11.07.2015 Views

Understanding Smart Sensors - Nomads.usp

Understanding Smart Sensors - Nomads.usp

Understanding Smart Sensors - Nomads.usp

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

156 <strong>Understanding</strong> <strong>Smart</strong> <strong>Sensors</strong>Membership1Very lowLowNormal High Very high0.70.50.300 65% 100%Pressure / vacuumFigure 7.4 Fuzzy logic membership functions for pressure control.of 0.7 assigned for “greatly.” For low pressure, the rule could be: If the pressureis low, then increase the power slightly, with a weight of 0.3 assigned for“slightly.”The final part of the fuzzy logic process is defuzzification. The defuzzificationprocess takes a weighted average to translate the fuzzy outputs into a singlecrisp value. Output membership functions are typically defined bysingletons. A singleton in an 8-bit system is defined as an 8-bit value that representsthe output value corresponding to the linguistic label of the outputsystem.To simplify the implementation of fuzzy logic on existing MCUs, a fuzzykernel can be developed by the MCU manufacturer [12]. A fuzzy kernel, orengine, is software that performs the three basic steps of fuzzy logic: fuzzification,rule evaluation, and defuzzification. As shown in Figure 7.5, thefuzzy-inference unit (FIU) receives system inputs and information from theknowledge base for each step of the process. All application-specific informationis contained in the knowledge base that is developed independently fromthe fuzzy-inference program.With fuzzy logic, applications that were once thought to be too complexto be practical are being easily controlled. Furthermore, the rules in a fuzzy systemoften hold true even if the operating parameters of the system change. Thattypically is not true in conventional control systems.In general, fuzzy logic does not require special hardware. The softwarerequirements can be low based on a limited number of rules to describe the system.For most control tasks, software running on standard processors canperform fuzzy logical operations. Specialized hardware does simplify codedevelopment and boost computational performance for low-cost applications.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!