close

Вход

Забыли?

вход по аккаунту

...Dynamic Model Inspecting with OPM Model

код для вставкиСкачать
Visual Dynamic Model
Inspecting with OPM ModelBased Simulation Environment
Yevgeny Yaroker,
Valeria Perelman,
Prof. Dov Dori
18 February 2015
Introduction



Our domain: Conceptual design phase in
the systems engineering lifecycle.
The decisions made during this phase are
the most critical to get right and hardest to
change.
Existing testing approaches operate on the
system’s detailed design, a stage which is


too technical for the customer to follow
may be very expensive to backtrack
Motivation


There is a need to detect problems starting at
early stages of the system development, when
detailed design does not yet exist.
Benefits include:




Debugging during earlier stages of system
development, finding system malfunctions before
investing in extensive code writing
Working on a more abstract model – keeping the model
simpler with concealed details.
Modeling with focus on satisfying requirements.
Predictability of changes/modifications in design
Model-Based Simulation
Frameworks

System model simulation frameworks
for detailed design:


Graphical presentation of process flows


Modelica, ARENA, Simulink, …
OPM Animation (OPCAT)
Model-based simulation case tools:


xUMLite – (xUML)
…
System testing and evaluation
approaches
formal verification
Formal
run-time verification
simulation
unit tests
integration
tests
metrics
prototyping
checklist
scenario
Nonformal
questionnaires
Conceiving &
alternatives
evaluation
Conceptual
modeling
Detailed
Design
Sub-systems Integration
Implementation
Time/
Phase
What makes the conceptual
model evaluating so complicated?


Human abilities to comprehend system dynamics,
based on numerous static diagrams, are limited
even when well-organized and holistic modeling
languages (ML) are used.
High level of abstraction, which is typical of
conceptual MLs, propagate notations ambiguity.
These factors lead to numerous human errors while:


constructing system models
reading conceptual models written by others
Research Goal
Develop and evaluate a visual dynamic
model inspecting with OPM model-based
simulation environment.
What is Object Process
Methodology (OPM)?




OPM is a comprehensive, generic systems
development and lifecycle support paradigm
OPM Integrates the system’s function, structure
and dynamics in a single, unifying model.
Complexity is controlled through recursive and
selective scaling (zooming) of objects and/or
processes to any desired level of detail.
The OPM model combines image and text:
 intuitive graphics
 a subset of natural English
OPM Basic Concepts



Object: A thing that exists or can exist physically
or logically
Process: A thing that transforms an object by
creating it or consuming it or changing its state
Objects and processes can be connected with
links, which can be structural (such as
aggregation, generalization) and procedural
(enabling, transformation, and event links)
OPM main links
Structural Links
Procedural links
Connect objects to objects
Connect objects to processes
Consumption
GeneralizationSpecialization
AggregationParticipation
Result
Effect
ExhibitionCharacterization
ClassificationInstantiation
Agent
Instrument
Uni- and Bidirectional
Tagged link
object
object
10
Solution Outline



Develop a model-based animation mechanism as an
extension of OPM using the infrastructure provided by
OPCAT
Define clear system behavioral rules in line with OPM
syntax and semantics
Design a software tool with a wide user profile:




having easy-to-use User Interface (UI),
not requiring a special technical background
providing efficient problem detection and reporting
mechanisms
Develop a flexible system to enable future extensions
Technical Requirements

Simulation/visualization workflow requirements:



Simulation shall follow OPM rules.
Instance-level simulation shall be enabled.
“DVD film” simulation mode requirement: Enable easy work through possible
scenarios inexperienced users:
Forward/backward simulation, step by step/continuous simulation,
pause/continue (relevant only for continuous mode), changing simulation
speed
Debugging functional requirements:






Capability to define breakpoints
Provide “lifespan” component which graphically describes the state of all the OPM
entities at any stage of simulation
Provide special “Debug Info” component which detects possible problems and notifies
user about them.
Capability to reproduce problematic scenarios
Simulation Typical View
Object with
existing
instance is
colored
Red token
runs on the
activated link
Process is
colored if it is
currently
executed
Simulation Main Controls

Main Toolbar – Controlling simulation flow





Starting/Stopping simulation.
Playing forward/backward.
Controlling simulation velocity.
Invoking simulation properties dialog.
Status Bar – Observing simulation status


Play mode.
Current timeline.
Status Bar
Main toolbar
Model Debugging Process



User can toggle breakpoint on a process.
The simulation runs till the breakpoint is
reached.
Reaching the breakpoint will pause the
simulation.
Debug Information

Notifying a user regarding the problems making
the further progress run impossible, such as:



A process invocation failure,
A required manual process activation,
“No future simulation events” situation.
Architecture – Main Constructs


Task Queue consists of Simulation Tasks.
A Simulation Task:





represents an atomic simulation activity
implements Command and Undo DPs.
Task Queue keeps the simulation working
schedule.
Rule defines the Simulation Tasks to be scheduled
upon some simulation events.
Scheduler uses rules to build the Tasks Queue.
Architecture: The Three Layered
Model
Tasks Queue Creation Layer
Tasks Queue Execution Layer
Plug-in Layer
Rules

Simulation Rule is defined for each OPM entity
and for each simulation event.



For instance, there is a rule for object’s instance
creation and its deletion.
An activated rule determines the Simulation
Tasks to be executed and the set of next Rules
to be scheduled.
For each rule there is a class implementing its
functionality.
Process Activation Rule
Example
Activation Conditions:


The elements linked to the process with one of the following links
should be active -



Instrument link.
Condition link.
…
Consequent Rules:




Process termination will be scheduled t time units after its activation
time, where t – is the process execution time.
Elements linked to the process with Consumption link will be
deactivated.
…
Consequent Tasks:




The process will be painted.
Elements linked to the process with Result link will be painted
progressively.
…
Evaluation Experiment






Two OPM models were prepared for two different example systems.
The systems are very similar in their size and complexity.
Structural and behavioral errors were inserted intentionally to the OPM
models.
Two groups of students were asked to find all the errors in the two
models.
Each group analyzed one system using solely static set of the model
diagrams, and another system using simulation tool.
Following table describes the experiment setup
Group A
Group B
System
Model 1
Static (manual) analysis
Analysis with the simulation
environment assistance
System
Model 2
Analysis with the simulation
environment assistance
Static (manual) analysis
Evaluation Results (1)


Group of 98 students made the experiments in pairs (49
pairs).
The experience of this group with OPM and system
modeling is average.
Behavioral errors found
(max = 5)
Structural errors found
(max = 5)
Static (manual)
analysis
2.12
1.08
Analysis with the
OPCAT simulation
environment
2.98
0.50
Evaluation Results (2)


Paired t-test results. N =49 pairs, 98 students.
T
P
Behavioral aspect
5.66
< 0.001
Structural aspect
-3.77
< 0.001
The students were also asked to give their general
impression about helpfulness of the tool. The average
mark is 4.81 on a 1-7 scale.
Summary



Through the research visual dynamic model
inspecting with OPM model-based simulation
environment was developed.
Results gathered using two experiments carried
out on a large group of students confirmed the
efficiency of the proposed solution.
Although the proposed solution is OPM-oriented,
the architecture and attitude could be reused to
implement simulation engines for other MLs
1/--страниц
Пожаловаться на содержимое документа