Real time adaptive parametric equalization of Ultrasonic Transducers

Parametric equalization is often used to achieve a desired response from an audio transmitter, but is rarely applied to ultrasonic transducer systems. The ability of a broadband ultrasonic transmission and reception system to adapt its frequency and time domain response to changing acoustic conditio...

Full description

Saved in:
Bibliographic Details
Main Authors: McSweeney, S.G., Wright, W.M.D.
Format: Conference Proceeding
Language:English
Subjects:
Acoustic applications
Adaptive equalizers
Artificial neural networks
Filters
Frequency response
Real time systems
Remote monitoring
Signal to noise ratio
Transmitters
Ultrasonic transducers
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Parametric equalization is often used to achieve a desired response from an audio transmitter, but is rarely applied to ultrasonic transducer systems. The ability of a broadband ultrasonic transmission and reception system to adapt its frequency and time domain response to changing acoustic conditions would be a distinct advantage in certain applications. Ultrasonic remote monitoring systems would benefit significantly from this ability, as signal levels could be minimized and consequentially the transmitter power consumption decreased. This work presents a real-time adaptive ultrasonic parametric equalizer using optimization driven Matlab code to control the coefficients of a switched capacitor filter network implemented in a Cypress PSOC (Programmable System On a Chip). In this work, adaptive parametric magnitude equalization of a through-transmission ultrasonic system using CUTs (Capacitive Ultrasonic Transducers) has been achieved in real time by tracking a desired SNR (signal to noise ratio) across the operational frequency spectrum. A Matlab general radial basis function (GRBF) artificial neural network (ANN) was developed to control the equalization filter coefficients based on the received frequency response data. The adaptive parametric equaliser adjusts the magnitude of the driving signal to maintain the desired SNR as closely as possible. The neural network was trained using PSO (Particle Swarm Optimization) back-propagation, based on a state space model of the system developed from frequency response data. The developed equalization circuitry, which is switched capacitor based and was fully implemented on the PSOC, is also described.
ISSN:1051-0117
DOI:10.1109/ULTSYM.2009.5441886