Badre Bossoufi 1,2, Mohammed Karim1, Silviu Ionita2, Ahmed Lagrioui1,

J. Electrical Systems 8-2 (2012): 236-248 Regular paper Nonlinear Non Adaptive Backstepping with Sliding-Mode Torque Control Approach for PMSM Motor

In this paper, the nonlinear non adaptive backstepping combined with the slidingmode torque and flux control approach for an permanent magnet synchronous motor drive is proposed. we present a new contribution of FPGAs for control of electrical machines. First, the adaptive backstepping control approach is utilized to obtain the robustness for mismatched parameter uncertainties. The overall stability of the system is shown using Lyapunov technique. The simulation results clearly show that the proposed not adaptive scheme can track the speed reference. Secondly, the Indirect Sliding Mode Observer (ISMO) for permanent magnet synchronous motor (PMSM) speed and rotor position is used to realize sensorless control, we detailed description of the structure by Indirect Sliding Mode of Permanent Magnet Synchronous Machine PMSM. Finally, some experimental results are demonstrated to validate the proposed controllers. The experimental results carried from a prototyping platform are given to illustrate the efficiency and the benefits of the proposed approach and the various stages of implementation of this structure in FPGA.
Keywords: Not Adaptive Backstepping control; Backstepping Design Technique; FPGAs; Permanent Magnet Synchronous Machine (PMSM); Lyaponov Stability; Robust Adaptive Control; Systems Generator; Reusability.

1. Introduction Three-phase Permanent Magnet Synchronous Motor (PMSMs) are strongly used in industry and consume more than 70% of industrial electricity. This is why considerable efforts and different searches are being done to improve their performances and their efficiency. The efficiency of electrical machine drives is greatly reduced at light loads, where the flux magnitude reference is held on its initial value. The loss minimization is realised using high-quality materials and excellent design procedures in the manufacturing process. Moreover, expert control algorithms are employed in order to improve machine performance. In this paper we are interested in two mode control for PMSM drive, the not adaptative backstepping and indirect sliding mode control (ISMO). The not adaptive backstepping approach offers a choice of design tools for accommodation of uncertainties nonlinearities. And can avoid wasteful cancellations. However, the not adaptive backstepping approach is capable of keeping almost all the robustness properties of the mismatched uncertainties. The not adaptive backstepping is a rigorous and procedure design methodology for nonlinear feedback control. The principal idea of this approach is to recursively design controllers for machine torque constant uncertainty subsystems in the structure and step back the feedback signals towards the control input. this approach is different from the approach of the conventional feedback linearization in that it can avoid cancellation of useful nonlinearities in pursuing the
* Corresponding author: B.Bossoufi, STIC Team, Faculty of Sciences Dhar El Mahraz, University of Sidi Mohammed Ben Abdellah Fez, Morocc, E-mail: badre_isai@hotmail.com 1 STIC Team, Faculty of Sciences Dhar El Mahraz, University of Sidi Mohammed Ben Abdellah Fez, Morocc 2 Center of Modeling and simulation of the systems, Faculty of Electronics, Communications, University of PITETI, Romania

