A Novel Impedance-Network-Based Electric Spring

An electric spring (ES) can well maintain the balance between supply and demand to compensate for the intermittent nature of small-scale renewable energy resources (RES). Despite its popularity, the second generation of ES (ES-2) is deemed to have a few practical problems. The most conspicuous one is the requirement of accurate dead-time control is in the ES circuit to avoid bridge shoot-through problem, which is necessitated by the series-connection of multiple voltage sources and/or converters to realize a wide voltage range. This however could cause output voltage waveform distortions. In this study, inspired by the Z-source network structure, we propose a novel ES topology with a specifically designed impedance network, i.e., an impedance-network-based (i.e., a network of passive devices such as inductors and capacitors) ES (IN-ES), which intrinsically has a wide voltage range and is immune to the bridge shoot-through issue (i.e., switch tubes on the same bridge arm of the inverter are turned ON/OFF at the same time). Detailed theoretical derivation, simulation and experimentation are conducted in this study, which verify the unique advantageous features of the proposed IN-ES, demonstrating a wide voltage operation range, undistorted waveforms and safe operations.


INDEX TERMS
In recent years, renewable energy resources such as wind and solar have been widely recognized as a favorable alternative to mineral-based energy resources, due to their environmentally friendliness and inexhaustible availability [1], [2]. Enabled by the rapid development of power electronics technology, which reduces the side effects of their intermittent nature, an increasing amount of RESs are being integrated with the power grid. However, this new energy industry is facing many problems and obstacles. For instance, the low efficiency of renewable energy power generation is currently the main problem restraining its development. In addition, renewable energy power generation would cause energy waste due to its poor power quality. Both problems are mainly caused by the intermittency of wind and solar RESs, which makes heavily RES-integrated grids difficult to balance load demands and power supply. One of the traditional solutions is to mitigate this problem by employing demand-side energy management techniques such as energy storage [3], real-time pricing [4], [5], load scheduling [6]- [8] and direct on-off load control [9]. In 2012, researchers proposed a novel hardware-based demand-side energy management strategy, i.e., the electric spring, trying to tackle the same problem. The invented ESs are able to reduce voltage and frequency fluctuations [10]- [12], improve power quality [12], [13], enhance demand-side energy management of the power grid [14], and diminish energy imbalance in built environments [15]. To date, three versions of ES have been proposed. The first generation of ES (ES-1) includes capacitors and power inverters as shown in Fig. 1(a) [16], which collectively can provide reactive power regulation to non-critical loads (i.e., loads that can work under a certain range of voltage levels) connected in series with the ES. The second generation of ES (ES-2) is also connected in series with non-critical loads as shown in Fig. 1(b), where ES-2 replaces the capacitor in ES-1 with a battery storage system [17], which makes it possible to compensate both active and reactive power. The third generation of ES (ES-B2B) consists of two half-bridge back-to-back converters as shown in Fig. 1(c), which, in addition to ES-2's functions, can extract power from the power grid [18].
The above-mentioned ES structures have been experimentally proven to be able to maintain the supply and demand balance in RES-integrated power systems. However, the DC side of the most popular ES-2 requires a high DC voltage to achieve a wide range of voltage compensation and operate in an environment with electromagnetic interference, causing shoot-through of inverter bridges and subsequent damages. To achieve a high DC voltage for ES-2, traditionally, a number of batteries or boost converters are employed and installed on the DC side to increase the DC voltage, thereby increasing the output voltage of the ES-2. High voltage gain boost converters may also be used, which can be achieved using various methods, e.g., [19] and [20]. However, using the above methods to increase the DC side voltage of ES may result in shoot-through of the inverter bridge. The traditional solution is to insert a dead-time in the drive signal to avoid shoot-through of the inverter bridge, but the insertion of the dead-time control will cause the distortion of the output voltage waveform [21], [22].
To solve the problems in the existing ES structures, expanding from the design ideas from [23]- [26]. In this paper, we propose a novel impedance-network-based ES, as shown in Fig.1(d), and the IN-ES topology is analyzed with reference to the analytical methods of quasi Z-source inverter topologies in [27]- [29]. The impedance network resembles a quasi-Z-network, which is connected to the DC side of the existing inverter. Compared with the second generation of electric spring, the proposed IN-ES is able to boost the DC side input voltage by the introduced impedance network and avoid the loss caused by the two-stage converter, which can easily achieve a wide range of voltage compensation with a lower DC input voltage. It does not need many batteries or boost converters to be installed on the DC side. So the size of the battery can be reduced, which attests cost-effectiveness. In other words, the function of a wide output range facilitates the design of power supply systems to achieve small size and low costs. In addition, with the specifically designed inductor in the IN-ES, inverter bridges are immune to shoot-through issues, which therefore does not require the conventional dead-time control, thus preventing waveform distortion of the output voltage. The IN-ES can also increase system stability.
In conclusion, the proposed IN-ES with the quasi-Z-network has high engineering and application values. To verify the efficacy of the proposed IN-ES structure, theoretical analyses, software simulation and hardware experimentation are conducted in this study, and the results agree well with the expected features of the proposed IN-ES.
The rest of the paper is organized as follows. In Section II, the topology, working principle and operation modes of the proposed IN-ES are illustrated, which is followed by the closed-loop control design in Section III. Simulation and experimentation are reported in Sections IV and V, respectively. Finally, this paper concludes in Section VI.

