Modeling Of Complex Dynamic Systems Research Articles

Analysis of biological networks

Analysis of biological networks

Static and dynamic complex models: comparison and application to chemical systems

PurposeThe purpose of this paper is to present research in the area of the modeling of complex systems using feed‐forward neural network.Design/methodology/approachApplications of multilayer neural networks with supervisor learning on the own simulator program wrote in Borland® Pascal Language. Series‐parallel identification method is applied. Tapped delay lines (TDL) in static neural networks for modeling of dynamic plants are used. Gradient and heuristic learning algorithms are applied. Three kinds of calibration of learning and testing data are used.FindingsThis paper illustrates that feed‐forward multilayer neural networks can model complex systems. Feed‐forward multilayer neural networks with TDL can be used to build global dynamic models of complex systems. It is possible to compare the quality both models.Research limitations/implicationsThe learning and testing data from real systems to tune neuronal models require use of calibrating these data to range 0‐1.Practical implicationsThe models quality depends on kind of calibration learning data from real system and depends on kind of learning algorithms.Originality/valueThe method and the learning algorithms discussed in the paper can be used to create global models of complex systems. The multilayer neural network with TDL can be used to model complex dynamic systems with low dynamics.

Uncertainty in structural dynamics: Experimental validation of a Wishart random matrix model

In the modeling of complex dynamical systems, high-resolution finite element models are routinely adopted to reduce the discretization error. Such an approach may fail to enhance the confidence in simulation-based predictions when the dynamical systems exhibit significant variability in the parameterization and description of the mathematical model due to parametric (data) uncertainty and model uncertainty. In contrast to parametric uncertainty, model uncertainty poses significant challenges as no explicit parameter is available a priori to characterize its behavior. In the current investigation, a set of sprung-mass oscillators randomly attached to a baseline system (namely, a homogeneous cantilever plate) gives rise to model uncertainty. Contrary to model parametric uncertainty traditionally tackled using stochastic finite element method (SFEM), the model uncertainty arising from randomly attached sprung-masses gives rise to a new variant of dynamical system for every sample. The feasibility of adopting a Wishart random matrix to represent such ensemble of dynamical systems derived from model perturbation is investigated in this paper. The experiments conducted on a vibrating plate with randomly attached sprung-mass oscillators demonstrate that the Wishart random matrix model may provide a reasonable representation of model uncertainty in linear dynamical systems in the medium and high-frequency region.

Rough intervals—enhancing intervals for qualitative modeling of technical systems

The success of model-based industrial applications generally depends on how exactly models reproduce the behavior of the real system that they represent. However, the complexity of industrial systems makes the construction of accurate models difficult. An alternative is the qualitative description of process states, for example by means of the discretization of continuous variable spaces in intervals. In order to reach the required precision in the modeling of complex dynamic systems, interval-based representations usually produce qualitative models, which are sometimes too large for practical use. The approach introduced in this paper incorporates vague and uncertain information based on principles of the Rough Set Theory as a way of enhancing the information contents in interval-based qualitative models. The resulting models are more compact and precise than ordinary qualitative models.

Reactive animation: realistic modeling of complex dynamic systems

Some complex systems are transformational, of an input-process-output type, repeatedly carrying out their prescribed work for each new set of inputs. A far more problematic class comprises large systems that are heavily controlor event-driven. Such systems (dubbed reactive because they react to various kinds of events, signals, and conditions in intricate ways) are often concurrent and distributed. Reactive animation, a patent pending technology, offers great promise as a way to enhance the visualization of system behavior, especially for reactive systems, which are a central part of most current computerized technologies.

Bias Distribution in Continuous-Time Transfer Function Estimates by Prediction Error Method

Estimation of simple transfer function models of complex dynamical systems is an important issue for robustness of all subsequent application stages of the models, such as control, fault detection etc. Here we consider an RLS approach to s-transfer function estimation based on prediction error minimization. An analysis of the algorithm determines the influence of the estimator design variables on the model bias distribution over frequency. A numerical example demonstrates this effect and provides guidelines for determining these to suit a given specification.

CAUSA—A TOOL FOR MODEL-BASED KNOWLEDGE ACQUISITION

CAUSA is a knowledge acquisition tool which supports the incremental modeling of complex dynamic systems during the whole knowledge acquisition task. It provides an environment for the modeling and simulation of dynamic systems on a quantitative level. The environment provides a conceptual framework which includes primitives like objects, processes, and causal dependencies, allowing for the modeling of a broad class of complex systems. Simulation allows for the quantitative and qualitative inspection and empirical investigation of the behavior of the modeled system. CAUSA is implemented in Knowledge-Craft and runs on a Symbolics 3640.

Modelling of complex dynamic systems

Modelling of complex dynamic systems

Digital emulation of analog computer techniques for the solution of kinetic systems

AbstractA low‐cost microcomputer and package of assembly language routines has been developed to emulate the structure and performance of a large analog computer. The advantages of the analog computer, as implemented in this scheme, include (1) a significant reduction in the programming effort involved in modeling complex dynamic systems and (2) the control of the simulation and model parameters in a completely interactive and flexible manner. The symbolic nomenclature and schematic representations involving devices, such as integrators, comparators, multipliers, and function generators, offers a powerful alternative to the more conventional numerical methods, that is, to provide very simply the solutions to large systems of differential equations. This approach invariably leads the user to a more thorough understanding of the dynamic character of the system. The technique is illustrated using a chemical kinetics example involving the simulation of laser‐induced fluorescence. The results of this work have provided an assessment of a systematic error that occurs when using induced resonance fluorescence to measure OH concentrations in the troposphere of the earth.

Loss and coupling loss factors of two coupled dynamic systems

The statistical Energy Analysis (SEA) has been used to define and estimate the responses of complex dynamic systems. Often a complex dynamic system is modeled by dividing it into a number of subsystems—called basic dynamic systems. Within the format of SEA, the description of the model is given in terms of the loss and coupling loss factors. But to define the model, means must be devised for determining the values of these factors. This paper examines experimental methods that can be used to identify the factors. Considerations are limited to a complex consisting of two dynamic systems, but the elements for an extension to higher forms of complexes are included. The loss and coupling loss factors are formulated in terms of standard measured quantities obtained from experiments which involve quasi-transient as well as steady-state conditions. The formalism is used to ascertain the sensitivity and accuracy with which the factors can be estimated. It is demonstrated that not all the factors can usually be estimated with equal accuracy; indeed, a factor may escape determination altogether. The results may thus guide experimental schemes and procedures used in modeling complex dynamic systems in the format of SEA.