Multivariate Data Analysis and Soft Sensors SCE2106
Course Objectives
Multivariate data analysis; design of factorial experiments. Multivariate data analysis including chemometric methods. Introduction to PCA, MLR, PCR and PLS-R. Artificial Neural Networks, Fuzzy Logics and soft sensors. Case studies of Soft sensors. The course is based on extensive use of applications.
Course Description
Multivariate Data Analysis: Experimental design, significant testing, factorial design, fractional factorial design, MK and response surface methods, regression, MLR, PCA (principal component analysis) multivariate calibration: PCR (principal component regression), PLS-R (partiel least squares regression), overview of multivariate data analytisis methods, Applications and examples. Computer assignments.
Soft sensors: Wherever several (even hundreds or thousands of) measurements are processed together, the soft sensor approach is a viable approach. Fusion of signals can be used for estimating new quantities without actually measuring them. Soft sensors are especially useful in multi sensor data fusion, where measurements of diverse characteristics and dynamics are fused to get an overall view of the system under study. Multi Sensor Data Fusion (MSDF), Soft sensor technology. Steps in soft sensor technology: data collection, data pre-processing, variable and time delay selection, calibration, and validation.
Learning Methods
Lectures, tutorials, computer simulations, student presentations and laboratory assignments.
Assessment Methods
Intermediate evaluations count 40% while an individual written final test counts 60% in the final grade. no study aids are permitted during the final test.
Minor adjustments may occur during the academic year, subject to the decision of the Dean
Publisert av / forfatter Unni Stamland Kaasin <Unni.S.KaasinSPAMFILTER@hit.no> - 03/03/2009