A Comprehensive Review of Hybrid Energy Storage Systems: Converter Topologies, Control Strategies and Future Prospects

The ever increasing trend of renewable energy sources (RES) into the power system has increased the uncertainty in the operation and control of power system. The vulnerability of RES towards the unforeseeable variation of meteorological conditions demands additional resources to support. In such instance, energy storage systems (ESS) are inevitable as they are one among the various resources to support RES penetration. However, ESS has limited ability to fulfil all the requirements of a certain application. So, hybridization of multiple ESS to form a composite ESS is a potential solution. While integrating these different ESS, their power sharing control plays a crucial role to exploit the complementary characteristics of each other. Therefore, this article attempts to bring the numerous control strategies proposed in the literature at one place. Various control techniques implemented for HESS are critically reviewed and the notable observations are tabulated for better insights. Furthermore, the control techniques are classified into broad categories and they are briefly discussed with their limitations. From the carried-out analysis, the challenges faced towards the implementation of HESS for standalone and grid connected microgrid systems are presented. Finally, the future directions are laid out for the researchers to carry out the research and implementation of HESS technologies. Overall, this article would serve as a thorough guide on various control techniques implemented for HESS including their features, limitations and real-time applications.

From the past few years, the growing concerns on environmental effects due depletion of fossil fuels has resulted in transition towards RES to satisfy the global energy demand. Eco-friendliness, scalability and other extensive features of RES has attracted its deployment in commercial, industrial, and residential sectors. This is further supported by the quick progression in power electronics [1], which aids the complete control of RES limited by stochastic natural conditions [2]. Moreover, RES have various limitations like poor load following, intermittent power generation and non-dispatchable nature. Due to these factors, their coordination in the grid system is a challenging task for efficient operation, particularly for high capacity systems [3]. The notable challenges faced during the integration of RES are voltage instability, load discrepancy, frequency fluctuation, poor power quality and load-following [4]. The integration of ESS is a promising solution to overcome these limitations and to facilitate stable operation of grid. The integration of ESS with RES into the microgrid can avoid power fluctuations, improve power quality, frequency regulation and enable additional ancillary services [5]. Therefore, various ESS technologies have been evolved in recent years which can be categorized as electrical, electrochemical, chemical and mechanical storage systems. The widely used ESS are SC, SMES, flywheel, pumped hydro storage, batteries, CAES and hydrogen tanks. Among these technologies, batteries are treated as one of the most significant and promising ESS for maintaining the stability of power system networks [6]. Furthermore, ESS in the off-grid system plays a vital role in managing the momentary power fluctuations and quality of power. Based on the analysis of various ESS, their important characteristics are listed in Table. 1. Meanwhile, the well-known different types of batteries along with their suitable applications, limitations, and features are presented in Table. 2. Furthermore, ESS plays a significant role in the advancement of electric energy systems and extending RES to power the remote locations of the globe [7]- [10].
The global energy storage market is increasing substantially, since the last decade. It is predicted that the ES market may increase more than 26 billion USD in yearly sales by the year 2022, with a compounded annual growth rate of 46.5% [11]. Countries like Australia, the USA, India, UK, Chile, Japan, Italy, South Korea, and Germany are concentrating on the leading technologies of battery storage systems. Besides that, the national policies and subsidies provided by different countries further accelerate the growth of ESS. The combination of RES and ESS can reduce the dependency on the energy import, improve resiliency and reliability of the system and also help move towards decarbonization of grid. Furthermore, the developing nations are moving towards smart cities to achieve their goals such as environmental sustainability, adequate power supply, efficient mobility and the adoption of electric vehicles, where RES and ESS play a critical role. The capacity of ESS is exponentially increasing, which doubled during the year of 2017 and 2018 to 8 GWh [12]. The current global installed capacity of ESS is shown in Fig. 1. In addition to that, the adoption of ESS for various applications is presented in Fig. 2.
It is worth noting that, every ESS has its own limitations which confines its range of application, since an ideal application demands both high energy and high power. However, ESS are limited either by their power or energy capacity. Thus, it is necessary to build a system with a combination of  two or more ESS to form HESS [16]. For example, batteries have features of low specific power, high specific energy, less life-cycle, less capacity of self-discharge and less cost/Wh. On the other hand, the SC exhibits less specific energy, more specific power, fast charging, longer lifetime and high self-discharge [17]. Thus, the combination of battery-SC can utilize the complementary properties of each other. This combination has been popular due to its homologous working principle, ample availability, and low initial cost.
The benefits of implementing HESS are (see Fig. 3): • Minimizes the initial cost contrasted with a single energy storage system ( due to the decoupling of power and energy, the secondary storage system has to cover only the average demand for power).
• Enhances overall system efficiency.
• Improves storage capacity and lifetime of plant (minimizes the dynamic stress of the secondary storage system and optimizes the operation). Due to these credible features of HESS, numerous researchers and industrial experts have focussed on the  [14], [15].  development of HESS technologies for integration of RES into the grid. Therefore, this article collates various control techniques implemented for the development of HESS. Since control techniques are treated as the heart of the system, this article presents a detailed analysis of various control techniques, their benefits and limitations along with the future perspectives.
The remaining sections of the paper are organized as follows: The different converter topologies used for the HESS are discussed in section II. The detailed discussion on conventional control techniques of HESS and its comparative analysis is listed in section III. The intelligent control methods for managing the power flow between HESS devices are given in section IV and section V deals with the applications of HESS and discussion on case studies of real-time HESS plants implemented worldwide. Furthermore, the challenges in installation of HESS and future research directions are given in section VI and conclusions are made in section VII.

