Work Study and Ergonomics

Assignment
Why is Soft Computing required in
engineering?
 Introduction:
Soft computing became a formal area of study in the early 1990's. Earlier computational approaches
could model and precisely analyze only relatively simple systems. More complex systems arising in
biology, medicine, the humanities, management sciences, and similar fields often remained intractable
to conventional mathematical and analytical methods. That said, it should be pointed out that
simplicity and complexity of systems are relative, and many conventional mathematical models have
been both challenging and very productive. Soft computing deals with imprecision, uncertainty,
partial truth, and approximation to achieve tractability, robustness and low solution cost. Components
of soft computing include:

 Neural networks (NN)
 Fuzzy systems (FS)
 Evolutionary computation (EC), including:
 Evolutionary algorithms
 Harmony search
 Swarm intelligence
 Ideas about probability including:
 Bayesian network
 Chaos theory
 Perceptron
Generally speaking, soft computing techniques resemble biological processes more closely than
traditional techniques, which are largely based on formal logical systems, such as sentential
logic and predicate logic, or rely heavily on computer-aided numerical analysis (as in finite element
analysis). Soft computing techniques are intended to complement each other.
Unlike hard computing schemes, which strive for exactness and full truth, soft computing techniques
exploit the given tolerance of imprecision, partial truth, and uncertainty for a particular problem.
Another common contrast comes from the observation that inductive reasoning plays a larger role in
soft computing than in hard computing.
Neural Networks:
Artificial neural networks are made up of interconnecting artificial neurons (programming constructs
that mimic the properties of biological neurons). Artificial intelligence and cognitive modelling try to
simulate some properties of neural networks. While similar in their techniques, the former has the aim
of solving particular tasks, while the latter aims to build mathematical models of biological neural
systems.
Applications:
The tasks to which artificial neural networks are applied tend to fall within the following broad
categories:
 Function approximation, or regression analysis, including time series prediction and
modelling.
 Classification, including pattern and sequence recognition, novelty detection and sequential
decision making.
 Data processing, including filtering, clustering, blind signal separation and compression.
Application areas of ANNs include system identification and control (vehicle control, process
control), game-playing and decision making (backgammon, chess, racing), pattern recognition (radar
systems, face identification, object recognition, etc.), sequence recognition (gesture, speech,
handwritten text recognition), medical diagnosis, financial applications, data mining (or knowledge
discovery in databases), visualization and e-mail spam filtering.

Fuzzy Control System:

A fuzzy control system is a control system based on fuzzy logic—a mathematical system that
analyzes analog input values in terms of logical variables that take on continuous values between 0
and 1, in contrast to classical or digital logic, which operates on discrete values of either 0 or 1 (true
or false).

The logic involved can deal with fuzzy concepts—concepts that cannot be expressed as "true" or
"false" but rather as "partially true". Fuzzy logic has the advantage that the solution to the problem
can be cast in terms that human operators can understand, so that their experience can be used in the
design of the controller. This makes it easier to mechanize tasks that are already successfully
performed by humans.

The first industrial application was a cement kiln built in Denmark, coming on line in 1975. Seiji
Yasunobu and Soji Miyamoto of Hitachi in1985 provided simulations that demonstrated the
superiority of fuzzy control systems for the Sendai railway. Their ideas were adopted, and fuzzy
systems were used to control accelerating, braking, and stopping when the line opened in 1987.
Japanese consumer goods often incorporate fuzzy systems. Matsushita vacuum cleaners use
microcontrollers running fuzzy algorithms to interrogate dust sensors and adjust suction power
accordingly. Hitachi washing machines use fuzzy controllers to load-weight, fabric-mix, and dirt
sensors and automatically set the wash cycle for the best use of power, water, and detergent.
As a more specific example, Canon developed an autofocusing camera that uses a charge-coupled
device (CCD) to measure the clarity of the image in six regions of its field of view and use the
information provided to determine if the image is in focus. It also tracks the rate of change of lens
movement during focusing, and controls its speed to prevent overshoot.
As another example of a practical system, an industrial air conditioner designed by Mitsubishi uses 25
heating rules and 25 cooling rules. A temperature sensor provides input, with control outputs fed to
an inverter, a compressor valve, and a fan motor. Compared to the previous design, the fuzzy
controller heats and cools five times faster, reduces power consumption by 24%, increases
temperature stability by a factor of two, and uses fewer sensors.
The US Environmental Protection Agency has investigated fuzzy control for energy-efficient motors,
and NASA has studied fuzzy control for automated space docking: simulations show that a fuzzy
control system can greatly reduce fuel consumption. Firms such as Boeing, General Motors, Allen-
Bradley, Chrysler, Eaton, and Whirlpool have worked on fuzzy logic for use in low-power
refrigerators, improved automotive transmissions, and energy-efficient electric motors.

