26
Business Informatics Group Institute of Software Technology and Interactive Systems Vienna University of Technology Favoritenstraße 9-11/188-3, 1040 Vienna, Austria phone: +43 (1) 58801-18804 (secretary), fax: +43 (1) 58801-18896 [email protected], www.big.tuwien.ac.at Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF Luca Berardinelli , Emanuel Maetzler, Tanja Mayerhofer, Manuel Wimmer INDIN 2016 July, 18-22, Poitiers, France

Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Embed Size (px)

Citation preview

Page 1: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Business Informatics GroupInstitute of Software Technology and Interactive Systems Vienna University of TechnologyFavoritenstraße 9-11/188-3, 1040 Vienna, Austriaphone: +43 (1) 58801-18804 (secretary), fax: +43 (1) [email protected], www.big.tuwien.ac.at

Integrating Performance Modeling in IndustrialAutomation through AutomationML and PMIF

Luca Berardinelli, Emanuel Maetzler, Tanja Mayerhofer, Manuel Wimmer

INDIN 2016 July, 18-22, Poitiers, France

Page 2: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

= discipline/domain

Elec

troni

c

Engin

eerin

g

Mecha

nica

l

Engin

eerin

g

Softw

are

Engin

eerin

g

Context: Need for integration in CPPS engineering process

2= documents = tool

• Cyber Physical Production System (CPPS) Engineering is a multidisciplinary engineering process

• Different stackeholders bring their own methodologies, tools and docs to carry out different activities during the process

• Data exchange is a critical issue within the multidisciplinary CPPS engineering

Mechanical EngineerElectronic EngineerSoftware Engineer… tools

docs

methodologiesdiscipline

RequirementsSpecification

Production SystemCommitment

stakeholders

Process

Page 3: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

= discipline/domain

Context: Cross-cutting Model-Driven Engineering perspective

3= documents = tool

RequirementsSpecification

Production SystemCommitment

• Stakeholders also model engineers:

• Documents are models• Tools become model-driven, capable to

process models • Discipline-specific methodologies become

model-driven

MDE Tools

model

methodologiesdiscipline: MDEMechanical Engineer

Electronic EngineerSoftware Engineer…

Mod

el E

ngin

eer

+

stakeholders

Process

Elec

troni

c

Engin

eerin

g

Mecha

nica

l

Engin

eerin

g

Softw

are

Engin

eerin

g

technology

Page 4: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Goal: Integrating Model-Driven Performance Modeling in CPPS

4

SYST

EM IN

TEGR

ATIO

NSYSTEM D

ESIGN

Production SystemRequirements

software engineeringelectronics

mechanicsintegration

integration

Earl

y Va

lidat

ion

Model Engineer

you are here

early stages of CPPS design&validation

for MDPE

Model-Driven Performance Engineering

(MDPE)via Queuing Network

Page 5: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Goal: Integrating Model-Driven Performance Modeling in CPPS

5

MDPEvia Queuing Network,

do I need it?

https://www.arenasimulation.com/industry-solutions/manufacturing-simulation-software

http://jmt.sourceforge.net/ a “generic” QN Solver

integration

PMIF

Page 6: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

SYST

EM IN

TEGR

ATIO

NSYSTEM D

ESIGN

Production SystemRequirements

software engineeringelectronics

mechanicsintegration

Earl

y Va

lidat

ion

Model Engineer

Goal: Integrating Model-Driven Performance Modeling in CPPS

you are here

integration

Both common representations for

data exchange in two different domains.

AML is a standardPMIF is an academic initiative

Why AML and PMIF?PMIF

Three possible Integration Mechanisms 6

Page 7: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

7

Models: AutomationML

modelPMIF

Emerging standard (IEC 62714) for tool data exchange in CPPS Engineering

Foundation for harmonizing engineering data coming from a heterogeneous tool network by means of a unified format and data model

XML based document (CAEX.xsd)

.XSD

integration tech.

Schema

Page 8: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

8

Models: AutomationML

modelPMIF

• We provided also a complementary XML-based serialization format for AML top level format (CAEX) based on Ecore

• Metamodeling is a key MDE enabling activity• Metamodels are language specification• Ecore is a metamodeling technology

integration tech.

.ecoremetamodel

.XSDSchema

Page 9: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

9technology

Models: AutomationML

…or in Ecore, new enabling technology for MDE

https://eclipse.org/modeling/emf/

Model Engineer

AML in 10 secs slide?

AML.xsdXSD

cpps_example.amlXML

«conforms to»

Model (hierarchical graph based structure)ab

cd

c1 c2

c2.1…

Hierachy prototypes

«cloned»

c1 c2roles

blue semantics

orange semantics

interfaces«typing»

Page 10: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

10technology

Models: AutomationML

…or in Ecore, new enabling technology for MDE

https://eclipse.org/modeling/emf/

Model Engineer

CAEX in 10 secs slide?

