DocumentCode
150305
Title
Self-commissioning of inverter dead-time compensation by multiple linear regression based on a physical model
Author
Bedetti, Nicola ; Calligaro, Sandro ; Petrella, Robert
Author_Institution
Gefran, Gefran, s.p.a., Varese, Italy
fYear
2014
fDate
14-18 Sept. 2014
Firstpage
242
Lastpage
249
Abstract
Dead-times and switch voltage drops represent the most important sources of distortion of the (average) output voltage in PWM inverters. Their effect is a function of the parameters of the drive system and of the operating conditions, and is often intolerable in many drives applications, thus requiring a proper compensation strategy. Many techniques are implemented in industrial drives and reported in literature, even very recently. Differently from standard approaches the proposed methodology is based on a detailed physical model of the power converter (including output capacitance), described by a small set of parameters. A novel self-commissioning identification procedure is proposed, adopting Multiple Linear Regression. The technique is tested on a commercial drive in comparison to state-of-the-art techniques. Also back-EMF estimation improvements in a PMSM sensorless drive system are shown to provide additional validation of the method.
Keywords
PWM invertors; PWM power convertors; electric potential; permanent magnet motors; regression analysis; synchronous motor drives; PMSM sensorless drive system; PWM inverters; back-EMF estimation improvements; compensation strategy; dead-time compensation; industrial drives; multiple linear regression; operating conditions; physical model; power converter; self-commissioning identification procedure; standard approaches; switch voltage drops; Capacitance; Equations; Insulated gate bipolar transistors; Inverters; Mathematical model; Phase distortion; Switches;
fLanguage
English
Publisher
ieee
Conference_Titel
Energy Conversion Congress and Exposition (ECCE), 2014 IEEE
Conference_Location
Pittsburgh, PA
Type
conf
DOI
10.1109/ECCE.2014.6953400
Filename
6953400
Link To Document