2018 IEEE International Conference on Renewable Energy and Power Engineering Distributed Generator in Grid-Connected Mode Using Improved Exponential Sliding Mode Control S. Sinan1, A. Elnady2 and M. Al-Shabi3 Electrical and Computer Engineering Department University of Sharjah 27272 Sharjah, United Arab Emirates E-mail: ssinan03@gmail.com1, anady@sharjah.ac.ae2, malshabi@sharjah.ac.ae3 considered to operate the DGS in both modes: Grid-connected mode and stand-alone mode. In this section the literature review is focused on the schemes used for the grid-connected mode because the distributed generation system in this paper is operated in a grid-connected mode. Voltage oriented control (VOC), also called voltage vector control, has been adopted to control the distributed generation systems in a grid-connected mode such that it can be used to inject active and reactive power based on the system conditions [4-6]. Voltage oriented control is an indirect scheme to control the output power since it controls active and reactive power through the direct and quadrature currents, respectively. The main controller in the control loops is the PI controller or PR controller. The good performance of the VOC is not guaranteed at different operating conditions. Another scheme used with the DGS is the sliding mode control (SMC) [7-10]. The main structure of this scheme is similar to the VOC because the power references is converted into current references. The current error, generated between the current reference and its corresponding feedback, is controlled by the SMC to operate the DG. Several versions of the 1st order SMC have been utilized to control the DGS. These versions are classified based on the adopted reaching law and the definition of the sliding surface, (sliding manifold). The SMC with a differential form of the sliding surface is employed in [7] to the control the DGS in a grid-connected mode. The integral form of the sliding surface is presented in [7-10]. A constant-rate reaching law, which is expressed by the signum function, is utilized for the DGS in [10]. A power-rate reaching law accompanied by a boundary layer is used in [7-9] to improve the reaching mode and to reduce its chattering at the sliding mode. All versions of the 1st order SMC suffer from chattering in its output power/current. This drawback triggers the utilization of some advanced sliding mode controllers to control the DGS [11], [12]. A 2nd order SMC, super-twisting technique, has been formulated and employed to control the distributed generator in a grid-connected mode and in stand-alone mode as presented in [11]. The super-twisting technique presents a better performance at the sliding mode in terms of the chattering. Its performance is compared to the 1st order SMC in [7]. The exponential SMC has been published before in [13-14], and it is proved that this version offers a better performance than the conventional SMC. This classical Abstract—This paper shows the operation of the distributed generation system using an innovative control scheme. The core of this distributed generation system is the distributed generator, which is based on the 5-level diode clamped inverter. This distributed generator is operated by an improved version of the exponential sliding mode control. This developed sliding mode control perfectly fits the power electronics based distributed generator to control the output injected power, and it gives the shortest reaching time with much less chattering at its output power. The performance of the whole suggested control scheme along with the improved sliding mode controller is verified using simulation results. Keywords—Exponential sliding mode control, distributed generation system, PD-PWM, and power control. I. INTRODUCTION The distributed generation system (DGS) has been proliferating in distribution systems because it supplies the distribution system with extra power needed for the operation of the distribution system when the distribution system is heavily loaded. The DGS is characterized by the generation capacity of its distributed generator (DG) that is mainly determined by the rating of the inverter circuit. The generation capacity of the DG ranges from few KVA to few MVA [1-3]. The penetration of the DGS in the distribution system is continuously increasing nowadays because it brings a lot of benefits to the distribution system when it’s tied to the distribution system. The DGS achieves