II. INTERCONNECTION TOPOLOGIES
The interconnection topology of HPS, HES dictates the control flexibility, dynamic performance, efficiency and lifetime of ESS. The HPS and HES can be connected to the system directly or through power converters. The direct connection offers a simple system architecture, low cost and control complexity. Albeit, the use of power converters ensures decoupled control of HPS and HES and offers enhanced power regulation. The HESS can be either connected to the DC bus or use a separate DC-AC converter to connect with the AC bus. The interconnection topologies can be classified into passive, semi-active and active. The selection of topology vastly varies based on the system requirement and the functions of energy management system. A critical analysis of interconnection topologies is presented in Table 3.

A. PASSIVE
Passive architecture is the simple approach to interconnect HPS and HES to the system (see Fig. 4). The ESS are directly connected together without employing power converters [18], [19]. The matching of voltage levels of ESS  with the DC bus voltage or the load voltage is a prerequisite for direct connection [20]. The passive architecture resembles parallel operating synchronous generators, where the load is shared based on the impedance ratio. Similarly, the load is shared based on the internal resistance of ESS and its output characteristics. Where the resistance is greatly affected by the temperature and the instantaneous state of charge.
For instance, in a battery and SC hybrid, the high power pulses are absorbed by SC due to its low impedance. Thus, SC is analogous to a low pass filter. Furthermore, the charging and discharging characteristics vary considerably which offers poor regulation of voltage and load. The power balance of the system is given by Eq. 1. Where P gen , P hess , P load are power generated, power exchanged by HESS (negative during power consumption) and power consumed by load respectively. The power exchanged by HESS is the sum of power contributed by HPS and HES as given by Eq. 2. P gen ± P hess = P load (1) P hess = P hps + P hes (2) Initially, this topology has been used for pulse loads [21], vehicular applications with battery-SC hybrid and later adopted to power system applications [22]. Lately, battery-SC with passive interconnection has been used to mitigate the intermittency of RES in isolated microgrids [23]. However, this topology did not gain much attention due to the following limitations: • No control flexibility and the power sharing varies based on the source impedance.
• ESS are vulnerable to cascaded failure during contingencies as they are directly connected to the system.
• The voltage of ESS should be strictly matched with the DC bus or load voltage.

B. SEMI-ACTIVE
This topology is the extension of passive topology with one power converter to control HPS. Where a bidirectional DC-DC converter is used to control the power exchanged by HPS along with an appropriate control algorithm (see Fig. 5). The power exchanged by HESS with this topology is given by Eq. 3. Where α denotes the controllability of HPS and determines the power share of HPS.
The peak power requirements of the system are satisfied by HPS and the remaining demand is met by HES. In a study, the semi active topologies with controlled SC and controlled battery have been analysed. The power converter of SC is oversized to handle the pulse power output. Whereas the topology with controlled battery has variations in DC link voltage [24]. So, relatively high energy storage is necessary to maintain the DC bus voltage whenever a HES is interfaced with a converter. Recently, a battery-SMES hybrid has been used to support the WES in an isolated microgrid. The SMES has been used to extend the lifetime of battery by absorbing high frequency power variations. Similarly, a battery-SC hybrid has been used to suppress the fluctuations of small WES. Where the SC has been interfaced using bidirectional DC-DC converter and the battery is directly connected to the DC bus [25]. Although it offers partial flexibility there are a few limitations as follows: • The DC bus voltage varies when a HPS is directly connected to the system.
• The DC-DC converter should be designed to handle the large power spikes when interfaced with a HPS.

