Beruflich Dokumente
Kultur Dokumente
development
of
rule
base
and
decision
making
system,
UNIT-VII
Fuzzification :
Properties:
Operations:
Relations:
Defuzzification methods:
Properties:
UNIT VIII
Operations:
Applications:
Relations:
Cardinalities:
Process identification:
Membership Functions:
Control:
fault diagnosis:
load forecasting:
Fuzzy logic applications:
Introduction:
Description of Fuzzy Logic
In recent years, the number and variety of applications of fuzzy logic have
increased significantly. The applications range from consumer products such
as cameras, camcorders, washing machines, and microwave ovens to industrial
process control, medical instrumentation, decision-support systems, and
portfolio selection.
human intuition. Furthermore, computing with words exploits the tolerance for
imprecision and thereby lowers the cost of solution.Another basic concept in
FL, which plays a central role in most of its applications, is that of a fuzzy ifthen rule or, simply, fuzzy rule. Although rule-based systems have a long
history of use in AI, what is missing in such systems is a mechanism for
dealing with fuzzy consequents and fuzzy antecedents. In fuzzy logic, this
mechanisms provided by the calculus of fuzzy rules. The calculus of fuzzy rules
serves as a basis for what might be called the Fuzzy Dependency and
Command Language (FDCL). Although FDCL is not used explicitly in the
toolbox, it is effectively one of its principal constituents. In most of the
applications of fuzzy logic, a fuzzy logic solution is, in reality, a translation of a
human solution into FDCL.A trend that is growing in visibility relates to the
use of fuzzy logic in combination with neurocomputing and genetic algorithms.
More generally, fuzzy logic, neurocomputing, and genetic algorithms may be
viewed as the principal constituents of what might be called soft computing.
Unlike the traditional, hard computing, soft computing accommodates the
imprecision of the real world. The guiding principle of soft computing is: Exploit
the tolerance for imprecision, uncertainty, and partial truth to achieve
tractability, robustness, and low solution cost. In the future, soft computing
could play an increasingly important role in the conception and design of
systems whose MIQ (Machine IQ) is much higher than that of systems designed
by conventional methods.Among various combinations of methodologies in soft
computing, the one that has highest visibility at this juncture is that of fuzzy
logic and neurocomputing, leading to neuro-fuzzy systems. Within fuzzy logic,
such systems play a particularly important role in the induction of rules from
observations. An effective method developed by Dr. Roger Jang for this purpose
is called ANFIS (Adaptive Neuro-Fuzzy Inference System). This method is an
important component of the toolbox.Fuzzy logic is all about the relative
importance of precision: How important is it to be exactly right when a rough
answer will do?In this sense, fuzzy logic is both old and new because, although
the modern and methodical science of fuzzy logic is still young, the concepts of
fuzzy logic relies on age-old skills of human reasoning.
Membership Functions:
A membership function (MF) is a curve that defines how each point in the input
space is mapped to a membership value (or degree of membership) between 0
and 1. The input space is sometimes referred to as the universe of discourse, a
fancy name for a simple concept.One of the most commonly used examples of a
fuzzy set is the set of tall people. In this case, the universe of discourse is all
potential heights, say from 3 feet to 9 feet, and the word tall would correspond
to a curve that defines the degree to which any person is tall. If the set of tall
people is given the well-defined (crisp) boundary of a classical set, you might
say all people taller than 6 feet are officially considered tall. However, such a
distinction is clearly absurd. It may make sense to consider the set of all real
numbers greater than 6 because numbers belong on an abstract plane, but
when we want to talk about real people, it is unreasonable to call one person
short and another one tall when they differ in height by the width of a hair.If
the kind of distinction shown previously is unworkable, then what is the right
way to define the set of tall people? Much as with the plot of weekend days, the
figure following shows a smoothly varying curve that passes from not-tall to
Membership Functions Fuzzy sets describe vague concepts (e.g., fast runner,
hot weather, weekend days).A fuzzy set admits the possibility of partial
membership in it. (e.g., Friday is sort of a weekend day, the weather is rather
hot).The degree an object belongs to a fuzzy set is denoted by a membership
value between 0 and 1. (e.g., Friday is a weekend day to the degree 0.8).A
membership function associated with a given fuzzy set maps an input value to
its appropriate membership value.
