BULLETIN OF THE POLISH ACADEMY OF SCIENCES TECHNICAL SCIENCES, Vol. 58, No. 4, 2010 DOI: 10.

2478/v10175-010-0070-y

VARIA

Combining the principles of variable structure, direct torque control, and space vector modulation for induction motor fed by matrix converter
W. JING-XIN and J. JIAN-GUO
College of Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China Abstract. This paper presents a novel direct torque control method for an induction motor fed by a matrix converter using the variable-structure sliding mode scheme with the reaching law, in which combined with the space-vector pulse width modulation to ensure a high-performance operation, both in steady state and under transient conditions. The novel variable-structure controllers are designed to provide the fast and accurate torque and ux control which replace the traditional hysteresis comparators. Some experimental results are shown to prove the accuracy and low-ripple operation of the proposed algorithm. Key words: direct torque control (DTC), induction motor, variable structure control, matrix converter, reaching law.

1. Introduction
The induction motor (IM) fed by a three-phase matrix converter (shown in Fig. 1) has received considerable interest because of the lack of intermediate dc-link which allows a compact design, the sinusoidal input current and output voltage, the adjustable power factor, and the four-quadrant operation [12]. The direct torque control (DTC) scheme has been widely used in induction motor drives which can provide the fast torque response, the simple control scheme without the coordinate transformation, and robustness against motor parameter variations [39]. The direct torque control scheme for an induction motor fed by a matrix converter was initially presented in [10]. The generation of the voltage vectors required to implement the DTC of an induction motor fed by a matrix converter with the unity input power factor was allowed. However, the conventional DTC algorithm using hysteresis comparators has some drawbacks: it generates large torque, ux and current ripples especially in a low speed range; the input lters are dicult to design because of the variable switching frequency. There exist several methods which are designed to improve the performance of the classic DTC for a conventional inverter drive system, but there are few methods to improve the performance for the matrix converter drive system [320]. The small voltage vectors of matrix converter were used in the direct torque control scheme for matrix converter drives to reduce the electromagnetic torque ripples [21], the torque hysteresis comparator is modied in order to distinguish between large and small torque errors, then a new look-up table for direct torque control using small vectors and large vectors of matrix converter is adopted. However, the direct torque control scheme using small vectors of matrix converter is also based on hysteresis comparators. The linear PI torque and ux control using SVM is investigated by Kyo-Beum Lee [1112].
e-mail:

The hysteresis comparators of torque and ux are replaced by PI controllers, however, the PI controllers are sensitive to the change of motor parameters, speed and load, and PI controllers have a lag phenomenon in the sinusoid response for certain frequencies [17].

Fig. 1. Schematic representation of matrix converter

The variable-structure control (VSC) is an eective method to overcome these problems. It features robustness to parameter variations of load disturbances, fast dynamic response, and simplicity of design and implementation [1520]. In this paper, a novel method of the direct torque control using a variable structure control scheme with the space-vector pulsewidth modulation is derived, which is characterized by a constant switching frequency, a fast dynamic torque response and low torque ripples. The eectiveness of the proposed scheme is demonstrated through experimental results.

[email protected]

657

W. Jing-Xin and J. Jian-Guo

2. Classic DTC for matrix converter drives

The principle of the classic DTC is based on hysteresis torque and stator ux control that directly selects one of the six nonzero and two zero voltage vectors generated by a conventional two level inverter, in order to maintain the estimated stator ux and torque within the hysteresis bands.
Table 1 Permit switch status list of matrix converter Switching conguration +1 -1 +2 -2 +3 -3 +4 -4 +5 -5 +6 -6 +7 -7 +8 -8 +9 -9 01 02 03 Closed switch SAa SBb SCb SAb SBa SCa SAb SBc SCc SAc SBb SCb SAc SBa SCa SAa SBc SCc SAb SBa SCb SAa SBb SCa SAc SBb SCc SAb SBc SCb SAa SBc SCa SAc SBa SCc SAb SBb SCa SAa SBa SCb SAc SBc SCb SAb SBb SCc SAa SBa SCc SAc SBc SCa SAa SBa SCa SAb SBb SCb SAc SBc SCc

controller which generates the torque reference value for the torque controller, then the reference value of the torque and the stator ux magnitude are compared with the estimated values, the outputs of three hysteresis comparators generate the proper voltage vector with voltage selection policy to restrict the torque, stator ux and the average value of the sine of the displacement angle i within their respective hysteresis bands.

