The design and analysis of fault diagnosis architectures using the mod
el-based analytical redundancy approach has received considerable atte
ntion during the last two decades. One of the key issues in the design
of such fault diagnosis schemes is the effect of modelling uncertaint
ies on their performance. This paper describes a fault diagnosis algor
ithm for a class of nonlinear dynamic systems with modelling uncertain
ties when not all states of the system are measurable. The main idea b
ehind this approach is to monitor the plant for any off-nominal system
behaviour due to faults utilizing a nonlinear online approximator wit
h adjustable parameters. The online approximator only uses the system
input and output measurements. A nonlinear estimation model and learni
ng algorithm are described so that the online approximator provides an
estimate of the fault. The robustness, sensitivity, stability and per
formance properties of the nonlinear fault diagnosis scheme are rigoro
usly established under certain assumptions on the failure type. A simu
lation example of a simple second-order system is used to illustrate t
he robust nonlinear fault diagnosis scheme.