Professional Documents
Culture Documents
Korea2010 1
Korea2010 1
Keywords: Linear induction motor – model predictive control, sliding mode control.
1. Introduction
employed in the electrical machine control systems [8-9].
Due to its several advantages such as high starting thrust,
alleviation of gears between motor and the motion devices, However, The use of PI controllers for speed control of
simple mechanical construction, no backlash and less induction machine drives is characterized by an overshoot
friction, and suitability for low speed and high speed during tracking mode and poor load disturbance rejection].
applications [1-4], the linear induction motor (LIM) has
been widely used in a variety of applications like as Because of rapid improvements in power electronic devices
transportation, conveyor systems, actuators, material and microelectronics, the field orientation concept has made
handling, pumping of liquid metal, sliding door closers, possible the high performance applications of induction
motor drives [9-10]. It has been applied successfully to the
curtain pullers, robot base movers, office automation, drop
LIM by aligning the d-axis of the primary current with the
towers, elevators,..etc.
secondary flux linkage.
The linear induction motor and traditional rotary induction
In the past few years, modern control techniques have been
motor have similar driving principles. However, the control
characteristics of the LIM are more complicated. This is used to control the speed and/or position of the induction
motor drives: direct torque control (DTC) technique [11 ],
attributed to the time varying motor parameters as a result
sliding mode control method [12-13 ], and linear quadratic
of change in operating conditions such as mover speed,
Gaussian (LQG) method [14 ]. Also, intelligent methods
temperature, and rail configuration. Moreover, there are
uncertainties existed in practical applications of the LIM [5- such as neural, fuzzy and genetic algorithm have been
7] which usually composed of unpredictable plant employed for this purpose [15-17]. In spite of the success of
parameter variations, external load disturbance, and un- the past methods as speed or position controllers in the
induction motor applications, new control techniques are
modeled and nonlinear dynamics. Furthermore, since the
needed to face the large uncertainties existed in the
operation of LIM involves two contact bodies, friction force
is inevitably among the forces of motion and results in application of the LIM which seriously influence the
steady state error, limit cycle and low bandwidth. In control performance especially at low speeds.
addition, the friction characteristics may be easily varied On the other hand, the MPC technique appears to be an
due to the change of normal forces in contact, temperature efficient strategy to control many applications in industry. It
and humidity. Therefore, the LIM drive system must has many advantages such as very fast response, and
provide high tracking performance, and high dynamic robustness against load disturbance and parameters
stiffness to overcome the above difficulties. For their uncertainty. Its straightforward design procedure is
simplicity and effectiveness, PI controllers are considered considered as the major advantage. Given a model of the
as the most widely used controllers which have been system, only an objective function incorporating the control
---------------------------- objectives needs to be set up. Additional physical
constraints can be easily dealt with by adding them as
A.A. Hassan, is with Faculty of Engineering, Dept. of Electrical inequality constraints, whereas soft constraints can be
Engineering, Minia University, Minia, Egypt.(email: aahsn@yahoo.com )
accounted for in the objective function using large penalties.
T.Hiyama is with Dept. of Electrical Engineering & Computer Science, Moreover, MPC adapts well to different physical setups and
Kumamoto University, Kumamoto, Japan. (email: hiyama@cs.kumamoto- allows for a unified approach [18-19].
u.ac.jp )
* T. H. Mohamed is with High Institute of Energy, South Valley In this paper, the speed control of the field oriented LIM
University, Egypt ( email : taherhie@yahoo.com ) drive has been developed based on MPC technique. The
Page 1 of 6
field orientation principle is used to decouple the mover 𝑖𝛼𝑠 , 𝑖𝛽𝑠 : 𝛼 − 𝛽 primary current components,
speed from the secondary flux amplitude. The MPC 𝑉𝛼𝑠 , 𝑉𝛽𝑠 : 𝛼 − 𝛽 primary voltage components,
technique law produces its optimal output derived from a 𝜎 : Leakage coefficient,
quadratic cost function minimization based on simplified h : Pole pitch,
and linearized LIM model. The technique calculates the 𝑛𝑝 : Number of pole pairs.
