Sunteți pe pagina 1din 10

Proceedings of the ASME 2009 International Design Engineering Technical Conferences &

Computers and Information in Engineering Conference


Proceedings of the ASME 2009 International Design Engineering Technical Conferences &
Computers and Information in EngineeringIDETC/CIE
Conference 2009
August 30 - September 2, 2009, San Diego, California,
IDETC/CIE 2009 USA
August 30-September 2, 2009, San Diego, USA

DETC2009-87202
DETC2009-87202

COMPARING GENERALIZED ORDER PID CONTROLLERS FOR NETWORKED


CONTROL SYSTEMS WITH RANDOM DELAYS AND DATA DROPOUTS

Inés Tejado, Blas M. Vinagre YangQuan Chen


Industrial Engineering School Center for Self-Organizing and Intelligent Systems (CSOIS)
University of Extremadura Utah State University
06071 Badajoz, Spain 4120 Old Main Hill, Logan
Email: itejbal,bvinagre@unex.es Email: yqchen@ece.usu.edu

ABSTRACT use of a multipurpose shared network to connect spatially dis-


The performance of a Networked Control System (NCS) can tributed elements results in flexible architectures and generally
be affected directly and indirectly by transmission time delays reduces installation and maintenance costs. Consequently, NCSs
and data losses in the communications network. In this paper, have been finding application in a broad range of areas. How-
we propose different control strategies, traditional and fractional ever, the insertion of the communications network in the feed-
order PID controllers, to control remotely an experimental plat- back loop makes the analysis and design of an NCS complex,
form, called Smart Wheel, at the Center for Self-Organizing and since it imposes additional challenges to the control system due
Intelligent Systems (CSOIS), Utah State University, and analyze to several issues: fixed or randomly distributed delays in trans-
their effectiveness by means of the jitter margin in simulation. mitting information between components, unreliable communi-
A simulation model of a communications network for different cation links with information losses (packet dropout), or the con-
nature of transmission delays, based on MATLAB/Simulink, is sideration of band-limited channels (see [2–4] for more informa-
presented and used for simulations. tion).
The goal of networked control systems is similar to other
kinds of control systems, i.e. to achieve stability, and provide
INTRODUCTION good closed loop performance. General procedures for designing
Future automation systems, an even those currently in use controllers for networked control include either modifying the
today, will to an increasing extent consist of a large number of network protocol to suit the system (as the controller may not
intelligent devices and control systems connected by local or take into account delay effects), or considering/compesating such
global communications networks [1]. Networked Control Sys- effects during controller design [5].
tems (NCSs) or Network Based Control Systems (NBCSs) are IP based NCSs networks have been shown to have non-
feedback control systems where sensors, actuators and comput- uniform distribution and multifractal nature, while these findings
ing elements are connected by means of a communications net- have not been effectively utilized in NCSs analysis and synthe-
work or other shared medium. So, the defining feature of an NCS sis (see [6] and the references therein). Our previous work on
is that control and feedback signals are exchanged among the NCSs [8–10] showed a communications network model for sim-
system components in the form of information packages through ulations which consisted of a packet losses model and a trans-
a network. mission delays model, that is, two separated models. Here, we
The study of NCSs brings together the historically separate present a whole model for the network traffic with packet losses
disciplines of communications networks and control theory. The and transmission delays, considering different kinds of distribu-

