Shunt and Series Conditioning Schemes for Hybrid Matrix Converter

Size: px

Start display at page:

Download "Shunt and Series Conditioning Schemes for Hybrid Matrix Converter"

Lindsey Francis
5 years ago
Views:

Shunt and Series Conditioning Schemes for Hybrid Matrix Conerter Ameer lanabi and Bingsen Wang Department of Electrical and Computer Engineering Michigan State Uniersity 2120 Engineering Building

1 Shunt and Series Conditioning Schemes for Hybrid Matrix Conerter Ameer lanabi and Bingsen Wang Department of Electrical and Computer Engineering Michigan State Uniersity 2120 Engineering Building East Lansing, MI 48824, USA Ahstract-This paper is focused on conditioning the oltage and current waeform quality of a hybrid matrix conerter that consists of a conentional nine-switch matrix conerter and an auxiliary back-to-back oltage source conerter. Upon critical ealuation of the existing methods for shunt and series compensation, the fundamental limitations for achieing superior results hae been identified. A new strategy based on power aeraging for obtaining the reference compensating current and oltage has been proposed. The effectieness of the proposed method has been erified by the simulation results for both the shunt and series compensation with concurrent presence of the harmonic components in oltages and currents. I. INTRODUCTION ybrid matrix conerters can potentially enable matrix H conerters in high power applications that conentional matrix conerters will not be able to attain [1]. It uses a conentional nine-switch matrix conerter in conjunction with an auxiliary back-to-back ac-dc-ac conerter that conditions the current and oltage waeforms on the input and output side of the matrix conerter as shown in Fig. 1. The matrix conerter processes the main power at low switching frequency to enable significant reduction of switching losses and to allow for adoption of high-power semiconductor deices such as integrated gate commutated thyristors (IGCTs) [2][3]. The auxiliary ac-dc-ac conerter is dedicated to improing the power quality at the input and output terminals of the matrix conerter by minimizing harmonic currents drawn from the source and harmonic oltages applied to the load. Essentially, the auxiliary back-to-back conerter functions as a shunt-andseries actie filter (AF). Seeral AF control techniques hae been presented in the literature [4,5,6]. Based on the operating principle, these techniques can be categorized into two groups. The first group of methods are based on instantaneous reactie power theory (IRPT) [7] and extract the reactie component of the power and the oscillatory component of the real power. The other methods are based on filtering techniques and extract the fundamental component of the current or oltage such as notch filter and fast Fourier transform (FFT) methods [8,9,10,11]. In this paper different AF control strategies are presented and critically ealuated. The limitations of the existing control approaches hae been clearly identified. ",,(I) i (t) ':,1 Main Matrix COllerter r-@" R L 1',(1) ;,(1) F L ;H VB Vcb* Vert) ;,(1) -co ic c -. r Veil * icc'" ic"* lea. * t " 1 Vdc AlIxiliw}' Back-lo-Back COllerter I ls4 115 \1' I 151, 151,\ IS3' \14' IS 5'\',. Fig. 1: Hybrid matrix conerter circuit. For ariable frequency applications, the adaptie approach is preferred [8]. Howeer, notch filter is only suitable for harmonic detection and can not extract the reactie components [9]. On the other hand, such adaptie approaches might hae the problem with conergence and robustness [10]. Other filtering techniques such Fourier-based methods are widely used. FFT requires high computational effort and discreet Fourier transform (DFT) requires synchronization method such as phase-locked-loop (PLL) [12]. Furthermore, they share the same limitation as adaptie notch filter of being unable to extract the reactie component. Therefore, only the analysis of adaptie notch filter is presented in the this paper. IRPT [4] is a time domain method that is based on the law of energy conseration. It utilizes the concept that the non-actie component of the oltage and current does not contribute to any energy transfer from the source to the load. IRPT method needs coordinate transformation to the synchronous reference frame. This coordinate transformation changes the oscillating AC ariables to a DC ariables and the harmonics appear in form of ripple on top of the DC signal. Although this approach proides extra filtering to the system, it requires a synchronization method such as PLL. The instantaneous reactie power theory seems to be a good solution in obtaining the reactie components, but it fails when both the current and the oltage contain harmonics. This T F /16/$ l6 IEEE 3703

