Statistical approach to endurance models - data processing using regression models

The paper presents a methodology for processing experimental data related to the issue of the electrical insulation system endurance, using the method of accelerated laboratory aging. Several possible approaches are proposed. Traditionally used is the phenomenological approach combined with mathematical regression of measured data. This method is complemented by confidence intervals to ensure statistical certainty of the results of the experiment statements. The paper also describes a statistical approach to endurance modeling, however similarly supplemented with statistical certainty by using confidence intervals. Statistical certainty based on confidence intervals brings very valuable information for the practical use of electrical insulation materials and systems, especially considering design and dimensioning.