Copyright JES 2010 on-line : journal/esrgroups.org/jes

J. Electrical Systems 8-2 (2012): 236-248

objectives of stabilization and tracking. A nonlinear backstepping control design scheme is developed for the speed tracking control of PMSM that has exact model knowledge. The asymptotic stability of the resulting closed loop system is guaranteed according to Lyapunov stability theorem. The speed variation of the PMSM is widely used in high-performance applications. The PMSM has very large power density, high power factor and high efficiency. In a highperformance control of PMSM, the information of rotor position and speed is very very important. In the speed control loop. For the field oriented control, the coordinate transformation have needs precise rotor position. Rotor position and speed can be measured by a shaft encoder or other type of sensors, in other case the speed is measured with an Encoder resolver connected to the PMSM machine drive. However, the presence of such sensors is not acceptable for cost, maintenance and reliability reasons. The concept of sensorless control was proposed in the 1970s and has been continually developed for PMSM rotor position and speed estimation. The basic principle of sensorless control is to deduce the rotor speed and position using various information and means, including direct calculation, parameter identification, condition estimation, indirect measuring and so on. The stator currents and voltages are generally used to calculate the information of speed and rotor position. The Indirect Sliding mode observer (ISMO) applied to PMSMs drive sensorless control has been studied since 1990s. Sliding mode is a non-linear concept, using some switching laws to control the system running in two structures, in order to converge to a sliding mode plane. For a sliding mode observer in PMSM, the aim is to estimate the unknown parameters by some measured components. Because of ISMOs low-sensitivity to precise model and parameters, it is robust and has good low-speed performance. The FPGA technology is now used by an increasing number of designers in various fields of application such as signal processing, telecommunication, video, embedded control systems, and electrical control systems. This last domain, i.e. the studies of control of electrical machines, will be presented in this paper [1]. Indeed, these components have already been used with success in many different applications such as Pulse Width Modulation (PWM), control of induction machine drives and multimachine system control. This is because the FPGA-based implementation of controllers can efficiently answer current and future challenges of this field. This paper presents the realization of a platform for not adaptative Backstepping control combined with the Indirect Sliding Mode Control of PMSM using FPGA based controller. This realization is especially aimed for future high performance applications. In this approach, not only the architecture corresponding to the control algorithm is studied, but also architecture and the ADC interface, Encoder interface and RS232 UART architecture [2]. Considering the complexity of the diversity of the electric control devices of the machines, it is difficult to define with universal manner a general structure for such systems. 2. PMSM model system In this paper, we apply the different algorithms control on a machine type PMSM (Permanent Magnet Synchronous Motor) [3], which consists of three stator windings and a rotor magnet. This motor is described by the following equation (Voltage, Flux, Torque),

B. Bossoufi et al: Nonlinear Non Adaptative Backstepping for PMSM Motor

d sd Vsd = rs .isd + dt . sq V = r .i + d sq . sd sq s sq dt = L .i + sd sd f sd = L .i sq sq sq d Ce = J . dt + f . Cr = p.

(1)

Where is the rotation's speed, p the Number of pairs of poles, J the moment of inertia, f the Coefficient of viscous friction, Cr the resistive torque, f the flux produced by the permanent magnet, Lsd and Lsq the d-q axis stator inductance, Vsd and Vsq the d-q axis stator voltage, rs the stator winding resistance and Ce the electromagnetic torque. 3. Nonlinear not adaptative backstepping approach The schematic diagram of the speed control system study with application of the nonlinear backstepping controller is shown in Fig.1. The parameters of the synchronous machine are given in the Appendix. In this section, we employ the nonlinear backstepping schemes to design the controllers for PMSM systems with angular velocity measurement.

Fig.1: System configuration of not-adaptive Backstepping Control

The Not Adaptive backstepping approach algorithm is control techniques that can linearize a nonlinear system such as the PMSM machine drive in the presence of uncertainties. Unlike other feedback linearization techniques, adaptive backstepping has the flexibility of keeping useful non linearitys intact during stabilization. The essence of backstepping is the stabilization of a virtual control state. Hence, it generates a corresponding error variable which can be stabilized by carefully selecting proper control inputs. These inputs can be determined from Lyapunov stability analysis [4]. It is obvious that the dynamic model of PMSM is highly nonlinear because of the coupling between the speed and the stator currents (equation (1)). According to the vector control principle, the direct axis current id is always forced to be zero in order to orient all the linkage flux in the d axis and achieve maximum torque per ampere.

