< back to main site


Modelling acoustic signals in the calibration of underwater electroacoustic transducers in reverberant laboratory tanks.

Robinson, S P; Harris, P M (1999) Modelling acoustic signals in the calibration of underwater electroacoustic transducers in reverberant laboratory tanks. NPL Report. CMAM 29

[img] Text

Download (730kB)


In the calibration of an underwater electroacoustic transducer, the stepped-sinusoidal signals that are commonly used are contaminated by transients due to the resonant behaviour of the transducer, as well as noise. In addition, when calibrations are undertaken in reverberant laboratory tanks of finite size, reflections from the tank boundaries often arrive before the steady-state response of the transducer can be observed directly. However, by use of suitable signal models, it is possible to predict the steady-state response from the initial transient-dominated part of the waveform.
The approach considered here is to model the free-time response of the device by a function consisting of a sum of complex exponential terms which are used to describe both the steady-state and resonant behaviour of the device. In this report, we review two classes of estimation method: linear prediction methods (such as Prony’s method and its variants), and nonlinear least-squares methods. Standard linear prediction methods are shown to be statistically biased and inefficient. This observation motivates the development of a linear prediction method in which proper account is taken of the error structure in the data. A nonlinear least-squares algorithm is also described based on a safeguarded Gauss-Newton algorithm that uses regularisation to address the ill-conditioning that is a property of the underlying problem.
Results are presented of using these methods to analyse simulated data generated to represent the measured response of a device. The effect of varying in a systematic manner the properties of the simulated data and device under test is investigated. Results are also presented for data obtained from measurements of a real transducer, and it is shown how these results may be used to establish the free-field sensitivity of the device. Finally, results are presented of using the information gained from the modelling of the transducer resonant behaviour to predict the transducer frequency response at frequencies other than those under test.
This report constitutes the deliverable of project 2.1d(ii) of the 1995-98 NMS Acoustics Metrology Programme.

Item Type: Report/Guide (NPL Report)
NPL Report No.: CMAM 29
Subjects: Acoustics
Acoustics > Underwater Acoustics
Last Modified: 10 Sep 2018 13:57
URI: http://eprintspublications.npl.co.uk/id/eprint/1092

Actions (login required)

View Item View Item