MEMBERSHIP FUNCTIONS
Definition: a membership function for a fuzzy set A on the universe of discourse
X is defined as A:X [0,1], where each element of X is mapped to a value
between 0 and 1. This value, called membership value or degree of
membership, quantifies the grade of membership of the element in X to the
fuzzy set A. Membership functions allow us to graphically represent a fuzzy set.
The x axis represents the universe of discourse, whereas the y axis represents
the degrees of membership in the [0,1] interval. Simple functions are used to
build membership functions. Because we are defining fuzzy concepts, using
more complex functions does not add more precision. Below is a list of the
membership functions we will use in the practical section of this tutorial.
Triangular function: defined by a lower limit a, an upper limit b, and a value m,
where a < m < b.
There are two special cases of a trapezoidal function, which are called Rfunctions and L-functions:
UNIT-VII
`Fuzzification Methods
1. Discrete case (numerical)
(X )
I
( Ai )
xi
Level NO.
-6
Range
Xn-3.2
NB NM NS ZE PS PM PB
1.0 0.3 0.0 0.0 0.0 0.0 0.0
-5
-3.2Xn-1.6
-4
-1.6Xn-0.8
-3
-0.8Xn-0.4
-2
-0.4Xn-0.2
-1
-0.2Xn-0.1
-0.1Xn+0.1
+0.1Xn+0.2
+0.2Xn+0.4
+0.4Xn+0.8
+0.8Xn+1.6
+1.6Xn+3.2
3.2Xn
Fuzzification Methods
2. Continuous case (functional)
NB
NM
ZE
g(x)
NS
PM
PS
PB
g(x)
(x)
( x) e
x c 2
)
(x)
( x)
1
xc
a
2b
Curse of dimensionality.
2. Tree partition
Each region can be uniquely specified along a corresponding decision
tree.
Exponentially increase in the number of rules.
More membership function for each input are needed.
3. Scatter partition
By covering a subset of the whole input space that characterizes a region
of possible occurrence of the input vectors.
Limit the number of rules to a reasonable amounts.
Antecedent
Consequent
(premise)
(conclusion)
R [ A B] C
R [ A ( x) B ( y )] c ( z )
Change of Error
Error
NB
PS
PM
PM
PB
PB
PB
PB
PM
PS
PO
NO
NS
NM
NB
NM
PS
PM
PM
PB
PB
PB
NS
PS
PM
PB
NM
NM
NO
PS
PS
PM
PB
PB
NB
NB
NB
NO
NO
PM
PB
PB
NB
NB
NB
NS
NS
NO
PM
PM
NB
NB
NB
MN
MN
NS
-
NB
NB
NB
MN
MN
NS
-
where all the entries of the matrix are fuzzy sets of error, change of error, and
control. The abbreviations used above mean:
PB : Positive Big
PM : Positive Medium
PS : Positive Small
PO : Positive Zero
NO : Positive Zero
NS : Negative Small
NM : Negative Medium
NB : Negative Big
B2
C1
Premise 3
x is and y is ,then z is
C2
Conclusion
z is
C ( A B) [( A1 B1 ) C1 ] [( A2 B2 ) C2 ]
( A B) [( A1 B1 ) C1 ] ( A B) [( A2 B2 ) C2 ]
C1 C2
Inference Mechanisms
1. Mamdani fuzzy inference model.
2. Larsen fuzzy inference model.
3. Tsukamoto fuzzy inference model.
4. Sugeno fuzzy inference model.
ci ( z ) i ci ( z )
i Ai ( x) Bi ( y )
c ( z ) c1 c2
[1 c1 ( z )] [ 2 c2 ( z )]
C1
ci ( z ) i ci ( z )
i Ai ( x) Bi ( y )
c ( z ) c c
1
[1 c1 ( z )] [ 2 c2 ( z )]
min or
product
A1
B1
C1
is) Ci y
i Ci ( z i )
A2
Z
B2
C2
1 z1 2 z 2
1 2
z1
z2
weighted average
z z
z
1
ri
f
First-order polynomial function : (x, y)
i = p ix + qiy + ri
Z0
1 f1 ( x0 , y0 ) 2 f 2 ( x0 , y0 )
1 2
at the constant.
