Search

This letter presents a data-based control approach to achieve high-performance trajectory tracking with Unmanned Aerial Vehicles (UAVs). We revisit an existing Iterative Learning Control (ILC) algorithm based on the notion that the performance of a system that executes the same task multiple times can be improved by learning from previous executions. While we will specifically refer to multirotor platforms for the experimental validation, the formulation can be applied to any dynamic system (including systems with underlying feedback loops). The novelty of this work is the introduction of a smoother to estimate the repetitive disturbance to improve the learning performance. This estimator must rely on an accurate system model that has been obtained through a black-box identification procedure using the Predictor-Based Subspace Identification (PBSID) algorithm. A Monte Carlo analysis has been carried out with the aim of showing the performance improvements and limitations of the proposed algorithm with respect to existing approaches. Finally, the proposed approach has been validated through experimental activities involving a small quadrotor performing an aggressive manoeuver.

This study focuses on the model identification problem of an unmanned helicopter. Two identification approaches are evaluated, a subspace method and a combined method. The work gives detailed analysis of time-domain and frequency-domain simulations. Results show that both approaches can match the time-domain and frequency-domain responses of the original data in a good degree. The combined method is reliable in extracting structured dynamic models.

Aerospace control education can significantly benefit from actual hands-on experience. In most cases, however, such experience can only be provided to students in small-scale project activities. In this paper the experience of the UAV Lab course, held at Politecnico for the first time in the fall of 2018, is presented and discussed. The course, aimed at an interdisciplinary group of students, covers the whole design cycle for a multirotor UAV, from conceptual design to in-flight validation, with specific reference to modelling, simulation, identification and control. The course has been conceived as an extra-curricular activity, so the emphasis is not on conventional lectures but rather on hands-on experience in hardware/software integration, data collection and analysis and flight testing. The paper presents the course syllabus and organisation and provides an overview of the obtained results and the feedback provided by the students.

System identification is now a well established approach for the development of control-oriented models in the rotorcraft field (see, e.g., the survey paper Hamel and Kaletka (1997), the recent books Tischler and Remple (2006), Jategaonkar (2006) and the references therein). Though the application to full scale rotorcraft is by now fairly mature, less experience has been gathered on small-scale vehicles, such as, e.g., quadrotors. This paper deals with the problem of characterizing the attitude dynamics of a variable-pitch quadrotor from data and presents the results obtained in an experimental identification campaign. More precisely, on-line and off-line methods have been considered and the performance of black-box versus grey-box models has been compared.

Accurate dynamic modeling of helicopter aeromechanics is becoming increasingly important, as progressively stringent requirements are being imposed on rotorcraft control systems. System identification plays an important role as an effective approach to the problem of deriving or fine tuning mathematical models for purposes such as handling qualities assessment and control system design. In this paper, the problem of deriving continuous-time models for the dynamics of a small-scale quadrotor helicopter is considered. More precisely, the continuous-time predictor-based subspace identification approach is adopted and the results obtained in an experimental study are presented and discussed.

Subspace model identification algorithms have become extremely popular in the last few years thanks to the ease with which they can provide consistent estimates for MIMO state space models in a non iterative way, exploiting a full parameterisation of the system matrices. The only, well known, downside of this approach is the impossibility to impose a fixed basis to the state space representation, and therefore the difficulties in recovering physically-motivated models. The problem considered in this paper is the one of recovering the numerical values of the physical parameters of a structured representation of the system starting from a fully parameterised identified model. It will be shown that this can be achieved without explicitly constructing the similarity transformation, for linear time-invariant systems, linear time-periodic systems and linear parameter-varying systems identified from a periodic scheduling sequence. Two numerical examples are presented to illustrate the approach.

Information technology (IT) systems must provide their users with prescribed quality of service (Qos) levels, usually defined in terms of application performance. Qos requirements are in general difficult to satisfy, since the system workload may vary by orders of magnitude within the same business day. To meet Qos requirements, resources have to dynamically be allocated among running applications, re-configuring them at run-time. To deal with resource allocation so as to manage system overload issues admission control and server virtualization are typically used: the former is a protection mechanism that rejects requests under peak workload, whereas the latter allows partitioning physical resources such as CPU and disks into multiple virtual ones. For designing effective controllers to ensure the desired Qos levels, a reliable model of the server dynamics is needed. The given systems, while retaining the time-varying nature that allows one to model workload variations. To address this issue, a constrained black-box subspace identification approach endowed with a novel structure selection is designed, and the performance of the identified models are assessed on experimental data.

