Chosen Model Structure Research Articles

On identifying the non-linear dynamics of a hovercraft using an end-to-end deep learning approach

We present the identification of the non-linear dynamics of a novel hovercraft design, employing end-to-end deep learning techniques. Our experimental setup consists of a hovercraft propelled by racing drone propellers mounted on a lightweight foam base, allowing it to float and be controlled freely on an air hockey table. We learn parametrized physics-inspired nonlinear models directly from data trajectories, leveraging gradient-based optimization techniques prevalent in machine learning research. The chosen model structure allows us to control the position of the hovercraft precisely on the air hockey table. We then analyze the prediction performance and demonstrate the closed-loop control performance on the real system.

Kernel-based regularized iterative learning control of repetitive linear time-varying systems

For data-driven iterative learning control (ILC) methods, both the model estimation and controller design problems are converted to parameter estimation problems for some chosen model structures. It is well-known that if the model order is not chosen carefully, models with either large variance or large bias would be resulted, which is one of the obstacles to further improve the modeling and tracking performances of data-driven ILC in practice. An emerging trend in the system identification community to deal with this issue is using regularization instead of the statistical tests, e.g., AIC, BIC, and one of the representatives is the so-called kernel-based regularization method (KRM). In this paper, we integrate KRM into data-driven ILC to handle a class of repetitive linear time-varying systems, and moreover, we show that the proposed method has ultimately bounded tracking error in the iteration domain. The numerical simulation results show that in contrast with the least squares method and some existing data-driven ILC methods, the proposed one can give faster convergence speed, better accuracy and robustness in terms of the tracking performance.

Gender Neutralisation for Unbiased Speech Synthesising

Machine learning can encode and amplify negative biases or stereotypes already present in humans, resulting in high-profile cases. There can be multiple sources encoding the negative bias in these algorithms, like errors from human labelling, inaccurate representation of different population groups in training datasets, and chosen model structures and optimization methods. Our paper proposes a novel approach to speech processing that can resolve the gender bias problem by eliminating the gender parameter. Therefore, we devised a system that transforms the input sound (speech of a person) into a neutralized voice to the point where the gender of the speaker becomes indistinguishable by both humans and AI. Wav2Vec based network has been utilised to conduct speech gender recognition to validate the main claim of this research work, which is the neutralisation of gender from the speech. Such a system can be used as a batch pre-processing layer for training models, thus making associated gender bias irrelevant. Further, such a system can also find its application where speaker gender bias by humans is also prominent, as the listener will not be able to judge the gender from speech.

GS-1. First 1000 days of life and non communicable diseases nexus

In the Prediction Error identification framework, it is crucial that the collected data are informative with respect to the chosen model structure to get a consistent estimate. In this work, we focus on the data informativity property for the identification of multi-inputs single-output ARX systems in closed-loop and we derive a necessary and sufficient condition to verify if a given multisine external excitation combined with the feedback introduced by the controller yields informative data with respect to the chosen model structure.

Data Informativity for the Closed-Loop Identification of MISO ARX Systems

In the Prediction Error identification framework, it is crucial that the collected data are informative with respect to the chosen model structure to get a consistent estimate. In this work, we focus on the data informativity property for the identification of multi-inputs single-output ARX systems in closed-loop and we derive a necessary and sufficient condition to verify if a given multisine external excitation combined with the feedback introduced by the controller yields informative data with respect to the chosen model structure.

Assessment of the parameter identifiability of population balance models for air jet mills

Air jet mills are ubiquitous breakage devices not only in the pharmaceutical industry, but also in food and the toner manufacturing industry. The popularity of air jet mills arises due to its self-classifying, non-contaminating, and non-degrading operation while also maintaining a narrow particle size distribution. A popular approach towards mathematically modelling comminution devices like the jet mill is the population balance model framework. Population balance model for breakage is a semi-empirical framework in which several parameters are estimated from the experimental data. In many cases the experimental data available is insufficient, or of bad quality to guarantee the estimation of unique parameters. Thus it is imperative to assess the identifiability of such models to ensure that the chosen model structure is identifiable, both structurally and practically. In this study, we analyse the identifiability of the population balance model developed for the air jet mill. The identifiability analysis is carried out on basis of in-silico data generated by assuming certain parameter values to represent the true process. Based on several scenarios tested with data containing variety of information, it is shown that the model is practically non-identifiable for the situations in which the particle size measurement is available only at the end of the milling. The model becomes identifiable once continuous time measurements are available. The same holds for a reduced model with only the five breakage parameters.

