Set Of Input Variables Research Articles

Evolving connectionist system versus algebraic formulas for prediction of renal function from serum creatinine

In clinical trials, equation 7 from the Modification of Diet in Renal Disease (MDRD) Study is the most accurate formula for the prediction of glomerular filtration rate (GFR) from serum creatinine. An alternative approach has been developed using evolving connectionist systems (ECOS), which are novel computing structures that can be trained to generate accurate output from a given set of input variables. This study aims to compare the prediction errors associated with each method, using data that reproduce routine clinical practice as opposed to the artificial setting of clinical trials. The methods were compared using 441 radioisotope measurements of GFR in 178 chronic kidney disease patients from 12 centers in Australia and New Zealand. All clinical and laboratory measurements were obtained from the patients' center rather than central laboratories, as would be the case in routine clinical practice. Both the MDRD formula and ECOS used the same predictive variables, and both were optimized to the study cohort by stepwise regression and training, respectively. Mean measured GFR in the cohort was 22.6 mL/min/1.73 m(2). The bias and precision of the MDRD formula were -3.5 mL/min/1.73 m(2) and 34.5%, respectively, improving to -1.2 mL/min/1.73 m(2) and 31.1% after maximal optimization of the formula to study data. The bias and precision of the ECOS were 0.7 mL/min/1.73 m(2) and 32.6%, respectively, improving to -0.1 mL/min/1.73 m(2) and 16.6% after maximal optimization of the system to study data. The prediction of GFR using ECOS was improved by accounting for the center from where clinical and laboratory measurements originated within the connectionist model. Algebraic formulas will be associated with greater prediction error in routine clinical practice than in the original trials, and machine intelligence is more likely to predict GFR accurately in this setting.

Local Sensitivity Analysis of a Numerical Model of Volcanic Plinian Columns through Automatic Differentiation

Numerical simulation of geological phenomena are nowadays widely used to understand them. A good example are the numerous codes developed to model mass motion in different processes. Such models are approximations to the real phenomena because the physical conditions are simplified or because the values of the parameters intervening in the modeling are known only approximately. For this reason, an obligatory step in the application of these models is the assessment of their sensitivity to changes in the parameters. Sensitivity analysis is also required to evaluate the influence of the different input variables in the outcome of the model. In general this analysis is carried out by repeated computation of the model to different sets of input variables and settings of the model. A more fruitful procedure to carry out such analysis can be achieved by automatic differentiation. In this paper we illustrate such approach through its application to the model of an eruptive plinian column. The results of the analysis allow to assess, in a systematic and efficient manner, the relevance of the different boundary values and parameters of the model. The analysis reveals the effects on velocity, temperature, gas fraction, radius and ultimately column height, by changes in their boundary value (i.e. at the vent). The effect of changes in the value of the gas constants, density of pyroclasts and specific heat of the erupted material is also evidenced. Alternatively, the results can be regarded as a mean to estimate the values of the variables for different set of boundary values without recurring to new runs of the model. Automatic differentiation is shown to be a convenient scheme to obtain derivatives of model variables. This type of analysis forms in addition the basis for further work in inversion schemes.

Fuzzy computing based rainfall-runoff model for real time flood forecasting

This paper analyses the skills of fuzzy computing based rainfall–runoff model in real time flood forecasting. The potential of fuzzy computing has been demonstrated by developing a model for forecasting the river flow of Narmada basin in India. This work has demonstrated that fuzzy models can take advantage of their capability to simulate the unknown relationships between a set of relevant hydrological data such as rainfall and river flow. Many combinations of input variables were presented to the model with varying structures as a sensitivity study to verify the conclusions about the coherence between precipitation, upstream runoff and total watershed runoff. The most appropriate set of input variables was determined, and the study suggests that the river flow of Narmada behaves more like an autoregressive process. As the precipitation is weighted only a little by the model, the last time-steps of measured runoff are dominating the forecast. Thus a forecast based on expected rainfall becomes very inaccurate. Although good results for one-step-ahead forecasts are received, the accuracy deteriorates as the lead time increases. Using the one-step-ahead forecast model recursively to predict flows at higher lead time, however, produces better results as opposed to different independent fuzzy models to forecast flows at various lead times. Copyright © 2004 John Wiley & Sons, Ltd.