468

J. Electrical Systems 8-2 (2012): 236-248

Lsq V disd r pisq + sd = s .isd + Lsd Lsd dt Lsd disq dt = Vsq f rs L p + .isq sd pisd Lsq Lsq Lsq Lsq

(2)

C d 3 p f = ( f isq + ( Lsd Lsq )isd isq ) + r dt 2 J J J The vector [x ] = [isd isq ]T choice as state vector is justified by the fact that currents

and speed are measurable and that the control of the instantaneous torque can be done comfortable via the currents isd and/or isq. And stator voltages as control variables u = Vsd Vsq T .

The principal objective of the backstepping controller is to regulate the speed of the PMSM drive to its reference value ref whatever external disturbances. We assume that the engine parameters are known and invariant. 3.1. Backstepping Speed Controller The first step is defined the tracking errors:

e = ref
The derivative of (3) is: de & & = & 1 3 p ( i + (L L )i i ) f C & = = e ref ref f sq sd sq sd sq r dt J 2 We define the following quadratic function:

(3) (4)

1 2 V1 = e 2

(5)

Its derivative along the solution of (5), is given by: & 1 3 p & =e e (6) & V 1 = e ref J 2 ( f isq + ( Lsd Lsq )isd isq ) f Cr Using the backstepping design method, we consider the d-q axes currents components isd and isq as our virtual control elements and specify its desired behavior, which are called stabilizing function in the backstepping design terminology as follows: i sdref = 0 (7) 2 = + + i ( f C J . k . e ) r sqref 3 p f With k is a positive constant Substituting (7) in (6) the derivative of V1:

& = k e2 0 V 1

(8)

3.2. Backstepping Current Controller We have the asymptotic stability of the origin of the system (1). We defined current following errors:

ed = isdref isd eq = isqref isq

with isdref = 0

(9)

Their dynamics can be written:

B. Bossoufi et al: Nonlinear Non Adaptative Backstepping for PMSM Motor

&sdref i &sd = &d = i e

Lsq V rs &sqref i &sq = 2 ( f + Cr + J .k .e ) + rs .isq + &q = i pisq sd e .isd Lsd Lsd Lsq Lsd 3 p f (10) p f L V + sd pisd + sd Lsq Lsq Lsd

To analyze the stability of this system we propose the following Lyapunov function:

1 2 2 2 V2 = (e + ed + eq ) 2
Its derivative along the trajectories (8), (9) and (10) is:

(11)

Vsd rs Lsq 2 2 2 & = 1 (e e & & & V + isq + + ed ed + eq eq ) = ke kd ed kqeq + ed [kd ed 2 Lsd Lsd Lsd 2 + + 3p 2(k J f ) 3 p f 3p eq + (Lsd Lsq )ed isq ke ) (Lsd Lsq )eisq ] + eq[kqeq + .( 2J 3 p f 2J 2J 3 p f 2J e Vsq Lsq + rs L isq + sd isd + f ] Lsq Lsq Lsq
3pL sd (Lsd Lsq)eisq 2J
(12)

The expression (12) found above requires the following control laws:

Vsd = kd Lsded +rsisd Lsq isq +

2Lsq(kJ f ) 3pf 3pf Lsq 3p ( ) Vsq = e L L e i k e isd + + q sd sq d sq f +kqL sqeq + 2J e +rsisq + Lsd 2J 3pf 2J
With this choice the derivatives of (13) become:

(13)

& = k e k e k e 0 V 2 d d q q
4. Nonlinear Indirect Sliding-Mode Control

(14)

4.1. Principle of Variable Structure Systems The principle of variable structure systems has been studied primarily in the Soviet Union. Subsequently, much research on power systems control field has been made elsewhere to complement the theoretical study and find some possible applications. The sliding mode control is a particular operating mode of variable structure system. Using this command has been limited longer due to oscillations to the phenomena of slip and limitations of the switching frequency of power switches [5]. There are different regulatory structures by sliding mode current in systems based on variable structure sliding mode control. Consider the following controlled system (15):

dy = h( x) + G ( x)u dt

(15)

Where u is the input vector of dimension m, y is the state function of dimension n, h is the state function describing the system evolution over time and G is a matrix of dimension n*m. For the synthesis of a regulatory structure by sliding mode, it is necessary to define initially the switching function S(y) of dimension m, where Si(y) is the ith switching function S(y).

