The need for accurate emissions measurements has coerced researchers into trying
to reconstruct the true transient emission signal from that measured by the
analyser. This paper discusses two such methods and examines the validity of
those methods by testing them with real-time emissions data. The first method is
the sequential inversion technique, which tries to reconstruct the input second
by second, based on the measured response at each second and the dispersion
characteristics of the analyser. The reconstruction was found to be accurate,
but there were some constraints associated with the dispersion characteristics
and the reconstruction failed if there was signal noise. The second method, the
differential coefficients method (DCM) of Ajtay and Weilenmann, reconstructs the
input signal by approximating the analyser input as a linear combination of the
output and the output derivatives. When tested with real-time data, the DCM
predicted the emission signal even when there was noise imposed on the signal.
While the DCM is clearly a better prediction technique, the accuracy of the DCM
is reduced when noise is added to the analyser input. The DCM, when coupled with
cross-correlation techniques, can be a powerful tool in retrieving
‘lost’ information associated with the measurement delays
and dispersion characteristics of the analyser.