the following benefits: x It regulates the voltage at the point of common coupling where it is installed. x It injects more active power to the distribution system under heavy loading conditions. x It increases the overall stability of the system. x It reduces the voltage drop and the power losses over the distribution feeders since the power direction is reversed in some feeders. The DGS entails a control algorithm that precisely controls the operation of the distributed generator in a fast and accurate manner. Many control schemes have been developed and employed for the DGS. These schemes are 978-1-5386-9365-0/18/$31.00 ©2018 IEEE 7 exponential SMC is modified in [15] to fit the operation of the power electronics circuits. Its control law is modified by adding one constant-proportional term to the discrete term of its control law. The most recent scheme to control the DGS is the direct power control (DPC) [16-18]. This scheme is originated from the direct torque control used to control the drive systems. The DPC adjusts the power directly not through the currents as mentioned before. The conventional DPC depends a hysteresis controller and the space vector modulation (SVM), which cause oscillation shown in the output power and variable switching frequency. Several modifications have been introduced to the DPC in order to improve its performance and minimize its drawback. A look-up-table is used to minimize the switching frequency [16]. Predictive control is merged with the DPC to improve the performance of the SVM and hysteresis controllers [17], [18]. The contribution of this paper is exemplified in using an improved exponential sliding mode control to accurately control the currents and consequently the injected power. This improvement gives a better performance than what is published in [13-14]. This paper has five sections, where the second section shows the formulation of the exponential sliding mode control. The third sections depicts the suggested control scheme along with the proposed controller. The fourth section displays the simulation results. The last two sections demonstrate the findings and appendix of this paper. where G 0 is a strictly positive offset less than 1, U is a strictly positive integer, and D is strictly positive. The formulas in (1-3) prove to be better than the conventional SMC in terms of the reachability time defined and proved in [13], [14]. The chattering is reduced at the sliding mode because at the beginning of this reaching mode the value of k s is large; so the magnitude of the udis tends to be . G0 While at the end of the reaching mode, the value of just derived using the Lyapunov stability criterion. The final expression of the ueq is given as, 1 1- The sliding surface in this research is defined by its integral form as given in (7), while the sliding surface in [13-15] is defined by its differential form, which does not guarantee the minimum steady-state error (For instance some results in [14] shows a clear steady-state error). This integral SMC is defined as, s udis (2) The contribution in [13-14] is the definition of the discrete input udis given as, (3) (4) signum function and N ( s) is defined by, ( )= + (1 − ) | | k N( s ) tanh( s ) (8) modified from what given in (6) to fit the state-space model of the system understudy since the state-space model is given as, x Ax Bu Fd (9) The continuous input u eq is modified to be where k1 , k2 and k are the positive constant, the sign is the traditional (7) The tanh function is selected because the signum function does not perfectly match the operation of the power electronics based systems like the system under study in this paper since it causes sudden sharp changes in its output and spontaneous deviation of the states from the sliding surface leading to possible unstable operation. While the tanh function is different since it does not produce these sharp changes. 3- The definition of the continuous input u eq is slightly (O1 O2 dt ) e sign ( s ) N (s) While in [15], the contribution is exemplified as, k u dis k1s 2 sign ( s ) N (s) (O1 O2 ³ dt ) e It is proved that the integral SMC gives less error than any other form for the sliding surface [7]. 