Effects of Simulation Length and Flexible Foundation on Long-Term Response Extrapolation of a Bottom-Fixed Offshore Wind Turbine

This paper deals with statistical and modeling uncertainty on the estimation of long-term extrapolated extreme responses in a monopile offshore wind turbine. The statistical uncertainty is addressed by studying the effect of simulation length. Modeling uncertainty is explored by evaluating the effects of considering a rigid and flexible foundation. The soil's flexibility is taking into account by considering the improved apparent fixity method. To identify the most relevant environmental conditions, the modified environmental contour method is used. The analysis focuses on the fore-aft shear force (FASF) and the fore-aft bending moment (FABM) at the mudline. The results show that using a simulation length of 10-min, does not provide sufficient accuracy. It was found that for the FASF, simulation lengths of at least 30-min are required to achieve an accuracy of about +/-5%. For the FABM, it was found that both the extrapolations made with 20-min and 30-min simulations achieved similar levels of accuracy of about 20%. Meanwhile, the results obtained from 10-min simulations reached deviations of about 40%. Finally, from the comparison made between a rigid and flexible foundation, it was found that the extrapolated responses exhibit maximum deviations up to around 5% and 10% for the FASF and the FABM, respectively. Also, for the FABM, it was observed that the consideration of a flexible foundation causes the critical wind speed to shift from 16.5 m/s (rigid) to 18 m/s (flexible).

Experimental and Modeling Uncertainty Considerations for Determining the First Item Ignited in a Compartment Using a Bayesian Method

Fire scene reconstruction and determining the fire evolution (i.e. item-to-item ignition events) using the post-fire compartment is an extremely difficult task because of the time-integrated nature of the observed damages. Bayesian methods are ideal for making inferences amongst hypotheses given observations and are able to naturally incorporate uncertainties. A Bayesian methodology for determining probabilities to items that may have initiated the fire in a compartment from damage signatures is developed. Exercise of this methodology requires uncertainty quantification of these damage signatures. A simple compartment configuration was used to quantify the uncertainty in damage predictions by Fire Dynamics Simulator (FDS), and a compartment evolution program, JT-risk as compared to experimentally derived damage signatures. Surrogate sensors spaced within the compartment use heat flux data collected over the course of the simulations to inform damage models. Experimental repeatability showed up to 4% uncertainty in damage signatures between replicates . Uncertainties for FDS and JT-risk ranged from 12% up to 32% when compared to experimental damages. Separately, the evolution physics of a simple three fuel package problem with surrogate damage sensors were characterized in a compartment using experimental data, FDS, and JT-risk predictions. An simple ignition model was used for each of the fuel packages. The Bayesian methodology was exercised using the damage signatures collected, cycling through each of the three fuel packages, and combined with the previously quantified uncertainties. Only reconstruction using experimental data was able to confidently predict the true hypothesis from the three scenarios.

Analyzing of Process Capability Indices based on Neutrosophic Sets

Process capability analysis (PCA) is an important statistical analysis approach for measuring and analyzing the ability of the process to meet specifications. This analysis has been applied by producing process capability indices (PCIs). \({C}_{p}\) and \({C}_{pk}\) are the most commonly used PCIs for this aim. Although they are completely effective statistics to analyze process’ capability, the complexity of the production processes based on uncertainty arising from human thinking, incomplete or vague information makes it difficult to analyze the process capability with precise values. When there is uncertain, complex, incomplete and inaccurate information, the capability of the process is successfully analyzed by using the fuzzy sets. Neutrosophic sets (NSs), one of the new fuzzy set extensions, have a significant role in modeling uncertainty, since they contain the membership functions of truth, indeterminacy, and falsity definitions rather than an only membership function. This feature provides a strong advantage for modeling uncertainty. In this paper, PCA has been performed based on NSs to overcome uncertainties of the process. For this purpose, specification limits (SLs) have been reconsidered by using NSs and two of the well-known process capability indices (PCIs) named \({C}_{p}\) and \({C}_{pk}\) have been reformulated. Finally, the neutrosophic process capability indices (NPCIs) named \({C}_{p}\) \(\left({\tilde{\stackrel{⃛}{C}}}_{p}\right)\) and \({C}_{pk}\) \(\left({\tilde{\stackrel{⃛}{C}}}_{pk}\right)\) have been derived for three cases that are created by defining SLs. Additionally, the obtained NPCIs have also been applied and confirmed on real case problems from automotive industry. The obtained results show that the NPCIs support the quality engineers to easily define SLs and obtain more flexible and realistic evaluations for PCA.

Uncertainty-Aware Few-Shot Image Classification

Few-shot image classification learns to recognize new categories from limited labelled data. Metric learning based approaches have been widely investigated, where a query sample is classified by finding the nearest prototype from the support set based on their feature similarities. A neural network has different uncertainties on its calculated similarities of different pairs. Understanding and modeling the uncertainty on the similarity could promote the exploitation of limited samples in few-shot optimization. In this work, we propose Uncertainty-Aware Few-Shot framework for image classification by modeling uncertainty of the similarities of query-support pairs and performing uncertainty-aware optimization. Particularly, we exploit such uncertainty by converting observed similarities to probabilistic representations and incorporate them to the loss for more effective optimization. In order to jointly consider the similarities between a query and the prototypes in a support set, a graph-based model is utilized to estimate the uncertainty of the pairs. Extensive experiments show our proposed method brings significant improvements on top of a strong baseline and achieves the state-of-the-art performance.