Modelagem dinâmica em espaço de estados de sinais de ultrassom para a estimação do tempo de trânsito utilizando filtro de Kalman.

Abstract

Ultrasound transducers are applied in several systems to measure flow velocity, distance and surface wear, by estimating the ultrasound travel time, from its emission in the transmitting transducer to its detection in the receiving transducer (time of flight), presenting high speed response and good accuracy. Techniques based on time difference and signal processing have been used to time of flight estimation in order to reduce uncertainties. The beginning of signal detection in the receiving transducer for time of flight estimation is hampered by the noise levels of the medium, whose amplitudes are confused with the amplitudes that represent the signal detection. A methodology for ultrasound time of flight estimation applied to wind speed measurement is proposed, based on the extended Kalman filter that uses an estimation of stretch of the ultrasound signal for time of flight indirect measurement, from measurements of the amplitude of the receiving transducer signal and the time of flight initialization value. For time of flight initialization, that is a necessary step to use the Kalman filter, was proposed a methodology that identifies a voltage threshold value of the receiving transducer signal (from which the initial time of flight is determined) to be used. Among the techniques that are used to measure the time of flight, its limitations refer to the measurement range restricted to a signal cycle, the need to find a mathematical function that describes the ultrasound signal or sections of it and middle noise levels. Based on this method, a computational model for the propagation of ultrasound waves was built and experimental procedures were performed using a wind tunnel, in order to apply the algorithm with the Kalman Filter in the time of flight indirect measurement. In simulation, the time of flight was obtained with the phase difference technique in the time domain after the use of Kalman filter as a signal filtering step, in scenarios with signal-to-noise ratio (SNR) values less than 35 dB, for which the conventional techniques of detection of threshold and phase difference in the time domain cannot be applied directly. The Kalman filter was used for time of flight indirect measurement, join as the methodology that identifies a proper voltage threshold value of the receiving transducer signal to be used, obtaining in simulation and experimental results a good performance of the proposed technique, with values of time of flight with accuracy comparable to those obtained with the difference technique in frequency domain, however without the restriction of measurement range limited to 2π (one signal cycle).