Reinforcement Learning Based H Control for Oil & Gas De-Oiling System

Research output: Contribution to conference without publisher/journalPosterCommunication

77 Downloads (Pure)

Abstract

A de-oiling system consisting of a set of gravity separators and hydrocyclonesis used to separate oil from water in O&G production, to ensure low OiW concentration in the discharge. PID is currently used for de-oiling system control, but it is not always effective to guarantee separation efficiency. Control has been verified its effectiveness comparing with PID controllers in our previous works. However, the current control is model-based, requiring a lot of work for system identification. Therefore, it is difficult to transfer the developed control algorithms into different industrial facilities. In this work, we aim to develop an automatic control generation method such that the de-oiling control can automatically learn the optimal control policy from its behaviour in an online manner, i.e., learning from data without requiring system identification.
Original languageEnglish
Publication date2019
Number of pages1
Publication statusPublished - 2019
EventKick-off: AI for the people - Rendsburggade 14, Aalborg, Denmark
Duration: 19 Nov 201919 Nov 2019

Other

OtherKick-off: AI for the people
LocationRendsburggade 14
Country/TerritoryDenmark
CityAalborg
Period19/11/201919/11/2019

Keywords

  • De-oiling
  • O&G production
  • automatic control generation method

Fingerprint

Dive into the research topics of 'Reinforcement Learning Based H Control for Oil & Gas De-Oiling System'. Together they form a unique fingerprint.

Cite this