Data-based Modelling Approach Research Articles

ARX-NNPLS Model Based Optimization Strategy and Its Application in Polymer Grade Transition Process

Since it is often difficult to build differential algebraic equations (DAEs) for chemical processes, a new data-based modeling approach is proposed using ARX (AutoRegressive with eXogenous inputs) combined with neural network under partial least squares framework (ARX-NNPLS), in which less specific knowledge of the process is required but the input and output data. To represent the dynamic and nonlinear behavior of the process, the ARX combined with neural network is used in the partial least squares (PLS) inner model between input and output latent variables. In the proposed dynamic optimization strategy based on the ARX-NNPLS model, neither parameterization nor iterative solving process for DAEs is needed as the ARX-NNPLS model gives a proper representation for the dynamic behavior of the process, and the computing time is greatly reduced compared to conventional control vector parameterization method. To demonstrate the ARX-NNPLS model based optimization strategy, the polyethylene grade transition in gas phase fluidized-bed reactor is taken into account. The optimization results show that the final optimal trajectory of quality index determined by the new approach moves faster to the target values and the computing time is much less.

Time Series Forecasting Of Water Quality Of River Godavari

The research and developmental activities for generating new water quality models provide a valuable source of knowledge in the fields of water resources and environmental engineering. Due to limited water quality data and high cost of water quality monitoring, the data based modeling approaches are being extensively used. This paper presents the forecasted monthly values of water quality parameters viz., pH, Water Temperature and Dissolved Oxygen of River Godavari at Rajahmundry water quality monitoring station, Andhra Pradesh, India. Time series methods of forecasting were used for the predictions of parameter values. The predicted values are compared with actual measured observations using statistical performance tests. The results indicated that the proposed models were performing well. The MAE, RMSE and MAPE values, for pH 0.249, 0.359and 3.228; for Water Temperature 0.94, 1.18 and 3.24 and for Dissolved Oxygen 0.56, 0.67 and 7.52, respectively. The forecasted water quality parameter observations for the study period (2009-2012) and the future period (2012-2015) showed seasonal trend pattern which indicated quality of water is greatly influenced by the seasonal inflows. The time series forecasting methods with allowable error limits can be used to assess the future water quality and for taking protection measure for conservation of the river.

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.

Reverse engineering of metabotropic glutamate receptor-dependent long-term depression in the hippocampus

This study focused on metabotropic glutamate receptor-dependent long-term depression (mGluR–LTD) in the hippocampus. This form of LTD is suggested to play a key role in learning, memory and the plasticity of behaviour. Recent advances have started to uncover the underlying mechanisms of mGluR-LTD [1]. However, it is not completely clear how these mechanisms are linked and it is believed that several crucial mechanisms still remain to be revealed. The two main objectives of this study were (i) to quantify the dynamics of mGluR-LTD responses by dynamic data-based models and (ii) to identify underlying dominant processes of mGluR-LTD by applying mathematical system identification methods. In recent years, more and more researchers advocate the use of a top-down modelling approach (reverse engineering) for improving the knowledge of biological systems [2,3]. The drug dihydroxyphenylglycine (DHPG) was used to induce mGluR-LTD in rat brain slices (table ​(table1).1). The drug was applied for different durations (5min, 15min, 2 hours) and in different concentrations (15mM, 30mM). In addition, also different sampling intervals (5min, 30s, 90s) were used. Table 1 Overview of the experiments For the modelling, discrete-time Transfer Functions (TF) models were used. The models described the relation between the DHPG application (input) and the long-term depression responses (output). All models were very accurate (all RT2-values higher than 0,94) and reliably estimated. For a 2 hours application of 30 µM DHPG sampled with a frequency of 1/30s, the time-constant of the mGluR-LTD response was 92s. Thus, the models for high sampling rate indicated that a sampling interval of 30s would be ideal to minimize information loss of the dynamics of mGluR-LTD responses. Interestingly, it was suggested that there are three dominant sub-processes underlying mGluR-LTD: one fast sub-process, one slow sub-process and an immediate sub-process. This study suggests that the dynamic data-based modelling approach can be a valuable tool for reverse engineering of mGluR-dependent LTD responses.

Iterative Data-based Modelling and Optimization for Rapid Design of Dynamic Processes

We consider an off-line process design problem where the response variable is affected by several factors. We present a data-based modelling approach that iteratively allocates new experimental points, update the model, and search for the optimal process factors. A flexible non-linear modelling technique, the kriging (also known as Gaussian processes), forms the cornerstone of this approach. Kriging model is capable of providing accurate predictive mean and variance, the latter being a quantification of its prediction uncertainty. Therefore, the iterative algorithm is devised by jointly considering two objectives: (i) to search for the best predicted response, and (ii) to adequately explore the factor's space so that the predictive uncertainty is small. This method is further extended to consider dynamic processes, i.e. the process factors are time-varying and thus the problem becomes to design a time-dependent trajectory of these factors. The proposed approach has been demonstrated by its application to a simulated chemical process with promising results being achieved.