A Proposed Structure for Coupling Tiled Surfaces with the Planetary Boundary Layer

Abstract A generalized coupling is proposed between atmospheric models and surface schemes (land and ocean). A set of input and output variables is defined for this purpose in such a way that it can be used by many current and future models, including mosaic or tile schemes. The basic concept is to pass atmospheric variables from the lowest model level and their relation to corresponding fluxes to the surface scheme. The surface scheme returns the fluxes. In this framework, there is no need for the atmospheric model to have detailed information about the surface. Only the result of the surface computations is needed; namely, the fluxes, which are applied as a boundary condition. The equations for fully implicit coupling are derived, and the relevance for numerical stability is demonstrated. It is also shown that fully implicit coupling in a tile scheme leads to more robust results than partially implicit coupling.

The influence of lifestyle on cardiovascular risk factors: Analysis using a neural network

The cardiovascular pathologies are the most common causes of death in the elderly patient. To single out the main risk factors in order to effectively prevent the onset of the disease, the authors experimented a special computerized tool, the neural network, that works out a mathematical relation that can obtain certain data (defined as output) as a function of other data (defined as input). Data were processed from a sample of 276 subjects of both sexes aged 26–69 years old. The output data were: high/low cholesterolemia, HDL cholesterol, triglyceridemia with respect to an established cut-off; the input data were: sex, age, build, weight, married/single, number of children, number of cigarettes smoked/day, amount of wine and number of cups of coffee. We conclude that: (i) a relationship exists, deduced from a neural network, between a set of input variables and a dichotomous output variable; (ii) this relationship can be expressed as a mathematical function; (iii) a neural network, having learned the data on a sufficiently large population, can provide valid predictive data for a single individual with a high probability (up to 93.33%) that the response it gives is correct. In this study, such a result is found for two of the three cardiovascular risk indicators considered (cholesterol and triglycerides); (iv) the repetition of the neural network analysis of the cases in question after a “pruning” operation provided a somewhat less good performance; (v) a statistical analysis conducted on those same cases has confirmed the existence of a strong relationship between the input and the output variables. Therefore the neural network is a valid instrument for providing predictive in a single subject on cardiovascular pathology risks.

Application of steady-state and dynamic modeling for the prediction of the BOD of an aerated lagoon at a pulp and paper mill: Part II. Nonlinear approaches

Neural networks can provide effective predictive models for complex processes that are poorly described by first principle models, such as wastewater biological treatment systems. In this paper multilayer perceptron (MLP) and functional-link neural networks (FLN) are developed to predict inlet and outlet biochemical oxygen demand (BOD) of an aerated lagoon operated by International Paper of Brazil. In Part I, predictive models for both inlet and outlet BOD for the aerated lagoon were developed using linear multivariate regression techniques. For the current case study, MLP networks are the best choice for the prediction models. When only a relatively small number of samples is available, substantial improvement in inlet and outlet BOD prediction is shown for both FLN and MLP modeling using a reduced input variable set that was generated using partial least squares (PLS). Thus, this paper provides a novel approach for developing PLS–FLN model structures.

Carbon mass fluxes of forests in Belgium determined with low resolution optical sensors

The primary objective of this paper is to describe the validation of a parametric model (C-Fix) designed to estimate the basic carbon mass fluxes of forests in Belgium. Most validation efforts in the literature are based on point measurements. Since landscapes in Belgium are quite heterogeneous, the spatial up-scaling of a point measurement to the level of a sensor pixel is a crucial issue. Process based models quite often have a large set of input variables, some of them hardly available or not measurable on a regional basis; on the other hand most of the process based models possess a prognostic capacity. The parametric C-Fix model estimates carbon mass fluxes from local, regional to continental scales. It ingests only a modest number of meteorological input variables, including satellite observations. Parametric models do not possess a prognostic capacity, but the spatial up-scaling by the use of remote sensing data is much more straightforward than with process models. In this paper, we describe the validation of C-Fix with eddy covariance NEP (net ecosystem production) measurements and further applied C-Fix for the mapping of the geographical distribution of carbon mass fluxes over the entire Belgian territory, using NOAA-AVHRR (1997) and SPOT4-VGT imagery (April 1998–March 1999). We combine a forest probability map, derived from NOAA data for 1997 with the mapped estimates of Belgian NEP to obtain forest NEP per image pixel. Forest NEP is validated regionally, with measurements of carbon exchange obtained at two Belgian Euroflux eddy covariance tower sites (the Brasschaat, ‘Inslag’ and Vielsalm, ‘Tinseubois' forest sites). A correlation analysis is performed for the estimated forest NEPs at both Euroflux sites and the NEP measured there, assumed to be primarily from forest. Finally, a correlation analysis with a process-based stand scale model (ASPECTS) is performed for the two forest sites, as a cross check on the validation results. Our results demonstrate that a parametric model, like C-Fix, provides a good basis to estimate the evolution and geographical distribution of the main constituents of the carbon budget of terrestrial ecosystems, in this study specifically forest ecosystems at the regional scale (Belgium).