Defuzzification Methods
1. Max criterion Method
Largest of Maximum.
Smallest of Maximum.
2.Mean of Maximum method (MOM).
n
Z MOM
i 1
zi
n
zi maximum value.
reaches
Z
Z
( z )zdz
( z )zd
or
Z COA
i 1
( zi ) zi
i 1
( zi
n = number of quantization levels of output c
Center of Height
C f
k k
fk
Ck : Core
f k : height
f1
C1
f2
C2
f3
C3
Middle of Max
middle
Examples of Fuzzy Control
Fuzzy car.
FLC hardware systems.
Fuzzy Automatic Train Operation (ATO) systems.
Fuzzy automatic container crane operations
Fuzzy logic chips and fuzzy computers.
Fuzzy washing machine.
Fuzzy vacuum cleaner.
Fuzzy air conditioner.
Fuzzy camcorder.
Parts List
(1) IBM or compatible personal computer equipped to run Microsoft Quick
BASIC. IBM is a registered trademark of IBM Corporation. Microsoft and
Quick BASIC are registered trademarks of Microsoft, Inc.
(2) Controller (see below).
(3) Signal conditioner (transistor amplifier to adjust levels as needed).
(4) Transistor - 2N3053.
(5) DC motor, 1.5 V to 3.0 V, 100 ma., 1100 Rpm to 3300 Rpm, and compatible
generator.
The above speed control system is low cost and suitable for learning at home
where being rigorously, mathematically correct is not required. It is important
to be aware that this speed controller is only an experimental controller to get
familiar with the fuzzy logic concept. It is not what engineers call a rigorous,
technically correct application of fuzzy logic. The difference is in the fact that
this approach does not add triangles to compute center of mass as specified by
Dr. Bart Kosko (Fuzzy Thinking, Chapter 10). Adding triangles can be done,
but is difficult and time consuming, however that is the way a truly
professional application would be designed. There are ICs that do it all and
commercially available fuzzy logic controllers that do everything correctly.
This fuzzy logic controller project was done under pressure of very limited
money available, resulting in an inexpensive approach. What is needed is an
analog to digital converter, which connects to a PC, and a digital to analog
output device from the PC to the transistors and DC motor-generator being
controlled. Often this is all in one plug-in card that goes inside the PC. Plug
in the A to D and D to A converter in the PC and write a program to measure
the input and control the output according to fuzzy logic principles. This
approach can be somewhat expensive and was not used in this case.
For this experiment, the controller was a 40-8 controller manufactured by
Prairie Digital Company. Click on the following Web page to see this controller.
http://www.prairiedigital.com/PDI_Website/PDI_Model40.htm (The author
has no connection at all with Prairie Digital.) The 40-8 controller is external to
the computer, connecting to the PC via a standard RS-232 serial port. The
serial port connects to the 40-8 controller via a serial cable. A BASIC program
is used to communicate with the 40-8 controller.
The Prairie Digital
instruction book has sample programs showing how to do this. Through a
BASIC program, you can read the analog voltage level on one of the 40-8 analog
input lines, then tell the 40-8 to output a pulse-width-modulated signal. By
controlling the pulse width of the 40-8 output, the average value of the output
is the equivalent of varying the level of the output in an analog fashion.
If not constrained by cost, a 12 bit, A to D unit should be used, rather than the
8 bit unit. This would provide improved control. This approach, using the 408 controller, is low cost, in the range of $100 to buy the 40-8. Purchasing the
items Prairie Digital offers to accompany the controller, that is the connector,
cables, software, etc., is recommended. It costs very little extra, but is well
worth it.
With regard to the other hardware, only low cost transistors, resistors,
capacitors, etc., were used for the signal conditioner providing input to the
motor-generator. The DC motor and DC generator were small, low power units
purchased from a surplus catalog. The motor output shaft was connected to
the generator input shaft with a small section of shrink insulation tubing;
cheap, simple and effective. The power supply for everything, including the
40-8, was a 12 Volt DC power supply removed from an old Apple computer.