Closed-loop identification of MIMO systems in the Prediction Error framework: Data informativity analysis

In the Prediction Error identification framework, it is essential that the experiment yields informative data with respect to the chosen model structure to get a consistent estimate. In this work, we focus on the data informativity property for the identification of multi-inputs multi-outputs systems in closed-loop and we derive conditions to verify if a given external excitation combined with the feedback introduced by the controller yields informative data with respect to the chosen model structure. This study covers the case of multiple-inputs multiple-outputs model structures that are natural generalizations of the classical single-input single-output model structures used in Prediction Error identification and the classical types of external excitation vectors, i.e., vectors whose elements are either multisine or filtered white noises.

Design of experiments for nonlinear system identification: A set membership approach

Design of Experiments (DoE) is an important step in system identification. Regardless of the chosen model structure and identification method, the DoE quality determines an upper bound on the accuracy of the identified models. One of the greatest challenges in this context is to design an experiment which gives the maximum information about the dynamics of the system of interest. In this paper, a novel DoE algorithm for input-constrained MISO nonlinear systems, based on set membership identification, is proposed. The DoE algorithm is aimed to minimize the so-called radius of information, a quantity giving the worst-case model error. Two numerical examples are presented, showing the effectiveness of the approach and its potential in view of real-world applications.

Data informativity for the open-loop identification of MIMO systems in the prediction error framework

In Prediction Error identification, to obtain a consistent estimate of the true system, it is crucial that the input excitation yields informative data with respect to the chosen model structure. We consider in this paper the data informativity property for the identification of a Multiple-Input Multiple-Output system in open-loop and we derive conditions to check whether a given input vector will yield informative data with respect to the chosen model structure. We do that for the classical model structures used in prediction error identification and for the classical types of input vectors, i.e., input vectors whose elements are either multisines or filtered white noises.

Data Informativity for the Identification of particular Parallel Hammerstein Systems

To obtain a consistent estimate when performing an identification with Prediction Error, it is important that the excitation yields informative data with respect to the chosen model structure. While the characterization of this property seems to be a mature research area in the linear case, the same cannot be said for nonlinear systems. In this work, we study the data informativity for a particular type of Hammerstein systems for two commonly-used excitations: white Gaussian noise and multisine. The real life example of the MEMS gyroscope is considered.

Analysis of the Informative Data of Industrial Data-Based Modelling

Modelling plays an important role in continuously operated plants as in chemical, metallurgical and aerospace fields, directly affecting subsequent researches of industrial processes. Rapid industrial development requires a high-quality model. Generally, external excitation is applied to the industrial process, ensuring the informative data for system identification. However, these processes are regularly operated to achieve a secure operation and to meet production objectives. For the safety and avoid violating product quality, external excitation can be forbidden or limited in the plant. Historical records are logged, which is natural to use them for analysis. These data are available at less, even no cost for identification. In this paper, based on the definition of information matrix and attenuating excitation, a detailed standard is aimed to propose to help extract informative data with respect to the chosen model structure to support system identification. The assessment of this standard is evaluated in a case, and the benefits are demonstrated, which ensures identification information enough, decreases information waste and achieves less, even no impact or cost of industrial processes.

A Dynamic Model for Indoor Temperature Prediction in Buildings

A novel dynamic model for the temperature inside buildings is presented, aiming to improve energy efficiency by providing predictive information on the heat demand. To analyse the performance and generalizability of the modelling approach, real measurement data was gathered from five different types of buildings. Easily available data from various sources was utilized. The chosen model structure leads to a minimal number of input variables and free parameters. Simulations with real data from five buildings, and applying the identical model structure showed that the average modelling error during the 28-h prediction horizon was constantly below 5%. The results thus demonstrate that the model structure can be standardized and easily applied to predict the indoor temperatures of large buildings. This would finally enable demand side management and the predictive optimization of the heat demand at city level.