1 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use



tions for these delays. ⎨ 1, if υ > 0
The rest of this paper is organized as follows. In Section 2, sign(υ) = 0, if υ = 0 (3)

main problems induced in an NCS due to the communications −1, if υ < 0
network are described and modeled. A network traffic model for
simulations is presented. Section 3 shows the description of the and α ∈ (0, 2], β ∈ [−1, 1], γ ∈ (0, ∞) and δ ∈ (−∞, ∞). These pa-
experimental platform used in this work. In Section 4, the pro- rameters are (see [5] and the references therein for more infor-
posed controllers are designed. Simulation results are included mation):
in Section 5 in order to verify the robustness of the designed
controllers. The final section draws the main conclusions of this - The characteristic exponent, α, which specifies the thick-
paper. ness of probability density function.
- The skewness index, β, which means the assymetry of the
function.
TRAFFIC MODEL OF A COMMUNICATIONS NETWORK - The scale, γ, expresses the dispersion of the distribution.
Network modeling is in many cases a prerequisite for control - δ is the location parameter and is a function of the mean or
design. Among the problems introduced by the communications median of the entire distribution.
network, random transmission delays and information losses are
described and modeled in this section in order to develop a traffic
model of the network for simulations considering different delay Information Losses
nature. Another important issue is the occurrence of information
losses or missing messages, either in the controller site or at
the plant site, which is referred to as the problem of data packet
Transmission Delays losses or dropout. Some of the packets get lost in the network due
The major drawback of NCS is that the network dynamics to interference, noise, node failures and packet collisions [12].
affect the stability of the control system due to the additional net- According to [14, 15], packet losses can be modeled as an
work delays. Two transmission delays can be considered, caused infinite delay.
by the network portion lying between the plant outputs and the
controller and the one lying between the controller and the plant
actuator [11]: sensor-to-controller delay (τ sc ), and controller-to- Simulation Model of the Network Traffic
actuator delay (τ ca ). The model of the traffic network presented here is a model
In practice, these delays are induced not only by the finite based on delay, τ network , and includes transmission delays and
speed allowed by the communication medium, but also for the information losses by the following form:
particular structure of the communication protocol, and they can
be of fixed or time-varying/random nature depending on the com- 
munications protocol used. Some studies consider the modeling τ, if there is not dropout
τnetwork = (4)
of delays or traffic in NCS. In the literature, there are several ∞, if there is dropout
models available for network delays depending on the network
type and protocols used: 1) constant delay, 2) random delay with
being τ the transmission delay with the corresponding distribu-
probability distributions [12, 13], and 3) random delay with mul-
tion mentioned above.
tifractal nature [5, 6]. These varying delays are often referred to
In order to consider the possible asymmetry of network de-
as jitter [12, 13]. Hence, different delay distributions are consid-
lays in simulation, referred to different values of τ ca and τsc , the
ered for simulations: uniform or gaussian distribution, multifrac-
network model includes two different blocks. These transmis-
tal nature or non-uniform distribution, and random nature.
sion time delays have different distributions (uniform, α-stable
α − stable distributions can be used to model network traffic
or random), whose values are between a tunable maximum and
[5, 7], whose characteristic function has this form:
minimum values.
  Following [16], the data network subject to data losses is
ϕ(υ) = exp jδϕ − γ|υ|α [1 + jβsign(υ)ω(υ, α)] (1) simulated as a random number generator with uniform distribu-
tion between 0 and 1, a zero order hold and a switch, where its
threshold is set to (1 − r), being r the transmission rate at the
where
network. During r percent of the time, a logic level 1 is applied
 to the control input of the switch, letting the present delay be
tan απ
2 , if α = 1 transmitted to its output. The rest of the time, the switch causes
ω(υ, α) = (2)
2
π log |υ|, if α = 1 the delay in the network be infinite [14, 15]. Therefore, with

2 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


this model, when a packet dropout is simulated, the present data
is the previous available data. On the contrary, the transmitted
data is the present data with the corresponding assigned delay. In K −Ls 0.1484 −0.592s
this model, maximum and minimum values of τ sc and τca can be P(s) = e = e (5)
τs + 1 0.045s + 1
tuned.