3. Modelling of induction motors

The circuit equations of induction motor in the synchronous speed rotating coordinate are given by (1) u = R i + dsd sd s sd 1 sq dt dsq u =R i + sq + 1 sd s sq dt , (1) 0 = Rr ird + drd (1 )rq dt 0 = Rr irq + drq + (1 )rd dt where usd , usq , isd , isq , sd , sq are stator voltages, currents and ux linkages in the synchronous speed rotating frame, ird , irq , rd , rq are rotor currents and ux linkages in the synchronous speed rotating frame, Rs is stator resistance, Rr is rotor resistance, 1 is synchronous angular velocity, is rotor electrical speed. The ux linkages equations of induction motor are given by (2) sd = Ls isd + Lm ird sq = Ls isq + Lm irq , (2) =L i +L i rd m sd r rd sd = Lm isq + Lr irq where Ls is stator self-inductance, Lr is rotor self-inductance, Lm is mutual inductance. Using stator currents as the state variables, the induction motor can be modelled by the following equation from (1) and (2) Lm Lm disd dt = Ls Tr sd + Ls sq Rs Lr + Rr Ls usd isd + (1 )isq + , Ls Lr Ls (3) disq Lm Lm dt = L T sq L sd s r s Rs Lr + Rr Ls usq isq (1 )isd + , Ls Lr Ls where = 1 L2 /Ls Lr is leakage coecient, Tr = Lr /Rr m is the rotor time constant. Equation (3) can be transformed into the stationary reference frame by using (4) f f = cos s sin s sin s cos s fd fq

Fig. 2. Block diagram of the DTC scheme with matrix converter

There are 21 possible switching congurations of matrix converter which can be used in DTC algorithm, these congurations are summarized in Table 1. From Table 1, it appears that the matrix converter can generate a higher number of output voltage vectors with respect to two-level inverter. So, DTC for matrix converter drives can control a further variable in addition to stator ux and torque. In [10], the average value of the sine of the displacement angle i between the input line-to-neutral voltage vector and the corresponding input line current vector has been chosen as a third variable. The schematic diagram of DTC for matrix converter drives is represented in Fig. 2. The outer loop contains a speed PI 658

(4)

where s is the angle between the d-axis and -axis.

Bull. Pol. Ac.: Tech. 58(4) 2010

Combining the principles of variable structure, direct torque control... And the model of the induction motor can by (5)(8) 1 1 dis = s + s dt Ls Tr Ls Rs Lr + Rr Ls is is + Ls Lr 1 1 dis dt = Ls Tr s Ls s Rs Lr + Rr Ls is + is + Ls Lr ds = u R i , s s dt be described 4.1. Sliding surfaces. The sliding surface S is designed so as to enforce the sliding-mode operation, we can dene up to two switching functions, the sliding surface is set as S = [S1 S2 ]T . S1 = T (t) + KT T ( )d T (0) (9) S = (t) + K ( )d (0) 2

u Ls

(5)

u Ls

(6) ds = u Rs is , dt where u , u , is , is , s and s are stator voltages, currents and ux linkages in the stationary reference frame. 3 P (s is s is ) (7) 2 where T is the estimated torque and P is the number of pole pairs = 2 + 2 , (8) s s T = where is the square of stator ux linkage.

where T = T T and = , T and are, respectively, the reference values of torque and square of stator ux, T and are estimated torque and square of stator ux. KT and K are control gains. The rst function corresponds to the control law of the electromagnetic torque while the second function denes the control law of the stator ux. The motion projections of the system on S subspace are derived by dierentiating the vector S dS1 = dT dT + K (T T ) = dT + K T T T dt dt dt dt dS2 = dt d d dt dt + K ( ) = d + K dt (10)

4. Variable structure control scheme

The variable structure control strategy is based on the design of discontinuous control signal that drives the system states towards special manifolds in the state-space [1619]. There are many dierent ways to control the parameters of each structure and to dene the switching logic [1520]. In this paper, the strategy is based on torque and square of stator ux variable structure controllers, the block diagram of the proposed strategy is shown in Fig. 3. The outer PI controller and the square of the stator ux generator produce the reference values of torque and square of stator ux, then the dierence between torque and square of stator ux reference values and estimated values are sent to the variable structure direct torque controller, the results of the controller are the control voltage vectors in the stationary frame.