Approximation of a Discrete Event Stochastic Simulation Using an Evolutionary Artificial Neural Network

A computer simulation model may be regarded as a mapping function bietween a set of input and output variables. Although simulation models are very popular experimentation tools, in many cases they are computationally expensive. Hence, it would be essential to have fast, accurate approximation of computer simulation. This paper examines the use of an evolutionary artificial neural network for approximating a lot size - reorder point inventory system simulation. The proposed approach was compared with a Backpropagation trained neural network and multiple linear regression models.

Artificial Neural Networks for Modelling of the Impact Toughness of Steel

The application of artificial neural networks (ANNs) to the prediction of the Charpy impact toughness of quenched and tempered (QT) steels and ferrous weld metals is examined in detail. It is demonstrated that the Charpy impact toughness can be accurately predicted using the selected input variables and their ranges of values. The capacity of ANNs to handle problems involving large sets of input variables is illustrated by a model developed to predict the impact energy of weld metal (WM) produced by flux cored arc welding (FCAW). The usefulness of ANNs for alloy design and process control is demonstrated through another model developed to predict the toughness of a QT structural steel as a function of composition and postweld heat treatment. Although comparison of the two models indicates that the trends in toughness with changes in Mn and B concentrations are in opposite directions for weld metal and QT steel, it is shown that these trends can be reconciled with reported experimental results and theoretical interpretations.

Fuzzy Neural Network for Predicting 4-CBA Concentration of PTA Process

A fuzzy neural network model has been developed to predict the 4-CBA concentration of the oxidation unit in PTA process. Several technologies are used to deal with the process data before modeling. Suitable input variables set has been selected according to prior knowledge, experience and fuzzy curve method. Dead time delay has been considered in the fuzzy neural network model. The simulation results show that the model of the fuzzy neural network is better than that of AMOCO in prediction precision.

High-resolution forest fire weather index computations using satellite remote sensing

Wildfires are important in regions dominated by forest, such as found in large parts of Canada. The principal objective of this study was to provide homogeneously distributed indices for the Canadian Fire Weather Index (FWI) System. The FWI was calculated using four sets of input variables: meteorological station measurements (OBS); weather forecast model output (SIM); meteorological station measurements and remote sensing estimations combined (SAT1); and weather forecast model output and remote sensing estimations combined (SAT2). Remote sensing parameterization of air temperature and relative humidity was performed. The air temperature and relative humidity reproduced showed good agreement with ground-based measurements (R2 = 0.77 and SE = 1.48°C; R2 = 0.73 and SE = 5%, respectively). For the FWI regionalized using this requirement, category SAT1 showed the best fit. Category SAT2 produced more precise results (0.09 to 2.19% of the normalized root mean square error) versus SIM.

Modeling wet‐gas annular/dispersed flow through a venturi

AbstractA theoretical model for gas–liquid annular/dispersed flow through a venturi meter is reported. It is based on an earlier model developed for venturi scrubbers. Changes implemented are based on new research and on the different physics between the two cases. The predictions of the model have been tested using information from recent experiments on venturi meters employed for measuring wet‐gas flows with a liquid volume fraction up to 10%. The model gives good predictions with appropriate values of a small set of input variables.