optimal control signal while respecting the given 𝑝 : Differential operator.
constraints over the output speed and force. The field
𝐹𝑒 : electromagnetic force,
oriented LIM drive along with the proposed MPC controller
𝐹𝐿 : external force disturbance,
has been tested under the motor parameters variation and
𝑀 : total mass of the moving element,
load disturbance. A comparison has been made between the
𝐷 : viscous friction and iron-loss coefficient
response of the MPC , SMC and the traditional PI
controllers. Simulation results proved that the proposed
3. Indirect Field Oriented LIM
controller can be applied successfully to control the speed
of the LIM drive and provide the best performance.
In the field oriented control method, the dynamics of the
The paper is organized as follows: Section 2 presents the highly coupled nonlinear structure of the induction machine
dynamic model of the linear induction motor. Indirect field becomes linearized and decoupled. The decoupled
oriented technique is described in section 3. General relationship is obtained by proper selection of state
consideration about MPC and its cost function are presented coordinates, under the hypothesis that the rotor flux is kept
in section 4. The implementation scheme of the LIM drive constant [17]. Therefore, the rotor speed is only
together with the MPC technique is described in section 5. asymptotically decoupled from the rotor flux, and is
Simulation results and general remarks are presented in linearly related to the torque current only after the rotor flux
section 6. Finally, the conclusions and future work are becomes in the steady state.
given in section 7.
The flux model of the LIM can be described in the d-q
synchronous frame as:
2. LIM Dynamic Model
𝐿𝑚 1 𝜋 𝑛𝑝 𝜋
The electrical dynamic model of the LIM is modified from 𝑝𝜆𝑑𝑟 = 𝑇𝑟
𝑖𝑑𝑠 − 𝑇𝑟
𝜆𝑑𝑟 +
𝑣𝑒 −
𝑣 𝜆𝑞𝑟 (6)
the traditional model of a three phase, Y-connected
induction motor in 𝛼 − 𝛽 stationary frame and can be 𝑝𝜆𝑞𝑟 =
𝐿𝑚
𝑖𝑞𝑠 −
1
𝜆𝑞𝑟 −
𝜋
𝑣𝑒 −
𝑛𝑝 𝜋
𝑣 𝜆𝑑𝑟 (7)
described by the following differential equations [20]: 𝑇𝑟 𝑇𝑟
𝑅𝑠 1−𝜎 𝐿𝑚 𝑛 𝑝 𝐿𝑚 𝜋 Where:
𝑝 𝑖∝𝑠 = − + 𝑖𝛼𝑠 + 𝜎 𝐿 𝜆𝛼𝑟 + 𝜎 𝐿 𝑣𝜆𝛽𝑠 +
𝜎𝐿𝑠 𝜎 𝑇𝑟 𝑠 𝐿𝑟 𝑇𝑟 𝑠 𝐿𝑟
1
𝑉 (1) 𝜆𝑑𝑟 , 𝜆𝑞𝑟 : 𝑑 − 𝑞 secondary flux components,
𝜎 𝐿𝑠 𝛼𝑠
𝑖𝑑𝑠 , 𝑖𝑞𝑠 : 𝑑 − 𝑞 primary current components,
𝑅𝑠 1−𝜎 𝐿𝑚 𝑛 𝑝 𝐿𝑚 𝜋 𝑣𝑒 = 2𝑓 : synchronous linear velocity ,
𝑝 𝑖𝛽𝑠 = − + 𝑖𝛽𝑠 + 𝜎𝐿 𝜆𝛽𝑟 − 𝜎 𝐿 𝑣𝜆𝛼𝑠 +
𝜎𝐿𝑠 𝜎 𝑇𝑟 𝑠 𝐿𝑟 𝑇𝑟 𝑠 𝐿𝑟 𝑓 :supply frequency.