CONTROLLERS DESIGN
DESCRIPTION OF PROTOTYPE SMART WHEEL AS- When applying the ideal PID algorithm in control of NCS,
SEMBLY AT CSOIS where varying time delays are present, one can show by using the
A stand-alone 3-axes robotic wheel assembly at CSOIS concept of jitter margin that the stability of the system cannot be
called Smart Wheel is the NCS plant taken into consideration guaranteed, at least in the sense of the jitter margin condition, for
for our experimental study. It is a self-contained robotic wheel any additional delays. In such case the jitter margin is zero. It
which has a steering axis, drive axis and z-axis, each of which should be noted that the jitter margin is a conservative criterion
can be actuated independently [17] as shown in Fig. 1. Smart and only sufficient, so the zero jitter margin does not necessarily
Wheel is equipped with steering and drive motors and has a lin- imply instability for additional delays [12, 20].
ear actuator for z-axis movement. It also has a power distribu- The aim of this work is to analyze and compare the robust-
tion unit, drive circuitry for the motors and actuators, encoders ness of three different PID controllers, a traditional PID designed
for drive and steering feedback and a microcontroller. from Ziegler-Nichols method, an optimal PID controller taking
into account tuning rules for varying time delay systems and a
fractional order PID, for the network based control of the Smart
Wheel in simulation.

PID Controller Using an Improved Ziegler-Nichols


Method
Traditional Ziegler-Nichols method is not enough to design
a PID controller which gets a proper system performance be-
cause of its intrinsic characteristics, so it is necessary to apply an
improved method. In [21], the authors propose the tuning rules
(7) and (8) for a PI controller with the following expression:

ki
PI(s) = k p + (6)
s

Figure 1. SMART WHEEL PICTURE AT CSOIS

0.15L + 0.35τ
kp = (7)
KL
Anyone with a computer connected to the network can con-
trol the Smart Wheel by communicating with it through serial
server. An internet camera (DLink - model DCS5300) with a
built in web server is directly connected to hub and is located 0.46L + 0.02τ
ki = (8)
near the Smart Wheel assembly. It has pan, tilt and zoom capabil- KL2
ities and sends streaming audio and video of the wheel’s motion
to the remote computer. The audio and video streams are inde- According to Eqn. (5), k p = 1.1901 and k i = 5.2533.
pendent of the controller data which uses the serial server. The
performance of the closed loop system is analyzed by plotting
the encoder data on the screen of the remote computer (see [18] Optimal PID Controller
for more information and the references therein). The tuning rules of an optimal PID controller (9) for varying
From results in [18, 19], the dynamical mathematical model time delay systems [12,19], referred to as OPID, are given by the
of the Smart Wheel is given by the following transfer function: expressions (10)-(12).

3 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


ki
PID(s) = k p + + kd s (9)
s


1 0.4τ − 0.04
kp = + 0.16 (10)
K L

Figure 2. BODE DIAGRAMS OF CONTROLLED SYSTEM USING THE


 PROPOSED CONTROLLERS
1 −0.11τ3 + 1.5τ2 − 1.5 0.35τ2 + 4τ + 50
ki = + (11)
100K L2 L
In order to use this controller in Simulink, an approximation
for the PIλ and PDµ parts have been obtained by polynomials
of order 2 using a simple MATLAB command, invfreqs, being
1
the controller expression (16). With regard to the PI λ part, the
kd = 0.4τ2 + 11τ (12) approximation has been done to keep the integral action by the
100K
form shown in (17), that is, the non-integer part of the integral
order.
Therefore, this is the optimal PID for the Smart Wheel:

0.069s4 + 1.725s3 + 2.941s2 + 1.085s + 0.039


5.4241 PI λ Dµ (s) = 102
PID(s) = 0.8277 + + 0.0334s (13) 10−4 s5 + 8.558s4 + 65.85s3 + 14.8s2 + 0.687s
s (16)

 λ
Fractional Order PIλ Dµ Controller λ λ1 s + 1 (λ1 s + 1)λ 1
In order to design the fractional order PI λ Dµ (FPID) based PI = = · (17)
s sλ−1 s
on the rules in [22], we consider the phase margin of the con-
trolled system applying the optimal PID and the crossover fre-
quency: 67.7 o at the frequency of 0.811 rad/s. Therefore, we Figure 2 presents the Bode diagrams of the controlled sys-
want to design the PI λ Dµ controller with Eqn. (14) from these tem using the proposed controllers. As we can see, the differ-
specifications. ences between the designed controllers in the frequency domain
are not meaningful. Obtained results are summarized in Tab. 1,
where ωc is the crossover frequency.