Flow routing models are an essential tool for water systems planning and management. In the last few years, advances in modelling capability, enhanced data availability, and increasing computational power have made the use of large, process-based, flow-routing models more common. However, the computational burden of such models still prevents their systematic application for operational purposes. This paper presents an alternative, data-driven approach to the identification of a flow-routing model both in one-step-ahead prediction and simulation mode, and demonstrates it by application to the Red River network, in Northern Vietnam. In particular, MIMO time-invariant models for the network are considered, their parameters are estimated using subspace model identification techniques and their performance is assessed with respect to the available data. Results show that the non-linear process of flow propagation is accurately reproduced both in prediction and simulation using very low order models.

The complexity of information technology (IT) systems is steadily increasing. System complexity has been recognized as the main obstacle to the further advancement of IT and has recently raised energy management issues. Control techniques have been proposed and successfully applied to design autonomic computing systems, i.e., systems able to manage themselves trading-off system performance with energy reduction goals. As users' behavior is highly time varying and workload conditions can change substantially within the same business day, the linear parametrically varying (LPV) framework proves particularly suitable for modeling such systems. In this paper, the identification of single-input-single-output and multiple-input-multiple-output state space LPV models for the performance control of autonomic web service systems is addressed. Specifically, subspace LPV identification methods are shown to yield accurate dynamic models for the considered application. Their effectiveness is assessed on experimental data measured on a custom implementation of a workload generator and micro-benchmarking Web service applications.

The nine regular and three brief papers in this special issue present the current state-of-the-art in the neighboring fields of Linear Parametrically Varying (LPV) modeling and system identification.

Black-box modelling of the Rainfall-Runoff Relationship (RRR) in urban drainage systems is not a new challenge for system identification, as the knowledge of the RRR is an essential tool in modelling and design of urban drainage networks. In the hydraulic practice the RRR is described through physical-based models, so-called conceptual models, which are completely characterised by a restricted set of design parameters. In previous work, a calibration procedure for conceptual models based on the estimation of black-box dynamic models for the RRR has been proposed, thereby shifting the problem from the actual tuning of conceptual models to the identification of black-boxmodels on the basis of rainfall and runoff measurements. In this paper the problem of identifying black-boxmodels for the RRR of small urban catchments is considered in a parametrically-varying setting. More precisely, both linear and nonlinear parametrically varying models are identified on the basis of experimental data for two small urban catchments and their performance is compared with previous results available in the literature.

In service oriented environments, service providers need to comply with the service level objectives (SLO) stipulated in contracts with their customers. Recently, this problem has begun to be formalized in terms of a constrained control problem, where SLOs are translated into set-points for the response time of the servers, possibly different according to the customer class. As the behavior of the server response time is highly time varying and the workload conditions substantially change within the same business day, the LPV framework seems very promising for modeling such systems. In this paper, the identification of multivariable LPV models for the performance control of Web services in virtualized hosting environments is addressed. Specifically, the suitability of subspace LPV identification methods is investigated and their performance assessed on experimental data.

Model identification of polynomial NARX models involves a lengthy and computationally intensive procedure for selecting the model structure among a possibly large set of candidate regressors. If the model structure is under-parameterized to reduce the burden of the model selection phase, unsatisfactory results are generally obtained. This inaccuracy problem can be somewhat circumvented by focusing the identification process on the obtainment of an accurate local model over a specific frequency range. Such frequency tailoring is achieved in the nonlinear modeling framework by direct error filtering, as opposed to the data pre-filtering practice adopted in the linear context. This work discusses the application of error filtering to classical NARX model identification methods. A simulation example is provided to show the performance of the proposed approach in deriving accurate local models, despite an under-parameterized model structure. It is also shown that a proper error filtering may increase the model accuracy in simulation with respect to available identification techniques.

Linear Fractional Transformations (LFTs) are a widely used modelling framework in modern control and identification. Deriving such models from physical first principles is a non-trivial and often error-prone process, if carried out manually. In this paper a new approach to reduced-order LFT modelling and identification starting from equation-based, object-oriented (O-O) descriptions of the plant (formulated using the Modelica language) and experimental data are presented. This approach allows to reduce the gap between user-friendly model representations, based on object diagrams with physical connections, block diagrams with signal connections, and generic differential-algebraic models, and the use of advanced LFT based identification and control techniques.