4.2. Variable structure control law. The second stage of the variable structure control law design is to select the stator voltage vector which has to be applied so that the torque and square of stator ux trajectories stay on their sliding surfaces. Substituting for T , and their derivatives using (5)(8) leads to dS = F + DU, (11) dt where calculation for derivatives of D and F is shown below F = 3 P 1 + Rs Lr + Rr Ls ( i i ) s s s s 1 2 Ls Ls Lr (is s + is s ) + KT T , F2 = 2Rs s is + 2Rs s is + K , 3 1 P s is 2 Ls 2s (13) (12)

3 1 P is s 2 Ls D= 2s

and U = [u u ]T . In this paper, the time-derivative terms of the system states are designed according to reaching law. The reaching law is a dierential equation which species the dynamics of a system states [20]. The form of the reaching law used is dS1 = k1 |S1 | 1 sgnS1 dt , dS2 (14) = k2 |S2 | 2 sgnS2 dt k1 > 0, k2 > 0, 0 < 1 < 1, 0 < 2 < 1,
Fig. 3. DTC of induction motor driven by Matrix converter based on variable structure control
Bull. Pol. Ac.: Tech. 58(4) 2010

where k1 , k2 , 1 and 2 are control gains, sgnS1 and sgnS2 are sign functions. It can be seen from Eq. (14) that the reach659

W. Jing-Xin and J. Jian-Guo

ing law forces the system state to reach the sliding surface at a variable speed, the convergence rate is faster when the system states are far away from the sliding surface (the values of |S1 | and |S2 | are large), while the convergence rate is slow when the system states are close to the sliding surface (the values of |S1 | and |S2 | are small), so the merits of high-rate convergence and low chattering can be achieved by using this reaching law. Substituting (14) in (11) leads to U = D1 F1 + k1 |S1 |1 sgnS1 F2 + k2 |S2 | 2 sgnS2 . (15)

The SVM algorithm is based on the selection of four active vectors which are listed in Table 2 and zero vectors that are applied for the whole cycle period. The duty cycles of the 4 active vectors can be calculated as follows with the absolute values [2] 2 cos(o /3) cos(i /3) 1 = (1)kv +ki m , cos i 3 (20)

2 cos(o /3) cos(i + /3) , (21) 2 = (1)kv +ki +1 m cos i 3 2 cos(o + /3) cos(i /3) 3 = (1)kv +ki +1 m , (22) cos i 3 2 cos(o + /3) cos(i + /3) 4 = (1)kv +ki m , cos i 3

The main shortcoming of the variable structure is the existence of high-frequency chattering, the high frequency components of the chattering are undesirable because they may excite un-modeled high frequency system dynamics and even result in unforeseen instability [17]. Using the smooth function instead of the switching function around the sliding surface can alleviate the problem, that is sgn(Si ) = Si , |Si | + >0 i = 1, 2. (16)

(23)

In order to prove the stability of variable structure control system, the control system should satisfy the Lyapunov stability theory, the Lyapunov function is selected as 1 T S S. (17) 2 The time derivative of V on the state trajectories is given by dV dS = ST . (18) dt dt The time derivative of V on the state trajectories can be written by the following equation with Eq. (14) and (18) V = dV = S1 (k1 |S1 |1 )sgnS1 S2 (k2 |S2 |2 )sgnS2 . (19) dt dV It can be seen from Eq. (19) that < 0, hence, the dt variable structure control system is stable according to the Lyapunov stability theory [17].

where o is the output voltage vector angle, i is the input current vector angle, and i is the input displacement angle, m is the voltage transfer ratio, kv is the sector number of output voltage vector, ki is the sector number of input current vector, the duty cycle for the zero vector is 0 = 1 (1 + 2 + 3 + 4 ).
a)

(24)

b)

5. Double space-vector pulsewidth modulation for matrix converter

A brief introduction of the double space-vector pulsewidth modulation for matrix converter is presented in this section. The double space-vector pulsewidth modulation for matrix converter is fully discussed in [2]. The double space-vector pulsewidth modulation algorithm for a matrix converter has the capability to achieve the full control of both the output voltage vector and the input current vector. Figure 4 shows the synthesis schematic of output voltage vector and input current vector of matrix converter, Fig. 5(a) and (b) show the output voltage vector v o and input current vector ii which have xed directions but variable magnitudes depend upon the instantaneous values of the input line-to-line voltages and output line currents, respectively. 660

Fig. 4. Synthesis schematic of output voltage vector and input current vector: a) synthesis of output voltage vector b) synthesis input current vector
Bull. Pol. Ac.: Tech. 58(4) 2010

Combining the principles of variable structure, direct torque control...

a)

classic DTC and proposed DTC, the sampling period is 90 s for the classic hysteresis band-based DTC, the average switching frequency of classic hysteresis band-based DTC drives is 1.3 kHz, and the sampling frequency is also 1.3 kHz for proposed DTC using space vector modulation.

b)