II. THE PROPOSED IN-ES
This section explains the topology, working principle and operation modes of the proposed IN-ES, and its superiority over its conventional counterparts.

A. TOPOLOGY OF IN-ES
As briefly mentioned in the introduction, typical topologies of existing ESs include half-bridge inverter, a full-bridge inverter, and a three-phase inverter. The ES-1 has the advantage of a simple structure and easy implementation, but it can only provide reactive power compensation [30]. With battery storage system, ES-2 has the capability to provide both active and reactive power compensation, which has been widely used for voltage stabilization and power factor correction [31]. The back-to-back converter topology enables ES-B2B to connect to the electric grid, and export and import electric power to and from the grid. In addition to active and reactive power compensation, ES-B2B can also suppress harmonics, improving power quality [18].  A switched boost inverter (SBI) that provides power to AC load is shown in Fig. 2. The SBI as ES application circuit with power grid and the equivalent model are shown in Fig. 3, it can be clearly seen that the voltage of the DC-side boost capacitor at this time may be disturbed by the energy rectified from the backward power flow of AC inverter bridge, causing the capacitor V C 1 damaged and then further make the boost converter failed. The circuit of the proposed IN-ES connected with power grid is shown as Fig. 4, which couples a switch-boost quasi-Z-network and DC load. The DC load will pave an energy path for bidirectional energy consumption at the boost stage, and further for grid terminal voltage control. When the critical load voltage is unstable, the proposed IN-ES will boost DC input voltage V in to compensate for active or reactive power of the power grid, which will stabilize the critical load Z c voltage.    As shown the shoot-through state time (last T 0 ) of the proposed IN-ES in Fig. 5, the diodes D 1 , D 2 and switch S 0 are on, the diodes D 0 , D 3 and D 4 are off. At the same time, current flows from capacitor C 1 to inductor, load and shoot-through switch S 0 , and the corresponding equations for the shoot-through state are (1) According to (1), we have When it works in the non-shoot-through state as shown in Fig. 6, during the non-shoot-through time T 1 , D 0 , D 3 and D 4 are on, while D 1 , D 2 and S 0 are off. Similarly, current flows from voltage source V in and inductors L 1 and L 2 to capacitor C 1 and the load, which results in equations as The switching interval between these two states is negligible, and thus one can assume voltage of inductors L 1 and L 2 , namely V L 1 and V L 2 are approximately equal, and one has The above two states operate alternatively during one switching period, i.e., T = T 0 + T 1 . According to the volt-second balance and ampere-second balance principle, we have If shoot-through duty ratio is then we can have The DC link voltage through the inverter bridge can be expressed as where The peak voltage can then be obtained as the boost factor due to the shoot-through state, and the peak voltage iŝ In Figs. 1(a)-(d), the grid reactive power is compensated by capacitor C s , whose stored energy can be expressed as where V es is the voltage across capacitor C s . According to (14), the magnitude of voltage V es can be controlled by the modulation ratio M and the shoot-through boost factor B. Under the condition that the rated voltage of the capacitor C s is not exceeded, according to (15), the energy stored in capacitor C s can be increased by changing modulation ratio M and shoot-through boost factor B, thus enhancing the reactive power compensation capacity of the C s . This theoretical analysis justifies the superiority of the proposed IN-ES over traditional ES structures and can provide a wider range of voltage control through reactive power compensation, owing to the controllable capacitor energy.
According to the aforementioned analyses, the DC link voltage of the inverter can be derived from the capacitor voltage V C 1 , and the DC link output power equals the capacitor's output power. In terms of the power constant principle, one can get where P is output power at the AC side, V es is the magnitude of the AC side output voltage and I Ls is the magnitude of the AC side output current. According to (16), it can be seen that there is a secondary low-frequency pulsation due to the output power of the inverter and the electrical energy provided by the DC side. Therefore, the twice low-frequency ripple also exists on the DC side, i.e., V C1 and V p have twice low-frequency ripples.