 λ  µ Table 1. SUMMARY OF SPECIFICATIONS IN THE FREQUENCY DOMAIN


λ λ1 s + 1 λ2 s + 1
PI D (s) = Kx
µ µ
(14) Strategy Gain Margin (dB) Phase Margin (o ) ωc (rad/s)
s xλ2 s + 1 PIZ−N 11.1 71.3 0.792
OPID 10.6 67.7 0.811
The obtained values for the controller parameters are: K = PIλ Dµ 9.08 70.2 0.641
11.923, x = 8.87·10 −4, λ = 0.887, µ = −0.476, λ 1 = 1.974 and
λ2 = 26.6. So, the designed fractional order controller has the
expression: SIMULATION RESULTS
Next, the effectiveness of the proposed controllers over a
network configuration, using the developed model of the net-
 0.887  −0.476 work traffic, will be illustrated through MATLAB/Simulink sim-
λ 1.974s + 1 26.6s + 1 ulations.
PI D (s) = 338.314
µ
s 0.024s + 1 In order to evaluate the performance of the closed-loop con-
(15) trol system, a criterion must be set. One candidate measure for

4 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Figure 3. SYSTEM RESPONSES FOR LOCAL CONTROLLERS Figure 4. SYSTEM RESPONSES FOR LOCAL CONTROLLERS WITH
JM = 1.457s
NCS is the jitter margin [12]. Therefore, we present simulation
results to analyze the system performance in two cases: consid-
ering local and remote controllers. In both, we focus on the mea-
sure of the jitter margin.
The jitter margin is defined as a function of the amount of
constant delay in the control loop, and it describes how much
additional time-varying delay can be tolerated before the loop
goes unstable [20].

Simulation Results for Local Controllers


Figure 3 shows the step responses of the Smart Wheel using
the proposed controllers locally.
Figure 5. SYSTEM RESPONSES FOR LOCAL CONTROLLERS WITH
Doing an analysis in the frequency domain and taking into
JM = 1.571s
account the definition of jitter margin, the condition (18) is veri-
fied at the crossover frequency:

PM·π
= JM·ωc (18)
180

where PM is the phase margin of the controlled system, in de-


grees, ωc is the crossover frequency, in rad/s, and JM is the jitter
margin, in seconds. So, the jitter margin can be calculated ap-
plying the following expression:

PM·π
JM = (19)
180ωc Figure 6. SYSTEM RESPONSES FOR LOCAL CONTROLLERS WITH
JM = 1.911s
Therefore, the jitter margin of the controlled system using
each designed controllers can be obtained from Tab. 1 as:
Table 2. JITTER MARGIN APPLYING THE PROPOSED CONTROLLERS These theoretical results can be contrasted by the simulation
Strategy Jitter Margin (s) results in Fig. 4, 5 and 6. It is seen that the best system perfor-
PIZ−N 1.571 mance is obtained using the fractional order PI λ Dµ controller, as
OPID 1.457 it can be deduced from the Tab. 2.
PIλ Dµ 1.911

5 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Figure 7. DIAGRAM OF NETWORKED SMART WHEEL

Simulation Results for Remote Controllers Figure 8. NETWORK TRAFFIC WITH UNIFORM DISTRIBUTION FOR
Here, we include simulation results considering remote con- DELAYS
trollers according to the scheme in Fig. 7 in different simulation
cases for packet losses.

