AbstractThis thesis examines the role that composite frequency test signals play in system identification. The advantages and disadvantages of composite frequency response analysers over conventional analysers are discussed. Two low peak-factor composite frequency test signals are reviewed which largely overcome the problems associated with frequency response analysers which employ more established test waveforms.
A novel, high performance frequency response analyser incorporating these two composite waveforms has been fully designed and developed. The instrument which makes use of a 256 point Cooley-Tukey Fast Fourier Transform, demonstrates the feasibility of producing an analyser based upon the new waveforms which has a performance equal to, or greater than existing commercial analysers.
The relative performance of the two test waveforms has been evaluated by applying them to a number of real systems and comparing the results with those obtained when using a commercial monotonic analyser. In addition to linear systems, tests have been carried out on systems exhibiting non-linear behaviour in an attempt to establish the ability of the instrument to minimise harmonic distortion in the measurement process.
The theoretical basis for the frequency response analysis has been developed and the potential problems associated with aliasing have been analysed. In order to avoid the complexity and expense associated with anti-aliasing filters however, no such filter has been incorporated within the instrument. In place of a filter, a new anti-aliasing compensation technique has been developed. This technique is largely based upon a software algorithm and offers significant advantages over existing analogue/digital filtering techniques. The theoretical basis for the technique is developed, and a method for its mechanisation within the developed instrument is presented.
This anti-aliasing algorithm has been incorporated within the analyser and has been applied to the frequency response measurement of known systems using a composite frequency test waveform. The results are compared to those obtained without using the algorithm to demonstrate the effectiveness of the new technique.
|Date of Award||Jul 1990|