National Instruments, www.natinst.com, sells a fuzzy logic system where the
fuzzy control action is accomplished by the software. National Instruments
applications engineers recommend one of their several analog/digital in,
The next three steps use a charting technique which will lead to a computer
program. The purpose of the computer program is to determine the voltage to
send to the speed controlled motor. One function of the charting technique is
to determine the "degree of membership" (see Ch. 1) of the Too slow, About
right and Too fast triangles, for a given speed. Further, the charting technique
helps make the continuous control feedback loop easier to visualize, program
and fine tune.
5. Associate the above inputs and outputs as causes and effect with a Rules
Chart, as in Figure 4, below. The chart is made with triangles, the use of
which will be explained. Triangles are used, but other shapes, such as bell
curves, could also be used. Triangles work just fine and are easy to work with.
Width of the triangles can vary. Narrow triangles provide tight control when
operating conditions are in their area. Wide triangles provide looser control.
Narrow triangles are usually used in the center, at the set point (the target
speed). For our example, there are three triangles, as can be seen in Figure 4
(three rules, hence three triangles).
6. Figure 4 (above) is derived from the previously discussed Rules and results
in the following regarding voltage to the speed controller:
a. If speed is About right then Not much change needed in voltage to the
speed controller.
b. If speed is Too slow then increase voltage to the speed controller to Speed
up.
c. If speed is Too fast then decrease voltage to the speed controller to Slow
down.
7.
Determine the output, that is the voltage that will be sent from the
controller/signal conditioner/transistor to the speed controlled motor. This
calculation is time consuming when done by hand, as we will do below, but this
calculation takes only thousandths of a second when done by a computer.
Assume something changes in the system causing the speed to increase from
the target speed of 2,420 Rpm to 2,437.4 Rpm, 17.4 Rpm above the 'set point."
Action is needed to "pull" the speed back to 2,420 Rpm. Intuitively we know
we need to reduce the voltage to the motor a little. The "cause" chart and
vertical speed line appear as follows, see Figure 5 below:
The vertical line intersects the About right triangle at .4 and the Too fast
triangle at .3. This is determined by the ratio of sides of congruent triangles
from Plane Geometry:
Intersect point / 1 = 11.6/29 = .4
Intersect point / 1 = 17.4/58 = .3
8. The next step is to draw "effect" (output determining) triangles with their
height "h" determined by the values obtained in Step 7, above. The triangles
to be drawn are determined by the rules in Step 6. Since the vertical 2,437.4
Rpm speed line does not intersect the Too slow triangle, we do not draw the
Speed up triangle. We draw the Not much change and the Slow down triangles
because the vertical speed line intersects the About right and Too fast
triangles. These "effect" triangles will be used to determine controller output,
that is the voltage to send to the speed control transistor.
The result is
affected by the widths we have given the triangles and will be calculated. See
Figure 6, below. The Not much change triangle has a height of .4 and the Slow
down triangle has a height of .3, because these were the intersect points for
their matching "cause" triangles; see Figure 4, above.
The way around this is to make the states "fuzzy", that is, allow them to change
gradually from one state to the next. You could define the input temperature
states using "membership functions" such as the following:
With this scheme, the input variable's state no longer jumps abruptly from one
state to the next. Instead, as the temperature changes, it loses value in one
membership function while gaining value in the next. At any one time, the
"truth value" of the brake temperature will almost always be in some degree
part of two membership functions: 0.6 nominal and 0.4 warm, or 0.7 nominal
and 0.3 cool, and so on.
The input variables in a fuzzy control system are in general mapped into by
sets of membership functions similar to this, known as "fuzzy sets". The
process of converting a crisp input value to a fuzzy value is called
"fuzzification".
A control system may also have various types of switch, or "ON-OFF", inputs
along with its analog inputs, and such switch inputs of course will always have
a truth value equal to either 1 or 0, but the scheme can deal with them as
simplified fuzzy functions that are either one value or another.