2 because of the oerlap of the frequencies of the current and the oltage harmonics. Therefore, it is incapable of obtaining the correct compensating current or oltage when harmonics are present in both the current and the oltage. To address these limitations of the aforementioned methods, a new control strategy is proposed. This proposed control method is able to effectiely obtain the correct actie component of current or oltage in cases where both the current and the oltage are non-sinusoidal and proide full control oer the power factor. The rest of the paper is organized as follows. Section II and III present the principle of the methods of adaptie notch filter and IRPT, respectiely. The proposed method is explained in section IV and is erified by the simulation results in section V. A comparatie ealuation of the three methods in Section V is followed by the conclusion in Section VI. II. ADAPTIVE NOTCH FILTER METHOD Adaptie notch filtering (ANF) is the technique that selectiely extracts of a harmonic component of certain frequency. It best suits the applications where the elimination of certain harmonics is required. Notch filter can estimate the salient parameters embedded in the signal such as the amplitude, and the signal frequency [8]. The dynamic behaior of the ANF can be characterized = by the following deferential equations. XU +i2w2x 2(w(u(t) - x ' ) (1) w ' = -,xw(u(t) - x ' ) where x is the integral of the fundamental component of the input signal u(t), w is the estimated frequency of the fundamental component; ( and, are adjustable real positie parameters that determine the accuracy and conergence speed of the ANF; u(t) is the signal from which the fundamental component is to be extracted. For a sinusoidal input signal, the system described by (1) and (2) has a unique periodical orbit located at ( ) ( (2) ':,;t,,:,;) ), (3) where the estimated component of the frequency w is identical to its actual alue WI, which is mathematically explained in [9]. A detailed implementation of the ANF is shown in Fig. 2. The same control circuit is used to obtain the fundamental component of the output oltage. III. INSTANTANEOUS REACTIVE POWER METHOD The instantaneous reactie power theory proides significant insight for understanding the power transferred from the source to the load and among the three phases. By elimination of the power component that does not contribute to the energy transmission from the source to the load, sinusoidal input current and sinusoidal output oltage can be achieed. The compensation system in instantaneous reactie power theory consist of only passie components and switches. Therefore, n(t) + OJ--f---- I---,---- OJ x Fig. 2: Detailed implementation of the adaptie notch filter. the net energy added or drown by the compensating system is zero. Pc = 0; Ps = PL = P (4) where Pc is the real power from the compensator, Ps and PL are the source aerage power, load aerage power, respectiely. The aerage power is gien by 1 it = = P Px - Tx t-tx p(t)dt. where Tx denotes the aeraging interal that can be zero, one fundamental cycle, one-half cycle, or multiple cycles, depending on compensation objecties and the energy storage capacity of passie components [13]; p(t) is the instantaneous real power. The instantaneous reactie power theory work can be explained from early definition of non-actie current by Fryze [14]. ip(t) = ( (, i) ). (t), iq(t) = i(t) - ip (t),, where ip is the actie current component, (t) and i(t) is the reference oltage and current; iq is the non-actie current component; (, i) is the inner product of the oltage and the current oer the interal {t - Tx, t} with respect to weighting factor equal one, (, ) is the inner product of the oltage and itself oer the interal {t - T x, t}; They can be expressed as follows = = (, i) Il(t). i(t)11 - = = (,) Il(t)112-1 it Tx t-tx 1 j.t Tx t-tx A. Actie Current on the Input Side V(T). i(t) dt = P V2 (T) dt = ; ms On the input side of the matrix conerter the source oltage is sinusoidal and the source current is not sinusoidal as it shown in Fig. 5 (a). They are expressed by the following equations (9) (5) (6) (7) (8) 3704

