Techniques for early warning of slight changes in systems and plants a
re useful for condition-based maintenance. In this paper we present an
approach for this problem. This approach is based on the so-called 'a
symptotic local' approach for change detection previously introduced b
y some of the authors. Its original principle consists in characterizi
ng a system via some identified model, and then to monitor its changes
using some data-to-model distance also derived from identification te
chniques. We show here that this method is of much wider applicability
: model reduction can be enforced, biased identification procedures ca
n be used, and finally one can even get rid of identification and use
instead some much simpler Monte-Carlo estimation technique prior to ch
ange detection. Experiments on AR models are reported and an example f
rom gas turbine industry is briefly discussed.