1
𝑉
𝜎 𝐿𝑠 𝛽𝑠
(2)
In an ideally decoupled induction motor, the secondary flux
𝐿𝑚 1 𝑛𝑝 𝜋 linkage axis is forced to be aligned with the d-axis, and the
𝑝 𝜆𝛼𝑟 = 𝑇𝑟
𝑖𝛼𝑠 − 𝑇𝑟
𝜆𝛼𝑟 −
𝑣𝜆𝛽𝑟 (3) field orientation conditions can be applied. It follows that:
𝑛𝑝 𝜋
𝑝 𝜆𝛽𝑟 =
𝐿𝑚
𝑖𝛽𝑠 −
1
𝜆𝛽𝑟 + 𝑣𝜆𝛼𝑟 (4) 𝜆𝑞𝑟 = 0 , and 𝑝𝜆𝑑𝑟 = 𝑝𝜆𝑞𝑟 = 0 (8)
𝑇𝑟 𝑇𝑟
Page 2 of 6
The electromagnetic force can be described in the d-q outputs and on the proposed optimal future control actions.
synchronous frame as [17]: The prediction has two main components : The free
response which being expected behavior of the output
𝐹𝑒 = 𝑘𝑓 𝜆𝑑𝑟 𝑖𝑞𝑠 − 𝜆𝑞𝑟 𝑖𝑑𝑠 (11) assuming zero future control actions, and the forced
response which being the additional component of the
Where 𝑘𝑓 is the force constant which is equal to: output response due to the candidate set of future controls.
For a linear systems, the total prediction can be calculated
3𝑛𝑝 𝐿𝑚 𝜋 by summing both of free and forced responses, reference
𝑘𝑓 = trajectory signal is the target values the output should attain.
2𝐿𝑟
The optimizer is used to calculate the best set of future
control action by minimizing the cost function J, the
With the implementation of the field oriented control,
optimization is subject to constraints on both manipulated
equation (11) can be rewritten using equations (8) and (9)
and controlled variables [21-22].
as:
The general object is to tighten the future output error to
zero, with minimum input effort. The cost function to be
𝐹𝑒 = 𝐾𝐹 𝑖𝑞𝑠 (12)
minimized is generally a weighted sum of square predicted
errors and square future control values, e.g. in the
Generalized Predictive Control (GPC) :
Where
𝑁2
𝐽 𝑁1, 𝑁2, 𝑁𝑢 = 𝑗 =𝑁1 𝛽 𝑗 𝑦 𝑘 + 𝑗 𝑘 − 𝑤(𝑘 + 𝑗) 2 +
𝐾𝐹 = 𝑘𝑓 𝐿𝑚 𝑖𝑑𝑠
𝑁𝑢 2
𝑗 =1 𝜆 𝑗 𝑢(𝑘 + 𝑗 − 1) (13)
If the d-axis primary current (flux current component) is
kept constant at the rated value, therefore the Where 𝑁1 , 𝑁2 are the lower and upper prediction horizons
electromagnetic force is directly proportional to the q-axis over the output, 𝑁𝑢 is the control horizon, 𝛽 𝑗 , 𝜆 𝑗 are
current; which can be realized via closed loop control. In weighting factors. The control horizon permits to decrease
this case, if the q-axis current (load current component) is the number of calculated future control according to the
rapidly changed in response to the load variation, this will relation: ∆𝑢 𝑘 + 𝑗 = 0 for 𝐽 ≥ 𝑁𝑢 .
be followed by a rapid change in the motor developed force 𝑤 𝑘 + 𝑗 represents the reference trajectory over the future
and the LIM will exhibit a high dynamic performance. horizon 𝑁.
