基于自适游离聚类的复合材料多源多态不确定性辨识研究

To solve the challenges of intertwining inherent and cognitive uncertainties, multi-source heterogeneity of parameters, and complex and difficult-to-distinguish uncertainty forms of multi-source factors such as materials, manufacturing, service environment, and structural response of composite materials, this paper proposes a multi-source polymorphism uncertainty identification method based on self-adaptation excursion clustering (SEC). Firstly, the uncertainty attributes are decoupled by combining the data scale, statistical characteristics, and physical constraints, and the clustering centers of intrinsic, cognitive, and mixed uncertainties are generated by combining the labeling attributes, and the uncertainty boundaries are clarified to achieve qualitative classification. Secondly, the free evaluation strategy of fuzzy information entropy was introduced to effectively deal with the boundary ambiguity of the representation model and adapt to the best uncertainty characterization model. Finally, the identification and classification verification of polymorphic uncertainty is carried out for the multi-source factors of composite materials and their structures, and the uncertainty identification of materials, manufacturing processes, service environment, structural response, and other factors is carried out with high precision and high efficiency, and the identification accuracy reaches 95.5% with the support of 50% labeled data, which improves the scientific of the subsequent quantitative analysis of multi-source uncertainty characterization of composite structures.