DESIGN AND ANALYSIS OF INTERCEPTOR DRAINS—ANALYTICAL SOLUTION

ABSTRACT An analytical solution is obtained for estimation of seepage from canal with symmetrically placed interceptor drains, as well as natural drainage, founded on infinite pervious soil medium. Interceptor drains are considered to be of finite width. However, the depth of water is neglected both in the main canal and interceptor drains. Integral equations are obtained by using Zhukovsky's function and Schwarz-Christoffel transformation. Algorithm is then developed to numerically solve the nonlinear integral equations to obtain the value of seepage discharges and profile of the free surface. Computer program has been developed to obtained results for any number of sets of input variables. Using this solution interceptor drains can be designed by considering the width of drains. Further, using this solution, nomographs can be prepared and effects of various non-dimensional ratios of parameters such as canal width, distance between canal and natural drainage, depth of drainage below canal level, location and width of interceptor drains can be determined.

Fluorescence spectroscopy and multivariate spectral descriptor analysis for high-throughput multiparameter optimization of polymerization conditions of combinatorial 96-microreactor arrays.

Selection of optimum process conditions in combinatorial microreactors is essential if the combinatorial synthesis process is to be correlated with the synthesis process on a more conventional scale and the materials are to have the desired chemical properties. We have developed a new methodology for the high-throughput multiparameter optimization of polymerization reaction conditions in arrays of microreactors. Our strategy is based on the application of nondestructive spectroscopic techniques to measure chemical properties of polymers directly in individual microreactors followed by the multivariate spectral descriptor analysis for rapid determination of the optimal process conditions. We have demonstrated our strategy in the high-throughput multiparameter optimization of process conditions in thin-film melt polymerization reactions performed in 96-microreactor arrays for combinatorial screening of new polymerization catalysts. The combinatorial polymerization system was optimized for the best processing parameters using a set of input variables that included reactant parameters (relative amounts of starting components and catalyst loading) and processing variables (reaction time, reaction temperature, and inert gas flow rate). The measured output parameters were the chemical properties of materials and reproducibility of the material formation in replicate polymerizations in microreactors. Spatially resolved nondestructive evaluation of polymer formation was performed directly in individual microreactors and provided information about the spatial homogeneity of polymers in microreactors. It showed to be another powerful indicator of the reproducible polymerization process on the combinatorial scale. Although the methodology described here was implemented for high-throughput optimization of polymerization conditions, it is more general and can be further implemented for a variety of applications in which optimization of process parameters can be studied in situ or off-line using spectroscopic and other tools.

Neural network-based identification of SMB chromatographic processes

In this contribution, the identification problem for the control of nonlinear simulated moving bed (SMB) chromatographic processes is addressed. For process control the flow rates of extract, desorbent, and recycle of the SMB process, and the switching time are the manipulated variables. But these variables influence the process in a strongly coupled manner. Therefore, a new set of input variables is introduced by a nonlinear transformation of the physical inputs, such that the couplings are reduced considerably. The front positions of the axial concentration profile are taken as model outputs. Multilayer feedforward neural networks (NN) are utilized as approximating models of the nonlinear input–output behavior. The gradient distribution of the model outputs with respect to the inputs is used to determine their structural parameters and the network size is chosen by the SVD method. To illustrate the effectiveness of the identification method, a laboratory scale SMB process is used as an example. The simulation results of the identified model confirm a very good approximation of the first principles models and exhibit a satisfactory long-range prediction performance.

Eutrophication research of West Lake, Hangzhou, China: modeling under uncertainty

The models such as the eutrophication ecosystem model of West Lake, Hangzhou (EEM), are always used to make policy decisions for eutrophication management. Thus it is important to know the uncertainty in the model predictions due to the combined effects of uncertainty in the full set of input variables, and the individual input parameters whose variations have the greatest effect on variations in model predictions. In this study, randomized methods based on Monte Carlo technique have been developed and applied to the model (EEM). The technique consists of parameter sensitivity analysis, randomly sampling from underlying probability distributions and multivariate regression analysis. With this technique, model uncertainties during modeling are clarified and their propagation evaluated. Results show that among the five input parameters selected for uncertainty analysis, the settling rate of algae SVS and water temperature TEM have the largest contribution to model prediction uncertainty of the model outputs (PC, PS and PHYT).

