Agent Based Monitoring: VTB as Simulation Agent

1
Agent Based Monitoring VTB as Simulation Agent
2004 VTB Users and Developers Conference 15-16
September 2004

• F. Ponci
• Dept. of Electrical Engineering
• University of South Carolina

2
Outline

• Introduction the agent-based monitoring and
  diagnostics
• The simulation agent
• VTB-LabVIEW environment
• Validation of device models
• Generation of data for training of diagnostic
  systems
• Implementation of diagnostics algorithms
• The simulation agent for power electronics system
  monitoring

3
Agent-based monitoring and diagnostics I

• Power electronics power system
• Flat control structure for maximum flexibility
  and reconfigurability

4
Agent-based monitoring and diagnostics II

• Situation awareness of a complex systems requires
  data from many points of the system
• Allowing for interaction between the measurement
  sections
• Including inaccessible points
• Agency-agent structure for the monitoring and
  diagnostic system

5
Agent-based monitoring

• Monitoring agents are measurement sections
  capable of
• Sharing data with other agents
• Requesting services from other agents
• Taking decisions
• Simulation agent is a monitoring agent able to
  share knowledge coming from the simulation of the
  entire system or of a subsystem

6
Agent-based diagnostics

• The diagnostic agent posses the capabilities of
  the monitoring agents
• The diagnostic agent can also evaluate the
  health status of the system or of one of the
  subsystems
• The diagnostic agent may utilize the shared
  measured or simulated data

7
The role of the simulation agent during system
operation

• In steady state conditions
• generation of virtual data to be compared with
  the real data to perform diagnostic actions
• In real-time
• Generation of virtual measurements of non
  accessible measuring points
• Out of the monitoring loop
• training and situation awareness training
• a posteriori misbehavior analysis
• In support of a decision making system
• what-if simulation scenarios

8
The role of the simulation agent during
incremental system prototyping

• Product Model validation
• Once a given part of the system is built, the
  validated Product Model has to take the place of
  Requirement Model within the simulation
• Remote testing of the equipment
• real equipment location far from design location
• Real equipment inaccessible for technical or
  proprietary reasons
• Testing and tuning of monitoring and diagnostics
  set-up and algorithms

9
VTB environment for the simulation agent

• Requirements for the environment of the
  simulation agent
• easy interface to data acquisition systems
• real-time capability
• high level of flexibility
• possibility of run time changes
• networking
• VTB as the natural choice for simulation
• VTB-LabVIEW as an example of implementation
  environment of the simulation agent

10
Simulation Agent Data Feeding

• Single-point mode
• For real time applications
• E.g. obtain virtual measures from inaccessible
  points during system operation
• Buffered-point mode
• For quasi real-time or non real-time applications
• E.g. a posteriori recreation of different
  scenarios for analysis of a misbehavior

11
VTB-LabVIEW interfaceModel Validation of a
Device I
Simulated and measured outputs
Physical System Active filter Input from power
supply
The input of the simulated system is the input of
the physical system
System Input
LabView Acquisition Platform Acquired data input
and output of the filter
12
VTB-LabVIEW interfaceModel Validation of a
Device II
Simulated and measured outputs (superimposed)
Physical System 1-phase transformer (no
load) Input from the mains
The input of the simulated system is the input of
the physical system
System Input
LabView Acquisition Platform Acquired data input
and output of the transformer
13
Data generation for training a diagnostic system
VTB simulated system
Wavelet processing
Data acquisition
Neuro-fuzzy system
Fault
Non fault
14
Diagnostics algorithm implemented in VTB
PoliMi Milan-Italy
Agency structure
USC Columbia SC-USA
System Manager Agent PC 131.175.14.8
Measurement section and drive control
Wavelet Unit Agent PC 129.252.22.202
Measurement section Data Acquisition and
Monitoring Agent
Fuzzy Unit Agent PC 129.252.22.215
Internet/Intranet (TCP/IP protocol)
15
Physical Experimental set-up I
Very low power (70W) ? Visible effects of the
non-linearities of the drive No-load operating
conditions Fixed duty-cycle
16
Physical Experimental set-up II
Differential voltage probe Active current probe
Digital scope with GPIB interface 100kHz sampling
frequency Buffer capacity 15,000 samples LabVIEW
PXI with network connection Target of execution
of a custom VI Interface PC VI Settings
17
Physical Experimental set-up III
PC with network connection VTB and LabVIEW
installed hosts the VTB schematic of the
system the VTB-LabVIEW interface model
PC with network connection LabVIEW
installed hosts LabVIEW VI that reads data from
the in-port, visualizes data may take action on
the system or on the measurement chain
18
VTB experimental set-up
FROM VTB TO Monitoring system ActiveX-based
VTB-LabVIEW interface block sends simulated data
to the Manager Agent
FROM Monitoring system TO VTB ActiveX-based
VTB-LabVIEW interface block imports measured data
as simulation
Signal controlled voltage source
19
Input voltage
Input voltage as measured in the system
Input voltage in VTB
20
Rectified voltage
Rectified voltage as measures in the system
and as simulated by VTB
21
The AC current as visualized by the Manager Agent
AC current as simulated by VTB and as measured in
the system
22
Future Directions

• Test of the simulation agent on complex power
  electronics physical systems
• Blue-collar agents must be upgraded to
  white-collar agents
• More interaction, in terms of data sharing and
  action requests, e.g. active measurement set-up
  changes request, e.g. data from specific
  inaccessible points
• Integration of existing diagnostics tools within
  the agent-based monitoring system
• Non-Intrusive Load Monitoring (NILM) developed by
  Steve Leeb at MIT (current collaboration)
• Integration of the agent-based monitoring system
  with the agent-based control system
• Implementation choices, data sharing (trade-off
  between redundancy that favors reconfigurability
  and flexibility and repetition that increases
  cost and maintenance)
• Collaboration with ESRDC Control Thrust partners

23
Conclusions

• The agent-based monitoring and diagnostics is the
  natural complement to flat, agent-based control
  structures
• The simulation agent plays a significant role in
  agent-based monitoring and diagnostics both in
  the prototyping and operation phases of the
  system
• A DC brushless drive as an example of use of the
  simulation agent
• Future developments call for widespread
  collaboration and more complex workbenches