C. EFFICIENCY ANALYSIS
Parasitic parameters in power electronics circuits can make the actual voltage gain deviate from the theoretical value. In this subsection, the efficiency of the proposed IN-ES converter is investigated. In order to obtain the practical gain, parasitic resistances of the power electronics components used in this study are taken into account. This includes diode forward voltage (V F ), diode conduction resistor (R D ), winding resistor of inductor (R LS ), equivalent series resistance of capacitor (R C ), the drain-source on-resistance of IGBT (R CE ), and the turn-on and turn-off time of IGBT denoted as t on and t off .
The currents flowing through the diodes can be obtained as and i D 1,2 = I L 1 , Converter shoot-through, 0, Converter non-shoot-through, (18) and the RMS value and average value of the current of diode can be deduced as and The total power loss in diode D k (0 ≤ k ≤ 4) is the summation of the power loss in R D and the power loss associated with the voltage source V D . Assuming that all diodes are identical, according to (19) and (20), the total power loss of the diodes can be obtained as Total conduction loss in the inductor due to their equivalent series resistance (ESR) is given by It can be easily seen from the above analysis that the current flowing through the capacitor can be represented as Converter non-shoot-through, The RMS value of the capacitor current is given by The total conduction loss in the buffer capacitors due to their ESRs can be obtained as It can be easily deduced from the above analysis that the current flowing through the IGBT switch S 0 to S 4 can be obtained as i S 1,2,3,4 = 2I L 1 , Converter shoot-through, I i , Converter non-shoot-through.
Hence the RMS value of the currents of the S 0 and S 1,2,3,4 can be obtained as Considering the switching loss and conduction loss of inverter bridge, the power loss in the IGBT is thus P S = P Q (con) + P Q (swi) = 5R CE I 2 S 0(RMS) + 4V CE I S 1,2,3,4 (RMS) (t on + t off )f 2 where P Q (con) and P Q (swi) are conduction and switching power losses of switches and V CE is the drain-source voltage of S 0 to S 4 being off. Then the total energy efficiency of the proposed converter can be calculated as, η = P P + P Loss = P P + P rL + P rC + P S + P D .

D. OPERATION MODES OF THE PROPOSED IN-ES
The proposed ES structure with a high-step-up ratio and immunity of bridge shoot-through has eight operation modes, which resembles ES-2, i.e., (i) inductive mode, (ii) capacitive mode, (iii) resistive mode, (iv) negative-resistive mode, (v) inductive plus resistive mode, (vi) capacitive plus resistive mode, (vii) inductive plus negative-resistive mode and (viii) capacitive plus negative-resistive mode. This section mainly introduces the voltage regulation function when the proposed IN-ES provides pure reactive power compensation (non-critical load is assumed to be a RL load).

