A robust geometric alignment scheme is proposed to monitor the status of flexible manipulators. The performance of the system is evaluated continuously and if the system is faulty, the fault is detected. isolated and the severity of the fault is estimated. Status of the system manifests as variations in a set of parameters termed the diagnostic parameters. The feature vector which is a vector formed of the coefficients of the system transfer function, is assumed to be a multilinear function of these diagnostic parameters and is estimated by performing a number of identification experiments; each diagnostic parameter is subject to some known perturbation and the feature vector is estimated by minimising a weighted frequency response error. The feature vector estimates are validated by verifying the multilinearity with respect to the diagnostic parameters. The influence of these diagnostic parameters on the feature vector is captured in a vector termed influence vector. The distance between (and the inner product of) the feature vector and the influence vector are analysed to diagnose faults. The proposed scheme is evaluated on simulated and actual control systems.