Experimental comparison of interval-based parameter identification procedures for uncertain ODEs with non-smooth right-hand sides

Noninvertibility in neural networks

Contemporary tasks of complex system simulation are often related to the issue of uncertainty management. It comes from the lack of information or knowledge about the simulated system as well as from restrictions of the model set being used. One of the powerful tools for the uncertainty management is ensemble-based simulation, which uses variation in input or output data, model parameters, or available versions of models to improve the simulation performance. Furthermore the system of models for complex system simulation (especially in case of hiring ensemble-based approach) can be considered as a complex system. As a result, the identification of the complex model's structure and parameters provide additional sources of uncertainty to be managed. Within the presented work we are developing a conceptual and technological approach to manage the ensemble-based simulation taking into account changing states of both simulated system and system of models within the ensemble-based approach. The states of these systems are considered as a subject of classification with consequent inference of better strategies for ensemble evolution over the simulation time and ensemble aggregation. Here the ensemble evolution enables implementation of dynamic reactive solutions which can automatically conform to the changing states of both systems. The ensemble aggregation can be considered within a scope of averaging (regression way) or selection (classification way, which complement the classification mentioned earlier) approach. The technological basis for such approach includes ensemble-based simulation techniques using domain-specific software combined within a composite application; data science approaches for analysis of available datasets (simulation data, observations, situation assessment etc.); and machine learning algorithms for classes identification, ensemble management and knowledge acquisition.

A semi-autonomous unmanned underwater vehicle (UUV), named LAURS, is being developed at the Laboratory of Sensors and Actuators at the University of Sao Paulo. The vehicle has been designed to provide inspection and intervention capabilities in specific missions of deep water oil fields. In this work, a method of modeling and identification of yaw motion dynamic system model of an open-frame underwater vehicle is presented. Using an on-board low cost magnetic compass sensor the method is based on the utilization of an uncoupled 1-DOF (degree of freedom) dynamic system equation and the application of the integral method which is the classical least squares algorithm applied to the integral form of the dynamic system equations. Experimental trials with the actual vehicle have been performed in a test tank and diving pool. During these experiments, thrusters responsible for yaw motion are driven by sinusoidal voltage signal profiles. An assessment of the feasibility of the method reveals that estimated dynamic system models are more reliable when considering slow and small sinusoidal voltage signal profiles, i.e. with larger periods and with relatively small amplitude and offset.

: The objective of this research was to develop techniques for the design of beneficial wave dynamics in jet engines. Our approach focused on symmetry-breaking concepts and stability analysis of systems with mistuning in the presence of feedback nonlinearities. The second part of this research focused on the identification of nonlinear dynamical system models from very high-dimensional data.

The performance of a continuous-time Recursive Least Squares (CRLS) and a discrete-time Recursive Least Squares (DRLS) algorithms are examined for the growth medium temperature control of a cooling batch bioreactor in which Saccharomyces cerevisiae growth at aerobic condition by using Continuous-time Generalised Predictive Control (CGPC) algorithm. MATLAB programme was utilized for recursive parameter identification algorithms (CRLS and DRLS). The success or otherwise of these algorithms are estimated using parameter norm criterion for the various order of models and several input signals. There is a considerable improvement of identification algorithms with the reduced order of models. It has been shown that the performance of a DRLS algorithm is as successful as the other recursive parameter identification of a continuous-time system model.

Learning from data plays a major role in understanding complex natural and engineered systems. System identification (SysID), as a data-driven modelling technique, provides a powerful tool for building dynamical system models. Building models from noisy small data is a challenging research question in many practical problems. This paper is concerned with parsimonious and transparent modelling of dynamical systems which are of high interest in many real applications. Sparse Bayesian learning (SBL), due to its ability to use prior information to generate sparse predictive models, is employed in this study to estimate models from small data. The performance of the proposed sparse Bayesian learning approach is tested using real-life data. Experimental results show that the SBL approach shows strong performance for solving small data modelling problems.

Identification and robustness analysis of nonlinear hybrid dynamical system concerning glycerol transport mechanism