Given "mappings" of input variables into membership functions and truth
values, the microcontroller then makes decisions for what action to take based
on a set of "rules", each of the form:
IF brake temperature IS warm AND speed IS not very fast
THEN brake pressure IS slightly decreased.
In this example, the two input variables are "brake temperature" and "speed"
that have values defined as fuzzy sets. The output variable, "brake pressure", is
also defined by a fuzzy set that can have values like "static", "slightly
increased", "slightly decreased", and so on.
This rule by itself is very puzzling since it looks like it could be used without
bothering with fuzzy logic, but remember the decision is based on a set of
rules:
All the rules that apply are invoked, using the membership functions and
truth values obtained from the inputs, to determine the result of the
rule.
This result in turn will be mapped into a membership function and truth
value controlling the output variable.
These results are combined to give a specific ("crisp") answer, the actual
brake pressure, a procedure known as "defuzzification".
give greater narrowing, while "somewhat" broadens the function by taking the
square root.
In practice, the fuzzy rule sets usually have several antecedents that are
combined using fuzzy operators, such as AND, OR, and NOT, though again the
definitions tend to vary: AND, in one popular definition, simply uses the
minimum weight of all the antecedents, while OR uses the maximum value.
There is also a NOT operator that subtracts a membership function from 1 to
give the "complementary" function.
There are several different ways to define the result of a rule, but one of the
most common and simplest is the "max-min" inference method, in which the
output membership function is given the truth value generated by the premise.
Rules can be solved in parallel in hardware, or sequentially in software. The
results of all the rules that have fired are "defuzzified" to a crisp value by one of
of several methods. There are dozens in theory, each with various advantages
and drawbacks.
The "centroid" method is very popular, in which the "center of mass" of the
result provides the crisp value. Another approach is the "height" method, which
takes the value of the biggest contributor. The centroid method favors the rule
with the output of greatest area, while the height method obviously favors the
rule with the greatest output value.
The example below demonstrates max-min inferencing and centroid
defuzzification for a system with input variables "x", "y", and "z" and an output
variable "n". Note that "mu" is standard fuzzy-logic nomenclature for "truth
value":
There are two input variables, temperature and pressure, and a single output
variable, the turbine throttle setting. The turbine's operation can be reversed,
so the throttle setting can be positive or negative. The fuzzy set mappings are
shown below:
Large negative.
Medium negative.
Small negative.
Zero.
Small positive.
Medium positive.
Large positive.
The output value will adjust the throttle and then the control cycle will begin
again to generate the next value.
BUILDING A FUZZY CONTROLLER
* Consider implementing with a microcontroller chip a simple feedback
controller:
A fuzzy set is defined for the input error variable "e", and the derived change in
error from the previous error input, "delta", as well as the "output", as follows:
LP:
SP:
ZE:
SN:
LN:
large positive
small positive
zero
small negative
large negative
If the error ranges from -1 to +1, with the analog-to-digital converter used
having a resolution of 0.25, then the input variable's fuzzy set (which, in this
case, also applies to the output variable) can be described very simply as a
table, with the error / delta / output values in the top row and the truth
values for each membership function arranged in rows beneath:
_______________________________________________________________________
-1 -0.75 -0.5 -0.25 0
0.25 0.5 0.75 1
_______________________________________________________________________
mu(LP)
0
0
0
0
0
0
0.3 0.7
1
mu(SP)
0
0
0
0
0.3 0.7
1
0.7 0.3
mu(ZE)
0
0
0.3 0.7
1
0.7 0.3
0
0
mu(SN)
0.3 0.7
1
0.7 0.3
0
0
0
0
mu(LN)
1
0.7 0.3
0
0
0
0
0
0
_______________________________________________________________________
This table is a little tricky to interpret. It simply defines each of the elements of
the fuzzy set ("LP" through "LN") in terms of the values ("mu") that they have
relative to the given values in the top row.
It is important to remember that for the error / delta variables, the top row
values are used to generate the mu values, while for the output variable, the
mu values are used to generate the top row values.
Suppose this fuzzy system has the following rule base:
rule
rule
rule
rule
1:
2:
3:
4:
IF
IF
IF
IF
e
e
e
e
=
=
=
=
These rules are typical for control applications in that the antecedents consist
of the logical combination of the error and delta signals, while the consequent
is a control command output.