C. ACTIVE
The active topology employs separate bidirectional DC-DC converter to control HPS and HES (see Fig. 6). This topology offers the highest possible controllability with the decoupled control of both the ESS. It facilitates the energy management strategy to exploit the complementary characteristics of HPS and HES. Furthermore, it accommodates the adoption of wide variety of control strategies. However, all these advantages come at the expense of increased power conversion losses and high cost of converters. The power exchanged by HESS with this topology is given by Eq. 4. The variables α and β represent the controllability of HPS and HES respectively. The control strategies determine these variables based on several factors like SOC, frequency of power variation and deterioration rate of battery.
By far, this topology has been widely utilized in the application of HESS in power system. This can be subdivided into parallel active and series active topologies. The parallel topology employs two separate set of converters for interfacing HPS and HES in parallel [26], [27]. Whereas the series topology has cascaded HPS and HES with power converter to decouple it from the DC bus [28]. The series topology is often neglected as it demands the power converter to be rated to the total power rating of HESS. So, parallel active topology has been widely used for power system applications. Various advantages of this topology are: • Improved flexibility with decoupled control of HPS and HES.
• A wide range of control strategies can be employed.  • The voltage levels of ESS are independent of system voltage.
• It has inherent fault tolerance capacity with converters decoupling ESS from the system

III. CLASSICAL STRATEGIES
Design and implementation of optimal controller is the most significant concern in HESS. The selection of suitable control technique for HESS relies upon an array of parameters.
To name a few, extension of storage lifespan, reduction of power intermittency, power quality, controller response time, expense of controller and structure of hybridization. Implementation of a suitable realistic controller technique is essential to accomplish consistent, effectual and safe operation of HESS. The implementation of EMS focus into two levels of architecture, a lower level control system which controls the DC bus voltage and maintains the current flow. Whereas the high level control focus on allocation of power, SOC monitoring and other objectives of the system. [29], [30]. The goals of implementing HESS for off-grid system include optimizing micro grid performance, improving stability of system and reducing the cost of operation (see Fig. 7). To achieve these goals, different techniques were proposed and they are classified in this article as classical and intelligent control strategies (see Fig. 8).

A. FILTRATION BASED CONTROL TECHNIQUES
The power transfer between ESS can be classified into HF and LF components. The HF components are the result of sudden variation in load or irregularities in power generation VOLUME 8, 2020  by RES. The LF components are the ones that occur during the regular behaviour of RES. HF components need ESS with rapid response time, whereas the LF components require ESS having high energy density [31]. FBC separates the power demand into low and high frequency components with help of filter circuit which leads to flattening of battery current variations. With this motivation, S.K. Kollimalla et. al., in [32] developed LPF based FBC EMS to control the charging levels of battery. The proposed controller has been evaluated to enhance the life span of battery. This method exhibits faster dynamic response and has minimised computational burden. The conventional HESS scheme along with regular FBC, novel FBC scheme proposed by authors in [32] are shown in Fig. 9. In addition to that, a FBC based on high pass filter is proposed by the authors of [33] are exhibited in Fig. 10.
The method based on rate limit controller for effective management of HESS is implemented in [27], [32], [34], [35]. This method is more efficient in controlling the battery charge and discharge conditions by producing optimal current reference signal for the HESS. It is less complex and well suited for isolated DC microgrids. Similar to the rate limit, multilevel control based EMS for HESS is proposed in [36], [37]. In this work, the authors considered different types of batteries to form a DC microgrid and used centralized controller. This centralized controller helps in enhancing system stability and consistency of power supply. It also has the features of SVR and autonomous SOC recovery.
With an aim to reduce the active power variations and to achieve better voltage regulation, MMC grid-tied HESS system has been proposed in [38]. In this approach, SC and batteries have been divided as upper and lower arms to decouple the power. The SC and batteries can mitigate the low and high frequency fluctuations, thereby achieving better reactive power regulation by SC. The proposed power decoupling strategy is shown in Fig. 11.