Model (hierarchical graph based structure)ab

cd

c1 c2

c2.1…

Hierachy prototypes

«cloned» interfaces

c1 c2roles

blue semantics

orange semantics«assign semantics to»

AML.xsdXSD

cpps_example.amlXML

«conforms to»

«typing»

Page 11: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

11

Models: AutomationMLAML Language Specification in Ecore

Eclipse Modeling Frameworkhttps://eclipse.org/modeling/emf/

IH

IE

RC

concepts for structure modelingconcepts for semantics modeling

Model Engineer

AML.ecoreECORE

cpps_example.amlXML

«conforms to»

Page 12: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

12

Representing a Production System in AutomationML

Pick and Place Unit Case Study• The PPU Production System produces

different types of tanks

• PPU resources are its stations

• Tank products are processed by resources in accordance with production processes

Page 13: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

13

Models: PMIF

• common representation for system performance model data to move QN models between analysis and simulation tools

• PMIF documents are serialized in XML• PMIF is specified both as XML Schema (.XSD)

and as Ecore Metamodel

integration tech.

modelPMIF

PMIF

.ecoremetamodel

.XSDSchema

workload(s)

=jobs

queuing network

Arrival Rate

Service Time Rate

Page 14: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

14

STACK

CRANE STAMP

CONVEYOR

RAMP

1 2

34567

2 32 43 24 54 64 7

Models: AutomationML and PMIF

Pick and Place Unit Case Study

21

Reso

urce

sPr

oces

sPr

oduc

ts

RAMP

RAMP

• The PPU Production System produces different types of tanks

• PPU resources are its stations

• Tank products are processed by resources in accordance with production processes

Page 15: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

15

Models: Performance Model Interchange Format (PMIF)

PMIF.ecoremetamodel

.XSDSchema

PMIF Language Specification in Ecore

Eclipse Modeling Frameworkhttps://eclipse.org/modeling/emf/

integration tech.

modelPMIF

Page 16: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

16

Models: Performance Model Interchange Format (PMIF)

STACK

CRANE STAMP

CONVEYOR

RAMP

Pick and Place Unit Case Study

RAMP

RAMP

• The PPU Production System produces different types of tanks

• PPU resources are its stations

• Tank products are processed by resources in accordance with production processes

PPU QueuingNetworkModel

Two models representing the same PPU system. How can I generate them?

Page 17: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: conceptual mappings between AML and PMIF

17

CPPS : {{Product}, {Process}, {Resource}}

PMIF

CPPSs can be described as PPR systems…

AML language concepts

PMIF language concepts

IE STACK

RC Resource(from AutomationML standard library)

SRC

Server«modeled in AML»

«modeled in PMIF»

Page 18: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: conceptual mappings between AML and PMIF

18

with service rate forIE STACK

RC(from AutomationML standard library)

SRC Process

Service Request

чServer

PMIF

CPPSs can be described as PPR systems…

AML language concepts

PMIF language concepts

«modeled in AML»

«modeled in PMIF»

CPPS : {{Product}, {Process}, {Resource}}

Page 19: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: conceptual mappings between AML and PMIF

19

on

Shown in the case study

IE STACK

RC(from AutomationML standard library)

SRC Product

Workload and arrival rate

чλ

PMIF

CPPSs can be described as PPR systems…

AML language concepts

PMIF language concepts

«modeled in AML»

«modeled in PMIF»

CPPS : {{Product}, {Process}, {Resource}}

Page 20: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: Overview of the PPU Case Study

20

integration PMIF

Linking Model

AML Native

Model Transformation

How to implement the conceptual integration?

Page 21: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

SYST

EM IN

TEGR

ATIO

NSYSTEM D

ESIGN

Production SystemRequirements

software engineeringelectronics

mechanicsintegration

Earl

y Va

lidat

ion

Model Engineer

Goal: Integrating Model-Driven Performance Modeling in CPPS

21

MDE Tools

model

3 integration mechanisms

PMIF

integration tech.

discipline: MDPEmethodologies

QN

you are here

Model-Driven Performance Engineering (MDPE)

integration

AML Native

PMIF

integration

Linking Model

PMIF

integration

Model Transformation

PMIF

.link

.transf.

integrationintegration

Page 22: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: PPU - Native AML Integration

22

integration

AML Native

PMIF

XML ID

AML.xsd PMIF.xsd

XML IDrefURI

XSD XSD

PPU.xml PPU_QN.xmlXML XML

languages

• Default integration choice when the integration should be carried out through XSD-based tech.

• AML ExternalDataConnector XML with refURI xs::anyURI

models

JAXP

Xerces

Integration coupled with AML docs.Not a MDE approach.

IE

Page 23: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: PPU - Linking Model

23

integration PMIF

AML.ecore PMIF.ecoreEcore

PPU.xml PPU_QN.xmlXML XML

languages

models

Linking.ecoreEcore Ecore

XML

LinkModel.xml

Linking Model

Linking language to create Link Models

Links are explicit and static. Manual approach that can be automated by identifing source and target patterns in AML and PMIF models.

API

IE

Page 24: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Integration: PPU – Model Transformation

24

integration PMIF

AML.ecore PMIF.ecore

Ecore

PPU.xml PPU_QN.xml

XML XML

languages

models

EcoretxtAML2PMIF.atl

Model Transformation

Mappings defined among languageconcepts (not their instances!)Links between models are dynamic, i.e., generated when ATL file is executed.

ATL is a transformation language available in Eclipse.

The integration is realized with rules with «from» and «to» patterns.ie.hasRole(‘Product’)

PMIF!Workload (

IEProduct2Workload(

«from» «to»

Page 25: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Conclusions

1. We presented three different integration mechanisms to support Model-Driven Peformance Engineering (PE) with Queuing Networks in early System Design phases of CPPS

2. We choose two languages, AutomationML (AML) and Performance Interchange File Format (PMIF), devised to ease the integration of heterogeneous tools (and then methodologies adopting these tools) in CPPS and PE domains

3. The integration mechanisms are tool-independent and independent from each other. However, each mechanism imposes some technological requirements on tools (e.g., Java-based libraries).

4. Each integration mechanism has pros and cons. Their adoption is left to the end user.

25

Page 26: Integrating Performance Modeling in Industrial Automation through AutomationML and PMIF

Future Work

1. Completing the AutomationML and PMIF Integration (PMIF2AML is still missing, can be implemented as Model Transformatiom)

2. Adopting Linking and/or Model Transformation intergration mechanisms in different model-driven methodologies (e.g., reliability analysis)

26