Constructing Distributed Hippocratic Video Databases for Privacy-Preserving Online Patient Training and Counseling

Digital video now plays an important role in supporting more profitable online patient training and counseling, and integration of patient training videos from multiple competitive organizations in the health care network will result in better offerings for patients. However, privacy concerns often prevent multiple competitive organizations from sharing and integrating their patient training videos. In addition, patients with infectious or chronic diseases may not want the online patient training organizations to identify who they are or even which video clips they are interested in. Thus, there is an urgent need to develop more effective techniques to protect both video content privacy and access privacy. In this paper, we have developed a new approach to construct a distributed Hippocratic video database system for supporting more profitable online patient training and counseling. First, a new database modeling approach is developed to support concept-oriented video database organization and assign a degree of privacy of the video content for each database level automatically. Second, a new algorithm is developed to protect the video content privacy at the level of individual video clip by filtering out the privacy-sensitive human objects automatically. In order to integrate the patient training videos from multiple competitive organizations for constructing a centralized video database indexing structure, a privacy-preserving video sharing scheme is developed to support privacy-preserving distributed classifier training and prevent the statistical inferences from the videos that are shared for cross-validation of video classifiers. Our experiments on large-scale video databases have also provided very convincing results.

Data-based mechanistic rainfall-runoff continuous-time modelling in urban context

This paper presents a data-based mechanistic modelling (DBM) approach to rainfall-runoff modelling based on the direct identification and estimation of continuous-time models from discrete-time series. It is argued that many mechanistic model parameters are more naturally defined in the context of continuous-time, differential equation models. As a results, there are advantages if such models are identified directly in this continuous-time form rather than being formulated and identified as discrete-time models. An illustrative example based on the analysis of rainfall-flow data from the Kerinou catchment in Brest demonstrates the relevance of the proposed modelling approach.

Nonlinear transfer function modelling of beach morphology at Duck, North Carolina

This paper presents a simple nonlinear data-based modelling approach for predicting the beach profile volume at Duck, North Carolina, USA. The state-dependent parameter form of the general transfer function (SDP TF) model is used to describe nonlinearity influencing these morphological data in two case examples. Case 1 investigates the nonlinearity associated with the dependency of wave forcing on the preceding beach volume. Case 2 investigates the ability to model the variables within the well-known diffusion equation for beach volume using this data-based approach. The results of this study show that the SDP TF approach can be used successfully to develop statistically robust models for describing nonlinearity in beach morphological systems. Furthermore, these models are shown to predict the beach volumes over both short (1 month ahead) and long (2 years ahead) time periods, and thus show great potential for practical applications in coastal zone management and engineering.

Simulation of sap flow in a beech tree by means of a dynamic data-based model as influenced by a solar eclipse

Responses of crucial plant processes such as transpiration and photosynthesis have often been modelled by mechanistic models or static regression models. The objective of this paper was to quantify the dynamics of sap flow rate in a beech tree to the rapid variations in short-wave solar radiation, air temperature and vapour pressure deficit using an alternative dynamic data-based modelling approach. In order to have enough dynamic information, data from the solar eclipse of 11 August 1999 were used. This permitted determination of the order of the sap flow dynamics, together with the relative contribution of short-wave radiation, air temperature and vapour pressure deficit. A multiple-input and single-output transfer function was used to simulate the sap flow rate responses in three branches at 22, 16 and 9 m accurately (R2 of 0.94, 0.86 and 0.90, respectively). The appropriate model structure was the same for the three branches and was characterised by second-order dynamics. An important advantage of the dynamic modelling approach presented was that it enabled the decomposition of the total sap flow rate response into partial responses to short-wave radiation, air temperature and vapour pressure deficit, respectively.

Data-based mechanistic modelling of three-dimensional temperature distribution in ventilated rooms filled with biological material

In this paper, a data-based mechanistic modelling approach was developed to real time monitoring and online adaptive control of three-dimensional temperature distribution in both an individual biological product and in a given movement. Boxes of potatoes were used as biological material. During the experiments, step inputs on an air inlet temperature were applied while airspace and potatoes temperature were recorded. The simplified refined instrument variable algorithm was used as a model parameter identification tool to obtain the best model order and parameters. By model compacting, several physically meaningful parameters were found to present the temperature distribution between the products and the airspace. A third order transfer function from the dynamic response of airspace from inlet air temperature with a high coefficient of determination R T 2 > 0.9 and a low standard error (SE < 0.01) explained the heat exchange in a system. Two physically meaningful parameters were found from the model parameters. A local volumetric fresh air concentration β 1 was defined to the temperature distribution in the airspace, and a local ‘cooling rate’ α 2 was presented for the product temperature distribution. The values of β 1 and α 2 existing in the model could be used to design a control system for real time monitoring and online adaptive control of three-dimensional temperature distribution in the process room.

