This work includes multivariate data analysis of process and sensor data by means of variable and spectral pre-processing and methods of variable selection for regression approaches of ultrasonic features (EPO, PLS, Kernel-PLS, SVR) as well as classification of NIR spectra (PLS-DA). Furthermore, adapted model population analysis for more robust models as well as algorithms of multivariate process control for process classification and in combination with swarm intelligence for sensor network control and process monitoring were studied.
«
This work includes multivariate data analysis of process and sensor data by means of variable and spectral pre-processing and methods of variable selection for regression approaches of ultrasonic features (EPO, PLS, Kernel-PLS, SVR) as well as classification of NIR spectra (PLS-DA). Furthermore, adapted model population analysis for more robust models as well as algorithms of multivariate process control for process classification and in combination with swarm intelligence for sensor network con...
»