|
Fuzzy Control
Fuzzy control is automatic control with rules rather than equations. An advantage of fuzzy control over
a conventional approach is that the control strategy is in words rather than an equation, it makes it
easier to understand for operators. A heating unit for a room, for instance, could be controlled in the
conventional way by an equation that adjusts the power in proportion to the deviation of the temperature from the desired temperature. A similar fuzzy controller could contain a rule
If temperature is low then turn heat up
The fuzzy rule improves readability of the control strategy at the expense of compactness. Fuzzy control is based on fuzzy logic, which in short is computing with words rather than numbers.
|
|
Block diagram of a fuzzy control system. The difference from a conventional control system lies in the rule base and the inference engine. An advantage
is that the end-user is able to read the rules. |
|
Fuzzy control is applied in consumer products such as washing machines, video cameras, and cars, as
well as in industry for controlling cement kilns, underground trains, and robots. Fuzzy logic (without
control) is also being applied within image processing, decision support, medical diagnosis, and in the financial sector.
Lotfi Zadeh introduced the idea of fuzzy sets in 1965 as a mathematical way to represent vagueness.
Fuzzy set theory is a generalisation of set theory since membership of a fuzzy set can be any real
number between zero and one, rather than just zero or one — the membership is gradual rather than crisp.
|
|
Membership functions describing a warm room. A crisp membership function defines the room as warm if the temperature is higher than 21 degrees (solid
curve), while a fuzzy membership function defines it as more or less warm (dashed curve). |
|
A fuzzy set is defined by means of a membership function which returns truth values between 0 and 1.
An element of the universe of discourse is a member to a degree, defined by the membership function.
The EUNITETRAIN server
The server is an educational server on the Internet concerning applications and technology related to
EUNITE, a European network of excellence. It serves as a learning central for students and professionals working with intelligent technologies. For example, through the server it is possible to
enroll in an online course in fuzzy control and access downloadable teaching material from an electronic library.
 31360 Fuzzy, Neural, And Adaptive Control
31361 Fuzzy Control (Internet Course)
EUNITETRAIN: courses, tutorials, demo, library
Projects
Publications
Teaching activities
Contact Jan Jantzen (homepage) <jj@oersted.dtu.dk>
|
