Composite laminates reliability assessment using diffusion process backed up by perspective forms of non-parametric kernel estimators

•Laminate composites reinforced by natural fibers and their reliability measures.•Complex reliability assessment of real testing records data on laminate composites.•Novel approach combining weighted regression, diffusion process and kernel estimates.•Modeling the hazard function using cooperating parametric and non-parametric models.•Practical applicability of suggested results on composites dependability performance. Numerous technical systems are modeled to identify dependability and safety measures. The modeling approach here is a real data driven approach. We work with the results obtained by assessing a progressive material—natural fibers reinforced laminate composites. The knowledge of certain properties, namely reliability performance, has not been sufficiently studied. Our approaches include the Wiener diffusion process, parametric and non-parametric estimations and mainly a hazard function which is estimated parametrically and non-parametrically by kernel estimators. We also demonstrate the significance between parametric and non-parametric estimations, the way they influence the selection of a resulting model and prospective practical applicability.