B. RULE BASED CONTROL TECHNIQUES
In general, RBC is based on the sequential decision making process pertaining to the control objective. The rules are defined either with system expertise or using mathematical models. Among various rule based approaches, thermostat is one of the simple and easier one. In this approach the high energy storage is constrained by the SOC limits. An enhanced method based on state machine control is introduced in [39], which can include numerous rules (to be characterized based on heuristic or experience of experts). In this work, the authors focused to implement a combination of fuel cell and battery with ratings (fuel cell power -1.2 kW and battery voltage 220 V DC, capacity 10 Ah). To optimize the power flow, system operation has been divided into 9 regions and accordingly rules have been framed to achieve effective control. Rule based techniques have been widely used due to its less computational burden, easy to implement and simple attribute [40]. RBC for off grid microgrid system with HESS is implemented in [41], [42]. In this work, the authors suppressed the fluctuations by filtering the wind speed and solar irradiation levels which is not practical in the real scenario. Thereby this method has limited implementation. RBC based on state machine control is shown in Fig. 12 C. DEAD BEAT CONTROL Deadbeat control works based on the model of the system. It generates the ratio of duty cycle to minimize error regulation in one control cycle. Thereby, it overcomes the state variable errors as well as effectively maintains the power sharing between ESS. Fast dynamic response and high control accuracy are the additional features of deadbeat controller. Furthermore, it also acquired the features of conventional controls like simple implementation and easier process involvement. In [43], the authors regulated the SC to respond for transient power demand and minimized the stress on battery to enhance its lifespan. The developed deadbeat controller is shown in Fig. 13.

D. DROOP CONTROL
A HPF based droop controller and VCD is incorporated together to regulate battery and SC in [44]. This hybrid controller is proposed to provide solutions for effective power sharing, avoid voltage fluctuations, and limit the SOC of battery and SC. Furthermore, in [44] they have implemented decentralized controller via VCD and VRD for SC and battery to limit low and high frequency components. Incorporation of SOC recovery loop is feature of this work, which smooths the dynamics of transient power sharing.
The other widely used type of droop controller for power management of HESS is the VID. It is developed with the combination of VCD and VRD controller. Due to this combination, the controller is able to manage high and low frequency power sharing among SC and battery. However, this method, fails to control voltage deviation. To nullify this effect, authors in [45] introduced SVR controller for batteries to manage the bus voltage. In addition to that, SVR is developed to act under fast recovery of SOC irrespective of the leakage current [45] and the control scheme is shown in Fig. 14. Furthermore, an EDC presented in [46]. This method is also developed with the combination of VRD and VCD to control the power flow of HESS in DC microgrid.  The authors focused to maintain constant power of battery by avoiding fluctuations of high frequency. The Fig. 15 shows the proposed control structure with the combination of VRD and VCD.
A dynamic droop control technique, with the consideration of frequency as a control function is proposed in [47]. The proposed system has been successfully implemented in the Uligam Island in Maldives. This proposed method, effectively manages the synergistic operation of two different kinds of ESS even though both (battery, SMES) has dissimilar integral behavior. The droop control is extensively used as it can control the different control units and generate precise droop values. The droop range selection for the battery and SMES is shown in Fig. 16.
The architecture of frequency control function used to propose dynamic droop controller is given in Fig. 17. Through rigorous research on droop control techniques, ref [48] proposed a novel ID control to improve the performance of VID. The proposed co-ordinated control between V-P and ID, fundamentally develops the HPF/LPF to acquire optimal dynamic power allocation between considered ESS. The dynamic power sharing can be accomplished at decentralized level which is the additional feature of ID. The control schemes proposed for ID and V-P are shown in Fig. 18 and Fig. 19 for a better insight.   The combination of cache control and adaptive droop is implemented in [49] to integrate PV and wind system with HESS. The cache control is preferred to combine dissimilar features of storage devices and an adaptive droop is used to achieve coordinated control of battery for long term operation. In addition to that, it also controls the online SOC, and schedules the SOC of BESS. The cache control scheme proposed is shown in Fig. 20.

E. SLIDING MODE CONTROL
Sliding mode control is a type of non-linear control which toggles between the control laws based on the state vector. Abeywardana et al., in [50] implemented improved SMC   technique by resolving the problem of variable switching frequency in conventional SMC. With this, authors minimized the power dissipation and design complexity of system. By combining the hysteresis control to the SMC an adaptive SMC is proposed in [51] with an aim of managing multimode HESS to overcome fluctuations in current. From the carried-out literature, the comparative analysis between conventional control strategies for HESS are tabulated in Table. 4.

IV. INTELLIGENT CONTROL TECHNIQUES
To overcome the limitations of classical control based EMS techniques, a real-time, intelligent based techniques such as adaptive PMP, meta-heuristics and MPC, are developed for the application of energy management in HESS. The systems implemented with these techniques were proven to have enhanced control performance, however, the time taken for the calculations at each cycle is high and the system cost is also high [57]. The detailed discussion on intelligent control techniques developed for HESS are discussed in following subsections.

