attempt to be widely used to equip manufacturing systems with

some degree of intelligence was the use of KBS. They seek to incorporate human

knowledge about an application area, usually elicited from experts in the particular

domain, so that the system can automatically replicate aspects of best practice. The
G.C.O.E.AMRAVATI 2
 ARTIFICIAL INTELLIGENCE
human knowledge is represented using the IF- THEN production systems or more

structured formats such as frames and semantic nets.

2.2 NNs: NNs are based on ideas about how the brain may work. Input stimuli (e.g.

the parameter values have been widely used in many classification and optimization

situations. History data are used to train the network, automatically determining the

most appropriate configuration of the hidden network.

2.3 Fuzzy logic: Fuzzy logic allows the representation and processing of uncertain

information such as linguistic statements, for example the customer desires a car, which

is fairly luxurious. Judging the degree not is used in mathematical modeling or target

setting.

2.4 GAs: GA use ideas from population genetics for solving complex global

optimization problems. A pool of potential candidate solutions evolves through

reproduction and mutation of the fittest and elimination of the least promising solutions

of each generation are made extinct.

2.5 CBR: In CBR, the intelligent component of the system contains a history of past

problems and the (successful) solutions, applied. Future problems can then be considered

through analogy with these past cases to home in rapidly on the most promising type of

solutions. A further step is incorporating machine learning though the updating of the

case-base with those for which the solutions suggested proved successful.

3. COMPONENTS OF AN INTELLIGENT MANUFACTURING

SYSTEMS:

G.C.O.E.AMRAVATI 3
 ARTIFICIAL INTELLIGENCE
The manufacturing
Process is a complex INTELLIGENT
PROCESS
One and can be dec- PLANNING INTELLIGENT
INTELLIGENT QUALITY
omposed into several DESIGN MANAGEMENT

components. Rao
INTELLIGENT
et al.(1993) decom- MANUFACTURING
INTELLIGENT
posed intelligent MAINTANANCE INTELLIGENT
AND DIAGNOSIS INTELLIGENT SCHEDULING
Control, systems into CONTROL

following components:

Fig.No.1: Components of an intelligent manufacturing

system.

Intelligent design, intelligent operation, intelligent control, intelligent planning and

intelligent maintenance. We modify this decomposition slightly to reflect the current

trends in the literature on intelligent manufacturing systems as shown in figure 1.

3.1 INTELLIGENT DESIGN: The importance of product design is

undeniable. A firms products or services are typically the primary source and focus of

contact with its customers, and the development of new designs plays a key role in

establishing and maintaining a competitive position for most firms. There are many

problems in design manufacturing systems.

3.2 FUZZY LOGIC: Optimizing the efficiency in cutting a sculptured surface using

numerically controlled machining techniques needs to carefully consider the relationship

between cutting edges and surfaces geometry. A fuzzy basis material removal

optimization approach is suggested by Ip (1998(to compensate the variation of cutting

speed due to the change of gradient on the sculptured surface in machining process.
G.C.O.E.AMRAVATI 4
 ARTIFICIAL INTELLIGENCE
Fuzzy modeling is used for the penicillin-G conversion processs. A linguistic fuzzy

model, which represents, the kinetic term of the conversion, is developed from

experimental data by means of fuzzy clustering.

3.3 CBR: The frequent used of past experience by human engineers when solving new

problems has led to an interest in the use of CBR to help automate engineering design.

Purvis and Pu (1998) developed a constraint based methodology for case combination.

The methodology is implemented in a CBR system called COMPOSER and has been

tested in two design domains: assembly sequence design and configuration design. Gao

et al (1998) used BR for mechanical plan systems design. Design plans are stored as the

actual cases in the CBR system. Cbr is used for the design of bar linkages and fixtures

design.

A product may have different ways of disassembly an experence is important

during disassembly if e want to satisfy goals such as part reuse, recycle or discard. Zeid

et al (1997) have proposed a CBR approach to solve design for disassembly (DFD)

problem.

3.4 HYBRID SYSTEMS: Chen et al (1998) have developed an integrated expert

system that consist of a knowledge base, a database, pattern-recognition, artificial NN

and GA modules for complicated chemical reaction systems used t prepare industrial

materials. The system has been used in many applications including the production of

alloy steel, synthetic rubber ceramic materials production and materials design of

composite materials, high temperature superconductors, and ceramic, semiconductors.

