Authors:
Wolfgang Mergenthaler
1
;
Jens Feller
1
;
Bernhard Mauersberg
1
and
Roger Chevalier
2
Affiliations:
1
FCE Frankfurt Consulting Engineers GmbH, Germany
;
2
Electricité de France and R&D, France
Keyword(s):
Multidimensional Response Surfaces, Nonlinear Optimization, Dynamic Programming, Gaussian Shapes, Adaptive Control, Statistical Learning.
Related
Ontology
Subjects/Areas/Topics:
Adaptive Signal Processing and Control
;
Informatics in Control, Automation and Robotics
;
Intelligent Control Systems and Optimization
;
Machine Learning in Control Applications
;
Optimization Algorithms
;
Optimization Problems in Signal Processing
;
Signal Processing, Sensors, Systems Modeling and Control
Abstract:
Technical processes, notably in the power transforming industries, generate a wealth of process data, commonly organized in a file with M records and 1 + n + m fields, i.e. a time stamp, followed by n independent and m dependent variables, summarized in the vectors x and y, respectively. Regardless of the availability of physical models it is interesting and often necessary to generate functional relationships between x and y from process data. The most prominent purpose is the optimization of certain performance indices under given constraints. This paper describes response surface estimation using Gaussian shapes along with finding optimal points on the surfaces to be used in machine control. The practical impact lies in the usability of this technique to increase machine efficiency on a broad industrial scale with its applications towards energy efficiency and climate protection.