This work addresses the identification of a control-oriented multivariable model of the structural response of a helicopter, with reference to the Agusta A109 helicopter. The estimated model will be used to design an active vibration control system to reduce the vibrations transmitted to the cabin by the main gearbox. The considered system consists of a set of piezoelectric actuators mounted at the interface between the gearbox and the cabin and accelerometers located on the cabin roof. Identification data are generated by independent excitation of each individual actuator. Model identification is carried out using a subspace identification algorithm.

Data pre-filtering is often used in linear identification to increase model accuracy in a specified frequency band, as pre-filtering is equivalent to a frequency weighting on the prediction error function. However, this interpretation applies only to a strictly linear setting of the identification problem. In this paper, the role of data and error pre-filtering in nonlinear system identification is analyzed and a frequency domain interpretation is provided, based on the use of Volterra series representations for the true system and the considered model structure. Simulation results illustrate the conclusions of the analysis.

The problem of the recursive formulation of the MOESP class of subspace identification algorithms is considered and two novel instrumental variable approaches are introduced. The first one leads to an RLS-like implementation, the second to a gradient type iteration. The relative merits of both approaches are analysed and discussed, while simulation results are used to compare their performance with one of the existing techniques. Copyright © 2004 John Wiley & Sons, Ltd.

A considerable amount of work has been dedicated in the past to the problem of the system identification of helicopter flight dynamics, while much less activity has been oriented to the goal of developing suitable identification procedures for rotor dynamics, mainly because of the difficulties associated with the task. This paper shows that subspace and optimization based identification techniques can be used to determine discrete-time linear parameter-varying models that have the potential to provide accurate descriptions for the (intrinsically time-varying) dynamics of a rotor blade. The identification techniques are presented and applied to simulated data generated by a physical model that describes the out-of-plane bending dynamics of a helicopter rotor blade.

In this paper a novel approach to the problem of star identification for spacecraft attitude determination based on data provided by a star camera is proposed. The algorithm is based on pattern matching ideas and provides improved certainty with respect to similar methods available in the literature. The achievable performance is demonstrated via simulation results.

A considerable amount of work has been dedicated in the past to the problem of identification of helicopter flight dynamics, both in the time and frequency domain, however, limited attention has been devoted so far to the application of subspace methods to the problem. The aim of this paper is to show that subspace based identification techniques can be used to determine accurate discrete-time and continuous-time linear models for helicopter dynamics. The identification techniques are applied to simulated data generated by a physical model that describes coupled rotor-fuselage dynamics for a realistic rotorcraft.

The problem of MIMO state space recursive identification is considered and analyzed using subspace model identification (SMI) techniques. In this paper the use of projection tracking techniques for the update of the observability subspace is proposed: existing results are used for the output error case and a novel instrumental variable (IV) projection tracking approach is proposed, to accommodate for arbitrary correlation of the disturbances. Simulation results show the performance achievable with the given algorithms.

This paper deals with the problem of robust control design for the attitude dynamics of a variable-pitch quadrotor. The proposed approach is based on structured H∞ synthesis applied to a black-box model of the quadrotor identified in dedicated bench-test experiments. An integrated procedure aimed at a fast and reliable deployment of the vehicle’s attitude control system was developed, covering identification and optimization-based tuning for the parameters of the on-board controller structure, based on desired closed-loop performance requirements. The goal is to demonstrate that the obtained attitude controller, achieved without risky and time consuming in flight identification test campaign, guarantees satisfactory performance in flight, in near hovering conditions.

The problem of estimating affine linear parametrically varying models is considered. An algorithm based on a Kalman filtering approach is proposed and its performance evaluated in a simulation study.

The problem of MIMO recursive identification is analyzed within the framework of subspace model identification (SMI) and the use of recent signal processing algorithms for the recursive update of the singular value decomposition (SVD) is proposed. To accommodate for arbitrary correlation of the disturbances, an instrumental variable (IV) approach is followed. In particular, recursive formulations for the subspace identification algorithms of the multivariable output-error state space (MOESP) class are given. A recursive algorithm for the identification of non-linear models of the Wiener type is also obtained.

The aim of the work presented in this paper is the evaluation of the performance of state-space subspace-based identification methods when applied to the identification of linear models for a flexible structure. A simulation study has been performed to this purpose, by making use of a high order model of a rotating flexible beam and three identification algorithms have been compared. The results show that the considered algorithms provide very similar performances and they all represents viable approaches to the identification problem.

