Limited Number Of Experimental Points Research Articles

Multifidelity data augmentation for data driven passive impact location and force estimation in composite structures under simulated environmental and operational conditions

Data driven methods for impact location and severity estimation have great potential for real life application due to their ability to construct meta models that are not affected by changes in structural complexity (e.g. stiffeners and cut-outs). However, to do so, an initial reference database containing known input and output pairs is required to form an accurate meta model. The requirement to collect this data (mainly in the form of experimental tests) is considered by many to be not feasible and hinders the application of data driven methods in large scale, real life structures. Here a new multifidelity approach is presented to reduce the “cost” of constructing the reference database necessary for data driven impact location and severity estimation methods. The proposed cokriging approach is used to combined “cheap” low fidelity FE (Finite Element) simulation data with a limited amount of accurate but “expensive” experimental data to construct a reference database that requires less experimental data sampling (thus lower “cost”) but with similar levels of accuracy compared to reference databases that were constructed from pure experimental data. Using previously developed impact location (kriging based localisation) and maximum force estimation (maximum impact force gradient method) methods, the effect of different reference databases with varying amounts of experimental data samples are tested. The results obtained on a CFRP (Carbon Fibre Reinforced Plastic) coupon and stiffened panel showed that using pure FE data for the reference database yielded poor results. By adding a limited number of experimental points, it was shown that the accuracy increases significantly approaching levels achieved using pure experimental data with significantly less samples. The wider context of data fusion is also explored, highlighting the possibility to combine various sources of data (including inspection history as well as data from sister assets) which can possibly be used to create a data driven framework that requires less initial data whilst also being robust and self-improving.

Dual-retention mechanism of dopamine-related compounds on monolithic stationary phase with zwitterion functionality

Seven retention models have been selected to describe a dual-retention behavior of ten dopamine-related compounds on polymer-based monolithic stationary phase with zwitterion sulfobetaine functionality. Regression quality, as well as a statistical significance of individual regression parameters, have been evaluated. Better regression performance showed two four-parameter models when compared to three-parameter models. On the other hand, limited number of experimental points disqualified statistical robustness of four-parameter models. Among three-parameter models, retention description introduced by Horváth and Liang provided comparable quality of regression at significantly improved robustness.Multivariate analysis of the best three-parameter models provided the description of physicochemical properties of dopamine precursors and metabolites. Principal component analysis and logistic regression allowed structural characterization of dopamine-related compounds based solely on regression parameters extracted from an isocratic elution data. Both polarity and type of functional groups has been correctly assigned for 3-methoxytyramine that has not been part of an evaluation study. Among applied dual-retention models, Horváth´s model, initially developed to describe a retention of ionic compounds on nonpolar stationary phases, provided robust regression of experimental data and allowed an extraction of structural characteristics of dopamine-related compounds

Naïve Bayes ant colony optimization for designing high dimensional experiments

In a large number of experimental problems, high dimensionality of the search area and economical constraints can severely limit the number of experimental points that can be tested. Within these constraints, classical optimization techniques perform poorly, in particular, when little a priori knowledge is available. In this work we investigate the possibility of combining approaches from statistical modeling and bio-inspired algorithms to effectively explore a huge search space, sampling only a limited number of experimental points. To this purpose, we introduce a novel approach, combining ant colony optimization (ACO) and naïve Bayes classifier (NBC) that is, the naïve Bayes ant colony optimization (NACO) procedure. We compare NACO with other similar approaches developing a simulation study. We then derive the NACO procedure with the goal to design artificial enzymes with no sequence homology to the extant one. Our final aim is to mimic the natural fold of 200 amino acids 1AGY serine esterase from Fusarium solani.