Fig. 5. Sector diagram of output voltage vector and input current vector: a) output voltage vector sector b) input current vector sector
Table 2 Switch status selection of matrix converter Input current 1 or 4 2 or 5 3 or 6 status 9 8 7 I 1 7 9 8 II or 4 3 2 1 III 0utput voltage 2 or 5 6 4 9 7 5 6 8 9 4 5 7 8 I II III IV 3 1 3 2 II or 6 6 5 4 III

Fig. 6. Experimental system

1 3 2 IV

3 2 1 I

4 6 5 IV

It should be noted that two of the values calculated by (20)(23) must be negative values, the negative value of the duty cycle means that the corresponding negative switching conguration has to be selected instead of the positive one. It can be seen from Table 2 that the switching status selection of matrix converter is 9, 7, 3 and 1 when kv = 1 and ki = 1, the values of 1 and 4 are positive while the values of 2 and 3 are negative calculated by (20)(23), so the nal switching status selection is +9, 7, 3 and +1 which can be seen in Table 1.

Figure 7 shows the estimated results of the steady state performance of the classic DTC at 500 rpm, Fig. 8 shows the estimated results of the steady state performance of the proposed variable-structure controlled DTC at 500 rpm. It can be seen from the results that the torque ripples are signicantly reduced by the proposed algorithm. Figure 9 shows the estimated result of torque response from 15 Nm-20 Nm, it can be seen that the torque response time of the proposed DTC is less than 1ms, which is almost the same as that of the classic DTC. Figure 10 shows the transient performance in a speed reversal process between the values 500 rpm.
a)

6. Experiments
Experiments are carried out to conrm the validity of the proposed scheme. The experimental setup of the proposed control system, shown in Fig. 6, consists of three-phase, 380 V, 50 Hz, 5.5 kW induction motor, control system using DSP (TMS320F2812) and FPGA (XC3S50AN-TQ144) with a 12bit A/D converter board, in which the DSP and A/D converter board completed the variable structure control algorithm, and the FPGA completed the four-step communication and pulses generation for matrix converter. The induction motor has the following parameter values: Rs = 1.517 , Rr = 1.483 , Ls = 0.174 H, Lr = 0.174 H, Lm = 0.1928 H. The parameter values of variable structure control system are: k1 = 9000, k2 = 9000, 1 = 0.5, 2 = 0.5, = 0.3, KT = 3.0, K = 4.0. In order to compare the performance between the
Bull. Pol. Ac.: Tech. 58(4) 2010

b)

Fig. 7. Estimated steady torque and ux at 500 rpm of classic DTC, Ts = 90 s, a) estimated torque, b) estimated ux

661

W. Jing-Xin and J. Jian-Guo a)

motor speed is compensated after 15 ms. Figure 12 shows an input phase source voltage and input current, the line current is in phase with the input phase voltage which allows unity input power factor operation. Figure 13 shows the output line voltage waveform at 900 rpm. The sampling time for Figs. 1013 is 150 s.
a)

b)

b)

Fig. 8. Estimated steady torque and ux at 500 rpm of proposed DTC, Ts = 770 s, a) estimated torque, b) estimated ux

a) Fig. 10. Speed and current response from +500 rpm-500 rpm, Ts = 150 s, a) speed response, b) motor phase current response a)

b)

b) Fig. 9. Torque response from 15 Nm-20 Nm, a) torque response of classic DTC, Ts = 90 s, b) torque response of proposed DTC, Ts = 770 s

The response to a speed command and phase current show the good dynamic behaviour. Figure 11 shows the experimental results of the response to a step load from 0 Nm to 15 Nm, once the step change on load torque has been applied, the motor speed is eected by the step load, then the outer speed controller raises the torque demand in order to keep track of speed reference value, it can be seen from Fig. 11 that the

Fig. 11. Experimental results of the response to the step load, Ts = 150 s, a) torque response to the step load, b) speed response to the step load

662

Bull. Pol. Ac.: Tech. 58(4) 2010

Combining the principles of variable structure, direct torque control...