Models Wagging the Dog: Are Circuits Constructed with Disparate Parameters?

In a recent article, Prinz, Bucher, and Marder (2004) addressed the fundamental question of whether neural systems are built with a fixed blueprint of tightly controlled parameters or in a way in which properties can vary largely from one individual to another, using a database modeling approach. Here, we examine the main conclusion that neural circuits indeed are built with largely varying parameters in the light of our own experimental and modeling observations. We critically discuss the experimental and theoretical evidence, including the general adequacy of database approaches for questions of this kind, and come to the conclusion that the last word for this fundamental question has not yet been spoken.

Neural networks in process life cycle profit modelling

Changes in operational environment of the process industry such as decreasing selling prices, increased competition between companies and new legislation, set requirements for performance and effectiveness of the industrial production lines and processes. For the basis of this study, a life cycle profit (LCP) model of a pulp process was constructed using different kind of process information including chemical consumptions and production levels of material and energy flows in unit processes. However, all the information needed in the creation of relevant LCP model was not directly provided by information systems of the plant. In this study, neural networks was used to model pulp bleaching process and fill out missing information and furthermore to create estimators for the alkaline chemical consumption. A data-based modelling approach was applied using an example, where factors affecting the sodium hydroxide consumption in the bleaching stage were solved. The results showed that raw process data can be refined into new valuable information using computational methods and moreover to improve the accuracy of life cycle profit models.

Forecasting the geomagnetic activity of the Dst index using multiscale radial basis function networks

The Dst index is a key parameter which characterises the disturbance of the geomagnetic field in magnetic storms. Modelling of the Dst index is thus very important for the analysis of the geomagnetic field. A data-based modelling approach, aimed at obtaining efficient models from limited input–output observational data, provides a powerful tool for analysing and forecasting geomagnetic activities including the prediction of the Dst index. In this study, the process of the Dst index is treated to be a structure-unknown system, where the solar wind parameter ( VBs) and the solar wind dynamic pressure ( P) are the system inputs, and the Dst index is the system output. A novel multiscale RBF (MSRBF) network is introduced to represent such a two-input and single-output system, where the Dst index is related to the solar wind parameter and the dynamic pressure, via a hybrid network model consisting of two submodels: a linear part that reflects the linear relationship between the output and the inputs, and a nonlinear part that captures the effect of the interacting contribution of past observations of the inputs and the output, on the current output. The proposed MSRBF network can easily be converted into a linear-in-the-parameters form and the training of the linear network model can easily be implemented using a forward orthogonal regression (FOR) algorithm. One advantage of the new MSRBF network, compared with traditional single scale RBF networks, is that the new network is more flexible for describing complex nonlinear dynamical systems.

Surrogate Modelling of Processes for Granular Material Thermal Treatment

Abstract Present advanced technologies creates increasingly complex engineering systems and the complexity of their engineering analysis advance at the same rate. In addition, the design and operation of complex engineering systems involves the integration of multiple disciplines and the resolution of conflicting objectives. Traditional parametric approach is inadequate to analyze these large-scale systems because of its computational inefficiency. Approximation techniques may be applied to build computationally inexpensive surrogate models for large-scale systems to replace expensive-to-run computer analysis codes. Numbers of approaches are existing. Response surface models are frequently utilized to construct surrogate approximations; however, they may be inefficient for systems having with a large number of variables. An alternative approach based on conceptual modelling for creating surrogate models is presented. A unified model for thermal processes is developed to compute thermal treatment of granular materials. The model does not require measurements but they can be useful for the model adjustment on specific situations or on its validation. A Databased modelling approach is used in order to keep at a minimum all the a priori assumptions on the physical mechanism driving the process formations. The motivation is to reduce computational expanse associated with use of simulation models needed to find good solutions. The model has been applied to replace of computationally expensive computer analyses.

Data-based mechanistic modelling approach to determine the age of air in a ventilated space

In literature, local mean age of air is used as an important index to evaluate indoor air quality in ventilated rooms. In this research, a data-based mechanistic approach is used to model the spatial–temporal mass distribution in an imperfectly mixed forced ventilated installation. A first-order transfer function model has proved to be sufficiently good in describing the mass transfer dynamics ( R t 2 = 0.987 ) of the system. Furthermore, it was possible to fully understand the physical meaning of the model parameter. The parameter is found to be an inverse of the age of air. This Data-Based Modelling approach proved to be more robust when dealing with measurement noise. Finally, the modelled age of air was validated with a classical step up determination of the age of air for experimental data. Good correlation ( R t 2 = 0.77 ) was found between both results, which proved the physical background of the model parameter.