2- The discrete term udis of the control law in this research is innovatively defined as, continuous input used to keep the states of the system on the sliding surface (sliding mode). In [13-15], the sliding surface is defined by its differential form as, k (6) The improvement in the exponential SMC in this paper is different from what has been published in [13-15] in the following aspects: where the udis is the discrete input, which is responsible for reaching mode used for transferring the states of the system from one sliding surface to another. The ueq is the ª ws B º ws Ax «¬ wx »¼ wx > @ ueq The exponential SMC has been introduced in [13-15]. The contribution of this exponential SMC is in the definition of its control law, which is expressed as, u udis ueq (1) u dis udis tends to be k . The other term of the control law is the ueq , which is becomes small; then the magnitude of the II. FORMULATION OF EXPONENTIAL SLIDING MODE CONTROL s s =− (5) 8 [ + − ] (10) The proof of the formula in (10) along with its stability analysis is given in the Appendix. After incorporating all aforementioned changes, the improved exponential SMC is ready to be merged with the proposed control scheme that will be explained in the next section. injected voltage of the distributed generator in the d-q frame Ed DG required , Eq DG required . These two voltages are used to generate a control signal whose magnitude and phase are defined as, M control signal III. PROPOSED CONTROL SCHEME ( Ed2 DG required Eq2 DG required ) / Vdc (12) The proposed control scheme is illustrated in Fig. 1 along with the model of the distribution system. In the grid-connected mode, the DGS is tied to the distribution system. Therefore, the DGS should be operated at a constant-power mode. The power is controlled through the injected currents. The power references are converted into current references as, Vq DG º ª P ref º ª I d ref º ª Vd DG 3 « » » « »« 2 2 Vd DG »¼ «¬Q ref »¼ «¬ I q ref »¼ 2(Vd DG Vq DG ) «¬Vq DG (11) where Vd DG ,Vq DG are the DGS voltages in the d-q frame. T control signal E tan 1( q DG required ) Ed DG required (13) This generated control signal is utilized to operate the in-phase disposition pulse width modulation (PD-PWM) [19] to operate the inverter based distributed generation. IV. SIMULATION RESULTS This section shows the results of the proposed control scheme of Fig. 1 for the DGS when it is connected to the distribution system. A. Power System Under Study and Distributed Generator I d ref , I q ref are the required injected currents (current The distributed system under is already depicted in Fig. 1, and its parameters are listed in Table 1. The parameters of the exponential sliding mode controller are given in Table 2. references) in the d-q frame. These current references are compared to the feedback currents I d DG , I q DG as shown in the Fig. 1. TABLE I: DISTRIBUTION SYSTEM PARAMETERS Description of Parameters Values Rating of the DGS 2.5 MVA Type of the transformer Step-up Rating of the transformer 2.5 MVA Turn’s ration of the transformer 5 Resistance of the tie feeder 0.9 Ω Inductance of the tie feeder 3 mH Rated voltage of Distribution system 6.6 kV Loads at distribution system Parameter k G0 D U Figure 1. Proposed control scheme with distribution system understudy. 50+31.4j TABLE II: CONTROLLER PARAMETErs Its value for active Its value for reactive power loop power loop 25 25 0.003 0.003 0.06 0.06 0.5 0.5 O1 1 1 O2 225 405 The topology of the distributed generator (DG) with its typical output voltage is illustrated in Fig. 2-a and Fig. 2-b, respectively. The typical voltage of the 5-level inverter and The current error is processed by the proposed exponential SMC formulated in the previous section to generate the input control law, which represents the required 9 the voltage after the transformer are illustrated in Fig. 2-b. The adopted topology of the distributed generator is the 5-level diode clamped inverter of Fig. 2-a, which is intentionally selected to reduce the injected harmonics compared to the 2-level or 3-level converter circuits. Figure 3a. Active power control using proposed exponential SMC scheme. Figure 3b. Reactive power control using proposed exponential SMC scheme. Figure 3. Power control using improved exponential SMC scheme in grid-connected mode. Figure 2a. Topology of 5-level diode clamped inverter per phase. The output of the control law for the suggested controller is the required injected voltages Ed DG required and Eq DG required by the distributed generator in the d q frame. These voltages are used to generate a control signal using inverse Park transformation. These voltages are portrayed in Fig. 4 to show the outputs