Currently there are many applications of Fuzzy Logic utilized by common household devices,
products, which most people are familiar with. A few products which have benefited from the
implementation of Fuzzy Logic are: camcorders with automatic compensation for operator
injected noise such as shaking and moving; elevators with decreased wait time, making
intelligent Fuzzy Logical decisions and minimizing travel and power consumption; anti lock
braking systems with quick reacting independent wheel decisions based on current and acquired
knowledge ; television with automatic colour, brightness and acoustic control based on signal and
environmental conditions; and finally, most importantly to this article, single loop temperature
and process controls.

There are literally dozens of applications in a standard vehicle which can deliver improved quality
and performance through fuzzy logic. Air conditioning, Anti-lock Braking Systems, Cruise Control,
Engine management, Transmission Systems and an array of new sensor based detection systems for
vehicle displacement have all been optimized through fuzzy.
Here are just few examples of how Fuzzy Logic has been applied in reality:
 Recognition of handwritten symbols with pocket computers (Sony)
 Flight aid for helicopters (Sugeno)
 Controlling of subway systems in order to improve driving comfort, precision of halting and
power economy (Hitachi)
 Improved fuel consumption for automobiles (NOK, Nippon Denki Tools)
 Single button control for washing-machines (Matsushita. Hitachi), Automatic motor-control for
vacuum cleaners with recognition of surface condition and degree of soiling (Matsushita)
 Prediction system for early recognition of earthquakes {Inst. of Seismology Bureau of
Metrology, Japan)

Chaos theory:
Chaos theory is a field of study in mathematics, physics, economics, and philosophy studying the
behavior of dynamical systems that are highly sensitive to initial conditions. This sensitivity is
popularly referred to as the butterfly effect. Small differences in initial conditions (such as those due
to rounding errors in numerical computation) yield widely diverging outcomes for chaotic systems,
rendering long-term prediction impossible in general.This happens even though these systems are
deterministic, meaning that their future behavior is fully determined by their initial conditions, with
no random elements involved.In other words, the deterministic nature of these systems does not make
them predictable. This behavior is known as determfinistic chaos, or simply chaos.

Chaotic behavior can be observed in many natural systems, such as the weather. Explanation of such
behavior may be sought through analysis of a chaotic mathematical model, or through analytical
techniques such as recurrence plots and Poincaré maps.

Applications

Chaos theory is applied in many scientific disciplines: mathematics, programming, microbiology,

biology, computer science, economics, engineering, finance, philosophy, physics, politics, population
dynamics, psychology, and robotics.
Chaotic behaviour has been observed in the laboratory in a variety of systems including electrical
circuits, lasers, oscillating chemical reactions, fluid dynamics, and mechanical and magneto-
mechanical devices, as well as computer models of chaotic processes. Observations of chaotic
behaviour in nature include changes in weather, the dynamics of satellites in the solar system, the
time evolution of the magnetic field of celestial bodies, population growth in ecology, the dynamics
of the action potentials in neurons, and molecular vibrations. There is some controversy over the
existence of chaotic dynamics in plate tectonics and in economics.