Validation of ERS scatterometer-derived soil moisture data in the central part of the Duero Basin, Spain

The objective of this study was to validate the soil moisture data derived from coarse-resolution active microwave data (50 km) from the ERS scatterometer. The retrieval technique is based on a change detection method coupled with a data-based modelling approach to account for seasonal vegetation dynamics. The technique is able to derive information about the soil moisture content corresponding to the degree of saturation of the topmost soil layer (∼5 cm). To estimate profile soil moisture contents down to 100 cm depth from the scatterometer data, a simple two-layer water balance model is used, which generates a red noise-like soil moisture spectrum. The retrieval technique had been successfully applied in the Ukraine in a previous study. In this paper, the performance of the model in a semi-arid Mediterranean environment characterized by low annual precipitation (400 mm), hot dry summers and sandy soils is investigated. To this end, field measurements from the REMEDHUS soil moisture station network in the semi-arid parts of the Duero Basin (Spain) were used. The results reveal a significant coefficient of determination (R2 = 0·75) for the averaged 0–100 cm soil moisture profile and a root mean square error (RMSE) of 2·2 vol%. The spatial arrangement of the REMEDHUS soil moisture stations also allowed us to study the influence of the small-scale variability of soil moisture within the ERS scatterometer footprint. The results show that the small-scale variability in the study area is modest and can be explained in terms of texture fraction distribution in the soil profiles. Copyright © 2004 John Wiley & Sons, Ltd.

Active control of the growth trajectory of broiler chickens based on online animal responses

The objective of the research reported here was to control the growth trajectory of broiler chickens during the production process based on an adaptive compact dynamic process model. More specifically, the daily feed supply was calculated, based on a model-based control algorithm, with the aim of following a previously defined target growth trajectory as close as possible. For the modeling of the dynamic growth response of broiler chickens to the control input, feed supply, an online parameter estimation was used. The developed control algorithm was able to grow the birds according to different target trajectories ranging from restricted (final BW of 1,800 g and 1,945 g in experiments 1 and 3, respectively) to compensatory growth trajectories (final BW of 2,400 g and 2,100 g in experiments 2 and 4, respectively). The mean relative error (MRE) between the different predefined target growth trajectories and the realized growth trajectories ranged from 3.7% to 6.0%. With a few exceptions, the numerical values of feed conversion ratio and mortality after wk 1 were lower and the values of uniformity index were higher in the controlled groups compared with animals fed ad libitum. As a conclusion, it can be stated that integration of dynamic data-based modeling approaches with new hardware and sensing techniques to measure information from the animals should make it possible to control broiler growth trajectories.

Functional nodes in dynamic neural networks for bioprocess modelling.

This contribution presents a novel method for the direct integration of a-priori knowledge in a neural network and its application for the online determination of a secondary metabolite during industrial yeast fermentation. Hereby, existing system knowledge is integrated in an artificial neural network (ANN) by means of 'functional nodes'. A generalized backpropagation algorithm is presented. For illustration, a set of ordinary differential equations describing the diacetyl formation and degradation during the cultivation is incorporated in a functional node and integrated in a dynamic feedforward neural network in a hybrid manner. The results show that a hybrid modelling approach exploiting available a-priori knowledge and experimental data can considerably outperform a pure data-based modelling approach with respect to robustness, generalization and necessary amount of training data. The number of training sets were decreased by 50%, obtaining the same accuracy as in a conventional approach. All incorrect decisions, according to defined cost criteria obtained with the conventional ANN, were avoided.

The influence of physical catchment properties on baseflow in semi-arid environments

The present paper examines the influence of physical catchment properties on baseflow in semi-arid environments. It relates the dimensionless quantity, baseflow index (BFI) derived from data-based mechanistic modelling approach to dimensionless indices representing topographic, geologic and climatic properties. The relationship is derived using a progression of techniques, from inspection and correlation analysis, through stepwise regression and finishing with multiple regression. The BFI has a strong relationship with climate and geology. Catchments with high rainfall or low evapo-transpiration underlain with granites or basalt tend to give high baseflow. Within the topographical parameters tested, drainage density index has strong relationship with BFI. The approach used in this paper provides a viable alternative for baseflow estimation on ungauged semi-arid catchments using drainage density, geology and climate indices as key factors.

A neural network approach to transmission line modelling and fault diagnosis of a hydraulic press control system

A databased modelling approach is developed for a transmission line using an artificial neural network (ANN). By embedding the pure delay properties of a lossless line, the ANN approach attempts to ensure that the key dynamic features of the real line are retained. The concept is utilized to investigate leakage fault conditions in an electrohydraulic pressure control system by considering network weight and bias changes as leakage faults are introduced at the beginning and at the end of the line. It is shown that one bias weight only is a significant indicator of the fault conditions.