An H Fault Estimation

Scheme of Wireless
Networked Control
Systems for Industrial
Real-Time Applications
Ying Wang, Steven X. Ding, Dongmei Xu, and Bo Shen
IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 22, NO. 6, NOVEMBER 2014

Presenter: Muhammad Royyan

Networked Systems Laboratory

Department of IT Convergence Engineering
Kumoh National Institute of Technology, South Korea

4/16/17 1
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 2
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 3
http://nsl.kumoh.ac.kr/
 Serendipity

4/16/17 4
http://nsl.kumoh.ac.kr/
 Serendipity

4/16/17 5
http://nsl.kumoh.ac.kr/
 Seminar History
Implementation of Kalman Filter and PID Controller for Inverted
Pendulum Robot (30 September 2016)

4/16/17 6
http://nsl.kumoh.ac.kr/
 Paper History
Domestic Conference
-
International Conference
-

4/16/17 7
http://nsl.kumoh.ac.kr/
 Paper History
International Journal
-

4/16/17 8
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 9
http://nsl.kumoh.ac.kr/
 Motivation and Contribution
Motivation
There is need for designing guaranteed high real-time
performance Wireless Sensor Networks.
Contribution
All the systems are modelled thus the parameters of WSN
gained as solution of Riccati difference equations which
essentially guarantee high real-time performance.
In this draft, model is developed as it still could solve with
arbitrary inputs, which means it could be directly applied to
Industrial Application.

4/16/17 10
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 11
http://nsl.kumoh.ac.kr/
 Related Works
No Main Idea Drawbacks
.
18 Robust scheduling algorithm When the network is under
with purpose of keeping the heavy load the network-induced
queue length in the network delay and packet losses still
close to an operating point. cannot be avoided even with
deterministic network behavior.
28 A unified measurement model
is used to depict the
transmission delay and
packet loss, and a least-
squares FD filter is designed
to estimate the system state
and fault on Internet-based
system.
29 H FE for time-varying system
with delayed measurements

4/16/17 12
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 13
http://nsl.kumoh.ac.kr/
 H
synthesize controllers achieving stabilization with guaranteed performance
(QOS of real-time in NCS).

4/16/17 14
http://nsl.kumoh.ac.kr/
 Sensor
System is represented by discrete periodic model integrated with periodic
information scheduling model.
k is number of period
j is current period

j = 0,1,2,T-1
is control input
is plant output
is actual input of plant
is part of filter resources

4/16/17 15
http://nsl.kumoh.ac.kr/
 Plant model

4/16/17 16
http://nsl.kumoh.ac.kr/
 Scheduler Model

Probability Density Function

4/16/17 17
http://nsl.kumoh.ac.kr/
 Plant model + Additive fault

Reading of Gyroscope when stationary

4/16/17 18
http://nsl.kumoh.ac.kr/
 Integrated Model

4/16/17 19
http://nsl.kumoh.ac.kr/
 H FE Algorithm

4/16/17 20
http://nsl.kumoh.ac.kr/
 H FE Algorithm
Suitable Model for H FE Algorithm
(according to lemma 1)

4/16/17 21
http://nsl.kumoh.ac.kr/
 H FE Algorithm

4/16/17 22
http://nsl.kumoh.ac.kr/
 H FE Algorithm

4/16/17 23
http://nsl.kumoh.ac.kr/
 H FE Algorithm

This algorithm is proofed optimal.

If you dont believe it, you can check the appendix

4/16/17 24
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Related Works
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 25
http://nsl.kumoh.ac.kr/
 Experimental Setup
FE algorithm experimented on WiNC platform with
integrated three-tank subsystem.
The three-tank subsystem is formulated into a discrete
periodic system when sensors are sampled with different
sampling rates.
The scheduler coordinating with this discretized system
model is also proposed both in the cases of no delay and
delay.
TDMA medium access method is adopted in WiNC to
ensure the deterministic transmission behaviors

4/16/17 26
http://nsl.kumoh.ac.kr/
4/16/17 27
http://nsl.kumoh.ac.kr/
 WiNC platform
Constructed to explore the control, FDI, fault tolerant
control, and communication issues based on real-time
reliable W-NCSs
The wireless network supports IEEE 802.11a/b/g
Remote controller, wireless communication facilities
optimized for industrial real-time wireless
communication.
Consist of two well-known laboratory benchmarks
three-tank system
inverted pendulum system

4/16/17 28
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Introduction
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 29
http://nsl.kumoh.ac.kr/
 Performance Evaluation
Step Explanation
1. State goal To implement H FE Algorithm Industrial Plant

.2. Performance metrics System output response, Estimated fault

3. System & workload parameters Model of system its self

4. Factor and value Transmission delay
5. Evaluation technique Simulated by MiNC.
6. Workload Model of system its self
7. Design experiments Three-tanks system simulation by Minc
platform
8. Analyze and interpret data Analyze output response real time

9. Present the result Using continuous Graph

4/16/17 30
http://nsl.kumoh.ac.kr/
 Performance Evaluation

4/16/17 31
http://nsl.kumoh.ac.kr/
 Performance Evaluation

4/16/17 32
http://nsl.kumoh.ac.kr/
 Performance Evaluation

4/16/17 33
http://nsl.kumoh.ac.kr/
 Performance Evaluation

4/16/17 34
http://nsl.kumoh.ac.kr/
 Performance Evaluation

4/16/17 35
http://nsl.kumoh.ac.kr/
 Overview
History
Motivation
Introduction
System model
Experimental Setup
Simulation performance
Conclusion and future work

4/16/17 36
http://nsl.kumoh.ac.kr/
 Conclusion
Conclusion
Network scheduling algorithm based on TDMA medium
access mechanism is expressed by state-space equations.
FE aims at dealing with periodic W-NCSs with scheduler,
which is called P S system.
It successfully make reliable real time system on Industrial
WSN.

4/16/17 37
http://nsl.kumoh.ac.kr/
 My Research
Fault Detection Isolation (FDI) and Fault Estimation (FE)
by NCS on Wireless Networked Control Systems for
Industrial Real-Time Applications.

4/16/17 38
http://nsl.kumoh.ac.kr/
 Thank you very much
for your attention
Questions/Answers Time

Networked Systems Laboratory

Department of IT Convergence Engineering
Kumoh National Institute of Technology, South Korea

4/16/17 39
http://nsl.kumoh.ac.kr/