3 ia(t) = ISj sin(wit - a) + ISh sin(wint + f3h) (10) where Vs j is the amplitude of the oltage fundamental component; Is j is the amplitude of the current fundamental component; ISh is the amplitude of the h-th order harmonic. The corresponding aerage power is p VSjlsj = 2 cosa (11) It can be erified that a(t), Ish sin(wint + f3h) only contain ac components. Therefore the corresponding aerage power is only a result of the fundamental components of the current ia and the oltage a(t). Likewise, substituting (9) and (10) in (8) yields VSj Va-rms = -2 (12) By substituting (9),(11) and (12) in (6), we can obtain the actie current component ip(t) = ISj cos (a) sin(wt) (13) By substituting (17), (18), and (19) in (14) we can obtain the actie component of the oltage VL jhjcosa + VLhhhcosah p () t = ' 2 ' ilj + l'lh 2 (hj sin(wot - a) + hh sin (wont + f3h)) (20) It is obsered from (20) that the actie component of the oltage is not sinusoidal. Therefore, the instantaneous reactie power theory is not able to achiee a sinusoidal load oltage. An FFT filter is used to obtain the fundamental component of the load current before it is used in the control loop [15]. This limitation of the instantaneous reactie power theory will be resoled by the proposed method in the next section. By expanding the same approach for a three phase system. we can define the compensation current and the compensation oltage as follows (21) This proes that the instantaneous reactie power theory is able to obtain the actie component of the current when the source oltage is sinusoidal and the source current contains harmonics. B. Actie Voltage on the Output Side On the output side of the matrix conerter, both load oltage and current include harmonics as shown in Fig. 6(a). Using a dual analogy to the actie current defined in (6), we can define the actie component p and the non-actie component Vq of the oltage as (, i), p(t) = - ( ' ' ) t (t), q(t) = (t) - p(t), (14) i,l. where (i, i) is the inner product of the current and itself oer the interal {t - T x, t} described by 1 i t. (i, i) = Il i(t)112 = - i2(t)dt = i;ms Tx t-tx (15) The load oltage A (t) and the load current i A (t) can be expressed by the following equations VA(t) = VLj sin(wot) + VL h sin(woht + f3h) ia(t) = hj sin(wot - a) + hh sin (wont + f3h) (16) (17) where VLj is the amplitude of the oltage fundamental component; VLh is the amplitude of the h-th order harmonic; hj is the amplitude of the current fundamental component; hh is the amplitude of the h-th order harmonic, the corresponding aerage power is and the corresponding rms current is l A-rms = where Vin = [a Vb c] T is the input oltage ector; i"t = (22) [i 1i ki 1 T is the fundamental output current ector; p is the oscillatory component of the real power; q is the reactie power. The control block diagrams for the input current compensation and the output oltage compensation are shown in Fig. 3(a). Both output current and oltage of the matrix conerter contain a significant amount of harmonics, in such cases the conentional IRPT control approach will not be able to obtain the correct actie oltage component. For such reason the fundamental component of the output current has to be extracted first before it is employed in the control circuit as shown in Fig. 3(b). IV. PROPOSED AV ERAGE POWER METHOD Aerage power compensation (APC) is based on the assumption that the total aeraged power drawn from the source is equal to the power obtained by the sinusoidal actie components of the current or the oltage. This assumption is mathematically represented in (23) for single-phase system, r (t)i(t)dt = r Jt-Tx Jt-Tx p(t)i(t)dt (23) where p is the actie component of the oltage. Assuming the the actie component of the oltage is sinusoidal, therefore, (24) We can estimate the amplitude of the actie oltage component by substituting (16) and (17) in (23), the amplitude A is determined by A ( ) ahiah cos(ah) = Vaj cos a +, (25) l aj From the actie oltage amplitude in (25), it can be (19) obsered that the actie component of the oltage p is not 3705

4: Aerage power compensation method control for (a) input current; (b) output oltage.