470

J. Electrical Systems 8-2 (2012): 236-248

4.2. Synthesis of a-indirect control by sliding mode The control law sliding mode must simultaneously satisfy the conditions of invariance and attractiveness. To do this the switching function S(y) must satisfies:

S ( y ) = 0 The invariance condition: . S ( y ) = 0

The attractiveness condition:

(16)

. S i ( y ) p 0 si S i ( y ) f 0 for (i = 0...m) . S ( y ) f 0 si S ( y ) p 0 i i

(17)

u* = ueq + uatt

These conditions lead to determining a new vector control: (18) The vector control given by equation (18) comprises two terms: - The first is the control vector specifying the equivalent control for the system to stay on the sliding surface. - The second is the vector control that ensures the attractive control system outside the sliding surface. It also requires the system dynamics starting from an initial point until it reaches the sliding surface (Fig.2) [6].

Fig.2: The state of Sliding Mode trajectory

The indirect control by sliding mode stator current vector of a PMSM machine ensures the calculation of direct and inverse components of the reference voltage vector expressed in the d-q plane. These components are then applied across the phases of the stator PMSM through an intermediate stage of the SVM. The development of such law control must satisfy simultaneously the conditions of invariance and attractiveness given by equations (16) and (17). The terms dq sq and dq sd are considered as terms of electromotive forces induced on the axis d and q and the expressions of the derivatives of the switching function S isd and S isq are given by equations (19) and (20). Thus, the indirect control by sliding mode d and q components of stator current can be carried out by taking on each axis d and q. Therefore, each component Vsd* and Vsq* is composed by two terms as shown in equation (18). The first term is the equivalent voltage vector which is active in steady state, while the second term is the voltage vector which is attractive assets in transition.
* Vsd = V sdeq + V sdatt * Vsq = Vsqeq + V sqatt

(19)

B. Bossoufi et al: Nonlinear Non Adaptative Backstepping for PMSM Motor

For trajectories currents isd and isq remained on their sliding surfaces (Sisd=0 and Sisq=0), apply the voltage vectors Vsdeq and Vsqeq on the axis d and axis q. These vectors can be calculated taking into account the following invariance conditions:
* S isd = ( i sd i sd ) = 0 . S isd = 0 * i sd = i sd S isd 1 dt = L (V sdeq rs i sd + dq sq ) = 0 d

(20)

* V sdeq = r s i sd dq sq = rs i sd dq sq
* i sq = i sq * S isq = (i sq ) 0 i = sq S isq . 1 (21) (V sqeq rs i sq dq sd ) = 0 = 0 S = isq dt L q * V sqeq = r s i sq + dq sd = rs i sq + dq sd Considering the derived of switching functions and the control formulas, the new components Vsd* and Vsq* will be: S * * Vsd = rs isd dq sq Ld isd = Vsdeq + Vsdatt dt (22) S V * = r i* + L isq = V + V sq s sq dq sd q sqeq sqatt dt From this attractive voltage vector system of reference voltage vector involves the

switching function derivative S isd and S isq . A structure of attractiveness is chosen at a constant speed and proportional action, which gives:

Vsdatt = Ld ( Ad sgn( S isd ) + K d S isd ) Vsqatt = Lq ( Aq sgn( S isq ) + K q S isq )

.
.

(23)

By applying the reference voltage vector given by the previous system, the result of the product of each of switching functions following system:

S isd and S isq its own derivative is given by the

(24)

. rs 2 2 S isd S isd = L S isd Ad S isd sgn( S isd ) K d S isd d . 2 2 S isq S isq = rs S isq Aq S isq sgn( S isq ) K q S isq L q

In resume:
* * Vsd isd dq sq Ld 0 Ad 0 sgn( Sisd ) K d 0 S isd * = rs * + + S ) (25) sgn( S ) + 0 ( 0 0 L A K V i q q isq q dq sd sq sq isq

For a given value of the reference current, the current i trajectory described is shown in Fig.3. The trajectory obtained is characterized by two stages: the attractiveness mode and sliding mode.