[6] R. Ortega, N. Barabanov, and G.E. Valderrama, Direct torque control of induction motors: stability analysis and performance improvement, IEEE Trans. on Automatic Control. 46 (8), 12091222 (2001). [7] N.R.N. Idris, Ch.L. Toh, and M.E. Elbuluk, A new torque and ux controller for direct torque control of induction machines, IEEE Trans. on Industry Applications 42 (6), 1358 1366 (2006). [8] L. Romeral, A. Arias, E. Aldabas, and M.G. Jayne, Novel direct torque control (DTC) scheme with fuzzy adaptive torqueripple reduction, IEEE Trans. on Industry Electronics 50 (3), 487492 (2003). [9] Y.-S. Lai, W.-K. Wang, and Y.-C. Chen, Novel switching techniques for reducing the speed ripple of AC drives with direct torque control, IEEE Trans. on Industry Electronics 51 (4), 768775 (2004). [10] D. Casadei and S. Giovanni, The use of matrix converter in direct torque control of induction machines, IEEE Trans. on Industry Electronics 48 (6), 10571064 (2001). [11] K.-B. Lee and F. Blaabjerg, An improved DTC-SVM method for sensorless matrix converter drives using an overmodulation strategy and a simple nonlinearity compensation, IEEE Trans. on Industry Electronics 54 (6), 31553166 (2007). [12] K.-B. Lee and F. Blaabjerg, Sensorless DTC-SVM for induction motor driven by a matrix converter using a parameter estimation strategy, IEEE Trans. on Industry Electronics 55 (2), 512521 (2008). [13] K.-B. Lee and F. Blaabjerg, Improved sensorless vector control for induction motor drives fed by a matrix converter using nonlinear modeling and disturbance observer, IEEE Trans. on Energy Convers. 21 (1), 5259 (2006). [14] G.S. Buja and M.P. Kazmierkowski, Direct torque control of PWM inverter-fed ac motors-a survey, IEEE Trans. on Industry Electronics 51 (4), 744757 (2004). [15] C. Lascu, I. Boldea, and F. Blaabjerg, Very-low-speed variable-structure control of sensorless induction machine drives without signal injection, IEEE Trans. on Ind. Appl. 41 (2), 591598 (2005). [16] C. Lascu, I. Boldea, and F. Blaabjerg, Direct torque control of sensorless induction motor drives: a sliding-mode approach, IEEE Trans. on Ind. Appl. 40 (2), 582590 (2004). [17] Z. Xu and M. F. Rahman, Direct torque and ux regulation of an IPM synchronous motor drive using variable structure control approach, IEEE Trans. on Power Electron. 22 (6), 24872498 (2007). [18] V.I. Utkin, Sliding mode control design principles and applications to electric drives, IEEE Trans. on Ind. Electron. 40 (1), 2336 (1993). [19] C. Lascu and A.M. Traynadlowski, Combining the principles of sliding mode, direct torque control, and space vector modulation in a high-performance sensorless AC drive, IEEE Trans. on Ind. Appl. 40 (1), 170177 (2004). [20] A. Naassani, E. Monmasson, and J.P. Louis, Synthesis of direct torque and rotor ux control algorithms by means of sliding-mode theory, IEEE Trans. on Ind. Electron. 52 (3), 785799 (2005). [21] C. Ortega, A. Arias, and J. Ballcells, The use of small voltage vectors of matrix converters in direct torque control of induction machines, EPE-PEMC 1, 314319 (2006).

Fig. 12. Input phase voltage and current waveforms, Ts = 150 s

Fig. 13. Output line voltage waveform at 900 rpm, Ts = 150 s

7. Conclusions
In this paper, the variable-structure direct torque control for induction motor fed by a matrix converter has been analyzed. An experimental system composed of an induction motor, a matrix converter and DSP with FPGA control board has been used to validate the proposed algorithm. The performance of the proposed algorithm is improved so that the torque ripples are reduced signicantly in comparison with the classic DTC, and also the transient torque response time is almost the same as that of the classic DTC. REFERENCES
[1] L. Huber and D. Borojevic, Space vector modulated threephase to three-phase matrix converter with input power factor correction, IEEE Trans on Industry Applications 31 (6), 12341246 (1995). [2] D. Casadei and S. Giovanni, Matrix converter modulation strategies: a new general approach based on space-vector representation of the switch state, IEEE Trans. on Industry Electronics 49 (2), 370381 (2002). [3] L. Joong-Hui, K. Chang-Gyun, and Y. Myung-Joong, A deadbeat type digital controller for the direct torque control of an induction motor, IEEE Trans. Power Electron. 17 (5), 739 746 (2002). [4] L. Yen-Shin and Ch. Jian-Ho, A new approach to direct torque control of induction motor drives for constant inverter switching frequency and torque ripple reduction, IEEE Trans. on Energy Convers. 16 (3), 220227 (2001). [5] N.R.N. Idris and A.H.M. Yatim, Direct torque control of induction machines with constant switching frequency and reduced torque ripple, IEEE Trans. on Industry Electronics 51 (4), 758767 (2004).

Bull. Pol. Ac.: Tech. 58(4) 2010

663