4 Pin, qin Reference Current s ---- Icomp i, Vo Aerage power Ploss "- --1 Aerage power (a) sine w,t) Aerage power (b) Fig. 3: Block diagrams for (a) determining the reference signals for compensation current and oltage using IRPT; (b) extracting the fundamental component of the output current. Fig. 4: Aerage power compensation method control for (a) input current; (b) output oltage. equal to the fundamental component, which is the fundamental reason that the ANF methods will not achiee maximum power transfer. A dual approach can be easily implemented to obtain the actie component of the input current. The block diagram of aeraged power method for the input current and the output oltage compensation are shown in Fig. 4(a) and (b), respectiely. V. SIMUL ATION RESULTS AND EVALUATION To ealuate the performance of the presented methods detailed simulation models hae been constructed based on the three phase hybrid matrix conerter shown in Fig. I. In each simulation model the control method is applied to compensate both the input current and the output oltage harmonics. The typical input and output current and oltage waeforms of the main matrix conerter are shown in Fig. 5(a) and (b). Fig. 5( c) shows the input source current after compensation using adaptie notch filter method, by which the fundamental component of the input current is calculated. The harmonic content is obtained by subtracting the fundamental component from the original current waeform. Although the notch filter is easy to implement, it is unable to correct the phase shift between the input oltage and the input current, Therefore, the ANF method is ineffectie when power factor correction is required. Another disadantage for ANF is its ineffectieness when both the oltage and the current contain harmonics such as the output oltage and current of the matrix conerter. In this case some of the power transferred to the load takes place at harmonic frequencies. Compensation to achiee output oltage equal to its fundamental component simply will not achiee maximum power transfer. The output oltage using ANF is shown in Fig. 5(d). The instantaneous reactie power theory allow for full control of the phase shift of the input current and subsequently a unity power factor could be achieed. Howeer, IRPT method is still unable to compensate the output oltage because both output current and oltage present harmonic. The only way to achiee sinusoidal output load oltage is to first obtain the fundamental component of the current before it is employed in the control circuit as shown in Fig. 3(b). The compensated input source current and the output load oltage using IRPT are shown in Fig. 5(e) and (t"). The aerage power method in contrast can proide unity power factor when compensating the input current. It is also able to obtain the actie component of the output oltage without any need of filtering the output current. The input source current and the output load oltage using APC are shown in Fig. 5(g) and (h). A summary of the comparatie ealuation of the methods is listed in Table 1. It is worth noting that the proposed APC methods features the lowest total harmonic distortion 3706

VI. CONCLUSION This paper has presented the preliminary results for the ealuation of existing approaches to shunt and series conditioning of the currents and oltages.