These rules can also be a little tricky to interpret. For example, what rule 1
says is that if the error value is the ZE fuzzy set element and the delta, or
change from last error, value is the ZE fuzzy set element, then the output is
also the ZE fuzzy set element. What is very tricky is that plugging actual
numeric values into this rule does not give, at least directly, an actual value for
the output setting.
The value for the output setting is defined by rule 1 as "ZE". Looking at the
fuzzy value table shows that "ZE" has its maximum value corresponding to the
value of "0" on the top row of the table, and so this rule always gives an output
setting of 0. Similarly, rule 2 always gives an output setting of "SN" or -0.5, rule
3 always gives an output setting of "LP" or 1, and rule 4 always gives an output
setting of "LN" or -1.
The thing to remember is that the actual output is a combination of the outputs
of these four rules, weighted by the values by the appropriate mu values
provided by the rule calculation, using the centroid calculation:
SUM( I = 1 TO 4 OF ( mu(I) * output(I) ) ) / SUM( I = 1 TO 4 OF mu(I) )
The result of this calculation gives the actual output setting.
* Now, suppose that at a given time we have:
error = 0.25
delta = 0.5
Then the corresponding mu values can be obtained by simply taking them from
the appropriate columns of the fuzzy set table:
_________________________________
error = 0.25 delta = 0.5
_________________________________
mu(LP)
0
0.3
mu(SP)
0.7
1
mu(ZE)
0.7
0.3
mu(SN)
0
0
mu(LN)
0
0
_________________________________
These values can then be plugged into the rules to give output values. Taking
the first rule:
= __________________
1.3
= -0.65 / 1.3
= -0.5
-- for the final control output. . Of course the hard part is figuring out what
rules actually work correctly in practice.
* If you have problems figuring out the centroid equation, remember that a
centroid is defined by summing all the moments (location times mass) around
the center of gravity and equating the sum to zero. So if X0 is the center of
gravity, Xi is the location of each mass, and Mi is each mass, this gives:
0 = ( X1 - X0 ) * M1 + ( X2 - X0 ) * M2 + ... + ( Xn - X0 ) * Mn
0 = ( X1 * M1 + X2 * M2 + ... + Xn * Mn ) - X0 * ( M1 + M2 + ... + Mn )
X0 * ( M1 + M2 + ... + Mn ) = ( X1 * M1 + X2 * M2 + ... + Xn * Mn )
( X1 * M1 + X2 * M2 + ... + Xn * Mn )
X0 = _____________________________________
( M1 + M2 + ... + Mn )
In our example, the values of mu correspond to the masses, and the values of
X to location of the masses.
"Fuzzy Controller
November 1995.
Challenges
8-bit
MCUs",
COMPUTER
DESIGN,
===========================================================
Overview of the artificial neural networks and fuzzy logic applications in
operational hydrological forecasting systems Damage due to flooding has
increase in many countries in the last years, and due to the global climate
change, which is now recognized as a real threat, an increase in the occurrence
of flooding events and especially of flash flooding events is likely to continue
into the future. In those conditions and because building new flood defenses
structures for defending vulnerable areas has serious financial implications,
the timely forecasting of floods is becoming more important for flood defense
and in general for water management purposes. The complexity of natural
systems and of hydrological processes that influence river levels evolutions
make the traditional modeling approaches, based on mirroring natural
processes with physically based equations very difficult. Despite the fact that
in the last decades the Operational Hydrological Forecasting Systems were
significantly developed, becoming more and more complex systems, ingesting
and processing in real time a great amount of data from automated
hydrometrical and meteorological stations networks and high resolution
gridded data from radars and satellites, together with the use of distributed
hydrological models, the warning and forecasts improvements are not very
significant, in many cases the performance of the new physically based
distributed models being comparable with the older conceptual lumped
models. The paper presents an overview of some alternative and
complementary modeling approaches, artificial neural networks and fuzzy logic
systems, possible applications for the improvements of the Operational
Hydrological Forecasting Systems, and presenting also some example of
rainfall-runoff modeling implementations. Artificial neural networks are widely
