[PPT] - Selecting Transients Automatically for the Identification of Models PowerPoint Presentation

SLIDE 1

Selecting Transients Automatically for the Identification of Models for an Oil Well

IFAC Workshop on Automatic Control in Offshore Oil and Gas Production

Antˆ

nio H. Ribeiro1

Luis A. Aguirre1

Laboratory of Modeling, Analysis and Control of Nonlinear Systems (MACSIN) Department of Electronic Engineering (DELT) Universidade Federal de Minas Gerais (UFMG)

Florian´

polis, 2nd IFAC Oilfield, 2015
A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 1 / 11

SLIDE 2

Introduction Oil Production Process

Offshore Oil Production Process

Introduction

Stacionary Production Unit; Riser; Flow Line; Manifold; Wet Christmas Tree; Wellhead; Gas Lift; Soft Sensors.

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 2 / 11

SLIDE 3

Introduction Oil Production Process

P&ID diagram

Introduction

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 3 / 11

SLIDE 4

Introduction Model Estimation

Time Series

Introduction

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 4 / 11

SLIDE 5

Introduction Model Estimation

Time Series

Introduction

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 4 / 11

SLIDE 6

Introduction Model Estimation

Time Series

Introduction

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 4 / 11

SLIDE 7

Introduction Model Estimation

Time Series

Introduction

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 4 / 11

SLIDE 8

Selecting transients automatically Goals and Objectives

Goals and Objectives

Selecting Transients Automatically

Define a quantitative way to measure how suitable a window is for system identification; Define a quantitative measure that can be used to discard windows where the output is not correlated with the input; Use these metrics to automatize the process of finding suitable windows for system identification.

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 5 / 11

SLIDE 9

Selecting transients automatically Dynamic Based Metric

Trajectory Matrix

Dynamic Based Metric

Trajectory Matrix

If you have a set of data containing values of y from k = 0 to k = m + n, you may write: y ∈ Rm and x ∈ Rn:

A =      y(n − 1) y(n − 2) . . . y(0) y(n) y(n − 1) . . . y(1) . . . . . . ... . . . y(m + n − 2) y(m + n − 3) . . . y(m − 1)      ;

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 6 / 11

SLIDE 10

Selecting transients automatically Dynamic Based Metric

Rank of Trajectory Matrix

Dynamic Based Metric

Deeply related with how strong a transient is; Equals the number of parameters one can estimate for an autoregressive model. Depends only on signal y evolution.

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 7 / 11

SLIDE 11

Selecting transients automatically Dynamic Based Metric

Effective Rank of Trajectory Matrix

Dynamic Based Metric

The effective rank r will be calculated as the maximum value of i, so that: σi − σi−1 > l

A. H. Ribeiro, L. A. Aguirre (UFMG)

Selecting Transients for Identification IFAC Oilfield 2015 8 / 11

SLIDE 12

Selecting transients automatically Dynamic Based Metric

Correlation Based Metric - Definition

Correlation Based Metric

The following scalar metric is proposed:

s =

τmax

τ=−τmax

|ρ(τ)| − p |τ| , (1)