Probabilistic fatigue estimation framework for aeroengine bladed discs with multiple fuzziness modeling

Abstract

To reduce the estimation errors of probabilistic fatigue lifetime caused by artificial cognitive factors, a probabilistic fatigue estimation framework with the consideration of multiple fuzziness (i.e., stress and strength) is proposed. In the presented framework, a fuzzy variable randomization-based fuzzy least squares support vector regression is proposed in the level of stress fuzziness, a fuzzy strength model with average stress calibration is established in the level of strength fuzziness, and the corresponding sampling-based probabilistic fatigue estimation framework is given. By regarding a typical compressor bladed disc with titanium-based superalloy as a case, the proposed framework is validated. Methods comparison shows that the proposed approach holds the highest estimation accuracy compared with the methods that do not consider fuzziness of stress or strength. The current efforts develop a novel approach to evaluate the probabilistic fatigue lifetime by fuzzy set theory, which also sheds a light on the reliability-based fatigue design of complex structures.

Keywords

• High-cycle fatigue
• Fuzzy reliability
• Bladed disc
• Surrogate model
• Support vector regression