Determination of burning velocities from spherically expanding H2/air flames

Laminar burning velocities of hydrogen/air mixtures can show discrepancies up to 30%, making chemical mechanism validation and improvement difficult. The source of uncertainties may come from different factors influencing at each processing and post-processing steps the final value. Considering a spherically expanding flame configuration, reflection on the accuracy of the formulations, used to derive the desired quantity, is proposed. Starting from the exact definition of the laminar burning velocity, two formulations – direct and indirect flame speeds formulations – are derived for spherical flames. Each single source of uncertainty involved in the formulations is pointed out. The emphasis is focused on a specific mixture at an equivalence ratio of 0.50, atmospheric pressure, and an initial temperature of 300K. This point represents the best tradeoff between low ratio of flame velocity and recording sampling rate and the occurrence of cellular flames (Le<1). An extensive experimental and numerical study (1D spherically expanding flames) of this mixture is carried out. As a result, the experimental laminar burning velocities determined by using the direct flame speed or the indirect flame speed formulae depict different values. However, when numerically determined, both formulae yield the same value. This paves the way to understand and identify the experimental error sources. Stretch and Lewis numbers effects (super-adiabatic temperatures) as well as radiation processes (burned gas motion) are studied. Nonetheless, they do not show to be the main source of uncertainty. The extrapolation procedure (linear or non-linear) according to the limited number of experimental points (rapid apparition of cellular structure) appears as the main factor influencing the discrepancy in laminar burning velocities.

Evolutionary Bayesian Network design for high dimensional experiments

Laboratory experimentation is increasingly concerned with systems whose dynamical behaviour can be affected by a very large number of variables. Objectives of experimentation on such systems are generally both the optimisation of some experimental responses and efficiency of experimentation in terms of low investment of resources and low impact on the environment. Design and modelling for high dimensional systems with these objectives present hard and challenging problems, to which much current research is devoted. In this paper, we introduce a novel approach based on the evolutionary principle and Bayesian network models. This approach can discover optimum values while testing just a very limited number of experimental points. The very good performance of the approach is shown both in a simulation analysis and biochemical study concerning the emergence of new functional bio-entities.

Reconstruction of cylinder pressure for SI engine using recurrent neural network

Cylinder pressure based engine control systems use variables deduced from cylinder pressure as a feedback input. Monitoring of cylinder pressure is possible through various intrusive and nonintrusive sensors but cost of these sensors limits their use in the engines of on-road vehicles. In the present work, a recurrent neural network (RNN) is proposed which can reconstruct cylinder pressure of spark ignition engine. The network uses instantaneous crankshaft speed and motored pressure as inputs. Initially, parameters of two-zone model are tuned at limited number of experimental points, so that cylinder pressure predicted by model matches to that of experimental results. Further, the tuned model is used to generate large number of training data. Validation has been carried out using experimental as well as simulated pressure trace. It has been found that RNN can reconstruct cylinder pressure with reasonably good accuracy.

Response to the “Comments to water-splitting activity of Photosystem II by far-red light in green plants” by H.-W. Trissl

Two results of our papers [1,2] should be distinguished, one being an observation that surprisingly long-wavelength quanta up to 780 nm support oxygen evolution from plant leaves, and the second being a peculiar spectral behavior of the quantum yield of the PSII electron transport, showing a local maximum at ∼745 nm. This is the quantum yield spectral dependence and its interpretation that has been questioned by H.-W. Trissl in his Comments. Several arguments were presented in favour of the absorption from thermally populated vibrational sublevels of PS II chlorophylls for explanation of the photosynthetic activity under extreme red light, without appellation to special far-red PSII chlorophylls, the presence of which was hypothesized in [1,2]. One of the arguments used was a good fit of our experimental oxygen evolution rates with the calculated PSII absorption spectrum shown in Fig. 1b of the Comments. This agreement is, unfortunately, deceptive, because only a limited number of experimental points and arbitrary normalization in a narrow spectral range was applied. When, however, the whole set of experimental points is presented, as shown in Fig. 1 of this letter, obvious discrepancies appear, clearly suggesting an extra absorbance with respect to PSII chlorophylls in the 720–760 nm region. Thus, while the proposed interpretation is certainly an option, which is also discussed in our works [1,2], the evidence given in the Comments is circumstantial, based upon an idealized computational model [3]. Amongst other things, this model, for example, uses the Kennard–Stepanov relation for the calculation of the fluorescence spectrum. Yet as noted in [4], the thermal equilibration assumed by Kennard–Stepanov (KS) is hardly ever seen in practice. The KS anomalies are

