Data-based Modeling Techniques Research Articles

Leakage-Flow Models for Screw Extruders.

Many theoretical analyses of extrusion ignore the effect of the flight clearance when predicting the pumping capability of a screw. This might be reasonable for conventional extruder screws with “normal” clearances but leads to errors when more advanced screw designs are considered. We present new leakage-flow models that allow the effect of the flight clearance to be included in the analysis of melt-conveying zones. Rather than directly correcting the drag and pressure flows, we derived regression models to predict locally the shear-thinning flow through the flight clearance. Using a hybrid modeling approach that includes analytical, numerical, and data-based modeling techniques enabled us to construct fast and accurate regressions for calculating flow rate and dissipation rate in the leakage gap. Using the novel regression models in combination with network theory, the new approximations consider the effect of the flight clearance in the predictions of pumping capability, power consumption and temperature development without modifying the equations for the down-channel flow. Unlike other approaches, our method is not limited to any specific screw designs or processing conditions.

Estimation of basis weight, ash content and moisture content in papermaking plants: A comparative study

The papermaking process is a typical nonlinear process with multiple input-output variables, so it is difficult to construct an accurate model for the process. Data-based modeling techniques may be used to establish a reliable paper plant model. In particular, the LSSVM (least-squares support vector machine) can be used to create a highperformance papermaking process model based on operation data. In this paper, we present a paper plant model that can predict three key output variables (basis weight, ash content, moisture content) with four input variables (stock flow, filler flow, speed, steam pressure) using LSSVM. The proposed LSSVM model is compared with other data-based models (the ANN (artificial neural network) model and the state-space model). The LSSVM model turned out to exhibit better estimation performance compared to others.

Using robust generalized fuzzy modeling and enhanced symbolic regression to model tribological systems

Tribological systems are mechanical systems that rely on friction to transmit forces. The design and dimensioning of such systems requires prediction of various characteristic, such as the coefficient of friction. The core contribution of this paper is the analysis of two data-based modeling techniques which can be used to produce accurate and at the same time interpretable models for friction systems. We focus on two methods for building interpretable and potentially non-linear regression models: (i) robust fuzzy modeling with batch processing and an enhanced regularized learning scheme, and (ii) enhanced symbolic regression using genetic programming. We compare our results of both methods with state-of-the-art methods and found that linear models are insufficient for predicting the coefficient of friction, temperature, wear, and noise-vibration-harshness rating of the tribological systems, while the proposed robust fuzzy modeling and the enhanced symbolic regression approaches, as well as the state-of-the-art regression techniques, are able to generate satisfactory models. However, robust fuzzy modeling and enhanced symbolic regression lead to simpler models with fewer parameters that can be interpreted by domain experts.

A benchmark for data-based office modeling: challenges related to CO2 dynamics

This paper describes a benchmark consisting of a set of synthetic measurements relative to an office environment simulated with the software IDA-ICE. The simulated environment reproduces a laboratory at the KTH-EES Smart Building, equipped with a building management system. The data set contains measurement records collected over a period of several days. The signals correspond to CO2 concentration, mechanical ventilation airows, air infiltrations and occupancy. Information on door and window opening is also available. This benchmark is intended for testing data-based modeling techniques. The ultimate goal is the development of models to improve the forecast and control of environmental variables. Among the numerous challenges related to this framework, we focus on the problem of occupancy estimation using information on CO2 concentration, which we treat as a blind identification problem. For benchmarking purposes, we present two different identification approaches: a baseline overparameterization method and a kernel-based method.

Model-based cell number quantification using online single-oxygen sensor data for tissue engineering perfusion bioreactors.

Online and non-invasive quantification of critical tissue engineering (TE) construct quality attributes in TE bioreactors is indispensable for the cost-effective up-scaling and automation of cellular construct manufacturing. However, appropriate monitoring techniques for cellular constructs in bioreactors are still lacking. This study presents a generic and robust approach to determine cell number and metabolic activity of cell-based TE constructs in perfusion bioreactors based on single oxygen sensor data in dynamic perfusion conditions. A data-based mechanistic modeling technique was used that is able to correlate the number of cells within the scaffold (R(2)  = 0.80) and the metabolic activity of the cells (R(2)  = 0.82) to the dynamics of the oxygen response to step changes in the perfusion rate. This generic non-destructive measurement technique is effective for a large range of cells, from as low as 1.0 × 10(5) cells to potentially multiple millions of cells, and can open-up new possibilities for effective bioprocess monitoring.

Data-based mechanistic modelling of indoor temperature distributions based on energy input

Energy used for building heating, ventilating and air conditioning contributes to a great share in the total energy consumption worldwide. Better understanding and management of energy distribution in those processes is essential for the improvement of process quality and efficiency of energy use. This paper presents the data-based mechanistic modelling approach which has been developed to model the dynamic indoor temperature distribution in an imperfectly mixed ventilated airspace based on energy input to the system. The combination of classical heat balance differential equations and the data-based modelling techniques for continuous-time system has brought a robust dynamic model suitable for model-based controlling and yet providing a profound insight at the energy and temperature distributions in ventilated systems. The effect of changing heat input on the temperature distribution inside a ventilated structure was studied. Dynamic response of indoor temperature to varying energy input could be explained by a second order transfer function model with a high coefficient of determination ( R 2 > 0.99), a low Young Identification Criterion (YIC < −2.3) and a low model standard error (SE < 0.028 °C). The physically meaningful model parameters as local heat load fraction γ and the coefficient of local temperature change h (°C J −1) were revealed. This modelling approach is very useful for future design of model-based predictive controller for zonal control of indoor temperature by the direct adjustment of heat load into ventilated structures.