One of the most successful applications of chaos theory has been in ecology, where dynamical
systems such as the Ricker model have been used to show how population growth under density
dependence can lead to chaotic dynamics.

Chaos theory is also currently being applied to medical studies of epilepsy, specifically to the
prediction of seemingly random seizures by observing initial conditions.

A related field of physics called quantum chaos theory investigates the relationship between chaos
and quantum mechanics. The correspondence principle states that classical mechanics is a special
case of quantum mechanics, the classical limit. If quantum mechanics does not demonstrate an
exponential sensitivity to initial conditions, it is unclear how exponential sensitivity to initial
conditions can arise in practice in classical chaos. Recently, another field, called relativistic chaos has
emerged to describe systems that follow the laws of general relativity.

The motion of N stars in response to their self-gravity (the gravitational N-body problem) is
generically chaotic.
In electrical engineering, chaotic systems are used in communications, random number

Swarm intelligence:
 Swarm intelligence (SI) describes the collective behaviour of decentralized, self-organized
systems, natural or artificial. The concept is employed in work on artificial intelligence. The
expression was introduced by Gerardo Beniand Jing Wang in 1989, in the context of cellular
robotic systems.
  SI systems are typically made up of a population of simple agents or boids interacting locally
with one another and with their environment. The agents follow very simple rules, and
although there is no centralized control structure dictating how individual agents should
behave, local, and to a certain degree random, interactions between such agents lead to the
emergence of "intelligent" global behaviour, unknown to the individual agents. Natural
examples of SI include ant colonies, bird flocking, animal herding, bacterial growth, and fish
schooling.

 The application of swarm principles to robots is called swarm robotics, while 'swarm
intelligence' refers to the more general set of algorithms. 'Swarm prediction' has been used in
the context of forecasting problems.

Applications

 Swarm Intelligence-based techniques can be used in a number of applications. The U.S.

military is investigating swarm techniques for controlling unmanned vehicles. The European
Space Agency is thinking about an orbital swarm for self assembly and interferometry.
NASA is investigating the use of swarm technology for planetary mapping. A 1992 paper by
M. Anthony Lewis and George A. Bekey discusses the possibility of using swarm intelligence
to control nanobots within the body for the purpose of killing cancer tumours.

 Crowd simulation

 Artists are using swarm technology as a means of creating complex interactive systems or
simulating crowds. Tim Burton's Batman Returns was the first movie to make use of swarm
technology for rendering, realistically depicting the movements of a group of bats using the
Boids system. The Lord of the Rings film trilogy made use of similar technology, known as
Massive, during battle scenes. Swarm technology is particularly attractive because it is cheap,
robust, and simple.

 Ant-based routing

The use of Swarm Intelligence in Telecommunication Networks has also been researched, in the
form of Ant Based Routing. This was pioneered separately by Dorigo et al. and Hewlett Packard
in the mid-1990s, with a number of variations since. Basically this uses a probabilistic routing
table rewarding/reinforcing the route successfully traversed by each "ant" (a small control packet)
which flood the network. Reinforcement of the route in the forwards, reverse direction and both
simultaneously has been researched: backwards reinforcement requires a symmetric network and
couples the two directions together; forwards reinforcement rewards a route before the outcome is
known (but then you pay for the cinema before you know how good the film is). As the system
behaves stochastically and is therefore lacking repeatability, there are large hurdles to commercial
deployment. Mobile media and new technologies have the potential to change the threshold for
collective action due to swarm intelligence (Rheingold: 2002, P175).

Soft Computing Applications in Mineral and Metal Industries

Soft Computing has many industrial application areas, also in mineral and metal processing.
Intelligent methods are used in software sensors to make the existing measurements more efficient or
to replace the non-existing measurements with software systems that form the measurement signals
e.g. from other existing measurements. Both fuzzy logic and neural networks have been used.
Another area is the monitoring of the measurement systems and assuring the good condition of the
system.