Time-varying parameter models for catchments with land use change: the importance of model structure

Abstract. Rapid population and economic growth in Southeast Asia has been accompanied by extensive land use change with consequent impacts on catchment hydrology. Modeling methodologies capable of handling changing land use conditions are therefore becoming ever more important and are receiving increasing attention from hydrologists. A recently developed data-assimilation-based framework that allows model parameters to vary through time in response to signals of change in observations is considered for a medium-sized catchment (2880 km2) in northern Vietnam experiencing substantial but gradual land cover change. We investigate the efficacy of the method as well as the importance of the chosen model structure in ensuring the success of a time-varying parameter method. The method was used with two lumped daily conceptual models (HBV and HyMOD) that gave good-quality streamflow predictions during pre-change conditions. Although both time-varying parameter models gave improved streamflow predictions under changed conditions compared to the time-invariant parameter model, persistent biases for low flows were apparent in the HyMOD case. It was found that HyMOD was not suited to representing the modified baseflow conditions, resulting in extreme and unrealistic time-varying parameter estimates. This work shows that the chosen model can be critical for ensuring the time-varying parameter framework successfully models streamflow under changing land cover conditions. It can also be used to determine whether land cover changes (and not just meteorological factors) contribute to the observed hydrologic changes in retrospective studies where the lack of a paired control catchment precludes such an assessment.

Noncollinear Two-Component Quasirelativistic Description of Spin Interactions in Exchange-Coupled Systems. Mapping Generalized Broken-Symmetry States to Effective Spin Hamiltonians.

We provide a consistent mapping of noncollinear two-component quasirelativistic DFT energies with appropriate orientations of localized spinor quantization axes for multinuclear exchange-coupled transition-metal complexes onto an uncoupled anisotropic effective spin Hamiltonian. This provides access to the full exchange interaction tensor between the centers of spin-coupled systems in a consistent way. The proposed methodology may be best viewed as a generalized broken-symmetry density functional theory approach (gBS-DFT). While the calculations provided are limited to trinuclear systems ([M3O(OOCH)6(H2O)3]+, where M = Cr(III), Mn(III), Fe(III)) with C3 symmetry, the method provides a general framework that is extendable to arbitrary systems. It offers an alternative to previous approaches to single-ion zero-field splittings, and it provides access to the less often examined antisymmetric Dzyaloshinskii-Moriya exchange interaction. Spin-orbit coupling is included self-consistently. This will be of particular importance for complexes involving 4d or 5d transition metal centers or possibly also for f-block elements, where a perturbational treatment of spin-orbit coupling may not be valid anymore. While a comparison with experimental data was indirect due to simplifications in the chosen model structures, the agreement obtained indicates the essential soundness of the presented approach.

Design and experiment of data-driven modeling and flutter control of a prototype wing

This paper presents an approach for data-driven modeling of aeroelasticity and its application to flutter control design of a wind-tunnel wing model. Modeling is centered on system identification of unsteady aerodynamic loads using computational fluid dynamics data, and adopts a nonlinear multivariable extension of the Hammerstein-Wiener system. The formulation is in modal coordinates of the elastic structure, and yields a reduced-order model of the aeroelastic feedback loop that is parametrized by airspeed. Flutter suppression is thus cast as a robust stabilization problem over uncertain airspeed, for which a low-order H∞ controller is computed. The paper discusses in detail parameter sensitivity and observability of the model, the former to justify the chosen model structure, and the latter to provide a criterion for physical sensor placement. Wind tunnel experiments confirm the validity of the modeling approach and the effectiveness of the control design.

Global sampling to assess the value of diverse observations in conditioning a real‐world groundwater flow and transport model

AbstractThe use of additional types of observational data has often been suggested to alleviate the ill‐posedness inherent to parameter estimation of groundwater models and constrain model uncertainty. Disinformation in observational data caused by errors in either the observations or the chosen model structure may, however, confound the value of adding observational data in model conditioning. This paper uses the global generalized likelihood uncertainty estimation methodology to investigate the value of different observational data types (heads, fluxes, salinity, and temperature) in conditioning a groundwater flow and transport model of an extensively monitored field site in the Netherlands. We compared model conditioning using the real observations to a synthetic model experiment, to demonstrate the possible influence of disinformation in observational data in model conditioning. Results showed that the value of different conditioning targets was less evident when conditioning to real measurements than in a measurement error‐only synthetic model experiment. While in the synthetic experiment, all conditioning targets clearly improved model outcomes, minor improvements or even worsening of model outcomes was observed for the real measurements. This result was caused by errors in both the model structure and the observations, resulting in disinformation in the observational data. The observed impact of disinformation in the observational data reiterates the necessity of thorough data validation and the need for accounting for both model structural and observational errors in model conditioning. It further suggests caution when translating results of synthetic modeling examples to real‐world applications. Still, applying diverse conditioning data types was found to be essential to constrain uncertainty, especially in the transport of solutes in the model.