1) VOLTAGE SUPPORT
As shown in Fig. 8, when the mains voltage V s = V nc + V es is lower than the nominal value V s−ref , the proposed IN-ES operates in the inductive status, i.e., absorbing reactive power from the grid. The phase of the output voltage V es leads its current I es by 90 • , as shown in Fig. 8.

2) VOLTAGE SUPPRESSION
As shown in Fig. 9, when terminal voltage is higher than the nominal value, the proposed IN-ES operates in the capacitive mode, and the output voltage V es of the IN-ES lags its current I es by 90 • . In this mode, IN-ES provides reactive power to the grid to decrease the magnitude of the terminal voltage.
Conclude that the proposed IN-ES can well stabilize the terminal fluctuation voltage both in voltage support and suppression to the nominal value.

III. CONTROL STRATEGY AND STABILITY ANALYSIS OF THE PROPOSED IN-ES A. CONTROL STRATEGY
In order to realize the advanced features of the proposed IN-ES, corresponding control purposes are listed as follow.
1) Control the modulation ratio M and the boost factor B through controlling the switches, so as to extend the voltage regulation range; and 2) Stabilize the terminal voltage at its nominal value. The control scheme shown in Fig. 10 is established in MATLAB/Simulink to achieve the above control objectives. With the battery storage system (denoted as V in ), the phase angle of the compensation voltage V es and the phase angle of the current, namely I es , flowing through the proposed IN-ES can produce complex power with an arbitrary angle, thereby stabilizing the voltage fluctuations. This control scheme uses a single-phase phase lock loop (PLL) to determine the phase angle of I es , and then changes the phase angle of output voltage in terms of required reactive power compensation. In Fig. 10, driving signals for switches S 0 ∼ S 4 are described.
In Fig. 10, a simple boost control method is utilized to achieve the DC side voltage. The method controls the ON and OFF of the DC side switch S 0 while inserting the shoot-through signal into the modulation signal of the traditional SPWM full bridge inverter, and finally balance DC side voltage. Detailed control waveforms is shown in Fig. 11 which is generated from Fig. 10.
(32) VOLUME 8, 2020 Since the perturbation of the input voltagev in negligible, the transfer function can be deduced as where Bode plots of the transfer functions G V C 1 (s) is shown in Fig. 12. It can be seen from the bode diagram that the phase margins of the transfer function are higher than 0, so the IN-ES is a stable system.
The IN-ES output port is represented by a variable voltage source V i , then the equivalent s-domain circuit diagram of Fig. 13 can be drawn. According to Norton's theorem, we can have According to the Kirchhoff's laws According to formula (36) where From formulas (35), (37) and (38), the transfer function of the system is where K pwm is the ratio of the DC voltage V p of the inverter to the amplitude of the triangular carrier. The Bode diagram of the transfer function of the system is shown in Fig. 14, and its phase angle margin is higher than zero, hence a stable system. Because the fluctuation of voltage V s will affect the stability of the system, in order to achieve a stable operation and improve the dynamic response, a voltage control strategy with a PI controller is used. The expression of the PI controller is It can be seen from Fig. 14 that the phase angle margin of the system is greatly increased after compensation, so this control strategy can ensure stable operations of the system.