An Automatic Depth Control System for Online Measurement of Spatial Variation in Soil Compaction, Part 2: Modelling of the Depth Control System

Soil compaction can be predicted in real-time by measuring the draught of a cutting tool, when the depth is held constant by the three-point hitch position control system of the tractor. A semi-analytical model was developed for the conversion of the voltage applied to the hydraulic subsystem of an agricultural tractor to the depth under the soil surface of an attached cutting tool (subsoiler) for online measurement of spatial variation in soil compaction. The model consisted of two distinct submodels: a model for the agricultural tractor based on a kinematic and dynamic analysis of the tractor in the XY plane and a 'grey box' model for the hydraulic subsystem. The order of this 'grey box' model was determined using a theoretical, mechanistic approach, while the parameters for the model structure have been determined using data-based modelling techniques. Both submodels have been coupled and successfully validated by comparing experimental results with those predicted by the model. It is concluded that the tractor dynamics are not excited by the activity of the hydraulic subsystem, when the subsoiler is attached. Therefore, a reduced model consisting of the hydraulic subsystem model multiplied by a conversion factor was found to be appropriate for the design of a depth controller for the soil compaction sensor.

Polynomial Behavioural Modelling of Switched Capacitor Amplifiers

A data-based modeling technique for amplifiers in switched capacitor circuits is presented. The measured output settling error of a switched capacitor (SC) amplifier is presented as a two-dimensional polynomial of the initial voltages in the input and output nodes of the amplifier. This fitted polynomial can then be used as a very compact behavioral discrete-time model that can replicate also the output-dependent errors of the amplifier. The procedure is illustrated by characterizing an existing amplifier and using the fitted error model of it in behavioral simulations of a pipeline A/D converter.

Heuristic, systematic, and informational regularization for process monitoring

Most data-based predictive modeling techniques have an inherent weakness in that they might give unstable or inconsistent results when the predictor data is highly correlated. Predictive modeling problems of this design are usually under-constrained and are termed ill-posed. This article presents several examples of ill-posed surveillance and diagnostic problems and regularization methods necessary for getting accurate and consistent prediction results. The examples include plant-wide sensor calibration monitoring and the inferential sensing of nuclear power plant feedwater flow using the following techniques: neural networks, nonlinear partial least squares techniques, linear regularization techniques implementing ridge regression, and informational complexity measures. © 2002 Wiley Periodicals, Inc.

Dynamic Modeling and Optimization of Micro-Hotplate Chemical Gas Sensors

The chemical process industries are facing tighter fugitive emission control standards. Cost efficient, portable, and sensitive sensors are needed to attack this leak detection problem in chemical plants. This paper reports a first step towards this aim, i.e. optimal tuning of micro-hotplate chemical gas sensors to recognize two similar vapors such as methanol and ethanol in air. Experiments are conducted by flowing gas over the sensor at a constant rate and concentration. Heater current pulse amplitudes (micro-hotplate temperatures) are varied randomly to generate a large database for training empirical models. After studying different data-based dynamic modeling techniques, wavelet networks (WNET) method proposed by Zhang and Benveniste was found to give the most accurate predictions for the methanol and ethanol responses.

Data-Based Modelling Techniques for Reciprocal Space of NMR Imaging

In this paper the application of data-based modelling techniques to a reciprocal space components of a Nuclear Magnetic Resonance (NMR) irnaging system is considered. These include conventional identification techniques based on Auto-Regressive Moving Average with eXogenous inputs (ARMAX), Non-Linear ARMAX (NARMAX) and identifications based on Artificial Intelligence (AI) techniques such as Artificial Neural Networks (ANN). The mean sum of squared errors and graphical fit are used to compare the optimum performance of each technique

A data-based damping modeling technique

Damping mechanisms exist in all vibration systems, but their nature is little understood and there is no systematic method for modeling general damping. This paper describes a novel damping modeling method (the Method of Energy Approximation, or MEA). This method is novel because it is a unique damping modeling method without assumed damping linearity; it is based on experimental data instead of physical principles; hence it is applicable to vibration systems of various materials and configurations; and it is suitable for vibration system transient control. Among the three quantities essential to an understanding of the dynamics of a vibration system, mass, stiffness, and damping, the last is the most complex and least understood. Therefore, with recent technology advances in such areas as composite materials and smart materials, the need for a good damping modeling method is more urgent than ever.

On data-based modelling techniques for fermentation processes

Six different modelling techniques were considered for the recombinant Escherichia coli fermentation process. These are Multiple Linear Regression (MLR), Principal Component Regression (PCR), Partial Least Squares (PLS), Auto-Regressive Moving Average with eXogeneous inputs (ARMAX), Non-linear ARMAX (NARMAX) and Artificial Neural Networks. The models use industrial on-line data from the process as input variables in order to forecast the concentrations of biomass and recombinant protein normally only available from off-line laboratory analysis. The models' performances are compared by prediction error and graphical fit using results obtained from a common testing set of fermentation data.