Feedback Control by Online Learning an Inverse Model

Waegeman, Tim; wyffels, Francis; Schrauwen, Benjamin

doi:10.1109/tnnls.2012.2208655

Cited by 53 publications

(41 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…P(t) is the covariance matrix that is the running estimate of the Moore-Penroose pseudo-inverse of (vv T + ηI) with η a regularization parameter Waegeman et al (2012) and P(0) denotes the initial value of P(t).…”

Section: Weight Updatementioning

confidence: 99%

Echo State Neural Network Based State Feedback Control for SISO Afine Nonlinear Systems

Mahmoud

Elshenawy

2015

IFAC-PapersOnLine

View full text Add to dashboard Cite

Echo state network (ESTN) is a new recurrent neural networks (RNN) with a simpler training method. Based on ESTN, this paper address a state feedback control algorithm for a class of perturbed SISO nonlinear systems in the affine form. The control algorithm is implemented without a prior knowledge of the nonlinear system. The network weights can be tuned on line by the Recursive Least Squares (RLS) method without off line learning phase needed. The convergence and the Bounded Input Bounded Output (BIBO) stability of the ESTN controller are proven. Moreover, all signals involved in the closed loop are proven to be exponentially bounded and then the stability of the system. We have used the tracking problem of one-link rigid robotic manipulator system as an example to verify the effectiveness of the proposed method.

show abstract

Section: Weight Updatementioning

confidence: 99%

Echo State Neural Network Based State Feedback Control for SISO Afine Nonlinear Systems

Mahmoud

Elshenawy

2015

IFAC-PapersOnLine

View full text Add to dashboard Cite

show abstract

“…In [19] and [20] we introduced a novel feedback controller which learns to control a plant (dynamical system) by online learning an inverse plant model based on real-time controlled plant-input/output pairs. In parallel, this preliminary model is used to actually control the system, producing a new plant-input output pair which gradually improves the inverse model.…”

Section: Controllermentioning

confidence: 99%

“…At the core of this feedback controller we use a RC-network to accommodate the inverse model. However, as described in [20], any dynamical system with a high dimensional state representation can be used to accommodate such model as well. In this paper we apply the same feedback controller (shown in Fig.…”

Section: Controllermentioning

confidence: 99%

Towards a Neural Hierarchy of Time Scales for Motor Control

Waegeman

wyffels

Schrauwen

2012

From Animals to Animats 12

Self Cite

View full text Add to dashboard Cite

Abstract. Animals show remarkable rich motion skills which are still far from realizable with robots. Inspired by the neural circuits which generate rhythmic motion patterns in the spinal cord of all vertebrates, one main research direction points towards the use of central pattern generators in robots. On of the key advantages of this, is that the dimensionality of the control problem is reduced. In this work we investigate this further by introducing a multi-timescale control hierarchy with at its core a hierarchy of recurrent neural networks. By means of some robot experiments, we demonstrate that this hierarchy can embed any rhythmic motor signal by imitation learning. Furthermore, the proposed hierarchy allows the tracking of several high level motion properties (e.g.: amplitude and offset), which are usually observed at a slower rate than the generated motion. Although these experiments are preliminary, the results are promising and have the potential to open the door for rich motor skills and advanced control.

show abstract

“…In Jordanou et al (2017), an on-line learning control framework based on ESNs was implemented for the control of an oil well. The controller, based on Waegeman et al (2012), is adapted on-line by the Recursive Least Squares algorithm to model the inverse plant dynamics, being able to successfully perform control tasks such as reference tracking and disturbance rejection. A key disadvantage of This work was funded in part by NTNU, Petrobras, and CNPq. the inverse model controller is the lack of a clear relation between the controller parameters and its effects on the system, such as in PID controllers.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear Model Predictive Control of an Oil Well with Echo State Networks

et al. 2018

View full text Add to dashboard Cite

In oil production platforms, processes are nonlinear and prone to modeling errors, as the flow regime and components are not entirely known and can bring about structural uncertainties, making the design of predictive control algorithms a challenge. In this work, an efficient data-driven framework for Model Predictive Control (MPC) using Echo State Networks (ESN) as the prediction model is proposed. Unlike previous works, the ESN model for MPC is only linearized partially: while the free response of the system is kept fully nonlinear, only the forced response is linearized. This MPC framework is known in the literature as the Practical Nonlinear Model Predictive Controller (PNMPC). In this work, by using the analytically computed gradient from the ESN model, a finite difference method is not needed to compute derivatives as in PNMPC. The proposed method, called PNMPC-ESN, is applied to control a simplified model of a gas-lifted oil well, managing to successfully control the plant, obeying the established constraints while maintaining setpoint tracking.

show abstract

Feedback Control by Online Learning an Inverse Model

Cited by 53 publications

References 38 publications

Echo State Neural Network Based State Feedback Control for SISO Afine Nonlinear Systems

Echo State Neural Network Based State Feedback Control for SISO Afine Nonlinear Systems

Towards a Neural Hierarchy of Time Scales for Motor Control

Nonlinear Model Predictive Control of an Oil Well with Echo State Networks

Contact Info

Product

Resources

About