5 VI. CONCLUSION This paper has presented the preliminary results for the ealuation of existing approaches to shunt and series conditioning of the currents and oltages. The critically ealuated methods are based on either instantaneous reactie power theory or fast Fourier transform. The main limitation for IRPT based method lies in its ineffectieness when the harmonics are concurrently present in oltage and current while the limitation for FFT based method is its inability to compensate the fundamental component. To address the limitations associated with the existing methods, a new method based on power aeraging has been proposed. The effectieness of the proposed method has been erified by the simulation results obtained from a detailed MathCAD/Simulink model. Furthermore, experimental work has been planned and the experimental results will be included in the final manuscript. REFERENCES [1] A. Janabi and B. Wang. "Hybrid matrix conerter based on instantaneous reactie power theory," in Proceedings of 41st Annual Conference of IEEE Industrial Electronics Society, Yokohama, Japan, No. 9-12, pp [2] B. Wang and G. Venkataramanan. "Six step modulation of matrix conerter with increased oltage transfer ratio," in Proceedings of 3J!h IEEE Power Electronics Specialists Conference (PESC), Jeju, Korea, June 18-22, pp [3] P. Wheeler, J. Rodriguez, J. Clare, L. Empringham, and A. Weinstein, "Matrix conerters: a technology reiew," IEEE Transactions on Industrial Electronics, ol. 49, no. 2, pp , [4] H. Akagi, Y. Kanazawa. and A. Nabae. "Instantaneous reactie power compensators comprising switching deices without energy storage components," IEEE Transactions on Industry Applications. ol. IA-20. no. 3, pp [5] F. Z. Peng, H. Akagi, and A. Nabae, "A new approach to harmonic compensation in power systems," in Record of the IEEE Industry Applications Society 23rd Annual Meeting, Pittsburgh, PA, Oct. 2-7, 1988, pp [6] F. Z. Peng and J. S. Lai, "Generalized instantaneous reactie power theory for three-phase power systems," IEEE Transactions on Instrumentation and Measurement, ol. 45, no. I, pp , [7] F. Z. Peng, H. Akagi, and A. Nabae, "A new approach to harmonic compensation in power systems-a combined system of shunt passie and series actie filters," IEEE Transactions on Industry Applications, ol. 26, no. 6. pp , [8] D. Yazdani. A. Bakhshai. G. Joos, and M. Mojiri, "A real-time selectie harmonic extraction approach based on adaptie notch filtering," in Proceedings of IEEE International Symposium on Industrial Electronics, Jun Jul pp [9] D. Yazdani, A. Bakhshai, G. Joos and M. Mojiri,"A Real-Time Three Phase Selectie-Harmonie-Extraction Approach for Grid-Connected Conerters" IEEE Transactions on Industrial Electronics. ol. 56, no. 10. pp , Oct [10] E. Laopa, P. Zanchetta, M. Sumner. and F. Cupertino. "Real-time estimation of fundamental frequency and harmonics for actie shunt power filters in aircraft electrical systems," IEEE Transactions on Industrial Electronics. ol. 56, no. 8, pp [II] M. Mojiri, and A. R. Bakhshai, "An adaptie notch filter for frequency estimation of a periodic signal," in IEEE Transactions on Automatic Control. ol. 49, no. 2, pp [12] L. Asiminoaei, F. BJaabjerg, and S. Hansen, "Detection is keyharmonic detection methods for actie power filter applications" IEEE Ind. Appl. Mag., ol. 13, no. 4, pp. 2233, Jul.lAug [13] F. Z. Pengo L. M. Tolbert and Zhaoming Qian. "Definitions and compensation of non-actie current in power systems," Power Electronics Specialists Conference, pesc IEEE 33rd Annual, Cairns, Qld , pp [14] S. Fryze, Actie, "Reactie, and Apparent Power in Non-Sinusoidal Systems," Przeglad Elektrot. no. 7, pp (in Polish) [15] P. Salmeron and S. P. Litran,"Improement of the Electric Power Quality Using Series Actie and Shunt Passie Filters" Transactions on Power Deliery, ol. 25, no. 2, pp , April I Methods Power factor control Obtaining the actie component of current and oltage Transient Response ( due to input oltage increase by by 0.3 pu). Complexly of Implementation. Total harmonic distortion (THD) Table I. A summary of the comparatie ealuation of the three methods. II ANF II IRP Unable to control the PF, Be- Can achiee unity PF by comcause it is only obtains the pensating all the reactie comfundamental component with its ponent q. phase shift. Only able to obtain the funda- Fail to obtain the actie commental component of the cur- ponent of load oltage, because, rent or the oltage. therefore it both load oltage and current fails to achiee maximum power contain harmonics. Therefore, it transfer when both the current is required to filter the load curand oltage contain harmonics. rent first before it is employed in the control circuit., Requires at least one cycle to current drops during the tranreach the steady state. sient period. Many design parameters need Obtaining the load current funto be optimized to achiee low damental component increases THD. Higher order than IRP and the computational efforts signif- APC. icantly. 7.8% 2.5% II APC Fully control the PF by selecting the required phase shift in the reference signal. Able to successfully obtain the actie component of the output oltage without any need of filtering the output current. Requires one cycle to reach the steady state without any oershoot in the current. Requires less computational effort. The same control circuit can be used for all cases. 2% 3707

j 500 g, 0.500 > M M M M M M M M J!1J1J1MJ!1J1M iji N N N N N N N N V V V V V V I M 0.02 0.04 0.06 0.

$oltage AB and current ia 300 rr-----,-----,--,----,----,-----,--,---, A 1\ A VV 300w-------- --"---"$ 1 (f) Output line oltage VAR using notch filter.

(g) Output line oltage AB using instantaneous reactie power theory.. 500VV V VVVVV V 0.04 0.06 0.08 0.

6 j 500 g, > M M M M M M M M J!1J1J1MJ!1J1M iji N N N N N N N N V V V V V V I M (a) Input line oltage Vab and current ia Timc(s) (e) Output line oltage AB and current ia 300 rr-----,-----,--,----,----,-----,--,---, A 1\ A VV 300w "---" (b) Input current ia using notch filter (f) Output line oltage VAR using notch filter. 300w "---" (c) Input current ia using instantaneous reactie power theory. (g) Output line oltage AB using instantaneous reactie power theory.. 500VV V VVVVV V (d) Input current ia using power aareging method. (h) Output line oltage VAB using power aareging method. Fig. 5. Input current ia and output line oltage V AB compensation using the three different methods. 3708

Hybrid Matrix Converter Based on Instantaneous Reactive Power Theory

IECON205-Yokohama November 9-2, 205 Hybrid Matrix Converter Based on Instantaneous Reactive Power Theory Ameer Janabi and Bingsen Wang Department of Electrical and Computer Engineering Michigan State University