ICS for multivariate functional anomaly detection with applications to predictive maintenance and quality control

Multivariate functional anomaly detection has received a large amount of attention recently. Accounting both the time dimension and correlations between variables is challenging due to existence different types outliers data. In context predictive maintenance quality control, data sets often contain number variables. However, most existing methods focus on small Moreover, in fields that have high reliability standards, detecting potential multivariate with as few false positives possible crucial. such context, adaptation Invariant Coordinate Selection (ICS) method from case particular interest. Two extensions ICS are proposed: point-wise global. For methods, choice relevant components together outlier identification interpretation discussed. A comparison made example avionics field control microelectronics field. It appears global selected can be recommended.