Randy Cronk
What do pension plans, elevators, and battery chargers all have in common? Each is an important application of fuzzy logic, a style of computing that is slowly and, according to experts, inevitably replacing the binary-based processing now found in many of the world's applications.
"Any application that's logic-based can be fuzzy-based," says David Brubaker, a fuzzy-logic consultant for Huntington Advanced Technology (Menlo Park, CA). "There's more fuzzy logic around than you probably think, from your appliances at home, to the car you drive to work, to the heating system in your building," he says. However, Brubaker adds that many fuzzy-logic designers won't discuss their projects, "either because they don't want to give away any competitive advantage or they are afraid of losing sales if their cu
stomers found out."
Such fuzziness about fuzzy logic has stifled the market, according to Bonnie Packert, cofounder and executive vice president of HyperLogic (Escondido, CA), maker of CubiCalc, one of the first general-purpose fuzzy-logic-design tools. Like many companies in the business, HyperLogic has experienced slow growth. "I think part of the reason is a lack of exposure for fuzzy, especially in the press," Packert says.
However, Market Intelligence Research (Mountain View, CA) says that worldwide fuzzy-logic revenues were $2 billion in 1994, a number that will reach $8 billion by 1998. The research firm also says that U.S. and European companies are now taking an active interest in the field, following successful commercial fuzzy projects in Japan.
Fuzzy logic provides a way for computers to deal with the ambiguities of the real world. Rather than using numbers that are always true or not, fuzzy applications can use numbers that are very true, slightly true, or somewhere in betwee
n.
FuziWare (Knoxville, TN, (800) 472-6183 or (615) 588-4144; fax (615) 588-9487) holds a patent on using fuzzy mathematics in a spreadsheet. With the company's FuziCalc product, which can import data from Windows-based spreadsheets, you can create scientific, technical, and financial-projection models that use fuzzy logic.
A simple example of an application of fuzzy arithmetic is revenue forecasting, in which you take historical data, such as a previous year's revenue, and project it for future years using three likely growth rates to derive a triangular fuzzy number. You then project revenue by multiplying the base revenue by (1 + the fuzzy growth rate). "Knowing the relative degree of belief across the range is critical information," says Karl Thorndike, president of FuziWare. "Encapsulating that information into a single fuzzy number keeps you from throwing away information each time you do a calculation."
Fuzzy Logic Can Illuminate Modeling and Design
screen_link (21 Kbytes)
The MathWorks' (Natick, MA, (508) 653-1415; fax (508) 653-2997) Fuzzy Logic Toolkit ($895, available for Macs, Windows, and Unix) combines fuzzy-logic modeling with engineering-design tools that integrate with the company's Matlab technical-computing environment and Simulink graphical simulation environment.
Getting Clear About Fuzzy