The selection of the sampling time plays an important role


in NCS. Traditionally, a small sampling time is chosen to ap-
proximate the continuous-time plant as closely as possible and
to enable accurate control. Nevertheless, in NCS, a small sam-
pling time causes high network load and an increasing risk of
network congestion, which results in longer delays and hence
lower performance. Thus the network delay and the sampling
time are coupled, and finding an optimal balance between the
two is a core requirement for achieving well performing and sta-
ble NCS [12].
Figure 9. NETWORK TRAFFIC WITH α-STABLE DISTRIBUTION FOR
In this case, we use the simulation model for the network
DELAYS
traffic presented above, with the following parameters for the
communications network: we suppose a symmetrical network
with the following time delays values: τ ca |max = τsc |max = 1.6h
and τca |min = τsc |min = 0.2h, where h is the sampling time (h =
20ms). It is reasonable to assume that the delays are distributed
on [0,bh], with b < 0.5 [13], so we consider a worse case. More-
over, firstly we set the parameter r as 0.9, that is, 90% of informa-
tion is transmitted through the network correctly and 10% of the
messages is missing. With regard to the delay distributions, we
simulate: a) an uniform or Gaussian distribution between τ max
and τmin ; b) an α − stable distribution with α = 1.1, β = 0, γ = 1
and δ = 0 (this distribution has been generated with the program
’stable.exe’ [23, 24]); and c) a random distribution with mean 1
and variance 0 between τ max and τmin . Therefore, network traf-
fic in these situations, including transmission delays and packet Figure 10. NETWORK TRAFFIC WITH RANDOM DISTRIBUTION FOR
losses, are shown in Fig. 8, 9 and 10, where packet loss is repre- DELAYS
sented as ’*’, at the top of the figures.

With these simulation parameters, the step responses con- the traffic model considered for the network (delay distribution
sidering the designed controllers remotely are depicted in considered).
Fig. 11, 12 and 13 for different delay distributions. As we can
see, problems induced by the communications network, that is, Hence, in order to measure the jitter margin of the controlled
transmission delays and packet losses, do not meaningfully af- system using remote controllers, we focus on α − stable distri-
fect the performance of the controlled system, independently of butions to model network traffic as same as recent researches and

6 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Figure 11. SYSTEM RESPONSES APPLYING REMOTE PID Figure 14. SYSTEM RESPONSES FOR REMOTE CONTROLLERS
ZIEGLER-NICHOLS CONTROLLER

Figure 15. SYSTEM RESPONSES FOR REMOTE CONTROLLERS


Figure 12. SYSTEM RESPONSES APPLYING REMOTE OPID CON-
WITH JM = 1s
TROLLER

trollers (see Fig. 3).

In Fig. 15, 16, 17 and 18 system responses are represented


using the same jitter margins as the local simulation. It is clear
that the system goes unstable in each case for a smaller value
of the jitter margin. Anyway, the best effectiveness of the con-
trolled Smart Wheel is obtained using the fractional order PI λ Dµ
controller.

Moreover, the proposed controllers effectiveness is verified


for different dropout cases, that is, for different values of the
transmission rate r in the network. Figure 19, 20, 21 and 22 show
Figure 13. SYSTEM RESPONSES APPLYING REMOTE FPID CON- the behavior of each controller over the system in these different
TROLLER information losses situations. It is observed the smaller value
of r, the smaller value of the jitter margin. So, the controlled
system goes unstable for a smaller value of the jitter margin as
long as the value of r decreases. On the other hand, an addi-
publications [5, 14] (multifractal nature of delays). tional delay is induced on the system response as time passes for
Figure 14 shows the Smart Wheel responses using remote higher dropout and jitter margin. Anyway, using the fractional
controllers. As we can observe, taking into account the men- order PIλ Dµ controller remotely, the Smart Wheel has the best
tioned before about the low influence of the communications net- behavior.
work, these responses are similar to the obtained using local con-

