Sound Speed Estimation Using Fuzzy Logic Approach for Outdoor Ultrasonic Applications

Abstract

AbstractThe accuracy of sound speed estimation has direct influence on performance of ultrasonic sensing applications, like those based on the distance measuring and the sound wavelength. The sound speed in air is affected by changes in the air properties, such as temperature, pressure, humidity, gas composition, and air turbulence. For many indoor applications, the variations of these properties are small, and the sound speed in air is considered as constant. For outdoor environments, the air characteristics have wide variations, which can not be neglected. In this case, a constant sound speed generates important errors of distance estimation. In addition, for low temperature values, the real distance is smaller than measured distance, and this can be a dangerous situation.In this paper, the variations of sound speed in air as a function of air properties are studied, along with their influence on the accuracy of ultrasonic sensing. The most important air characteristic is air temperature, while the air pressure and relative humidity affect the sound speed, especially at high temperature values. The influence of CO2 concentration on the sound speed is very small. Fuzzy rules are generated for sound speed values used in outdoor applications.Also, fuzzy estimation of sound speed is studied, using expert rules generated from the sound speed model. Different fuzzy systems were tested, with various membership functions and fuzzy rules. The selection was made based on the relative error and the mean square error of the fuzzy output, compared with the output of sound speed model. Accurate estimation of sound speed is obtained. The output surface and the relative error of the selected fuzzy estimator are also presented.KeywordsMembership FunctionMobile RobotFuzzy RuleSound SpeedOutdoor EnvironmentThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.