Use of genetic algorithms to select input variables in decision tree models for the prediction of benthic macroinvertebrates

Predicting freshwater organisms based on machine learning is becoming more and more reliable due to the availability of appropriate datasets, advanced modelling techniques and the continuously increasing capacity of computers. A database consisting of measurements collected at 360 sampling sites in non-navigable watercourses in Flanders was applied to predict the absence/presence of benthic macroinvertebrate taxa by means of decision trees. The measured variables were a combination of physical–chemical (temperature, pH, dissolved oxygen concentration, conductivity, total organic carbon, Kjeldahl nitrogen and total phosphorus), structural (granulometric analysis of the sediment, width, depth and flow velocity of the river) and two ecotoxicological variables. The predictive power of decision trees was assessed on the basis of the number of Correctly Classified Instances (CCI). A genetic algorithm was introduced to compare the predictive power of different sets of input variables for the decision trees. The number of input variables was reduced from 15 to 2–8 variables without affecting the predictive power of the decision trees significantly. Furthermore, reducing the number of input variables allowed to ease the identification of general data trends.

Implications of the Higgs boson searches on different soft SUSY-breaking scenarios

We investigate the Higgs boson sector of the Minimal Supersymmetric Standard Model (MSSM) in the framework of the three most prominent soft SUSY-breaking scenarios, mSUGRA, mGMSB and mAMSB. For each scenario, we determine the parameters at the electroweak scale from the set of input variables at higher energy scales (depending on the specific scenario) and evaluate the Higgs boson properties. The latter are based on results obtained within the Feynman-diagrammatic approach by taking into account the complete one-loop and the dominant two-loop contributions. The maximum value of the mass of the lightest neutral CP -even MSSM Higgs boson, m h , is determined in the three scenarios, and the behavior of the Higgs couplings to fermions and gauge bosons is investigated. Restrictions on tan β and on the set of higher-energy scale parameters are derived from the lower limits arising from the Higgs search at LEP2. We furthermore discuss the regions of parameter space in the three scenarios compatible with interpreting the excess observed at LEP2 as a Higgs signal, m h =115 +1.3 −0.9 GeV. The case where the events observed at LEP2 could originate from the production of the heavier neutral CP -even Higgs boson is also considered. The implications of a possible Higgs signal at 115 GeV for SUSY searches at future colliders are briefly discussed for each of the three scenarios.

NEURAL NETWORK-BASED IDENTIFICATION AND MPC CONTROL OF SMB CHROMATOGRAPHY

In this contribution, the identification and control of nonlinear SMB-chromatographic processes are discussed. Instead of using the physical manipulated process variables, the flow rates of extract, desorbent, and recycle, and the switching time directly, a new set of input variables (β-factors) is employed as control inputs to reduce input/output couplings. A new measure of the front positions of the axial concentration profiles is used as outputs. Multi-layer neural network models are identified for this nonlinear MIMO system. The identified model is used in a model predictive control algorithm. In this algorithm a parameter varying linear model is employed which avoids the on-line computation of the nonlinear optimization problem. The simulation results show that the identified model gives a very good approximation of the process models and the LPVMPC scheme has a good control performance.

Thermal expansion of PM ferrous materials

Simulation-based battery design encounters the difficulty of high computational cost. This paper presents a systematic approach based on the artificial neural network to reduce the computational burden of battery design by several orders-of-magnitude. Two neural networks are constructed using the finite element simulation results from a thermo-electrochemical model. The first neural network serves as a classifier to predict whether a set of input variables is physically feasible. The second neural network yields specific energy and specific power. Both neural networks are validated using extra finite element simulations out of the training data. With a global sensitivity analysis using the neural network, we quantify the effect of input variables on specific energy and specific power by evaluating large combinations of input variables, which is computationally prohibitive for finite element simulations. Among all parameters, the applied C-rate has the largest influence on specific power, while the electrode thickness and porosity are the dominant factors affecting specific energy. Based on this finding, we generate a design map that fulfills the requirements of both specific energy and specific power. Inparticular, we highlight the value of neural network in handling the non-linear, complex and computationally expensive problem of battery design and optimization.