Unsupervised fault detection of forest harvester head functions

Abstract It is difficult to come up with working solutions for the fault detection of forest harvesters due to the extreme variability in their operating conditions. This text studies the feasibility of implementing an automatic fault detection scheme based on an existing index hierarchy system. This index computation system estimates the performance of different subsystems of the harvester with dimensionless numbers. These indices can be used as feature vectors for traditional detection methods like the T 2 -statistic or methods based on clustering. The implemented methods are compared by using them on artificial simulated fault data.