Professional Documents
Culture Documents
Abstract-- This work deal with the problem related to the system and the fuzzy controller are implemented in the
energy recovered in an induction motor drive. During Matlab/Simulink® program .
deceleration , a regenerative braking control strategy allows
the recovery of the mechanical energy stored in an inertial
mass. This process involves the conversion of the kinetic energy II. MODEL OF DE SYSTEM
into electric energy. The parts that compose the system are a Fig. 1 shows the schematic model of lector-mechanical
squirrel cage induction motor variable frequency drive and
system emulating the electric car. This general scheme
controller and the mechanical system (flywheel) representing
the inertia of the automobile. In order to test the performance
shows the involved components, which are: induction
of the intelligent control, it is compared with a conventional motor, flywheel, disc brakes, ball bearings, and shaft.
control scheme. The conventional control scheme is used to Bearings
generate the samples of the input and output signals, which
allows the identification of the dynamics of the system (plant).
Also, it is used as a reference for the performance of the
intelligent control. shaft
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
disq (9a)
Vrq Z Zr Lmisd Lm Z Zr Lr ird Rr Lr d
irq
The transformation equations to translate from three phases dt dt
to two axis abcÆdq0 are the following ones: Vr 0 Rr Lrl dtd ir 0 (10a)
In order to obtain the mathematical representation for
ªd º ª0 3 3 º ªa º ªa º ª 0 2 2º ª d º (1) stator dq0 model (Table 1) of the IM, is required that the
«q » 1« » « » 1«
1 2»» «« q »»
1 1 » ««b »» «b » 3 speed of the equal stator to zero for MR Ȧ=0 is selected,
« 2«
« » 2
3«
¬« 0 ¼» ¬1 1 1 »¼ «¬ c ¼» ¬« c ¼» ¬« 3 1 2»¼ «¬ 0 »¼ and like the variables of voltage of Vs0=Vr0=0, as well as
where the voltages in the Vrd=Vrq=0 rotor, then the currents
is0=ir0=0, the matrix of equations are following [3],[5],
[9]:
ªVs d º ª Rstat dtd Ls 0 d
dt Lm 0 º ªIsd º
« » « 0 Rstat dt Ls
d d » « » (14) 0 Lm
«Vs q » « dt » « Isq »
« 0 » « dtd Lm Zr Lm Rrot dtd Lr Zr Lr » « Ir d »
« » « »« »
¬ 0 ¼ «¬ Zr Lm d
dt Lm Zr Lr Rrot dtd Lr »¼ «¬ I r q »¼
where
Ls Lm Lsl , Lr Lm Lrl
The variables Vsd, Vsq and Vs0 are generated by [5], [9],
[10], [11]:
Vsc Vsb (15)
V sd
3
Vsq
2Vsa Vsb Vsc (16)
3
Vsa Vsb Vsc (17)
Fig. 2. IM equivalent circuit for d-q axis Vs 0
3
The equations of the electrical system are have been taken where Vsa, Vsb and Vsc are the three-phase excitation:
from [5],[9], [10]. The equations for the stator are:
dO (2) Therefore the variables voltage amplitude (Amp) and
Vsd Rs isd sd ZOsq synchronous speed (Ȧsinc), will be generated by the
dt
d Osq control.
Vsq Rs isq ZOsd (3) The generated torque Te based on the currents:
dt
L I I I I
3P (18)
d Os 0 (4) T
Vs 0 Rs is 0 e
2 2
m sq rd sd rd
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
The parameters of the analytical model of IM are taken from Bearings Bearings
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
controller and the slogan, normally contributed by the driver
car, who stops in serious case the reference speed (Ȧref).
V. CONTROL STRATEGIES
In the previous sections Ȧe has been defined like the control
variable of output the variables of input (Amp), (Ȧsinc) and
Fig. 9. Diffuse control implementation of the electro-mechanical
(TL), being these signals used to generate the controllers, as system.
one is in Fig. 8.
a) b)
Fig. 8. System (it plants) and control. Fig. 10. Diffuse control training data, a) IM training, b) mechanical
brake training.