A. MODEL PREDICTIVE CONTROLLER (MPC)
MPC is a plant process optimization technique that forecasts the effect of future control decisions on the changing state of the plant. It is an efficient and systematic technique used to deal with restricted multivariate control problems typically in process control of industries [58]. The greatest benefit of MPC is that it enables optimization of the current time frame while preserving the future time frames. MPC is likely to predict future incidents and can thus take corrective steps [59]. In [60] the authors implemented an off-grid based wind/PV HESS via MPC to observe the SOC, load variations and levels of hydrogen tanks. The simulation findings reveal that the projected battery capacity and total device performance in comparison to the state controller system have improved by 12.23% and 14.65% [60]. Furthermore, to control the discrete/continuous dynamics in HESS system, MPC with MLD system is proposed in [61]. By using MPC, future control inputs and future system responses can also be predicted and it exhibits optimal control scheme. It has the capability of controlling numerous control variables in a large scale system. The control scheme of MPC proposed by authors in [62] is shown in Fig. 21. A hybrid system combination of fuel cells and SC are controlled with a MPC in [63]. In this work, the SC are interlinked directly in parallel with load and the fuel cell is connected to the load via DC-DC converter. The main limitations of the work carried-out by the authors are listed as follows: authors did not permit the gradual current changes to the considered power sources, the performance of the system may be affected due to rapid current changes as only the first order model of SC is considered. While designing the controller no actual DC-DC converter is considered. Moreover, the system performance has been evaluated only through simulation studies. Its major drawback in the reliance of MPC on mathematical model of the system.
MPC strategy is proposed in [64] with a combination of SC and batteries as energy storage devices. The main features of the work implemented are: the proposed MPC technique ensures the SOC of battery and SC are maintained within the limits, it gives unique steps to deal with the controller. On the other hand, the limitations exist in this work are, since by using of two state space models for two converters, the system is highly computationally insensitive as opposed to classical MPC. In addition to that, this method is not capable of regulating DC bus voltage. By extending the same concept given in [64], authors in [65], proposed less complex controller for effective management of HESS. Even though this proposed method, behaves less complex, the consideration DC bus voltage as constant will not be applicable in real-time DC microgrid system. Therefore, the limitations has to be addressed for its implementation in microrgrids. The direct control of converters using MPC based EMS for HESS are shown in Fig. 22.

B. NEURAL NETWORK AND FUZZY LOGIC
In search of the trade-off between complexity and efficiency, other suitable methods for real-time EMS are proposed based on fuzzy logic, neural networks and RL [67]- [69].
Smart control strategies like FLC and ANN are way more effective than the conventional control strategies [70], [71]. Smart control strategies can enhance the system dynamic behavior without requiring the exact system model, yet they don't guarantee optimum performance [72].
ANN is a mathematical model which was developed with the desire to recognize and process parallel data. The ANN consists of several machine neuron layers. These layers are divided into three types; namely input, hidden, and output layers. Thanks to its non-linear and adaptive mechanisms, generalization abilities and design independence with regard to parameter systems, hence ANN is ideal for use in control systems [73], [74]. However, due to its 'black box' nature and the instructional issue in the network [73], [74], ANN lacks rules to define the structure (cells and layers). In a study, an ANN control system is designed to learn how to efficiently exploit system resources over time by adjusting the energy storage system's control strategy to change power production. However, the main disadvantage of a control strategy based on ANN is that the learning and tuning process needs historical data [64]. In [69] an ANN-based control strategy for an off-grid wind / PV power network with a HSS / battery HESS is proposed. The ANN-based management strategy uses a feedback loop to learn from new experiences and data in a new way. Compared with conventional methods, the proposed system has quick response [69]. However, the computational complexity depends on the number of data sets used to train the ANN [64].
On contrary to ANN, FLC is understandable and insensitive to parameter variations. In addition to that, the FLC does not need an exact program and training method model. The FLC algorithm is based on the rules and MF. In [75], the authors proposed FLC for an HSS / battery HESS wind / PV power off-grid network. The proposed control strategy aims to minimize the cost of HSS and regulate level of hydrogen and SOC of battery. The results show that the proposed program optimizes the consumption costs by achieving a cumulative savings of 13 per cent over the regular program with RBC [75]. Indeed, FLC's MF are typically calculated by a trial and error process. This approach takes time and does not ensure optimal results [70], [71], [73], [74]. As the number of variables increases, the process of MF optimization becomes more complex. A multi-mode fuzzy based controller is introduced by authors to overcome the limitations of haar wavelet method which fails to consider the impacts of SOC [54]. The proposed multi-mode fuzzy is verified under long and short term scenarios. As a short term ac side performance is observed and in long term scenario, the problems of battery aging and efficiency are improved. The aforementioned goals are achieved by allocating FLC to three power sharing modes.
The NFC takes advantage of the inference ability of FLC, learning and process data in parallel features of ANN. The NFC applies neural learning rules for defining and optimizing MF of FLC. This methodology increases the accuracy of the fuzzy model in a limited period of time. In [76] the REPS control strategy was launched with the ANFIS [76]. MF of FLC are tuned and configured using the back-propagation adaptive error process. In the large scale operation (one-year simulation), the proposed system is compared with the RBC. The results of the simulation show that the two systems can provide the necessary energy while holding the SOC battery within operating limits. The program proposed strategy increases the performance by offering higher efficiency for battery and hybrid system [76].