472

J. Electrical Systems 8-2 (2012): 236-248

Fig. 3: Indirect Sliding Mode Control applied to a PMSM

The dynamics of the system controlled by such control depends essentially on the coefficients choice of the attraction mode matrices K and A (Fig.3). The theory of indirect sliding mode is appropriate to systems whose control is discontinuous.
5. FPGA-Based Implementation of an robust control syetem

5.1. Development of the Implementation There are several manufacturers of FPGA components such: Actel, Xilinx and Alteraetc. These manufacturers use different technologies for the implementation of FPGAs. These technologies are attractive because they provide reconfigurable structure that is the most interesting because they allow great flexibility in design. Nowadays, FPGAs offer the possibility to use dedicated blocks such as RAMs, multipliers wired interfaces PCI and CPU cores. The architecture designing was done using with CAD tools. The description is made graphically or via a hardware description language high level, also called HDL (Hardware Description Language). Is commonly used language VHDL and Verilog. These two languages are standardized and provide the description with different levels, and especially the advantage of being portable and compatible with all FPGA technologies previously introduced [7]. In this paper an FPGA XC3S500E Spartan3E from Xilinx is used. This FPGA contains 400,000 logic gates and includes an internal oscillator which issuer a 50MHz frequency clock. The map is composed from a matrix of 5376 slices linked together by programmable connections. The simulation procedure begins by verifying the functionality of the control algorithm by trailding a functional model using Simulinks System Generator for Xilinx blocks. For this application, the functional model consists in a Simulink timeis discretired model of the No adaptative Backstepping algorithm associated with a voltage inverter and PMSM model. 5.2. Prototyping platform To test the FPGA based controller, a prototyping platform for the control of a Permanent magnet Synchronous Machine was assembled (Fig.4).

B. Bossoufi et al: Nonlinear Non Adaptative Backstepping for PMSM Motor FPGA Inverter (IGBT) PMSM

LOAD

Fig.4: Prototyping platform control

6. Experimental results

6.1. Software Simulation In this section, we show the great improvement of performance PMSM machine, at first the results of not adaptive backstepping control is analyzed. After, the performance of the indirect sliding mode control is discussed and compared to that before.The characteristics for the PMSM motor simulated in this experiment are follows: rs=0.412 , p=4, LSd=3,24 mH, LSq=3,28 mH, J=0,0001473 Kg,m2, For a trajectory ref=150 rad/s at 0s, isdref=4.5A and Cr=7N.m.the following (Fig.5) shown the performance of the input output linearization control.

(a)

(b)

(c)

(d)

Fig.5: (a) Speed response trajectory, (b) Electromagnetic Torque Ce reponse, (c) d-q axis current without uncertainties, (d) d-q axis Flux without uncertainties

For test the further trajectory, the speed reference was made variable. The reference of the direct component of current was set to zero. For a trajectory ref (=150rad/s at 0s, =100rad/s at t=0.4s, =250rad/s at t=0.8s), the following (Fig.6) shown the performance of the PMSM.

474

J. Electrical Systems 8-2 (2012): 236-248

(a)

(b)

(c)

(d)

(e)

(f)

Fig.6: (a) Speed response trajectory, (b) Error Speed response, (c) abc Axis Stator Current, (d) dq Axis Stator Current, (e) dq Axis Stator Flux, (f) Electromagnetic Torque Ce

In the Fig.7 are shown the simulation result of indirect control by sliding mode stator when the current is applied between +Isn and -Isn (E=380V, frequency PWM=5 KHz, Kd= Kq = Ad= Aq= 320). It should be noted that the dynamics of the transient is lower than that obtained with other commands. However, the current has better quality control in steady state with fewer oscillations. These coefficients allow controlled quantities to follow their references to an error of about 99% on the PMSM parameters (Ld, Lq and rs). is1 is2 is3

