Conditional distribution modeling for estimating and exploiting uncertainty in control systems

Randa Herzallah, David Lowe

Research output: Working paperTechnical report

Abstract

This paper presents a general methodology for estimating and incorporating uncertainty in the controller and forward models for noisy nonlinear control problems. Conditional distribution modeling in a neural network context is used to estimate uncertainty around the prediction of neural network outputs. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localize the possible control solutions to consider. A nonlinear multivariable system with different delays between the input-output pairs is used to demonstrate the successful application of the developed control algorithm. The proposed method is suitable for redundant control systems and allows us to model strongly non Gaussian distributions of control signal as well as processes with hysteresis.
Original languageEnglish
Place of PublicationBirmingham
PublisherAston University
ISBN (Print)NCRG/2003/010
Publication statusUnpublished - 27 May 2003

Fingerprint

Neural networks
Control systems
Multivariable systems
Gaussian distribution
Dynamic programming
Hysteresis
Controllers
Uncertainty

Keywords

  • uncertainty in the controller and forward models
  • noisy nonlinear control problems
  • Conditional distribution modeling
  • neural network
  • dynamic programming problem
  • nonlinear multivariable system
  • redundant control systems
  • non Gaussian distributions
  • control signal
  • hysteresis

Cite this

@techreport{506aea169e7644fbb4f15494d4af791e,
title = "Conditional distribution modeling for estimating and exploiting uncertainty in control systems",
abstract = "This paper presents a general methodology for estimating and incorporating uncertainty in the controller and forward models for noisy nonlinear control problems. Conditional distribution modeling in a neural network context is used to estimate uncertainty around the prediction of neural network outputs. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localize the possible control solutions to consider. A nonlinear multivariable system with different delays between the input-output pairs is used to demonstrate the successful application of the developed control algorithm. The proposed method is suitable for redundant control systems and allows us to model strongly non Gaussian distributions of control signal as well as processes with hysteresis.",
keywords = "uncertainty in the controller and forward models, noisy nonlinear control problems, Conditional distribution modeling, neural network, dynamic programming problem, nonlinear multivariable system, redundant control systems, non Gaussian distributions, control signal, hysteresis",
author = "Randa Herzallah and David Lowe",
year = "2003",
month = "5",
day = "27",
language = "English",
isbn = "NCRG/2003/010",
publisher = "Aston University",
type = "WorkingPaper",
institution = "Aston University",

}

Conditional distribution modeling for estimating and exploiting uncertainty in control systems. / Herzallah, Randa; Lowe, David.

Birmingham : Aston University, 2003.

Research output: Working paperTechnical report

TY - UNPB

T1 - Conditional distribution modeling for estimating and exploiting uncertainty in control systems

AU - Herzallah, Randa

AU - Lowe, David

PY - 2003/5/27

Y1 - 2003/5/27

N2 - This paper presents a general methodology for estimating and incorporating uncertainty in the controller and forward models for noisy nonlinear control problems. Conditional distribution modeling in a neural network context is used to estimate uncertainty around the prediction of neural network outputs. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localize the possible control solutions to consider. A nonlinear multivariable system with different delays between the input-output pairs is used to demonstrate the successful application of the developed control algorithm. The proposed method is suitable for redundant control systems and allows us to model strongly non Gaussian distributions of control signal as well as processes with hysteresis.

AB - This paper presents a general methodology for estimating and incorporating uncertainty in the controller and forward models for noisy nonlinear control problems. Conditional distribution modeling in a neural network context is used to estimate uncertainty around the prediction of neural network outputs. The developed methodology circumvents the dynamic programming problem by using the predicted neural network uncertainty to localize the possible control solutions to consider. A nonlinear multivariable system with different delays between the input-output pairs is used to demonstrate the successful application of the developed control algorithm. The proposed method is suitable for redundant control systems and allows us to model strongly non Gaussian distributions of control signal as well as processes with hysteresis.

KW - uncertainty in the controller and forward models

KW - noisy nonlinear control problems

KW - Conditional distribution modeling

KW - neural network

KW - dynamic programming problem

KW - nonlinear multivariable system

KW - redundant control systems

KW - non Gaussian distributions

KW - control signal

KW - hysteresis

M3 - Technical report

SN - NCRG/2003/010

BT - Conditional distribution modeling for estimating and exploiting uncertainty in control systems

PB - Aston University

CY - Birmingham

ER -