Constraints over the control signal, the outputs and the
4. Model predictive control control signal changing can be added to the cost function:
5. System configuration
Page 4 of 6
reference and actual speeds, d-q secondary flux
components, 3-phase primary currents, developed force and 1.5
reference speed actual speed
the load force. The load force is assumed to be stepped 1
from 350 N to 700 N at t = 0.5 second. It has been noticed
(m/s)
that the reference and actual speeds are aligned and good 0.5
secondary flux is equal to the set value (0.0568 wb) while 0.15
the actual q-axis flux is kept zero during the simulation
0.1
period. This means that the field orientation condition has
(wb)
been realized which leads to high dynamic performance 0.05
q-component d-component
drive. The figure reports also that the developed force 0
follows the increase of the load disturbance. Similarly, the 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
primary phase currents respond quickly to the speed and
20
load variations.
10
(A)
0
2
reference speed -10
actual speed
(m/s)
1 -20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time, s
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.1
Fig. 4 Dynamic responses of the proposed system under
parameters mismatch condition.
(wb)
0.05
q-component d-component
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.15
20 MPC
(A)
0
0.1
-20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1000 SMC
(m/s)
0.05
(N)
500
PI
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1000
(N)
500 -0.05
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time, s
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Fig. 5 MPC response versus SMC and PI responses at low
Time, s
speed.
Fig. 3 Dynamic responses of the proposed system at load Thirdly, the tracking performance of the MPC controller is
disturbance.
investigated at low speed (0.1 m/s) and compared with the
response of both the SMC and PI controllers. The load
Secondly, the robustness of the LIM with the MPC force is assumed to stepped from 350 N to 700N at t = 0.5
controller is investigated during parameters uncertainty. second. Figure (5) shows the MPC response as well as the
In this case, the secondary resistance is increased by 25% in SMC and PI responses under such case of study. It has
the LIM model, while it is kept at its nominal value in both been noticed that with the MPC controller, the reference
of the controller and the slip calculator. Also, the mover and actual speeds are aligned and good tracking
mass is increased by 50% only in the motor model. performance has been achieved even at the instant of load
Figure. (4) depicts the response of the MPC controller in disturbance. This is because the MPC provides feedback
this case of uncertainty at no load (FL = 0N). It has been compensation for the load disturbance. In contrast, both the
indicated that very fast response has been achieved using SMC and PI controllers need a period of time in order to
the MPC controller. Also, the waveforms of the the primary attain the steady state value either from start or after the
currents are free of any ripples. In addition, the actual q- load disturbance took place. Although the SMC controller
axis and d-axis indicate that the field orientation condition provides fast speed response but it suffers from the
has been realized. existence of chattering in the control effort.
Page 5 of 6
8. Conclusions [11] LascuC., I. Boldea, and F. Blaabjerg, “A modified
direct torque control of induction motor sensorless drive”
This paper investigates the successful application of the IEEE Trans. Ind. Application, Vol. 36, pp.122-130, 2000.
MPC technique to control the speed and force of the linear [12]R. J. Wai, “Adaptive sliding mode control for induction
induction motor drive. The MPC technique calculates the servomotor drive”, IEE Proc.- Electr. Power Appl., Vol.
optimal control signal while respecting the given 147, No. 6, November 2000.
constraints over the actual speed and electromagnetic force. [13] Wen-Jieh Wang and Jenn-Yih Chen, ” A new sliding
The field orientation principle is used to asymptotically mode position controller with adaptive load torque
decouple the mover speed from the secondary flux. The estimator for an induction
complete nonlinear dynamic model of the system has been motor”, IEEE Trans. On Energy Conversion, Vol. 14, No.3,
described in the stationary frame. Digital simulations have September 1999,
been carried out in order to validate the effectiveness of the pp.413-418.
proposed scheme. The proposed controller has been tested [14] K.J. AstrÖm-B.J.Wittnmark,”adaptive control system
through mismatched parameters and load force disturbance design’,Book, Adisson Wesily publishing, 1995.