Diagnostics and fault detection systems are classical applications. Compared with algorithmic-
based fault detection the biggest advantage is that they give possibilities to follow human’s way of
fault diagnosing and handle different information and knowledge in a more efficient way.
Applications vary from troubleshooting of hydraulic and electronic systems to vibration analysis. The
essential feature in the future is combining of Soft Computing methods with mathematical models and
other conventional tools and finally their integration with total condition monitoring systems.

In production scheduling applications, systems are targeting to the higher throughput, the keeping
of delivery times and good end-product quality. The functions vary from the allocation of customer
orders to the selection of the routing in the mill according to product requirements. Two main
possibilities are available (Leiviska 1992): integrating Soft Computing with conventional algorithmic
scheduling or replacing existing systems with new ones. Combined systems that utilise fuzzy logic,
neural networks and genetic algorithms are used, because one approach alone can seldom lead to even
feasible solution.

Mining and Mineral Processing

Meech (1995) gives an extensive survey on the state-of-the art and future of intelligent methods in
mining industry. He also gives an exhaustive list of existing references (for instance, McDermott et al.
1992; Eggert et aI. 1994; Benford and Meech 1992; Lacouture and al.1991; Edwards and Mular 1992;
Poirier et al.1993; Meech and Harris 1992; Rakocevic and Meech 1995 and Kumar et al.1993). In
mining industry, Expert Systems are used but they are poor at mathematical problem solving; they
cannot learn and are usually not adaptable; generalisation and summarisation skills are essentially
non-existent. Speed issues will also arise in real-time applications. The methodologies that permit
adaptation are fuzzy logic and neural networks. For instance, (Ruokonen 1996; Laine et al.1995) are
describing the Expert System for ore-type classification that utilises neural networks (SOM). Fuzzy
control of ventilation processes in coal mines has been reported by (Chojcan
et al.1995).
Another area in mining industry that lends itself to Soft Computing is the pit design and scheduling
problems. They have been dealt with in two articles by (Thomas 1996a,b). Actually there are two
problems connected: Ultimate Pit Determination and Extraction Sequencing (Scheduling). They are
not independent, but should be solved together. Several optimisation methods have been used in
solving the separate problems, as surveyed by Thomas. Of these only Tolwinski and Underwood's
approach (Tolwinski and Underwood 1992), genetic algorithms (Denby and Schofield, 1994, 1995)
and simulated annealing are capable of dealing the overall optimisation problem. Also here the
combination of different approaches seems necessary.

SAG and FAG milling has been a conventional area for expert system applications because of the
complicated relationships between required power in grinding and the mill load (Paris and Cipriano
1991; Perry and Hall 1994; Bourassa and Lanthier 1999). Early applications of neural network exist
also for ball mills (Thibault et al. 1991; Aamentet al.1993).

Expert Control in flotation is already widely used (Cipriano et al.1991; Lacouture et al.1991). A
newer application combining image processing and expert systems is presented later in this book.
Cipriano et al. (1997) and van Deventor et aI. (1997) also report this kind of applications. Newer
publications refer to the application of:

• Fuzzy logic control (Carvalho and Durao 1999; Osorio et al.1999; Bergh et al.1998; Carvalho and
Duma 2000)

• Neural networks (Cortez and Durao 1995; Durao and Cortez 1995; Banolacci and Bouajila 2000)

• Genetic algorithms (Karr 1995)

in modelling and control of the flotation plant.

The FLC has been developed for a rotary dryer and it s control performance to compensate for the
variations in the output moisture of solids has been examined by experiments with the pilot plant
dryer in University of Oulu (Koskinen and Yliniemi 1996). Step changes in different input variables
as in the input moisture of solids and in mass flow of solids have been made. FLC has been tuned off-
line based on a trial and error procedure by changing the rules and the mapping of membership
functions. Experiments show that FLC combined with the PI-controller is able to maintain the output
moisture of solids as desired. Also fuzzy modelling with linguistic equations has been used to help the
tuning of FLC (Koskinen et al.1998). Also a hybrid neural/JPI-controller has been tested and found
able to compensate for disturbances in the input moisture. The neural network controller itself is an
s.c. direct inverse controller that utilises the inverse process model. The network has been trained by
back propagation algorithm in Matlab's Neural Network Toolbox. An adaptive fuzzy controller for
this process will be introduced later on in this book.

