Classification is a key task in dynamic systems monitoring and control. It enables to detect the current configuration of the system from a list of possible configurations that have been used in the learning step of the algorithm design. Classification can therefore be used in fault detection and isolation, but also in automatic detection of events and situations. This detection can then be fed to other algorithms (controller or state estimators) in order to better achieve their task.DiagSign is an advanced tool that enables a fast prototyping of supervised classification algorithms. It is model-free since it is based on the search for discriminant signatures that may be created from sensor measurements. The signatures are created based on a recently developed framework by the author.This book is a user manual of DiagSign.

It describes in an example based step-by-step way, how DiagSign can be used to rapidly derive a classification algorithm starting from available experimental or simulation based data files.