7 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Figure 16. SYSTEM RESPONSES FOR REMOTE CONTROLLERS Figure 18. SYSTEM RESPONSES FOR REMOTE CONTROLLERS
WITH JM = 1.457s WITH JM = 1.911s

Figure 17. SYSTEM RESPONSES FOR REMOTE CONTROLLERS


WITH JM = 1.571s
Figure 19. SYSTEM RESPONSES FOR REMOTE CONTROLLERS
WITH DIFFERENT RATE TRANSMISSION

CONCLUSIONS
This paper has focused on the networked control of the ex- to be a versatile and valuable tool to deal with networked sys-
perimental platform Smart Wheel applying traditional and frac- tems.
tional order control strategies in order to analyze the effects
of the communications network by means of the jitter margin.
Specifically, a PID controller tuned by improved Ziegler-Nichols ACKNOWLEDGMENT
method, an optimal PID controller for time-varying systems and This work has been supported by the Spanish Ministry
a fractional order controller have been proposed. Moreover, a Research Grants DPI2005-07980-C03-03 and TRA2008-06602-
simulation model of a communications network considering dif- C03-02/AUT.
ferent nature of delays is presented.
The behavior of the system applying the proposed con-
trollers remotely is adequate, however the best effectiveness is REFERENCES
obtained using the fractional order PI λ Dµ , since the jitter mar- [1] Vatanski, N., Georges, J.P., Aubrun, C., Rondeau, E., and
gin of the controlled system is higher than using the rest of con- Jämsä-Jounela, S.L., 2009, “Networked Control With De-
trollers. Besides, the way of tune the PI λ Dµ controller is so easy. lay Measurement and Estimation”. ScienceDirect, Control
On the other hand, experimental results are needed in order to Engineering Practice, 17, pp. 231-244.
verify the proposed controllers. [2] Hristu-Varsakelis, D., and Levine, W.S., ed., 2005,
Further research is needed on obtaining new and better ap- Handbook of Networked and Embedded Control Systems,
proximations of the fractional order controller. Birkhäuser.
To sum up, the fractional order PI λ Dµ controller has proved [3] Hespanha, J.P., Naghshtabrizi, P., and Xu, Y., 2007, “A

8 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Figure 20. SYSTEM RESPONSES FOR REMOTE CONTROLLERS Figure 22. SYSTEM RESPONSES FOR REMOTE CONTROLLERS
WITH DIFFERENT RATE TRANSMISSION (WITH JM = 1.457s) WITH DIFFERENT RATE TRANSMISSION (WITH JM = 1.911s)

Traffic Modeling Using α−stable Self-similar Processes”.