of the suggested controller. Figure 2b. Voltages of distributed generator and inverter per phase. Figure 2. Topology of the distributed generation with its output voltage waveforms. B. Performance of Control Scheme The performance of the presented control scheme along with the developed controller is depicted in Fig. 3-a for active power and in Fig. 3-b for reactive power for some arbitrary power references shown in red. Figure 4. Output of control law for the proposed controller. To prove the advantageous performance of the suggested control scheme, its output is compared to the exponential 10 SMC (classical exponential SMC) given in [13-14] and formulated in (1-5) for the same parameters defined in Tables 1 and 2. The active and reactive power tracking performance is given in Fig. 5-a for active power and Fig. 5-b for reactive power. Also, the output of the classical exponential sliding mode controller is given in Fig. 6. C. Impact of Distributed Generation System on Distribution System The impact of the DGS on the distribution system is exemplified in this section since the DGS injects active and reactive power based on the requirements of the distribution system. Fig. 7 clarifies this beneficial impact such that the DGS supplies the loads with its power and the surplus is injected to the upstream system; in this figure the negative power means injected power and vice versa. In Fig. 7 for t < 5 s, the injected power of the distributed generation system PDGS is not enough to supply the whole power required by the load Pload , then the distribution system supplies power PDS to the loads as well; meaning that both the distributed generation system and distribution system supply the loads. For 5 s < t < 10 s, the PDGS is increased such that the whole loads are supplied from the distributed generation system and the surplus is absorbed by the distribution system. In this duration, the PDS becomes positive, which means the Figure 5a. Active power control using classical exponential SMC. upstream system absorbs active power. For t > 10 s, the PDGS is decreased such that both the distributed generation system and distribution system are supplying loads. In this duration, the PDS is reversed to be negative, which means that the distribution system is supplying active power. A similar performance can be obtained for the reactive power. Eventually, the DGS helps the distribution system supply power, which fetches a better performance for the distribution system as described in the introduction section of this paper. Figure 5b. Reactive power control using classical exponential SMC. Figure 5. Power control using classical exponential SMC scheme in grid-connected mode. Comparing Fig. 3 to Fig. 5 and Fig. 4 to Fig. 6 reveals that the improved version, presented in this paper, has a better transient performance with much less overshoot. On the other hand, the classical exponential SMC shows a slightly less chattering at some duration especially for the reactive power performance for time from 20 s to 30 s. Figure 7. Performance of active power for distribution system, distributed generation system and loads. V. CONCLUSION This paper introduces a tangible improvement of the exponential SMC to enhance its reaching mode compared to the classical exponential SMC. Several modifications have been introduced to the classical exponential SMC in this research. These modifications have improved the overall performance of the control scheme and alleviate the drawbacks of classical exponential SMC such as steady-state Figure 6. Output of control law for the classical exponential controller in [13-14]. 11 error and compatibility of the discrete input of the control law with power electronics-based systems. The suggested control scheme is applied on the distributed generation system to operate it in a grid-connected mode. The simulation results prove that the improved version outperforms the classical exponential SMC in terms of reaching mode and overall tracking performance. [7] [8] APPENDIX [9] This section shows the proof of the formula given in (10) of Section II. The distribution system of Fig. 1 can be modeled by a state-space equation as, x Ax Bu Fd s 1 O2 ³ [10] I ref dq I DG dq e, edq To guarantee the stability of the system in the sliding mode, the Lyapunov criterion should be fulfilled. Therefore, the Lyapunov function is selected to express the Euclidian distance between the state variables and the