AI-augmented goal achievement

D. W. Russell

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

The notion of goals in the context of real-world process-control is, more often than not, linked with a minimization of complex cost-functionals which are derived from mathematical postulates that depend upon the topography of hypersurfaces described by state-variables. However, in poorly-defined systems there are usually no plausible identification models available on which to apply such elegant techniques, other than notorious linearizations that are both approximate and ontologically unsatisfying. This paper shows how the goal of reducing oscillatory, transient behavior in any system can be achieved by augmenting a Proportional, Integral and Derivative (PID) schema with an AI paradigm that is in no way dependant on any model of the system. Results from a simulation of a second-order control system are included to show how the augmentation of this traditional PID controller by an AI-driven signature-table can significantly reduce transient oscillations in response to random step inputs and to thus achieve an important real- world goal. Commentary is made on how the seemingly negative factors of risk, uncertainty and failure are turned into positive contributors in the learning process.

Original languageEnglish (US)
Title of host publicationApplications of Artificial Intelligence in Engineering
EditorsD.E. Grierson, G. Rzevski, R.A. Adey
PublisherPubl by Computational Mechanics Publ
Pages971-982
Number of pages12
ISBN (Print)1851667873
StatePublished - 1992

Fingerprint

Derivatives
Linearization
Topography
Process control
Identification (control systems)
Control systems
Controllers
Costs
Uncertainty

All Science Journal Classification (ASJC) codes

  • Software

Cite this

Russell, D. W. (1992). AI-augmented goal achievement. In D. E. Grierson, G. Rzevski, & R. A. Adey (Eds.), Applications of Artificial Intelligence in Engineering (pp. 971-982). Publ by Computational Mechanics Publ.
Russell, D. W. / AI-augmented goal achievement. Applications of Artificial Intelligence in Engineering. editor / D.E. Grierson ; G. Rzevski ; R.A. Adey. Publ by Computational Mechanics Publ, 1992. pp. 971-982
@inproceedings{3423689aab3d4b698ca587b7b55ef9ee,
title = "AI-augmented goal achievement",
abstract = "The notion of goals in the context of real-world process-control is, more often than not, linked with a minimization of complex cost-functionals which are derived from mathematical postulates that depend upon the topography of hypersurfaces described by state-variables. However, in poorly-defined systems there are usually no plausible identification models available on which to apply such elegant techniques, other than notorious linearizations that are both approximate and ontologically unsatisfying. This paper shows how the goal of reducing oscillatory, transient behavior in any system can be achieved by augmenting a Proportional, Integral and Derivative (PID) schema with an AI paradigm that is in no way dependant on any model of the system. Results from a simulation of a second-order control system are included to show how the augmentation of this traditional PID controller by an AI-driven signature-table can significantly reduce transient oscillations in response to random step inputs and to thus achieve an important real- world goal. Commentary is made on how the seemingly negative factors of risk, uncertainty and failure are turned into positive contributors in the learning process.",
author = "Russell, {D. W.}",
year = "1992",
language = "English (US)",
isbn = "1851667873",
pages = "971--982",
editor = "D.E. Grierson and G. Rzevski and R.A. Adey",
booktitle = "Applications of Artificial Intelligence in Engineering",
publisher = "Publ by Computational Mechanics Publ",

}

Russell, DW 1992, AI-augmented goal achievement. in DE Grierson, G Rzevski & RA Adey (eds), Applications of Artificial Intelligence in Engineering. Publ by Computational Mechanics Publ, pp. 971-982.

AI-augmented goal achievement. / Russell, D. W.

Applications of Artificial Intelligence in Engineering. ed. / D.E. Grierson; G. Rzevski; R.A. Adey. Publ by Computational Mechanics Publ, 1992. p. 971-982.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - AI-augmented goal achievement

AU - Russell, D. W.

PY - 1992

Y1 - 1992

N2 - The notion of goals in the context of real-world process-control is, more often than not, linked with a minimization of complex cost-functionals which are derived from mathematical postulates that depend upon the topography of hypersurfaces described by state-variables. However, in poorly-defined systems there are usually no plausible identification models available on which to apply such elegant techniques, other than notorious linearizations that are both approximate and ontologically unsatisfying. This paper shows how the goal of reducing oscillatory, transient behavior in any system can be achieved by augmenting a Proportional, Integral and Derivative (PID) schema with an AI paradigm that is in no way dependant on any model of the system. Results from a simulation of a second-order control system are included to show how the augmentation of this traditional PID controller by an AI-driven signature-table can significantly reduce transient oscillations in response to random step inputs and to thus achieve an important real- world goal. Commentary is made on how the seemingly negative factors of risk, uncertainty and failure are turned into positive contributors in the learning process.

AB - The notion of goals in the context of real-world process-control is, more often than not, linked with a minimization of complex cost-functionals which are derived from mathematical postulates that depend upon the topography of hypersurfaces described by state-variables. However, in poorly-defined systems there are usually no plausible identification models available on which to apply such elegant techniques, other than notorious linearizations that are both approximate and ontologically unsatisfying. This paper shows how the goal of reducing oscillatory, transient behavior in any system can be achieved by augmenting a Proportional, Integral and Derivative (PID) schema with an AI paradigm that is in no way dependant on any model of the system. Results from a simulation of a second-order control system are included to show how the augmentation of this traditional PID controller by an AI-driven signature-table can significantly reduce transient oscillations in response to random step inputs and to thus achieve an important real- world goal. Commentary is made on how the seemingly negative factors of risk, uncertainty and failure are turned into positive contributors in the learning process.

UR - http://www.scopus.com/inward/record.url?scp=0026981730&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0026981730&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0026981730

SN - 1851667873

SP - 971

EP - 982

BT - Applications of Artificial Intelligence in Engineering

A2 - Grierson, D.E.

A2 - Rzevski, G.

A2 - Adey, R.A.

PB - Publ by Computational Mechanics Publ

ER -

Russell DW. AI-augmented goal achievement. In Grierson DE, Rzevski G, Adey RA, editors, Applications of Artificial Intelligence in Engineering. Publ by Computational Mechanics Publ. 1992. p. 971-982