The problem of black box identification of a linear MIMO model for a distillation column is considered and two identification algorithms are compared, namely the prediction error minimization method for the estimation of I/0 models and the N4SID subspace method for the estimation of state space models. A reasoned comparison of the two identification methods on the basis of experimental data shows that the state space method can provide a more accurate representation of the plant than the prediction error method.

The problem of black box identification of a linear MIMO model for a distillation column is considered and the N4SID subspace based algorithm for the identification of linear state space models is applied to a set of experimental data. Model order selection and experiment design are based on a simulation study, performed on a non linear model of the plant. The comparison of the response of the identified model with the response of the plant shows a very satisfactory degree of accordance.

In this study, the authors present an overview of closed-loop subspace identification methods found in the recent literature. Since a significant number of algorithms has appeared over the last decade, the authors highlight some of the key algorithms that can be shown to have a common origin in autoregressive modelling. Many of the algorithms found in the literature are variants on the algorithms that are discussed here. In this study, the aim is to give a clear overview of some of the more successful methods presented throughout the last decade. Furthermore, the authors retrace these methods to a common origin and show how they differ. The methods are compared both on the basis of simulation examples and real data. Although the main focus in the literature has been on the identification of discrete-time models, identification of continuous-time models is also of practical interest. Hence, the authors also provide an overview of the continuous-time formulation of the identification framework.

The current state-of-the-art in the fields of control-oriented LPV modelling and LPV system identification is surveyed and the potential synergies between the two research areas are highlighted and discussed. Indeed, a number of methods and tools for the development of LPV models from nonlinear systems and for the identification of black-box LPV models from input/output data have been derived, in a rather independent way, in different research communities. The relative merits of analytical and experimental methods for the derivation of LPV models, as well as possible combinations of the two approaches, are analysed and eventually evaluated on a case study based on the modelling of a thermo-fluid system.

The problem of rotorcraft system identification is considered and a novel, two step technique is proposed, which combines the advantages of time domain and frequency domain methods. In the first step, the identification of a black-box model using a subspace model identification method is carried out, using a technique which can deal with data generated under feedback; subsequently, in the second step, a-priori information on the model structure is enforced in the identified model using an H ∞ model matching method. A simulation study is used to illustrate the proposed approach.

This paper deals with the problem of model identification in continuous-time using subspace techniques. More precisely, a recently presented continuous-time predictor-based subspace identification algorithm which relies on a system transformation using the Laguerre basis is considered and a bootstrap-based approach to the problem of quantifying the variance error associated with the identified models is proposed.

This work analyzes the performance of several black box nonlinear model identification techniques for input–output models with polynomial nonlinearities on a benchmark identification problem. The case study, proposed in Schoukens, Suykens, and Ljung (2008) , concerns a nonlinear SISO electronic system with a Wiener–Hammerstein structure, originally documented in Vandersteen (1997) . The objective being the obtainment of an accurate simulation model, capable of replicating the dynamic behavior of the system without using past measured output data, various output-error approaches have been tested and compared with more standard equation-error techniques. The provided analysis shows that excellent modeling performance can be obtained with these methods even without explicitly taking into account the block structure of the nonlinear system.

The problem of parameter estimation with both time and frequency-domain data is considered and a maximum likelihood approach to the problem is derived. In addition, a procedure to perform a coordinated input design in the frequency and time domain is proposed. This procedure is applied to the identification of the parameters of a model for the dynamics of a rotor blade, showing how the proposed co-design allows to achieve sensibly higher accuracy in the estimation of the system parameters.

Accurate dynamic modelling of helicopter aeromechanics is becoming increasingly important, as progressively stringent requirements are being imposed on rotorcraft control systems. System identification plays an important role as an effective approach to the problem of deriving or fine tuning mathematical models for purposes such as handling qualities assessment and control system design. In this paper the problem of deriving continuous-time models for the dynamics of a small-scale quadrotor helicopter is considered. More precisely, the continuous-time predictor-based subspace identification approach is adopted and the results obtained in an experimental study are presented and discussed.

This paper presents a new approach for the identification of the dynamical model of flexible manipulators. The structure of the identified model, chosen as a descriptor LPV model, is derived from the original non-linear equations. A set of experiments around different configurations is involved, which is suitable for an accurate measurement of the tip of the manipulator by video camera. The final estimation step is global, allowing the direct identification of the global model based on the collection of local experimental data. In an output error context, a genetic algorithm is used for the minimization of the identification criterion. As a case study, a robotic arm with two flexible segments is considered. Identification results based on simulations including noise show the effectiveness of the approach.