at both high and low speeds. The system response is [15] Faa-Jeng Lin, and Rong-Jong Wai,"Hybrid control
evaluated by comparing with other traditional control using recurrent fuzzy neural network for linear induction
schemes. Simulation results show that the proposed MPC motor servo drive", IEEE Trans. On Fuzzy Systems, Vol. 9,
controller response has many advantages; very fast No.1, Feb. 2001, pp.102-115.
response, robustness against parameter uncertainties and [16] Faa-Jeng Lin, Rong-Jong Wai, Wen-Der Chou, and
load changes, well tracking of speed trajectory at all speeds Shu-eng Hsu,"Adaptive backstepping control using
and has almost no current and force ripples. recurrent neural network for linear
induction motor drive", IEEE Trans. On Industrial
References Electronics, Vol. 49, No.1,Feb. 2002, pp.134-145.
[17] Faa-Jeng Lin, Hsin-Jang Shieh, Kuo-Kai Shyu, and
[1] I. Takahashi, and Y. Ide," Decoupling control of thrust Po-Kai Huang,"On-line gain tuning IP controller using real
and attractive force of a LIM using a space vector control coded genetic algorithm", Electric Power System Research
inverter", IEEE Trans. Indust. Appl, Vol. 29, No.1, 1993, 72 , 2004, pp. 157-169.
pp.161-167. [18] Thomas J., D. Dumur, J. Buisson and H. Gueguen.
[2] I. Boldea, and S. A. Nasar,"Linear electric actuators and Model Predictive Control for Hybrid Systems under a State
generators", Cambridge University Press, UK, 1997. Partition based MLD Approach (SPMLD). International
[3] Z. Zhang, T. R Eastham, and G.E. Dawson,"Peak thrust conference on informatics in control, automation and
operation of linear induction machines from parameter robotics ICINCO’04, Vol. 3, pp. 78-85, Setúbal, 2004.
identification", Proc. of IEEE IAS, 1995, pp. 375-379. [19] A. A. Hassan , J. Thomas, " Model Predictive Control
[4] G. Bucci, S. Meo, A. Ometto, and M. Scarano,"The of Linear Induction Motor Drive", 17th IFAC World
control of LIM by a generalization of standard vector Congress, Seoul, Korea, July 6-11, 2008.
techniques", Proc. Of IEEE IAS, 1994, pp. 623-626 [20] Faa-Jeng Lin, and Rong-Jong Wai,"Robust control
[5] A. Gastli, "Compensation for the effect of joints in the using neural network uncertainty observer for linear
secondary conductors of a linear induction motor", IEEE induction motor servodrive", IEEE Trans. On ower
Trans. On Energy Conversion, Vol. 13, No.2, June 1998, Electronics, Vol. 17, No.2, March 2002, pp.241-251.
pp. 111-116. [21] Clarence W. De Silva “ Mechatronic systems : devices,
[6] A. Gastli, "Improved Field Oriented Control of an LIM design, control, operation and monitoring”, book published
Having Joints in its Secondary Conductors", IEEE Trans. by crc press, Taylor & Francis Group, 2008.
On Energy Conversion, Vol. 17, No.3, Sept. 2002, pp. 349- [22] E. F. Camacho, and C. Bordons, “ Model Predictive
355. Control”, Book, published by Springer-Verlag London
[7] G.H. Abdou, and S. A. Sherif," Theoritical and limited 1999.
experimental design of LIM in automated manufacturing
systems", IEEE Trans. Indust. Appl, Vol. 27, No.2, 1991,
pp.286-293.
[8] C. M. Liaw, and C. W. Tseng,"High erformance speed
controller for voltage source inverter fed induction motor
drives", IEE Proc.-B, Vol.139, No. 3, May 1992, pp. 220-
226.
[9] C. M. Ritter, and J. L. Silvino, "An alternative
sensorless field orientation method", IEEE Trans. On
Energy Conversion, Vol. 14, No.4, Dec. 1999, pp. 1335-
1340.
[10] D. W. Novotony and T. A. Lipo," Vector control and
dynamics of ac drives", Oxford, U.K.:Clarendon, 1996
Page 6 of 6