A. Conventional control
1). PI control for MI With it matrices of n×m are obtained data, where the n -
For the design of the control of IM, the variables of control rows correspond to the moments of sampling in which
are: the amplitude of the voltage (Amp) and the synchronous the data and the m are taken - columns correspond to the
speed (Ȧsinc), see Fig. 8. From the error generated between antecedents and the consequent one. The time of
the speed of reference and the mechanical speed, the sampling T is constant during the stage of the data
following equations are deduced: acquisition.
(24)
³
Amp K P (Zref Ze ) K I (Zref Ze ) dt K1Ze Once obtained the signals of entrance and exit of the
(25) plant generated with aid of the conventional control,
Wsinc K P (Zref Ze ) K I ³ (Zref Ze ) dt Ze
ANFIS® was used to train or to tune the intelligent
Where the KP and KI are the gain of control PI, to calculate control. Parallel simulations become of the conventional
the gains the method of Ziegler and Nichols is used. control and diffuse control, with systems (plant)
From Amp and Ȧsinc the three-phase excitation is generated, identical, comparing their performance.
as it is possible to be observed in the equations (14-17).
2) Hybrid State Transition Kernel (NHTE)
2). Control of the mechanical brake
The brake will be driven when MI this decelerating, as long The approach considers the function of transition of
as it fulfills following the conditional ones for the states to define the process of data acquisition with
application of the pair to the brake disc intentions of identification and control [7][8][15][16].
The system to control has X(t) states of dimension N
d Z ref dt 0 Pot _ elec ! 0 error 0 TL
and U(t) entrances of P dimension, to see figure 11a.
dȦref/dt: acceleration The control and the measurement of the state update
Pot_elec: electrical power with periods of sampling of T seconds.
error: difference between speed of reference and mechanical Since access to a component of X is only had, we say
speed. X0(t), is used an observer to consider the no visible
Once action the brake takes a proportional control (P), to components of state X.
generate the torque, from the error generated between the Once collected the data of the plant, it is possible to
speed of reference and the mechanical speed, where torque obtain, by means of the NHTE an inverse plant control.
TL is: With the previous thing it is possible to generate U(t)
TL K F Zref Ze (27) controls that make journey to the plant from their initial
KF = gain of the control state X(t) to a following state X(t+T) dice, to see Fig.
11b.
B. Intelligent control The limitation of control NHTE is that the wished state
cannot be very far from the present state. The advantage
The diffuse control of IM by means of method NHTE will is that the wished state exactly obtains in T seconds
consider generating the control signals which are (Amp), (more or less noise and errors of measurement). If we
(Ȧsinc) (Fg. 9). The brake diffuse control will be used to want to reach "a distant" state of the present state, we
generate the torque (TL) as it is in Fig. 9. must do it in several small steps of T seconds each one.
For the obtaining of the diffuse control, provided generated At the time of using controller NHTE, in X(t) the real or
exit and input data’s are taken to the system (plants) of the considered state is fed on the plant. In X(t+T) the
same one. The system (it plants) is excited so that it follows wished state is fed, where at least one of the
a reference signal, this must cover all the dynamic range that components comes from the slogan signal as it is in the
could happen in operation, to see Fig. 10. figure 11c. Like result, the neuronal network or the
diffuse control will respond with the U(t) control that
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
we must apply to the plant during the interval [t, t+T) [Ruiz TABLE 3
1999]. FORMULAS OF POWER IN AN INDUCTION MOTOR
Powers electrica of entrante Pot ele Pot elec = Vs*Is
Losses in copper of the stator
Pot estat =3Is2Rstat
Pot estat
The losses in the nucleus
Pot núcleo=3((Is+Ir)2)*Rm
Pnúcleo.
Power of the air gap Pot entreh = Pot elec – Pot
Pot entreh. estat - Pot nucleo
Resistivas losses of the rotor
Pot rot =3Is2Rrot
a) b) Pot rot.
Mechanical power of exit
Pot mec = Te*Ȧe.
Pot mec.
