Introduction to Adaptive Control

Landau, I. D.; Lozano, R.; M’Saad, M.

doi:10.1007/978-0-85729-343-5_1

Cited by 18 publications

(38 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A considerable amount of effort has been made by academic and industrial researchers into the online estimation of vehicle parameters that are either prohibitively difficult or expensive to measure directly. The most usual approaches are to use estimation based on the Recursive Least Squares algorithm [2] [3] [4] or Kalman Filtering [2][3] [4] and many make use of data fusion from more than one source.…”

Section: A Brief Overview Of the Current State Of The Art In Vehicle mentioning

confidence: 99%

Gradient and Mass Estimation from CAN Based Data for a Light Passenger Car

Wragge-Morley¹,

Herrmann²,

Barber³

et al. 2015

SAE Int. J. Passeng. Cars – Electron. Electr. Syst.

View full text Add to dashboard Cite

We present a method for the estimation of vehicle mass and road gradient for a light passenger vehicle. The estimation method uses information normally available on the vehicle CAN bus without the addition of extra sensors. A composite parameter estimation algorithm incorporating a nonlinear adaptive observer structure uses vehicle speed over ground and driving torque to estimate mass and road gradient. A system of filters is used to avoid deriving acceleration directly from wheel speed. In addition, a novel data fusion method makes use of the regressor structure to introduce information from other sensors in the vehicle. The dynamics of the additional sensors must be able to be parameterised using the same parameterisation as the complete vehicle system dynamics. In this case we make use of an Inertial Measurement Unit (IMU) which is part of the vehicle safety and Advanced Driver Assist Systems (ADAS). Therefore, a method using some filtering and supervisory logic is employed to give a sensible update behaviour for the vehicle mass estimation algorithm. The main function of the supervisor is to reject the mass estimate produced by unsuitable available data due to unmodelled loss forces. Good estimation results are obtained from data from a vehicle which was also fitted with some additional instrumentation including GPS sensors and a high quality IMU for scientific verification purposes.

show abstract

Section: A Brief Overview Of the Current State Of The Art In Vehicle mentioning

confidence: 99%

Gradient and Mass Estimation from CAN Based Data for a Light Passenger Car

Wragge-Morley¹,

Herrmann²,

Barber³

et al. 2015

SAE Int. J. Passeng. Cars – Electron. Electr. Syst.

View full text Add to dashboard Cite

show abstract

“…The model of the plant is obtained by system identification of the secondary path (for details on the system identification of the model of the active suspension see [24], [25], [26], [27], [28]). The controller to be designed is a RS-type polynomial [29], [30] controller (see Fig. 3).…”

Section: A Plant Representation and Controller Structurementioning

confidence: 99%

“…The problem is, in fact, an on-line adaptive estimation of parameters in presence of noise [29], [20]. Equation 24is a particular case of identification of an ARMAX model.…”

Section: (24)mentioning

confidence: 99%

Rejection of narrow band unknown disturbances in an active suspension system

Constantinescu

Rey

Landau

2007

2007 European Control Conference (ECC)

View full text Add to dashboard Cite

The rejection of unknown disturbances of unknown or varying frequencies is one of the main problems in active vibration control. To achieve the rejection of the disturbances (at least asymptotically) without measuring them, a feedback solution can be considered. This paper discusses a feedback adaptive control approach for active vibration systems in the presence of unknown narrow band disturbances, with the model of the plant considered known (e.g. obtained by standard system identification). As the controller should incorporate the model of the disturbance, the rejection of unknown disturbances raises the problems of adapting the internal model of the controller and redesigning the controller in real-time. A direct adaptive control scheme based on the internal model principle and the use of the Youla-Kucera parameterization is proposed. This approach is comparatively evaluated with respect to an indirect adaptive control scheme based on the estimation of the disturbance model. The evaluation of the performance of the adaptive algorithms is illustrated in real time on an active suspension system.

show abstract

“…The assumption that there exist ideal weights such that the approximation property holds is very much like various similar assumptions in adaptive control [1,29], including Erzberger's assumptions and linearity in the parameters. The very important difference is that in the NN case, the universal approximation property always holds, while in adaptive control such assumptions often do not hold in practice, and so they imply restrictions on the form of the systems that can be controlled.…”

Section: Multilayer Neural Networkmentioning

confidence: 99%

“…Most of the early approaches used standard backpropagation weight tuning [56] since rigorous derivations of tuning algorithms suitable for closed-loop control purposes were not available. Many NN design techniques mimicked adaptive control approaches, where rigorous analysis results were available [1,29,12], proposing NN feedback control topologies such as indirect identification-based control, inverse dynamics control, series-parallel techniques, etc.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear Network Structures for Feedback Control

Lewis

1999

Asian Journal of Control

124

View full text Add to dashboard Cite

A framework is given for controller design using Nonlinear Network Structures, which include both neural networks and fuzzy logic systems. These structures possess a universal approximation property that allows them to be used in feedback control of unknown systems without requirements for linearity in the system parameters or finding a regression matrix. Nonlinear nets can be linear or nonlinear in the tunable weight parameters. In the latter case weight tuning algorithms are not straightforward to obtain. Feedback control topologies and weight tuning algorithms are given here that guarantee closed‐loop stability and bounded weights. Extensions are discussed to force control, backstepping control, and output feedback control, where dynamic nonlinear nets are required.

show abstract

Introduction to Adaptive Control

Cited by 18 publications

References 0 publications

Gradient and Mass Estimation from CAN Based Data for a Light Passenger Car

Gradient and Mass Estimation from CAN Based Data for a Light Passenger Car

Rejection of narrow band unknown disturbances in an active suspension system

Nonlinear Network Structures for Feedback Control

Contact Info

Product

Resources

About