<div class="section abstract"><div class="htmlview paragraph">High fidelity mathematical vehicle models that can accurately capture the dynamics of car suspension system are critical in vehicle dynamics studies. System identification techniques can be employed to determine model type, order and parameters. Such techniques are well developed and usually used on linear models. Unfortunately, shock absorbers have nonlinear characteristics that are non-negligible, especially with regard the vehicle's vertical dynamics. In order to effectively employ system identification techniques on a vehicle, a nonlinear mathematical shock absorber model must be developed and then coupled to the linear vehicle model. Such an approach addresses the nonlinear nature of the shock absorber for system identification purposes. This paper presents an approach to integrate the nonlinear shock absorber model into the vehicle model for system identification.</div><div class="htmlview paragraph">Three empirical mathematical models are proposed that describe the relationship between shock velocity and shock force. These models are used to predict the force behavior of the shock absorber when the shock absorber velocities ranged from 0 to 1.5m/s, in both compression and extension direction. Empirical shock absorber models can be quickly developed using experimental data and integrated into the larger vehicle model for complete system identification. The shock velocity calculated from the vehicle model is the input to the shock absorber model and the shock force is the feedback to the vehicle model. The influence of nonlinear shock absorber models on system identification for the half car model was analyzed. The result shows that the type of shock absorber model did not have much influence on vehicle system identification using simulated data generated from the model. Furthermore, the system identification method, integrating the nonlinear shock absorber dynamics model into the vehicle model, works well for the simulated data. Future work will focus on the use of experimental data.</div></div>

An alternative method was proposed for identification of linear continuous time system with noisy output. Using Gaussian modulating filters, by numerical integration, a discrete equivalent identification model that was parametrized directly in continuous time model parameters was obtained, and the LS estimation algorithm was developed. The LS estimator is biased even with white measurement output noise, and the bias is determined by variance of output noise. On the basis of obtained output measurement noise variance, the bias was compensated and the consistent estimates of system parameters were given. Simulation which compares the standard LS with BCLS algorithm indicates that the proposed algorithm is an efficient method for noisy output continuous time system identification.

Conventional methods for estimating the transfer function models based on deterministic perturbations of the input such as step, pulse and sinusoidal changes have not always been successful because the response of the system may be masked by uncontrollable disturbances, collectively referred to as noise. Particularly in multisensor-integrated systems, the effect of noise from the sensors is not negligible and so the information on the system model and noise model is important for the accurate control of the integrated systems. In this paper, a statistical method for estimating the transfer function models with superimposed noise is presented. Similarity between the auto-correlation function of linear filter system driven by white noise and the impulse response function of dynamic system enables us to develop a method of identifying the order of dynamic systems. In order to identify the systems with superimposed noise, this paper derives R-, S-, and GPAC array which provide us with pretty clear information on the identification of dynamic systems with noise. From the derived arrays, we can obtain the order of transfer function model and also the order of noise model represented by an ARMA process. Finally, an example of order identification of transfer function model and noise model is presented to show effectiveness of the proposed algorithm.

Buildings, consuming over 70% of the electricity in the U.S., play significant roles in smart grid infrastructure. The automatic operation of buildings and their subsystems in responding to signals from a smart grid is essential to reduce energy consumption and demand, as well as improve the resilience to power disruptions. In order to achieve such automatic operation, high fidelity and computationally efficiency building energy forecasting models under different weather and operation conditions are needed. Currently, data-driven (black box) models and hybrid (grey box) models are commonly used in model based building control operation. However, typical black box models often require long training period and are bounded to weather and operation conditions during the training period. On the other hand, creating a grey box model often requires long calculation time due to parameter optimization process and expert knowledge during the model structure determining and simplification process. An earlier study by the authors proposed a system identification approach to develop computationally efficient and accurate building energy forecasting models. This paper attempts to extend this early study and to quantitatively evaluate how the most important characteristics of a building energy system: its nonlinearity and response time, affect the system identification process and model accuracy. Two commercial building: a small-size and a medium-size commercial building, with varying chiller nonlinearity, are simulated using EnergyPlus in lieu of real buildings for model development and validation. The system identification method proposed in the early study is applied to these two buildings that have varying nonlinearity and response time. Adaption of the proposed system identification method based on systems’ nonlinearity and response time is proposed in this study. The energy forecasting results demonstrate that the adaption is capable of significantly improve the performance of the system identification model.

Model recognition and validation for an off-road vehicle electrohydraulic steering controller

Simulation of system models containing zero-order causal paths—I. Classification of zero-order causal paths

In Systems Biology, there is a growing need for simulation and optimization tools for the prediction of the phenotypical behavior of microorganisms. In this paper, an open-source software platform is proposed to provide support for research in Metabolic Engineering, by implementing tools that enable the simulation and optimization of dynamic metabolic models using ordinary differential equations. Its main functionalities are related with (i) phenotype simulation of both wild type and mutant strains under given environmental conditions and (ii) strain optimization tackling tasks such as gene knockout selection or the definition of the optimal level of enzyme expression, given appropriate objective functions. The central carbon metabolism of E. coli was used as a case study, to illustrate the main features of the software.KeywordsCentral Carbon MetabolismSystem Biology Markup LanguageDihydroxyacetone PhosphateStrain OptimizationMain FunctionalityThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

System identification in the presence of trends and outliers using sparse optimization