This paper deals with the problem of model identification in continuous-time using subspace techniques. More precisely, a recently presented continuous-time predictor-based subspace identification algorithm which relies on a system transformation using the Laguerre basis is considered and a recursive counterpart is developed.

Linear Fractional Transformations (LFTs) are a widely used modelling framework in modern control and identification. Deriving such models from physical first principles is a non-trivial and often error-prone process, if carried out manually. In this paper a new approach to reduced-order LFT modelling and identification starting from equation-based, object-oriented (O-O) descriptions of the plant (formulated using the Modelica language) and experimental data are presented. This approach allows to reduce the gap between user-friendly model representations, based on object diagrams with physical connections, block diagrams with signal connections, and generic differential-algebraic models, and the use of advanced LFT based identification and control techniques.

A recursive scheme for the identification of SIMO Hammerstein models is presented. In the proposed scheme, first the Markov parameters of the system are determined, by a least squares support vector machines regression through an over-parameterisation technique. Then, a state-space realisation of the system is retrieved using a recursive subspace identification method. Simulation results are provided to demonstrate the effectiveness of the algorithm.

The current state-of-the-art in the fields of control-oriented LPV modelling and LPV system identification is surveyed and the potential synergies between the two research areas are highlighted and discussed. Indeed, a number of methods and tools for the development of LPV models from nonlinear systems, for the symbolic realisation and manipulation of LFT representations and for the identification of black-box LPV and LFT models from input/output data have been derived, in a rather independent way, in different research communities. The relative merits of analytical and experimental methods for the derivation of LPV models, as well as possible combinations of the two approaches, are analysed and eventually evaluated on a case study based on the modelling of a thermo-fluid system.

Service providers need to comply with the service level objectives stipulated in contracts with their customers while minimizing the operating costs of the physical infrastructure, mainly related to energy costs. The problem can be effectively formalized using system identification and control theory: service levels are translated into set-points for the response times of the hosted applications and performance are traded-off with energy saving objectives based on suitable models for server dynamics. In this paper, the identification of LPV models for the performance control of Web services is addressed. The suitability of subspace LPV identification methods is investigated and their performance assessed on experimental data.

Published in Automatica (ISSN: 0005-1098) Publisher: Elsevier Science; International audience; The convergence properties of recently developed recursive subspace identification methods are investigated in this paper. The algorithms operate on the basis of instrumental variable (IV) versions of the propagator method for signal subspace estimation. It is proved that, under suitable conditions on the input signal and the system, the considered recursive subspace identification algorithms converge to a consistent estimate of the propagator and, by extension, to the state-space system matrices.

The problem of MIMO recursive identification is considered within the framework of subspace model identification (SMI). In this paper the use of recent signal processing algorithms for the recursive update of the SVD is proposed, and an instrumental variable (IV) approach is followed to accommodate for arbitrary correlation of the disturbances.

A novel class of linear time-varying models is proposed for nonlinear system identification purposes. These models are linear in the parameters, which are time-varying according to a nonlinear dynamic law. A specific parameter tuning algorithm is presented, which is based only on input/output measurements, but which also provides an estimate of the time-varying behaviour of the parameters. So, a Linear Parameter Varying (LPV) model is obtained to which is possible to apply the robust control techniques for LPV systems. Finally, some interesting relations between this model class and the Local Model Network (LMN) family are discussed in the framework of the LPV systems obtained by convex combination of Linear-Time-Invariant (LTI) systems.

The application of computational statistics provides simple algorithms for assessing model uncertainty in subspace identification.

A novel class of hybrid linear/nonlinear models is proposed for nonlinear identification purposes and a specific parameter tuning algorithm is presented. The connections between this model class and the Local Model Network family are outlined. Finally, the capabilities of this model in identification problems is outlined by means of simulations.

With the purpose of evaluating different model identification strategies and of assisting in the validation of active control designs, a simulation program for the dynamics of the main rotor of the AGUSTA A109c helicopter has been developed (based on a general-purpose multibody code). The program has been applied to the testing of various system identification algorithms, both for time-invariant and time-periodic models of the response of the rotor to perturbations in the control inputs, in simulated experiments.

Some results on the analysis of the fault detection problem in a subspace identification framework are presented and two approaches are proposed, exploiting an existing perturbation analysis of subspace methods.