C. OPTIMIZATION BASED METHODS
In addition to the above discussed FLC and ANN based controllers few other modern optimization based control techniques were also implemented for effective management of HESS. As the bio-inspired algorithms have capability of handling multi objective functions effectively and provide optimal solution which is proved in various fields of applications [77], [78]. Therefore researchers introduced the combination of FLC with optimization algorithms for effective framing of rules, which are discussed as follows.
PSO is a computational approach that optimizes the problem by iteratively trying to develop a solution for a specific objective. In [79], the authors proved that the PSO optimized FLC can be used for an HSS / Battery HESS off-grid PV / Wind power network. The MF of FLC are calculated by PSO algorithm and the network proposed is combined with an optimized FLC system. Results from the simulation show that the O & M and LPSP costs of the proposed system are 57% and 33% lower than optimized systems [79]. With the motivation of PSO+FLC technique, WCA is proposed which is enlivened by the normal water cycle. In [80] off-grid HESS using WCA with FLC is proposed. MF of FLC are optionally tailored using WCAs to reduce LPSP and O & M costs. The combination of FLC with LPF control strategy is implemented in [53], where the LPF is used to attenuate high dynamic components of battery and FLC was focused to manage peak current of battery. To get better performance of the FLC, the membership functions are optimized via PSO. The control scheme proposed in [53] are presented in Fig. 23.
The proposed system has been successfully implemented and tested in real-time rural house hold applications. Chia Y Y et.al., in [81] proposed a predictive energy management system based on SVM for battery / SC HESS. The predictive SVM-based strategy correctly predicts the demand with a precision of 100% and a classification time of 0.004866. With a load prediction, the proposed system reduces the battery stress and prolongs the battery life by operating the SC at 200 ms before demand for maximum power takes place [81]. However, the amount of SC power is uncontrollable. The other innovative predictive management technique for a battery HESS is proposed for off-grid wind power system [68]. Based on the predicted wind and load profile, the strategy adapts the operation of the batteries and FC so that the network does not face blackout when the wind is weak or the energy reserves are inadequate [68]. However, the proposed predictive analysis approach has a pre-determined threshold which limits its application.
A hybrid on-grid HESS using predictive control is developed in [82]. The proposed system forecasts daily and weekly photovoltaic power output, wind speed, electric demand, electricity hourly price data and ambient temperature. The daily / weekly FLC is optimized by using SFL and PSO according to the predicted daily and weekly parameters. The simulation findings reveal that 9.27 per cent of SOC for weekly SFL-optimized FLCs, regular SFL-optimized FLCs, weekly PSO-optimized FLCs and regular PSO-optimized FLC are up by 17.8% and 16.89% respectively and weekly operating costs are raised by weekly SFL-optimized FLC, daily SFL-optimized FLC, weekly PSO-optimized FLC, and daily PSO-optimized FLC. The other optimization based EMS for HESS using MIAD is introduced in [83]. In this work, the authors proved that the proposed algorithm has less iterations to accomplish the objectives. The considered objectives are reducing fluctuations in battery current flow, diminishing the energy losses produced by SC.