IEEE Transactions on Communications, 49(7), pp. 1203-
1214.
[8] Tejado, I., Vinagre, B.M., and Suárez, J.I., 2008, “Effects
of a Communication Network on the Longitudinal and Lat-
eral Control of an AGV”. In: Proceedings of 2008 IEEE In-
ternational Symposium on Industrial Electronics, pp. 2084-
2089.
[9] Tejado, I., Vinagre, B.M., Suárez, J.I., and de Pedro, T.,
2008, “Bounded Control Strategies for Minimizing the Ef-
fects of the Communications Network on Lateral Control
of an AGV”. In: Proceedings of 14th IEEE Mediterranean
Electrotechnical Conference.
Figure 21. SYSTEM RESPONSES FOR REMOTE CONTROLLERS
[10] Romero, M., Tejado, I., Suárez, J.I., Vinagre, B.M., and de
WITH DIFFERENT RATE TRANSMISSION (WITH JM = 1.571s)
Madrid, A.P., 2009, “GPC Strategies for the Lateral Con-
trol of a Networked AGV”. In: Proceedings of 5th IEEE
International Conference on Mechatronics.
Survey of Recent Results in Networked Control Systems”. [11] Ray, J., 2004, “Networked Control System Co-simulation
In: Proceedings of the IEEE, 95(1), pp. 138-162. for Co-design: Theory and Experiments”. Ph.D. Thesis,
[4] Gómez-Skarmeta, A.F., Martinez-Barberá, H., and Tomás- Case Western Reserve University.
Balibrea, L.M., 2002, “Mimics: Exploiting Satellite Tech- [12] Eriksson, L., 2008, “PID Controller Design And Tuning
nology for an Intelligent Convoy”. Intelligent Transporta- In Networked Control Systems”. Ph.D. Thesis, Helsinki
tion Systems, pp. 85-89. University of Technology.
[5] Mukhopadhyay, S., Han, Y., and Chen, Y.Q., 2009, “Frac- [13] Nilsson, J., 1998, “Real-time Control Systems With De-
tional Order Networked Control Systems and Random lays”. Ph.D. dissertation, Lund Institute of Technology.
Delay Dynamics: A Hardware-In-The-Loop Simulation [14] Zaborovsky, V., and Meylanov, R., 2001, “Informational
Study”. In: Proceedings of the 2009 American Control Network Traffic Model Based on Fractional Calculus”. In:
Conference. Proceedings of the 2001 International Conferences on Info-
[6] Peng, C., Yue, D., Tian, E., and Gu, Z., 2009, A Delay Dis- tech and Infor-net, 1, pp. 58-63.
tribution Based Stability Analysis and Synthesis Approach [15] FRACTEL, Inc., 2002, “New Traffic Model on
for Networked Control Systems. Journal of the Franklin In- the Base of Fractional Calculus”. See also URL
stitute, In Press, Corrected Proof, Available online 23 Jan- http://www.fractel.ru/papers.shtm.
uary 2009. [16] Ostertag, E., and Carvalho-Ostertag, J., 2006, “Inverted
[7] Kasaridis, A., and Hatzinakos, D., 2001, “Network Heavy Pendulum Stabilization in an NCS Scheme with Packet

9 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Losses”. In: Proceedings of the 7th Portuguese Conference
on Automatic Control.
[17] Flann, N., and Moore, K., 2000, “A Six-wheeled Omnidi-
rectional Autonomous Mobile Robot”. IEEE Control Sys-
tems Magazine, 20, pp. 53-66.
[18] Bhambhani, V., Han, Y., Mukhopadhyay, S., Luo, Y., and
Chen, Y.Q., 2008, “Experimental Study on the Real-
time Smart Wheel Networked Speed Control System Using
Hardware-in-the-Loop Control”. Submitted to Communi-
cations in Nonlinear Science and Numerical Simulation.
[19] Bhambhani, V., Han, Y., Mukhopadhyay, S., Luo, Y., and
Chen, Y.Q., 2008, “Random Delay Effect Minimization on
a Hardware-in-the-Loop Networked Control System Using
Optimal Fractional Order PI Controllers”. In: Proceedings
of the IFAC Conference on Fractional Differentiation and
Its Applications.
[20] Cervin, A., Lincoln, B., Eker, J., Årzén, K.E., and But-
tazzo, G., 2004, “The Jitter Margin and Its Application in
the Design of Real-Time Control Systems”. In: Proceed-
ings of the 10th International Conference on Real-Time and
Embedded Compunting Systems and Applications.
[21] Åström, K.J., and Murray, R.M., 2008, Feedback Systems.
An Introduction for Scientists and Engineers. Princeton
Universtiy Press, Princeton.
[22] Monje, C.A., 2006, “Design Methods of Fractional Order
Controllers for Industrial Applications”. Ph.D. Thesis, Uni-
versity of Extremadura.
[23] Nolan, J.P., John Nolan’s Stable Distribution Page. See
URL http://academic2.american.edu/˜jpnolan.
[24] Nolan, J.P., 2001, “Maximum Likelihood Estimation and
Diagnostics for Stable Distributions”. Lévy Processes: The-
ory and Applications, Birkhäuser, pp. 379–400.

10 Copyright 
c 2009 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/30/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

S-ar putea să vă placă și