Automated generation of feedforward control using feedback linearization of local model networks

An effective but yet simple approach is introduced to automatically attain a dynamic feedforward control law for non-linear dynamic systems represented by discrete-time local model networks (LMN). In this context, feedback linearization is applied to the generic model structure of LMN and the resulting input transformation is used as model inverse. This general and automated approach for model inversion is applicable even when the overall model complexity may be high. Thus, by representing a non-linear dynamic system by an LMN and applying the proposed feedforward control law generation, a dynamic feedforward control for such a non-linear system can be found automatically with the knowledge of measured input–output data only. However, when feedback linearization is considered, the stability of the internal dynamics plays a key role. This paper analyses the occurring internal dynamics for LMN, which directly result from the chosen model structure in identification, and discusses the effects on the transformed system. Finally, the effectiveness of the proposed data-driven feedforward control is demonstrated by a simulation example as well as by an actual application to the pre-distortion of a microelectromechanical systems (MEMS) loudspeaker with electrostatic actuation.

Pertussis models to inform vaccine policy

Pertussis remains a challenging public health problem with many aspects of infection, disease and immunity poorly understood. Initially controlled by mass vaccination, pertussis resurgence has occurred in some countries with well-established vaccination programs, particularly among adolescents and young adults. Several studies have used mathematical models to investigate drivers of pertussis epidemiology and predict the likely impact of different vaccination strategies. We reviewed a number of these models to evaluate their suitability to answer questions of public health importance regarding optimal vaccine scheduling. We critically discuss the approaches adopted and the impact of chosen model structures and assumptions on study conclusions. Common limitations were a lack of contemporary, population relevant data for parameterization and a limited understanding of the relationship between infection and disease. We make recommendations for future model development and suggest epidemiologic data collections that would facilitate efforts to reduce uncertainty and improve the robustness of model-derived conclusions.

Computationally Efficient Globally Linearizing Control of a CSTR using Nonlinear Black Box Models

The main objective of this work is to develop computationally efficient Global linearization based control (GLC) schemes, which are suitable for control of nonlinear processes with fast dynamics. Models used for the controller synthesis are discrete time block oriented nonlinear black box state space models identi=ed directly from the input-output perturbation data. The chosen model structures facilitate construction of closed form solutions to the unconstrained GLC formulations. The efficacy of the proposed control formulations is evaluated by conducting simulation studies on a benchmark continuously stirred tank reactor (CSTR) system which exhibits input multiplicity behavior. Analysis of the simulation results reveals that the proposed GLC formulations are able to achieve a signi=cant reduction in the average computation time without compromising on the closed loop performance.

ROBUST, COMPARABLE POPULATION METRICS THROUGH COLLABORATIVE PHOTO-MONITORING OF WHALE SHARKSRHINCODON TYPUS

The formulation of conservation policy for species that are rare and migratory requires broad cooperation to ensure that adequate levels of standardized data collection are achieved and that the results of local analyses are comparable. Estimates of apparent survival rate, relative change in abundance, and proportions of newly marked and returning individuals can inform local management decisions while highlighting corresponding changes at other linked research stations. We have applied computer-assisted photo-identification and mark-recapture population modeling to whale sharks Rhincodon typus at Ningaloo Marine Park (NMP), Western Australia, to create a baseline trend for comparison with other regional aggregations of the species. We estimate several ecological parameters of interest, including an average apparent survival rate of 0.55 yr(-1) for sharks newly marked (new) and 0.83 yr(-1) for sharks captured in multiple seasons (philopatric). The average proportion of philopatric sharks is found to be 0.65 of the total population, and we derive an average population growth rate of 1.12 yr(-1) for them. Our analysis uncovered significant heterogeneity in capture and survival probabilities in this study population; our chosen model structures and data analysis account for these influences and demonstrate a good overall fit to the time-series data. The results show good correspondence between capture probability and an available measure of recapture effort, suggesting that unmodeled systematic effects contribute insignificantly to the model fits. We find no evidence of a decline in the whale shark population at NMP, and our results provide metrics of value to their future management. Overall, our study suggests an effective approach to analyzing and modeling mark-recapture data for a rare species using computer-assisted photo-identification and opportunistic data collection from ecotourism to ensure the quality and volume of data required for population analysis.