sliding surface defined as, 1 V 2 [11] sT s [12] For stability of the system in the sliding mode, the derivative of the above Lyapunov function should be less than zero, which leads to s 0 that can be rewritten as, ̇= ̇+ 2 ̇= ̇ − ̇+ ̇=− − =− −1 [ − + [13] 2 + 2 =0 − 2 ] [14] REFERENCES [1] W. El-Khattam, and M. M. A. Salama, “ Distributed Generation Technologies, Definitions and Benefit,” Journal of Electric Power System Research,” vol. 71, pp. 119-128, 2004. [2] T.Ackermann, G. Andersson, and L. Söder, “ Distirbuted generation: a definition,” Electric Power System Research, Elsevier, vol. 57, no. 3, pp. 195-204, 2001. [3] G. Pepermansa, J. Driesenb, D. Haeseldonckxc, R. Belmansc, and W. D’haeseleerc, “ Distributed generation: definition, benefits and issues, “ Energy Policy, Elsevier, vol. 33, no. 6, pp. 787-798, 2005. [4] R. Teodorescu and F. Blaabjerg, “ Proportional-resonant controller, a new breed of controllers suitable for grid-connected voltage-source converter,” Proc. of Optimization of Electrical and Electronic Equipment, vol. 3, pp. 9–14, 2004. [5] W. Xuehua, R. Xinbo, B. Chenlei, P. Donghua, and X. Lin, “ Design of PI regulator and feedback coefficient of capacitor current for grid-connected inverter with an LCL filter in discrete-time domain, “ IEEE Energy Conversion Congress and Exposition, USA, pp. 1657–1626, 2012. [6] F. Gao, and M. R. Iravani,” A control strategy for a distributed generation unit in grid-connected and autonomous modes of [15] [16] [17] [18] [19] 12 operation,” IEEE Transactions on Power Delivery, vol. 23, no. 2, pp. 850–859, 2008. A. Elnady, “ Newly developed 1st order sliding mode of power and voltage control of multilevel inverter based distributed generator”, International Journal of Power and Energy Systems, ACTA Press, vol. 37, no. 4, 2018. J. Hu, L. Shang, Y. He, and Z. Q. Zhu,” Direct active and reactive power regulation of grid-connected DA/AC converter using sliding mode control approach,” IEEE Trans. on Power Electronics, vol. 26, no. 1, pp. 210-222, 2011. L. Shang, D. Sun, and J. Hu,” Sliding mode based direct power control of grid connected voltage source inverters under unbalanced network conditions,” Proc. of IET on Power Electronics, vol. 4, no. 5, pp. 570-579, 2010. A. M. Bouzid, J. M. Guerrero, A. Cheriti, M. Bouhamida, P. Sicard, and M. A. Benghanem,” Survey on control of electric power distributed generation systems for microgrid applications,” Journal of Renewable and Sustainable Energy Reviews, Elsevier, vol. 44, pp. 751-766, 2015. D. H. Phan and S. Huang, “ Super-twisting sliding mode control design of three-phase inverter for micro-grid distributed generation systems,” Journal of Control, Automation and Electric Systems, vol. 27, no. 2, pp. 179–188, 2016. A. Elnady, and S. Sinan,” An improved second-order sliding mode control for the distributed generation system in standalone and grid-connected modes,” International Transactions on Electric Energy Systems, vol. 27, no. 11, pp. 1–10, 2017. C. Fallaha, M. Saad, H. Kanaan, ” Sliding mode control with exponential reaching law applied on a 3 DOF modular robot arm,” Proc. of The European Control Conference, pp. 4925-4931, Greece, 2007. C. Fallaha, M. Saad, H. Y. Kanaan, K. Al-Haddad, ”Sliding mode robot control with exponential reaching law, “ IEEE Trans. on Industrial Electronics, vol. 1, no. 2, pp. 1-9, 2010. S. M. Mozayan, M. Saad, H. Vahedi, H. Fortin-Blanchette, and M. Soltani, ”Sliding mode control of PMSG wind turbine based on enhanced exponential reaching law,” IEEE Trans. on Industrial Electronics, vol. 63, no. 10, pp. 6148-6158, 2016. D. Zhi, L. Xu, B. W. Williams, “ Improved direct power control of grid-connected DC/AC converters,” IEEE Trans. on Power Electronics, vol. 24, no. 5, pp. 1280-1292, 2009. L. Shang, “Sliding-mode-based direct power control of grid-connected wind-turbine-driven doubly fed induction generators under unbalanced grid voltage conditions,” IEEE Trans. on Energy Conversion, vol. 27, no. 2, pp. 362-373, 2012. X. Liu, D. Wang, and Z. Peng, “ Predictive direct power control for three-phase grid-connected converters with online parameter identification,” International Transactions on Electrical Energy Systems, vol. 27, pp.1-21, 2017. A. T. Alsakhen, A. M. Qasim, B. Qaisieh, A. Elnady, “ Experimental and simulation analysis for the 5-level diode clamped inverter,” Proc. of International Conference on Electric Power and Energy Conversion Systems, pp. 1-4, 2015.