Dynamic performance degradation prediction of composite laminates remains a major challenge in reliable structural health assessment because the degradation process is high-dimensional, nonlinear, and strongly affected by material variability and data noise. Existing point prediction methods can estimate an expected degradation value, but they usually cannot quantify the uncertainty required for trustworthy maintenance decisions. Meanwhile, many interval prediction strategies treat point estimation and interval estimation as independent tasks, which weakens the intrinsic information sharing between prediction accuracy and uncertainty reliability. To address these limitations, this paper proposes a Fuzzy Inference System inspired point-interval joint framework for high-fidelity composite performance degradation prediction. Natural Gradient Boosting (NGBoost) is first adopted to establish a probabilistic degradation mapping and generate initial point and interval prediction results. Then, a Fuzzy Inference System (FIS) is introduced to adaptively calibrate the uncertainty interval and improve the final prediction accuracy. Grid search is further used to optimize key model hyperparameters. Experimental and finite-element degradation datasets of composite laminates are used for validation. The results show that the proposed framework can achieve accurate point prediction while maintaining reliable and compact prediction intervals, providing an uncertainty-aware basis for composite structural health assessment.

Composite laminates, performance degradation prediction, point-interval joint prediction, fuzzy inference system, uncertainty quantification