Towards a Paradigm for Activity Modeling

1
Towards a Paradigmfor Activity Modeling
Jason Garrett

• SMC2000 WP-2.9
• Modeling Control Hybrid Systems 2
• October 11, 2000, Hermitage D, 510 pm

2
Motivation, DuPont

• Give operators the ability to use information
  obtained from sensors in the plant and allow
  individuals the opportunity to use this
  information for decision making
• Real-time and historical information is obtained
  through a Process Monitoring and Control (PMC)
  database
• Chemical simulation capabilities accessible via
  remote supercomputer execution
• Signal Processing should be allowed at any
  integration point
• Remote clients should be easy to instantiate,
  completely configurable, and allow for multiple
  paths of bidirectional communication

3
Presentation Overview

• System Integration Problem
• General Case
• Our specific case
• Model Integrated Computing (MIC)
• Explanation
• Relevance to our Problem
• MIC -gt Activity Modeling Tool
• Our Specific Run-Time Environment
• Conclusions

4
System Integration, General
Classic Problem data need to be exchanged
through systems not sharing a common
interface Classic Solution GLUE code in Run-Time
Environment
5
System Integration, Specific
PMC
Simul.
AMT
SigProc.
GUI
neoClassic Problem frequent reconfiguration
system addition distanced operation manager Our
Solution Activity Modeling through Model
Integrated Computing
6
Model Integrated Computing
The Multigraph Architecture
Application
Metaprogramming
Application
Environment
Domain
Interface
Evolution
Evolution
App.
App.
App.
1
2
3
MIPS
Environment
Model Builder
Meta-Level
Translation
7
MIC Relevance

• Model Integrated Computing builds Model
  Integrated Program Synthesis (MIPS) environments
  based on a multi-aspect, containment hierarchy
  with different types of associations
• Associations such as connections, references,
  sets, must have higher level intentions
• Strict use of these associations allows for
  unambiguous interpretation of data to synthesize
  target environments
• Implication infrastructure standardization is
  inherent, solving a key system integration
  problem

8
MIC Relevance II

• All graphic components (syntax) and their
  associations (semantics) are assigned meaning
• MetaModeler is put in charge of making Systems
  conform to infrastructure and capturing necessary
  characteristics

9
MIC Relevance IIIThe Proposition
Refine our needs into a block structure
representing the specific systems (PMC, GUI,
Sim. SigProc.) and define visual characteristics
at the highest level Result an infrastructure
linked through a run-time kernel with subsystem
comps. capable of contrib. via dir. conns.
10
MIC -gt Activity Modeling Tool
Capture characteristics of individual systems at
meta-level, infuse these into the domain-specific
environment Solves reconfiguration eased
through finding essentials
11
Our RT Environment
12
Our RT Environment, GUI
Java Applet configured dynamically from model
information, allows for bidirect. communication
with data pipeline
13
Conclusions

• Model Integrated Computing application to System
  Integration problem in specific case of Chemical
  Plant processing
• Solves problems of
• Abstracting modeler from details of system
• Allowing for easy reconfiguration
• Infrastructure defn for addition of new comps.
• Places intrinsic difficulties on meta-modeler and
  gives domain user closer interaction
• Provides framework for numerous system
  integration problems

14
Towards a Paradigmfor Activity ModelingJason
Garrett, Akos Ledeczi, Frank DeCaria
Funding support from the Dupont Chemical
Corporation, Old Hickory Plant, TN

• SMC2000 WP-2.9
• Modeling Control Hybrid Systems 2
• October 11, 2000, Hermitage D, 510 pm