The fluidised bed granulation process has been modelled using artificial neural networks (Ramet et
al.1996).The independent input variables were the inlet air temperature, the atomising air pressure and
the amount of the binder solution. The responses used were mean granule size and granule friability.
Several fully connected multi-layer neural networks were tested. The training algorithm was a
modified back propagation algorithm designated as the Weigend weight eliminator.

The behaviour of the powder bed height inside the fluidised bed process is difficult to model and its
automatic control using conventional methods is questionable. The behaviour of materials is uncertain
and the granulation mechanisms are complex and unclear. The process is also highly non-linear and
there are severe difficulties in measuring the bed height. The bed height control is, however,
significant for the efficient operation of the whole process. This all makes it a good candidate for
fuzzy logic control. A prototype fuzzy logic controller was developed and tested in a laboratory scale
apparatus (Ramet et al. 1996).

Neural networks have also been applied in reconciling inconsistent process data (Aldrich and van
Deventer 1994, 1995) and as a soft sensor for particle size estimation (Du et aI.1997).

Manufacturing Automation and Robotics

A. General View

The term intelligence has been frequently used in this field since robotic technologies that mimic
human thinking and behaviour of bio-systems have been developed. Contemporary intelligence is
sometimes considered to be interactive information processing among human beings, environment,
and artificial objects. Intelligence is defined as human-like information processing and adaptation to
environment by leaning, evolution, and prediction in order to survive. The use of structured
intelligence by soft computing for intelligent robots has been considered. However, the interaction
with human beings is also important. Recently, emotional robots that interact with human beings have
attracted much interest by researchers. KANSEI (emotion, feeling) information processing has
become popular in Japan. This technology is needed for the development of human-friendly robots.
Other technologies, e.g., fuzzy associative memory and chaotic computation have also been used for
developing human-friendly robots (intelligent robots, welfare robots). Soft computing is widely used
in this field.

B. Application Fields

Soft computing has been used in the construction of intelligent robots and manufacturing systems and
for solving non-linear and uncertain problems in the fields of hands and manipulators, mobile robots,
multiagent robots, welfare robots, emotional pet robots, and manufacturing systems. It is expected
that soft computing will play an increasing role in the realization of human-friendly systems in the
future.

C. Hands and Manipulators

Lin et al. developed compact fuzzy decentralized controllers including sensor fusion schemes and
introducing human skills through communication lines for a five-finger robot hand with 17 degrees of
freedom. 17 potentiometers, 18 tactile sensors, and 17 actuators were installed in the design. Digital
signal processor (DSP) chips were used to implement the proposed schemes. Kiguchi et al. developed
position/force control using the fuzzy vector method with a fuzzy neural network that received aid
from an intelligent task planner. This is an intelligent control method because soft computing is used.
Fukuda et al. presented the design of an intelligent robotic system based on soft computing. An
architecture of a robot system with structured intelligence was developed by them.

The structured intelligence is explained such that perception and action functions are represented in
the horizontal axis, and reactive motion, skilled motion, primitive motion planning, and motion
planning functions as hierarchical levels are represented in the vertical axis, interacting with the
environment. The role of SC is central in this intelligent robot system.

D. Mobile Robots

Baranyi et al. proposed an improved vector field-based guiding model as an extension of the
potential-based guiding model. A simplified neuro-fuzzy approximation algorithm was applied to the
realization of models for the guidance of mobile robots.

