31
© Copyright 2014 OSIsoft, LLC. Presented by Optimising Performance By Integrating OSIsoft PI AF ® and MATLAB ® Into A Standard Monitoring Framework Dr J P Barnard

Anglo American Platinum Centre for Process Monitoring

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

© Copyr i gh t 2014 OSIso f t , LLC.

Presented by

Optimising Performance

By Integrating

OSIsoft PI AF® and MATLAB®

Into

A Standard Monitoring Framework

Dr J P Barnard

© Copyr i gh t 2014 OSIso f t , LLC.

Integrating OSIsoft PI AF and MATLAB

• Operational Performance Monitoring

• PI AF – MATLAB Integration

• Case Study 1: Smelter Sensor Report

• Case Study 2: Concentrator Aggregation Report

• Future Developments

2

© Copyr i gh t 2014 OSIso f t , LLC.

• Process Engineers develop

reporting in MATLAB

• End-users view reports in

Web browser

• Operators detect sensor

anomalies

Solution Results and Benefits

Integrating PI AF and MATLAB

Business Challenge

• Create specialised process

reports

• Empower non-programmer

engineer

• Integrate PI Asset

Framework (PI AF) and

MATLAB

• Generate reports in

MATLAB

3

Centre for Process Monitoring

Department of Process Engineering

Stellenbosch University

South Africa

4

Opportunities from Data

• Change point detection

• Process fault diagnosis

• Equipment condition monitoring

• Decision support

Solutions from Data

Knowledge from Data

5

Operational Performance Monitoring

Faulty

Sensors

Limit

Excursions

Aggregate

Production

Stats

PI System

Concentrator

Smelter

6

OPM – Specialised Reporting

Instrumentation

Faulty Instrument

Limit Excursion

Performance Aggregation

Production Output

Production KPI

7

OSIsoft PI AF

MATLAB

Specialised Analysis & Reporting

8

PI AF + MATLAB = Specialised Reporting

PI A

F

Asset Tree

Template-based

Friendly GUI

PI Integration

.NET SDK API

MA

TL

AB

Interactive development

Vectorised

Specialised Toolboxes

Rich Visualisation

OOP, Exceptions, Events

.NET Integration

Application deployment

9

PI AF – MATLAB Integration

• Objectives

• Architecture

• PI AF Elements as Units and Sensors

• PI AF Attributes as Process Values

• PI AF Database Map

• MATLAB PI AF SDK Integration Layer

10

PI AF – MATLAB Objectives

• Empower non-programmer engineer

• Develop Analysis & Reporting in MATLAB

• Avoid redesigning the wheel

• Leverage best of each technology stack

• Give PI AF Objects MATLAB Array Behaviour

11

PI AF-MATLAB Architecture PI AF Import Layer

AFMap.xlsx

AF

Pars

er

MultiLevelMap im

po

rtP

lan

t

op

m.A

FE

lem

en

t

OP

MC

on

fig

ura

tio

n

OPMUnit

OPMSensor

OPMAnalysis

AFTree

\\aaa\bbb\ccc

AFTable opm.AFTable

12

PI AF-MATLAB Architecture

AFCollection

(SDK)

AFCollection

(MATLAB)

AFArray

(MATLAB)

AFObject

(SDK)

AFObject

(MATLAB)

AFElement

(SDK)

AFElement

(MATLAB)

AFAttribute

(SDK)

AFAttribute

(MATLAB)

AFTable

(SDK)

AFTable

(MATLAB)

System.Object

(.NET)

Object

(MATLAB)

DataTable

(.NET)

DataTable

(MATLAB)

PI AF SDK Integration

Layer

• MATLAB array behaviour

• Reference by String or

Index

• Collection to cell array

• Find object

• Find element

• Find attribute

• DataTable to cell or

struct

• Find table

• Table name list

• Column name list

• DataTable.select

13

AFElements as Units & Sensors

OPMUnit

Property

PI AF

Map

PI AF Definition

Template

Property

Path

Procedure

Tables

Procedure

OPMSensor

Property

OPM Class

Category

Procedure

14

AFAttributes as Process Values

SensorData

Time

Value

OPMSensor PIAFImporter

PV: opm.AFAttribute Data

Time Span

PI Data

Archive

AFAttribute

15

OPM Analyses

OPMAnalysis

OPMUnit

OPMSensor

AFTable

PI AF

Map

Item

16

Data

Episode

PI AF Table

Data

Episode

OPMSession

OPMAnalysis OPMAnalysis Reports

PI AF Tree

The Plan Comes Together

Run

17

Case Study 1: SPAT Report

• PGM Smelter Analysis

• Sensor Fault Detection

• Temperature Anomaly Detection

• Import PI AF tree of Smelter

• Import Process Data via PI AF Attributes

• Run Analysis & Reporting in MATLAB

18

Instrument Source

Oc

cu

rre

nc

es

Fa

ilu

re S

tatu

s

Type

OPM Analysis in MATLAB

Report as HTML Page

19

Drill-Down

20

PI Coresight

• Better interaction than

MATLAB HTML Report

• Zoom

• Pan

21

Case Study 2: Conc Report

• PGM Concentrator Aggregation Analysis

• Import Subset of Process Data via PI AF Attributes

• Aggregate Performance Data in MATLAB

• Export Aggregate Statistics to TXT & SQL

• Import Aggregate Statistics via PI Interface for

Relational Database (RDBMS via ODBC)

22

23

24

Future Developments

• PI AF Attribute Templates for Unit KPIs

• PI AF Attribute Templates for KPI aggregation

• PI AF Attribute Templates for On-line Fault Detection

• PI Event Frames

• R&D on Fault Detection, Identification

• Implementation of R&D

25

Acknowledgements

• OSIsoft

• Anglo American Platinum

• OPTI-NUM solutions

• Stellenbosch University

26

© Copyr i gh t 2014 OSIso f t , LLC.

Questions

Please wait for the

microphone before

asking your

questions

State your

name & company

© Copyr i gh t 2014 OSIso f t , LLC.

Brought to you by

© Copyr i gh t 2014 OSIso f t , LLC.

Please don’t forget to… Complete the online survey for

this session

eventmobi.com/emeauc14

Share with your friends

#UC2014