Multivariate SPC using radial basis functions

Advances in computing and instrumentation have led to a major increase in the data logging capabilities of many modern chemical plants, which has in turn led to enhanced interest in dimensionally-reducing statistical techniques such as principal component analysis (PCA). The paper analyses linear PCA, and proposes a nonlinear extension of this methodology using a series of radial basis function (RBF) neural networks. Sample applications showing the benefits of using such a scheme are given, including a fault detection scenario on a validated model of an industrial overheads condenser and reflux drum plant.