IV. SIMULATION STUDY
In order to verify the effectiveness of the proposed IN-ES, in this section we conduct simulations in MATLAB/Simulink environment, which is compared to the popular ES-2 under different input voltages. Simulation parameters are shown in TABLE 2. The simulation employs the control strategy shown in Fig. 10 to control the circuit in Fig. 4 for verification.   As shown in Figs. 15∼18, in the simulation, voltage V s suddenly changes to 215V or 225V when t = 2s, which is lower or higher than its nominal value 220V. Under this condition, the proposed IN-ES and ES-2 start the control duty at t = 5s. Their control results are plotted in Figs. 15∼18, respectively. We can see that the proposed IN-ES and ES-2 are able to regulate V s to the nominal value.According to the formula V es = M V in of the output voltage of the full-bridge inverter, the DC side voltage of  In addition, as shown in Fig. 19(a), when the proposed IN-ES is not regulating the terminal voltage (t = 0 ∼ 5s), it can provide a constant DC voltage. At t = 5s, the IN-ES starts operating in voltage support mode and the DC side capacitance voltage is 294.5 ± 1V. This fluctuation is caused  by the output power on the AC side and the turn-on voltage drop of the diode and IGBT transistors. Because the voltage fluctuations sit within the 0.4% range, the IN-ES is considered to be able to provide stable DC voltage while absorbing reactive power from the grid. Similar conclusion can be drawn for IN-ES working in voltage suppression mode as in Fig. 19(b), where the DC side capacitance voltage is 294.5±1.7V (below 0.6% of the DC component) when IN-ES is providing reactive power to the grid.  From the simulation results shown in Figs. 15∼19, we can see that the proposed IN-ES is able to provide a higher compensation voltage to regulate the terminal voltage to its nominal value, hence a wider voltage control range. It also suggests that the ES-2 is not able to cope with such voltage deviations when the input DC voltage source is 100V. It is thus safe to conclude that the proposed IN-ES has a superior voltage regulation capability over the existing ES-2 for compensating the terminal voltage deviations.

V. EXPERIMENTAL VERIFICATION
In order to further verify the validity of the proposed IN-ES, a hardware prototype is established in this paper, which is shown in Fig. 20. The parameters used in the experiment are shown in TABLEs 3 and 4 for IN-ES and peripheral devices, respectively.
As shown in Figs. 21(a) and (b), when ES-2 operates in voltage support (inductive) mode, it is able to stabilize the terminal voltage from 28.9V to 30V, and the compensation  voltage V es of the ES-2 is 16V. Similarly, ES-2 is able to regulate the terminal voltage from 31.1V down to its nominal VOLUME 8, 2020 value, with the V es being 14.4V. Through further experiments we know that if the terminal voltage is to be stabilized under ±1.1V, ES-2 requires the DC input voltage to be at least 23V, which is greater than the input voltage, and ES-2 cannot satisfy this condition.
On the other hand, we can see from Figs. 22(a) and (b), the capacitor voltage V C 1 and the inverter bridge DC side voltage V p can be increased from 15V to 40V. When the proposed IN-ES regulates the terminal voltage, it is able to stabilize the voltage of critical load at its nominal value 30V. Compared with the traditional ES-2, its DC side voltage only requires 15V, and adjusting the shoot-through duty d can further reduce the DC input voltage requirement. According to (16) and the results in Fig. 23, we can observe that when IN-ES conducts voltage control, voltages V C 1 and V p have the twice low-frequency ripples. Conclude that IN-ES is superior to ES-2.

VI. CONCLUSION
In this study, we have proposed a novel impedancenetwork-based electric spring structure, and we have conducted thorough analysis, simulation and experimentation, which verified the efficacy of the proposed IN-ES. Unlike previous versions of electric springs, the proposed IN-ES has an additional switch-boost quasi-Z-network to the DC side. Compared with the ES-2, this structure has been proven to be able to reduce the requirement of the DC side voltage level and corresponding requirements for battery system, increase input-to-output voltage ratio, and expand electric spring's voltage compensation range. Furthermore, without needing dead-time control, the proposed IN-ES is intrinsically immune to the inverter bridge shoot-through issue. All these features are absent in the existing ES topologies. He is currently a Professor with the School of Automation, Guangdong University of Technology, Guangzhou. His current research interests include intelligent control systems, network systems, and signal processing. VOLUME 8, 2020