used as an effective approach for handling non-linear and noisy data,
especially in situations where the physical processes relationships are not fully
understood and they are also particularly well suited to modeling complex
systems on a real-time basis. Fuzzy logic is a generalization of Boolean logic
implementing the concept of partial truth or uncertainty, so within the fuzzy
set theory an element can have a gradual membership to different sets. To
describe system behavior with fuzzy logic, you need to define fuzzy sets, fuzzy
rules or so called IF-THEN rules and apply a fuzzy inference scheme. The
generation of a fuzzy forecast model can be based both on experts knowledge
and historical data. In conclusion, both artificial neural networks and fuzzy
logic modeling systems offer the potential for a more flexible, less assumption
approach to hydrological processes, and they have already been demonstrated
as successfully substitutes for the classical rainfall runoff models, and also
as tools for the real time updating of hydrological forecasting models and
especially for the multimodal approach. Keywords: hydrological forecasting
model, artificial neural network, fuzzy logic, operational hydrological forecasting
systems Operational hydrological forecasting systems On a global scale, floods
account for over 65% of people affected by natural disasters and they are the
most damaging of all natural disasters. Better forecasting floods and with a
larger lead time, is the main sustainable way of adapting to and managing such
disasters. Operational hydrological forecasting systems, which link state of the
river catchments, river discharges and water levels, recorded precipitations and
weather forecasts, can be used to respond to floods as they occur and to reduce
their costs in term of lives, property and other damages.
Current flood forecasting and warning systems have several limitations, such
as, insufficient lead-time to provide accurate flood warnings, inadequate spatial
and temporal resolution of the real-time rainfall observations and forecasts for
flood producing storm, little integration of different sources of forecast
information. Moreover their ability in considering the uncertainties in
estimating and forecasting precipitation and flood discharges is very limited.
The desirable characteristics of a good flood forecasting system are:
Timeliness: The lead-time is the time between making a forecast of an event
and its occurrence, if sufficient lead-time is available and the predictions are
accurate then evacuation, even of relatively large numbers of people may be
possible. The increase of the lead-time is mainly limited by the availability of
reliable quantitative precipitations forecasts, but also can be limited by the
hydrological models or forecasting methodologies that are implemented.
Accuracy: Is usually related to the correctness the forecasts of the magnitude
and time of the flood peak and of the resulting levels. In special situations, it
may relate to the forecasts of the complete hydrograph of the flood. The more
accurate the forecast the better flood control/modification and damage
mitigation measures can be implemented. Reliability: Can be associated with
Carling A., 1992: Introducing Neural Network. SIGMA PRESS - Wilmslow, United
Kingdom. Michael A. Arbib, 1995: The Handbook of Brain Theory and Neural
Networks. THE MIT PRESS, Massachusetts Institute of Technology, United States
of America.
Toderean G., Costeiu M., Giurgiu M., 1995: Artificial neural networks, Blue Press,
Cluj-Napoca.
Romania
Dumitrescu D., Hariton C., 1996: Neural networks, theory and applications, Ed.
Teora, Bucharest,
Romania.
French, M.N., Krajewski, W.F. and Cuykendall, R.R., 1992: Rainfall forecasting in
space and time
using a neural network, Journal of Hydrology, 137, 1-31.
Govindaraju, R.S., Anmala, J., Nedunuri, K.V. and Koelliker, J.K., 1995: Use of
neural networks in
predicting the nonlinear rainfall-runoff relationships for small watersheds, paper
presented at the
European Geophysical Society General Assembly, Hamburg, 1995 and
abstracted in the conference
proceedings.
Liong, S.Y., Nguyen, V.T.V., Chan, W.T. and Chia, Y.S., 1994: Regional Estimation
of Floods for
Ungaged Catchments with Neural Networks. In Cheong, H-F., Shankar
Maier, H.R. and Dandy, G.C., 1996: The use of artificial neural networks for the
prediction of water
quality parameters, Water Resources Research, 32, 4, 1013-1022.
Wilby, R.L., Abrahart, R.J., Dawson, C.W., 2003: Detection of conceptual model
rainfall-runoff