In [7], Yu Ming extends the consideration of performance deterioration to components for which a model with physical parameters is not available. Such components may be sensors and actuators or monitored subsystems for which only measurements are known. In order to quantify the severity of nonparametric faults, he introduces an efficiency factor 0 < f) < I, where fi = 0 denotes a failure while p = 1 indicates no fault. Accordingly, 1 - jS is a measure for the severity of a fault. For sensors, a fault is captured by multiplying the normal measurement by p, an actuator fault is taken [Pg.226]

Mitchell and Koper [64] have involved the density-functional theory to determine the parameters necessary for the construction of an off-lattice model with no freely adjustable parameters for Br electrodeposition on Au(lOO). [Pg.849]

In this design there are only six distinctly different factor combinations. Thus, there are no degrees of freedom for lack of fit when fitting a two-factor FSOP model with six parameters. There are three degrees of freedom for purely experimental uncertainty because of the four replicate experiments at the center point. [Pg.304]

When this procedure is applied to LHHW models, caution should be exercised in claiming a model to be the best. The need for this increases with the number of unknown parameters, since several models may then show almost equal convergence. In general, it is advisable to confine the methods to models with no more than three or four unknown parameters. Furthermore, it should be ensured that all data points are of equal precision. [Pg.878]

Numerical simulations have been made possible by the availability of experimental data on a well-characterized geometry and with accurate concentration measurementst The simple geometry of ceramic monoliths is essential for accurate numerical modelling with no independent adjustable parameters such as tortuosities or effective diffusions within porous media. The only adjustable parameter, the peak diffusion at the moving contact line, will be eliminated when an acceptable, simple flow model of split-ejection streamlines is available. [Pg.93]

The fit of the experimental data to Eq. (6.48) is very satisfactory, as illustrated in Eig. 6.22(a), where the solid lines were recalculated here using the Bell-Limbach model, with the parameters included in Table 6.4. This result also means that there is no substantial decrease in the zero-point energies of the two protons in the cis-intermediate states as compared to the initial and final trans-states, as this would increase the HD/DD isotope effect beyond the value of 2 as was illustrated in Fig. 6.14(c). [Pg.177]

Even though many different covariates may be collected in an experiment, it may not be desirable to enter all these in a multiple regression model. First, not all covariates may be statistically significant—they have no predictive power. Second, a model with too many covariates produces models that have variances, e.g., standard errors, residual errors, etc., that are larger than simpler models. On the other hand, too few covariates lead to models with biased parameter estimates, mean square error, and predictive capabilities. As previously stated, model selection should follow Occam s razor, which basically states the simpler model is always chosen over more complex models. [Pg.64]

In general these models are able to fit asymmetrical data sets but require the use of added parameters (thereby reducing degrees of freedom). Also, some of the parameters can be seriously correlated (see discussion in [2, 3, 8]). Most importantly, these are empirical models with no correspondence to biology. [Pg.293]

In Fig. 1, a comparison can be observed for the prediction by the honeycomb reactor model developed with the parameters directly obtained from the kinetic study over the packed-bed flow reactor [6] and from the extruded honeycomb reactor for the 10 and 100 CPSI honeycomb reactors. The model with both parameters well describes the performance of both reactors although the parameters estimated from the honeycomb reactor more closely predict the experiment data than the parameters estimated from the kinetic study over the packed-bed reactor. The model with the parameters from the packed-bed reactor predicts slightly higher conversion of NO and lower emission of NHj as the reaction temperature decreases. The discrepancy also varies with respect to the reactor space velocity. [Pg.447]

Conformational characteristics of PTFE chains are studied in detail, based upon ab initio electronic structure calculations on perfluorobutane, perfluoropentane, and perfluorohexane. The found conformational characteristics are fully represented by a six-state RIS model. This six-state model, with no adjustment of the geometric or energy parameters as determined from the ab initio calculations, predicts the unperturbed chain dimensions, and the fraction of gauche bonds as a function of temperature, in good agreement with available experimental values. [Pg.53]

Thus, at high I, the pair population is a considerably smaller fraction of the total OH population than the initial fraction given by a Boltzmann distribution at the flame temperature. For example, for the nominal values of 14 and 0.4 A for Oq and Oy, the infinite-intensity fraction is < 1% of the total while the zero-intensity value is 4%. This result is generally valid for the entire range of parameters inserted into the model, which represent physically realistic energy transfer rates. However, the precise numerical values depend sensitively on the actual parameters inserted. These facts form the central conclusions of this study (4). A steady state model with no dummy level and a different set of rate constants and level structure (5) shows some similar features. [Pg.144]

To test whether one can differentiate between a two-site discrete model and a dual distribution function, we calculated intensity Stern-Volmer plots for a two-component model as a function of R. These are also shown in Figure 4.13. What is remarkable is that even for the quite wide R = 0.25, there is no experimentally detectable difference between two discrete sites and two continuously variable distribution of sites. Only when one gets to R = 0.5 does the data deviate noticeably. However, even though the shape has changed, it is still well fit by a dual discrete site model with different parameters. [Pg.99]

© 2019 chempedia.info