E. Multiagent Robots

Ishiguro et al. developed an architecture for behaviour arbitration based on artificial immune
networks. Antigens and antibodies in the artificial immune network were used as agents for
environment description and decision-making for action, respectively. Each antibody (agent) was
assigned by fuzzy inference (if–then, if not–then).An action decision was made following the action
of the highest density agent. The complex system was optimized by EC, NN, reinforcement learning,
introduction of meta-knowledge and other. This was an example of the so-called reactive distributed
artificial intelligence and cognitive distributed artificial intelligence. Soft computing was used in the
design of reactive and cognitive artificial intelligences. Katagiri et al. used fuzzy inference and
random search learning, which was devised for controlling interactive behaviours of a group of
multiagent robots. Less computation time was needed compared with that for other multiagent robot
systems.

F. Welfare Robots

Kohata et al. used fuzzy associative memories and chaos computing to construct human friendly
multiagent robots for the welfare industry. Since this was a parallel computation method, the parallel
processing algorithm was implemented on A-NET (Actors NETwork). Ushida et al. proposed a fuzzy-
associative memory-based knowledge construction method for application to a human-machine
interface, and they demonstrated that it is an important tool for investigating human-friendly welfare
robots, and a possible alternative for developing intelligent robots. Takagi et al. used soft computing
techniques (FL and interactive EC) for hearing impairment compensation and physical rehabilitation.
In this field, the evaluation of system performance is subjective and depends on individuals (human
beings). KANSEI (emotion, feeling) information processing, which is popular in Japan, is suitable
also for this purpose.

G. Emotional Pet Robots

Kubota et al. have been using an emotional model for the development of evolutional pet robots. The
structured intelligence described in Sections VII-A and VII-C (perception learning and behaviour
learning) is based on soft computing. An emotional model based on feelings (moods) reflected by the
external environment is defined for the pet robot to perform tricks through interaction with its owner.
The owner teaches the pet robot how to perform tricks. Atkeson et al. have been studying human
behaviour using humanoid robots. Interdisciplinary technologies of brain science including artificial
neural networks and soft computing (wavelet transformation and others) have been widely used to
analyze human behaviour.

H. Manufacturing Technologies

Iokibe et al. proposed a fault diagnosis method using chaos computing for chaotic time series
analysis, fuzzy reconstruction of chaos state trajectories and separation of white noise from the
trajectories. Their method was applied to fault diagnosis for rotating machine parts, and it was found
that the

use of this method could greatly reduce the required computational time. Djordjevich et al. developed
a system for monitoring tool wear condition using neuro-fuzzy computing. The feed cutting force was
estimated by an adaptive neuro-fuzzy inference system based on the measurement of servo motor feed
currents. Yen et al. proposed a wavelet-based feature extraction method for monitoring vibration
conditions of dynamic systems. In their method, symptom vectors extracted by wavelet
transformation were fed into the inputs of the neural network classifier. Kim developed fuzzy logic
controllers (FLC) using a reconfigurable field programmable gate array (FPGA) system. The FLC
was partitioned into many temporally independent functional modules. Each module was installed
individually on the FLC automatic design and implementation system. Each implemented module
formed a downloadable hardware object that was ready to configure the FPGA chip.
 Conclusion

As can be understood from the topics discussed, soft computing is a term applied to a field which is
characterized by the use of inexact solutions to computationally-hard tasks, for which an exact
solution cannot be derived. Soft computing differs from conventional computing in that, unlike hard
computing, it is tolerant of imprecision, uncertainty, partial truth, and approximation. In effect, the
role model for soft computing is the human mind. The guiding principle of soft computing is: Exploit
the tolerance for imprecision, uncertainty, partial truth, and approximation to achieve tractability,
robustness and low solution cost. In many ways, soft computing represents a significant paradigm
shift in the aims of computing - a shift which reflects the fact that the human mind, unlike present
day computers, possesses a remarkable ability to store and process information which is pervasively
imprecise, uncertain and lacking in categoricity. The successful application of soft computing
suggests that the impact of soft computing will be felt increasingly in coming years. Soft computing is
likely to play an especially important role in science and engineering.
 References

 www.en.wikipedia.org

 http://www.omega.com/techref/fuzzylogic.html

 Fuzzy logic for engineering applications - Robert Babuska