Behavioral Modeling of Motor Drive Systems Efficiency

Publication Year:
2016
Usage 137
Downloads 104
Abstract Views 33
Repository URL:
https://opencommons.uconn.edu/gs_theses/927
Author(s):
Ulatowski, Artur
Tags:
motor drive systems; efficiency modeling; data driven methods; electric motor drive systems
artifact description
While power losses and efficiency of single components of electric motor drive systems were researched extensively, the overall system efficiency has been rarely approached. The wide-spread applications of electric machines make them the largest electrical energy consumers, thus the efficiency of motor drive systems is of fundamental effect on global energy consumption. Analytical efficiency modeling of motor drive systems is a very difficult task due to complex interactions in a motor drive system (electrical, mechanical, magnetic, thermal, etc.). In this thesis, a novel method of applying black-box approaches for motor drive efficiency modeling is presented. The proposed methodology is flexible and can be applied to any motor drive system where input and output powers and the control parameter can be measured or accurately estimated. The black-box methods do not model specific physical phenomena governing the system, but simply model the input-output relationship. The black-box modeling approach uses actual physical efficiency measurements of an experimental motor drive system. Such measurements include all possible interactions within the system, thus making the model comprehensive. Different methods of behavioral comprehensive black-box modeling approaches are presented and their accuracy is evaluated. Models are further benchmarked by assessing their ability to predict the value of the control parameter for maximum-efficiency operating points. One of the behavioral models is implemented on an experimental control platform and used to predict the control parameter for optimal efficiency operation. Finally, the developed models are scrutinized by comparing them to the current analytical knowledge.