Fluid dynamics of bubbles in liquid

Results gathered from the literature on the dynamics of bubbles in liquid are correlated by means of a formulation traditionally employed to describe the dynamics of isometric solid particles. It is assumed that the shape of the bubble depends, by means of the Eotvos number, on its diameter and on the gas-liquid surface tension. The analysis reported herein includes the dynamics of the isolated bubble along with wall and concentration effects. However, the effects of gas circulation in the bubble, which result in terminal velocities higher than those of a rigid sphere, are not being considered. A limited number of experimental points are obtained employing a modified version of the Mariotte flask which permits the precise measure of bubble volume. A classic bubble column is also employed in order to measure gas holdup in the continuous phase. Experiments were carried out employing air, with distilled water, potable water, water with variable amounts of surfactant and glycerin as the liquid phase.

Model for grain boundary sliding and its relevance to optimal structural superplasticity Part 5 – A unique numerical solution and its reliability

A few of the constitutive equations used to describe the mechanical response of superplastics are examined. The problems associated with the estimation of the values of the model constants are discussed. Attention is focused on the model for grain boundary sliding controlled optimal structural superplasticity recently proposed. A numerical procedure that allows the unambiguous determination of the material constants of this model is presented and validated by analysing the experimental data pertaining to a few systems. It is shown that this method allows the determination of the material constants with a limited number of experimental points. A comparison of the predictions with the experimental results shows that this model accurately describes optimal superplastic flow, for Regions I and II up to the point of inflection in the sigmoidal ln (stress)–ln (strain rate) curve. The advantages of the present method over others available in the literature are noted.

Estimation of Reactivity Ratios from Multicomponent Copolymerizations

Abstract The composition of the copolymer formed from n monomers in addition polymerization can be expressed in terms of the monomer feed composition and n(n - 1) binary reactivity ratios, according to the familiar simple copolymer model. Reactivity ratios are determined experimentally from cor-responding feed and monomer compositions in binary co-polymerizations. This article reports methods for deriving such reactivity ratios directly from multicomponent polymerization data. Analytical solution of the multi-component copolymer equations is not feasible because of the limited number of experimental points and experimental uncertainty in the copolymer composition. Computer-assisted procedures have been developed to estimate re-activity rates by optimizing the fit of predicted and experimental copolymer compositions, given the monomer feed composition and preliminary values of the reactivity ratios. All n(n - 1) reactivity ratios are adjustable. The methods are demonstrated for styrene/methacrylonitrile/ a-methylstyrene, butadiene/styrene/2-methyl- 5-vinyl- pyridine and acrylonitrile/methyl methacrylate/& methylstyrene systems. Binary reactivity ratios predict ternary copolymer compositions generally well in these cues. Reasons are suggested why reactivity ratios from multicomponent experiments may not match the corresponding parameters from binary copolymerizations.

The application of deconvolution techniques to dielectric data

The viability of a deconvolution procedure for the analysis of dielectric data is examined. A deconvolution procedure is applied to the complex permittivity ϵ(v) in order to derive a distribution function G(v) which represents the molecular processes occurring in solution. It is shown that with test data, subject to the normal experimental error, the deconvolution technique can be used to resolve a double dispersion for G(v), which is not evident from the experimental results. The deconvolution procedure does not require a model for the number of dispersions, unlike the least-squares-fitting methods which require a model for G(v) and the implicit assumption of the number of dispersions. The effect of a limited number of experimental points, each subject to error, taken over a finite frequency range on the distribution function G(v) is investigated, with particular reference to minimizing the possibility of spurious information in G(v).