(a)
Sliding Mode Attractiveness Mode Sliding Mode

B. Bossoufi et al: Nonlinear Non Adaptative Backstepping for PMSM Motor

(b)

(c)

(d)

(e)

The implementation of the indirect control by sliding mode on FPGA devices is characterized by a reduced operation time. The Fig.8 shown the experimental results of Indirect Sliding Mode PMSM with the FPGA platform are shown. Update frequency for this implementation is 20kHz. All results were extracted from the FPGA by the ChipScope tool of Xilinx.

Fig.7: (a) Stator current, (b) Instead of the current vector in the plane (, ), (c) Quadratic Current isq, (d) Speed Rotor, (e) Torque electromagnetic

(a) (b) (c) Fig.11: (a) Stator current locus for ISMC, (b) abc-axis current in the PMSM, (c) d-axis and q-axis current in the PMSM

The Experimental results show the performance of PMSM machine, using two approaches control nonlinear (Backstepping and sliding mode). This two control algorithms show the robustness and effecacit of the system.
7. Conclusion

In this paper a robust continuous approachs Nonlinear not Adaptative Backstepping Control and Indirect Sliding Mode Control strategy for permanent-magnet synchronous motor (PMSM) drive systems is presented. The FPGA based implementation is detailed, a bench test was realized by a prototyping platform, the experimental results obtained show the effectiveness and the benefit of our contribution and the different steps of implementation for the control FPGA.
Acknowledgment

We thank all those who contributed to make this work, including my teachers in my home laboratory STIC Team and the Center of Modeling and simulation of the systems, my wife and all my friends for their support. This work was performed in the Montefiore Institute, University of Liege (ULG) in Belgium, mobility funded by AVERROES (Erasmus Mundus Action 2).

476

J. Electrical Systems 8-2 (2012): 236-248

References
[1] [2] [3] [4] [5] [6] [7] [8] Melicio, R; Mendes, VMF; Catalao, JPS Modeling, Control and Simulation of Full- Power Converter Wind Turbines Equipped with Permanent Magnet SynchronousGenerator vol 5, pg 397-408 Mar - Apr 2010. J. J. Chen and K. P. Chin, "Automatic flux-weakening control of permanent magnet synchronous motors using a reduced-order controller," IEEE Trans. Power Electron., vol. 15, pp. 881-890, Sept. 2000. M. Rodic, K. Jezernik, Speed Sensoless Sliding Mode Torque Control of Induction Motor , IEEE Trans on. Industrial Electronics, February 2002. Hisn-Jang Shieh and Kuo-Kai Shyu, Non Linear Sliding Mode Torque Control With Adaptive Backstepping Approach For Induction Motor Drive, IEEE Transactions on Industrial Electronics, Vol. 46, N 2, April 1999, pp. 380-388. D. Zhang and H. Li, A stochastic-based FPGA controller for an induction motor drive with integrated neural network algorithms, IEEE Trans. Ind. Electron., vol. 55, no. 2, pp. 551561, Feb. 2008. C. F. Jung and J. S. Chen, Water bath temperature control by a recurrent fuzzy controller and its FPGA implementation, IEEE Trans. Ind. Electron., vol. 53, no. 3, pp. 941949, Jun. 2006. E. MONMASSON, and M. Cirstea FPGA Design Methodology for Industrial Control Systems A Review, IEEE Trans Ind. Electron.., vol.54, no. 4, pp.1824-1842, August. 2007. B.BOSSOUFI, M.KARIM, S.IONITA, A.LAGRIOUI, Low-Speed Sensorless Control of PMSM Motor drive Using a NONLINEAR Approach BACKSTEPPING Control: FPGA-Based Implementation Journal of Theoretical and Applied Information Technology JATIT, pp154-166, Vol. 36 No.1, 29th February 2012.