1. Lifespan Modeling of Low Voltage Machines Insulation Materials
- Author
-
Marie Chabert, David Malec, Pascal Maussion, Antoine Picot, Institut National Polytechnique de Toulouse - Toulouse INP (FRANCE), Centre National de la Recherche Scientifique - CNRS (FRANCE), Université Toulouse III - Paul Sabatier - UT3 (FRANCE), Université Toulouse - Jean Jaurès - UT2J (FRANCE), Université Toulouse 1 Capitole - UT1 (FRANCE), and Pistre, Karine
- Subjects
Engineering ,Logarithm ,[SPI] Engineering Sciences [physics] ,Stress (mechanics) ,Traitement des images ,Traitement du signal et de l'image ,Electrical and Electronic Engineering ,Analysis of variance ,Synthèse d'image et réalité virtuelle ,Films ,Twisted pairs ,business.industry ,Design of experiments ,Electrical engineering ,Modeling ,Function (mathematics) ,Structural engineering ,Vision par ordinateur et reconnaissance de formes ,Intelligence artificielle ,Accelerated aging ,Test set ,Partial discharge ,Lifespan estimation ,Electrical insulation ,Response surface ,business ,Low voltage - Abstract
This paper deals with the modeling of insulation material lifespan in a partial discharge regime. Accelerated aging tests are carried out to determine the lifespan of polyester-imide insulation films under different various stress conditions. The insulation lifespan logarithm is modeled as a function of different factors: the electrical and frequency stress logarithms and an exponential form of the temperature. The model parameters are estimated on a training set. The significance of the factors is evaluated through the analysis of variance (ANOVA). In a first step, the design of experiments method (DoE) is considered. The associated lifespan model is linear with respect to the factors. This method is well known for reducing the number of experiments while providing a good accuracy. In a second step, the response surface method (RSM) is considered. This method takes also into account some second order terms and thus possible interactions between the stress factors. Performance of the two methods are analyzed and compared on a test set.
- Published
- 2015