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
signals of speed of reference (Ȧref) and the mechanical mechanical speed in the deceleration of the system
speed (Ȧe) of the system. has (plant).
x The torque (Te) generated by IM is observed in the Fig. x To the stored being a kinetic energy in the inertia
19, it is controlled by the intelligent control. It is exerted steering wheel, and to want to reduce its speed
with greater torque, because the control in each ¨T, the drastically, the control will present/display the
mechanic speed Ȧe § Ȧref. necessary signals of control for the pursuit of the
x Part of the problematic one of the intelligent control in reference signal (figs. 18a, 18b and 18c), this will
the decelerations, must to that in the training data the depend of which so fast it is the deceleration that at its
decelerations was not considered expositive. moment is taken.
x Another point can be is that the sets that were taken for x The collected data of training for the diffuse control
the tuning of the control are not sufficient for the diffuse variable (TL) single represent a small portion of an
control, since three sets were considered for each acquired training (fig. 18c).
antecedent, this is reflected in the oscillations that the
Fig. 13. IM torque (Te), and brake torque(TL). Fig. 14. Stator electrical power (Pot_elec) and
Fig. 12. Reference (Ȧref), mechanical (Ȧe), and
mechanical power (Pot_mec) comparison.
synchronous speed (Ȧsinc).
Fig. 15a. Reference (Ȧref) and IM mechanical Fig. 15b. Zoom during the acceleration transient. Fig. 15c. Zoom during the deceleration transient.
speed (Ȧe) for conventional and diffuse control.
Fig. 16. IM generated torque (Te) for Fig. 20. Electrical power (Pot elec(t)) in the Fig. 21. Mechanical power (Te*Ȧe)
conventional and intelligent control. stator.
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.
[6] R.L. Mott, “Design of Elements of Machines”. Second [13] K.S. Narendra, K. Parthasarathy, “Identification and
Edition. Prentice Hall Hispanoamericana Editorial, S. A. Control of Dynamical Systems Using Neural Networks”.
Macmillan Inc., USA. 1992 IEEE Transactions on neural networks. Vol 1. No. 1.
[7] R. Canales, R. Barrera. “Analysis of Dynamic Systems and March 1990 pp. 4-27.
Automatico Control”. Editorial Limusa. México. 1980. [14] J. Ruiz, R. Castillo. “Self-Tuning Fuzzy Identification
[8] R. Castillo, “Tuning of diffuse controller based on the and Control Using a Hybrid State Transition Kernel”,
method of descendent gradient”, Tesis Cenidet mayo de1999, CIMCA’99. International Conference on Computational
Cuernavaca, Morelos. Intelligence for Modelling. Viena, Austria, 17 a 19
[9] D. W. Novotny, T.A. Lipo. “Vector Control and Dynamic of febrero de 1999. Aparece en Concurrent Engineering
AC Drives”. Oxford Science Publications, Oxford. (1996). Series, Vol. 55, pp. 204-209, IOS Press 1999.
[10] Ch.M. Ong. “Dynamic Simulation of Electric Machinery. [15] J.E. Shigley, Ch.R.Mischke. “Design in Mechanical
Using”. Matlab/Simulink©. Indiana. Prentice Hall PTR Engineering”. Sixth Edition. Mc Graw Hill Editorial.
Editorial. 1998. D.F. México. 2002.
[11] P. Vas. “Vector Control of AC Machines”. Clarendon Press- [16] E. Umei. “Dynamics of Systems and Control”. Editorial
Oxford. 1990 Thomson Learning. México, D. F. 2001.
[12] M.J. Nandayapa, A.A. Hernández, " Proving stand for [17] T. Wildi. “Electrical Machines, Drives and Power
systems of electrical propulsion with regeneration of Systems”, Fourth Edition, Prentice Hall, Inc., 2000.
energy", Tesis Cenidet. Cuernavaca, Morelos. Enero de 2003.
Authorized licensed use limited to: CAMBRIDGE UNIV. Downloaded on October 03,2022 at 13:22:48 UTC from IEEE Xplore. Restrictions apply.