Lee et al (1999) have developed fuzzy non- linear programming modes to

optimize machining operations. It uses fuzzy login together with traditional

G.C.O.E.AMRAVATI 5
 ARTIFICIAL INTELLIGENCE
mathematical programming to make a more flexible maker. The model also uses an NN

model, which can be used to assess the machinability of the machining operations. The

output from the fuzzy non-linear programming model provides the input for the NN

model. A material design system has been developed, utilizing mathematical modeling

and knowledge-based approaches by Shivathaya and Fang(1999). The KBS generates

about 15-30 different target compositions for steelmaking for each customer order. Fuzzy

logic is applied in the system for the design of steel plates, to rank the alternative target

compositions for steelmaking according to the degree to which they will satisfy

customers requirements.

4. FUZZY MODELLING FOR AIR FUEL RATIO CONTROL:

This section outlines the development and use of a fuzzy model applied to a

simulated engine air fuel ratio (AFR) control system. Accurate control of AFR relies on

knowledge of airflow into the engine and the dynamics, which affect injected fuel within

the inlet manifold. The system is non-linear with operating point dependent dynamics

and possesses a time delay, which varies, as a function of engine speed and inlet manifold

air mass flow rate. Whilst satisfactory control of the system is achieved in steady-state

using conventional PID control, there exist difficulties when encountering transient

conditions, such as sudden acceleration/ deceleration and loaded/unloaded conditions.

Essentially, a feed forward element is introduced as a base controller which makes used

of a fuzzy model; this adjusts the fuel injection rate according to the current operating

conditions of engine crankshaft speed and inlet manifold pressure, with the air flow into

the engine being estimated using a fuzzy model. The fuzzy model is identified using a

clustering technique applied to input-output data and the identified model is used to set
G.C.O.E.AMRAVATI 6
 ARTIFICIAL INTELLIGENCE
the feedforward fuel injection pulse width signal to obtain the desired AFR. An

advantage of the approach is that measurements of inlet manifold pressure and manifold

airflow rate are not required. Rather, use is made of a fuzzy model-based scheme which

requires measurements of the throttle plate angle, denoted >> & cir; and the factor >>

only, where >> = actual AFR/desired AFR; with the control objective being to maintain at

unity, thereby achieving the desired stoichiometric ratio. The quantity >> is measured

using a linear exhaust gas oxygen (EGO) sensor mounted in the exhaust system. Figure 2

shows the general control scheme where AB V, tim and tic denote the air by-ass valve

control signal and the contributing to the fuel injection signal it form the feedforward and

feedback controllers respectively. It should be noted that the feedback controller, giving

due consideration to the system time delay, is based on the Smith predictor method. The

set-pint of unity and fig 2 Control scheme using PID controller and feedforward fuzzy

model for base level control in the measured value of >> are denoted >> SP and >> e

respectively.

ABV a N ABV
WARDFEED ORF a m it
BASE CONTRO-
N -LLER
+ ti SYSTEM m

+
sp e
_ FEEDBACK
+ CONTROLER tic

Fig.No.2 General Control System for Air Fuel Ratio Control

One of the major limitations of employing feedback control regulating >> is that the

closed loop includes a transport delay. It is further compounded by the fact that this

delay varies as a function of both engine speed and mass air flow rate. The results
G.C.O.E.AMRAVATI 7
 ARTIFICIAL INTELLIGENCE
achieved using the fuzzy model based feedforward element, shown in fig 2. are

encouraging and indicate that such a fuzzy model-based approach can be effectively

applied to transient control of >>. Transient conditions are simulated when the engine

crankshaft speed is fixed at 2,500 rpm and the position of the throttle plate angle is

varied.

5. CONCLUSION:

This paper has highlighted following areas:

Introduction to AI techniques in design and components of intelligent

manufacturing systems

A case study of Fuzzy modeling for air fuel ratio control where fuzzy

methodologies have shown distinct advantages over conventional mode based

strategies, for control and decision making in two different applications. The

results to date are encouraging, indicating potential for implementation of fuzzy

logic methodologies in practice.

6. REFERENCES:
Intelligent systems I manufacturing: current developments and future prospects
by-:
Farid Meziane, Sunil Vadera, Khairy Kobbacy, Naqthan Proudlove
Applications of AI in design and manufacturing systems by-:
Dr. Ken Brown and Mrs. Pat Fothergill
Intelligent systems for optimization and control by-:
K.J.Burnham, O.C.L. Haas and D. J. G. JamesW

G.C.O.E.AMRAVATI 8