D. UNIFIED CONTROLLER
A unified EMS for the DC microgrid with HESS is developed by Tummuru et al., in [84]. The proposed technique VOLUME 8, 2020 proved the extension of life span of SC and battery under various operating conditions. The performance achieved by the proposed system gives superior than the EMS implemented in [85] with respect to voltage regulation, charge and discharge rates of battery and processing time. The similar type of unified controller given in [86] has features of faster dynamic voltage regulation, effective power sharing under any sort of disturbances, reducing fluctuations in rate of charge/discharge of battery, and power quality enhancement. The developed unified controller is presented as a flowchart in Fig. 24. Supervisory power management for HESS with hybrid microgrid is implemented by authors in [87], [88]. In these works, the authors have focused on cooperation between grid and microgrid. In addition, the adequacy of HESS to deal with various conditions from the PV and grid were also analyzed. Another control technique named FJC is proposed in [26] for control of HESS in DC microgrids. In this system, the efficiency of the system is improved by the use of SC to cater for the unmet load by batteries.
FFSVM to manage the power flow between batteries and EDLC is proposed in [89]. An asymmetrical cascaded multilevel converter is used to manage the HESS. In each mode of operation, either battery or EDLC is chosen to give or receive the power and other storage devices are used to maintain the output voltage. As per the DC and AC voltage reference the FFSVM produces switching times and sequences of the state vector.
A hierarchical control strategy to control power flow between SMES and batteries is proposed in [55]. Empirical mode decomposition tool is used for signal analysis and to achieve effective power allocation to eradicate the time delay unlike traditional methods. The structure of hierarchical control is shown in Fig. 25. The usage of SMES is increasing in HESS than SC due to its features such as efficiency high, low self-discharge, long life span and high power rating levels. Based on the complete analysis carried-out form the literature, the different intelligent control techniques used for power management of HESS for the application of microgrids are tabulated in Table. 5.

V. CASE STUDIES AND APPLICATIONS A. REAL-TIME CASE STUDIES
A variety of HESS combinations are common in simulation software's such as HOMER and MATLAB, real life applications are not as common though. A lot of pilot and demonstration projects are currently taking place in different parts of the world. Limited information is known about these projects as the companies and governments involved don't provide detailed information about these projects until noticeable progress is achieved. Most of the real life HESS applications are for renewable energy sources integration, frequency control and back-up power. They are considered to enhance the operation of the grid and to save money for the clients by lowering energy wastage and improving efficiency. A few examples of real-time scenarios of HESS implementation are: States was installed by Duke Energy in 2016 to handle peak demand response, load shifting and support. The hybrid SC-battery HESS provides multiple service applications, extended operational life, rapid response, real-time solar smoothing and load shifting. Benefit of this combination is that it provides long term grid support and easily integrates renewable installations [91].
• A 11.5 MW Li-ion, NaS battery HESS commenced operation in 2018 in Varel Niedersachsen, Germany. The energy service provider EWE and its Japanese partners, Japanese industrial development agency NEDO, Hitachi Chemical, NGK Insulators and Hitachi Power Solutions worked on the unique project. The purpose of the HESS combination was to balance out frequency fluctuations from renewable energy sources in the regional electricity network. Li-ion battery will act as a high power provider with its ability of quick discharge while the NaS battery will act as a high energy provider which is suitable for long term storage [92].
• A 1 MWh VF/Li-ion battery HESS was developed for Monash University, Australia by redT energy storage solution in 2018. The largest behind the meter Commercial and Industrial ESS to be installed worldwide. The main purpose of the HESS is for storing and dispatching energy from multiple sources including a 1MW solar panels [93].
• A 20 MW brand new concept of power to heat/battery storage was developed in Bremen Germany by AEG power solutions. Energy is stored in a battery and an electrical heating system which are then connected to a converter. Its main function is frequency control for power operations and renewable energy sources integration [94].
An overview of world wide real-time implementation of HESS are tabulated in Table. 6.

B. APPLICATIONS
HESS is predominantly deployed in the following sectors Power sector, Transport sector and Renewable energy sector. The applications of HESS in electric power system include grid stabilisation, frequency regulation, backup power and renewable energy sources integration.

1) POWER SECTOR
HESS in the power sector is used for ancillary services like FR or VR and backup power for the grid. Traditionally single ESS were used for such services but due to their limited capability in terms of energy, power density and dynamic response, HESS became the better alternative [66]. Battery/SC and battery/flywheel are the most common combinations of power sector HESS. They are used in microgrids where lifecycle of batteries is increased due to less stress from charging and discharging. Overall HESS application in the power sector is for grid stability, FR, back up and power quality services.

