A quantitive fault diagnosis system for backfill hydrocyclones using m
odelling and parameter estimation was developed. The system was evalua
ted using simulation, thereby allowing arbitrary process variable faul
t levels and combinations of faults to be introduced into the system,
as well as controlled levels of random noise in the process output mea
surements. The fault diagnosis system determines that point in the sea
rch space that minimises the difference between the measured process o
utputs and the process outputs predicted by previously determined proc
ess models by simultaneously adjusting the model parameters. At the mi
nimum, estimates of the unmeasurable process variables are provided by
the model parameters. A gradual increase in the feed - 75 micron frac
tion was considered, as well as a simultaneous increase in both the vo
rtex finder and spigot diameters. Faults could clearly be identified b
y observing the trends in inferred variable values over time. Subjecti
ve decisions as to the level of incipiency of faults can be reduced by
performing quantitative alarm analysis on these values. Both non-para
metric and parametric alarm analysis was evaluated and it was found th
at a combination of these methods would enable timeous detection of bo
th gradual and abrupt faults. The use of quantitative fault diagnosis
to detect incipient faults in process parameters could be of great val
ue in maintaining production levels, planning maintenance and optimisi
ng system performance.