2) TRANSPORT SECTOR
Peak power demand from sudden accelerations in EV draws a lot of energy from batteries, which means that the duration of power supply will be reduced and lifecycle of the battery will also be lowered. Increasing the battery capacity to deliver more power can solve the problem but this will be expensive and will add more weight and volume. Alternatively the HESS combination of a battery/super-capacitor or battery flywheel will effectively manage large fluctuations from sudden accelerations and regenerative breaking [97], [98]. The battery performance and lifetime will also be increased due to the HESS combination by reducing the charging/discharging rate and battery stress [99]. HESS is also used in electric trains to flatten fluctuations from various acceleration and deceleration cycles. They can be installed on the train or railway substation. They are utilized to reuse regenerative energy from vehicle breaking so that large power is feedback to the source during deceleration and power will be drawn during train acceleration [100].

3) RENEWABLE ENERGY SECTOR
HESS in renewable energy sector is mainly used to improve the power quality and system efficiency of renewable energy sources. HESS aids integration of intermittent renewable sources into the grid by providing long term energy balancing, short term power quality and frequency regulation services. It is used instead of a single ESS to meet peak power demand in some systems [100]. A HESS is a better solution in terms of cost effectiveness, practicality and durability for complete system implementation [5]. It is used to increase the storage and system lifetime by optimising lifespan and reducing dynamic stress on the high energy storage unit [101]. The noticeable observations and applications of HESS were tabulated in Table 7.

VI. CHALLENGES AND FUTURE DIRECTION
The investments on ESS is showing an increasing trend in the recent years. It is primarily dictated by the RES and microgrid infrastructure development in the present system. The batteries and SC are widely used for grid connected systems and EV applications due to its easy accessibility. In addition to that, the conventional ESS like pumped hydro storage and CAES were also used in wide range of applications. However, to improve the successful development of HESS technologies, the following challenges should be encountered: • Creating awareness among public for enhancement of HESS technologies towards achieving the reliable power supply to the consumer.
• Providing support on infrastructure development, installation and regular maintenance should be guaranteed to increase the installation capacity of HESS.
• Providing incentives for investment in HESS technologies is necessary which will bring attention towards the growth of HESS in the market.
• The upgradation of poor energy markets and weak grids is required which will attract the power distribution company towards deployment of HESS.
• The universal guidelines for HESS selection and its operating procedures should be improved according to the present technological development.
• The research on HESS technology is stagnant on laboratory scale with only theoretical perspective, so, development of projects to promote commercialization and industrialization of HESS technology is necessary.
• It is advisable to come up with HESS solutions with the support from researchers, economical advisors, electricity companies, consumers and social organizations.
The advancement of HESS needs modernization and breakthrough in long lifespan, minimal cost and high security. In addition to that, the following recommendations are worth addressing in near future: • An intelligent EMS controller with superior performance of HESS will facilitate the adoption of smart grids or microgrids in near future. So, HESS can enable the development of microgrids ranging from medium scale to large scale towards the transition of smart-grid.
• The innovative approach of ''energy internet'' for future energy supply and distribution system will highly depend on the performance, flexibility and reliability of HESS [108], [109]  • Development of new battery technologies namely aluminium, lithium-air, sodium-ion and graphene are need to replace with current batteries with substantial development in lifespan and performance [110], [111].
• Advancements in battery technologies will further increase the adoption rate of HESS for grid connected applications [112].
• Novel combination of ESS across different medium (mechanical, thermal) will widen the options of HESS for various applications. For instance, hybridization of fast responding high power ESS with high energy ESS like CAES, thermal energy storage and pumped hydro storage can be developed.
• The development of new control strategy for HESS via multi objective optimization by taking into consideration of technical and economical limitations is yet to be targeted.
• There exist inadequate research works have been carried-out using predictive controllers for implementation of HESS, with the improvement of these techniques for HESS, penetration of renewable energy can be further increased.
• As authors proposed in [113], a reconfigurable ES bank to change the configuration of battery bank dynamically to minimize the load on each battery and enhances the performance of HESS and give long life span. Performing research on these type of system bringing these into regular usual market is challenge in the present world.

VII. CONCLUSION
In this article, an extensive review of various control strategies implemented for HESS is presented. The different control strategies used for autonomous and grid connected microgrids are elucidated. The recently published articles on HESS are critically reviewed and noticeable characteristics of each control technique are summarized in tables. In addition to that, a detailed discussion of the pros and cons of intelligent and conventional control techniques are also presented.
As an added value, HESS adopted in real-time applications throughout the globe is analyzed and discussed. Finally, a pathway for the future researchers to carry out their research in this area is laid out, and the main challenges faced towards the implementation of HESS is presented. Therefore, this article helps the researchers and practitioners to have a complete idea on control technologies implemented for better coordination of ESS forming the HESS, thereby driving the penetration of RES into the grid.