An Effective Control Approach of Hybrid Energy Storage System Based on FLC/Grey Wolf Optimisation

In the modern era, the integration of renewable energy sources (RES) has bolstered the autonomy of urban energy infrastructures, reducing reliance on distant sources and grids. Batteries serve as a vital bridge between power supply and fluctuating load demands within RES systems. However, the unpredictable nature of RES behavior and varying load requirements often subject batteries to repeated deep cycles and irregular charging patterns. These cycles diminish the battery’s lifespan and escalate replacement costs. This study presents an innovative control strategy for a Solar-Wind model featuring a Battery-Supercapacitor Hybrid Energy Storage System. The objective is to prolong the battery’s operational lifespan by mitigating intermittent strain and high current demands. In contrast to conventional methods, the proposed control approach incorporates a Low-Pass Filter (LPF) and a Fuzzy Logic Controller (FLC). Firstly, the LPF minimizes the oscillations in battery consumption. Simultaneously, the FLC optimizes the high current demand on the battery while vigilantly monitoring the supercapacitor’s charge levels. Moreover, Grey Wolf Optimization (GWO) is employed to fine-tune the FLC’s membership functions, ensuring optimal peak current attenuation in batteries. The effectiveness of the proposed model is benchmarked against standard control techniques, namely Rule- Based Controller and Filtration-Based Controller. Comparative analysis reveals that the proposed method substantially reduces peak current and high power requirements of the battery. Consequently, this enhances the utilization of the supercapacitor, significantly augmenting the battery’s operational life. The results demonstrate a remarkable improvement over conventional systems, emphasizing the potential of this approach in optimizing energy storage systems for sustainable, long-term performance.

Introduction

In Renewable Energy Systems (RES), batteries play a crucial role in bridging the power gap between supply and varying load demands. However, due to the unpredictable nature of RES and the fluctuating power requirements of loads, batteries often undergo frequent deep cycles and irregular charging patterns, leading to a shortened lifespan and increased replacement costs [1]–[3]. To mitigate these challenges, Hybrid Energy Storage Systems (HESS) have emerged as a viable solution, reducing battery strain, capacity, and overall system costs [4].

Managing the power dynamics of both batteries and supercapacitors (SC) is essential to maximize energy consumption and conservation in the Battery-SC model. Aiming to increase battery lifespan by reducing peak current needs and constant stress is a common objective in HESS deployment [4], [5]. By minimizing peak current, voltage drop is lessened, enhancing battery efficiency and reducing overheating and dynamic stress [5], [6].

While fuzzy logic controllers (FLC) are widely used in Battery-Hydrogen Storage Systems (HSS), their application in RES with Battery-SC Systems is relatively unexplored. FLC, known for its simplicity and insensitivity to system changes, minimizes battery capacity loss significantly when compared to battery-only systems [7]. However, the challenge lies in optimizing FLC’s membership functions (MFs), traditionally done through time-consuming trial-and-error methods [8].

In this research, a novel control model for a PV-Wind power system with Battery-SC is proposed. The control approach aims to reduce the battery’s fluctuating active strain and high current demands while ensuring optimal state of charge for the supercapacitor (SOCsc) [8]. Employing a Sugeno-type FLC, known for its efficiency and robustness, the model adjusts the power flow between the battery and SC based on real-time power needs and SOCsc. To enhance FLC’s performance, the Grey Wolf Optimization (GWO) algorithm is employed, optimizing the FLC’s membership functions to decrease high battery current.

Simulink is utilized to compare the proposed system’s performance against traditional systems, using a rural household load profile. The results showcase the efficiency of the proposed control model, demonstrating its potential in significantly improving the performance and lifespan of Battery-SC systems. This paper presents the Simulink models of the suggested system, highlights the GWO optimization results, and compares the outcomes with conventional systems, shedding light on the promising future of optimized Battery-SC RES applications.

System Modelling and Components

A solar panel, a wind turbine, and a battery (lithium-ion) are all part of the proposed hybrid system, which also includes an SC. As indicated, all of the elements are coupled to a voltage unified DC bus [9].

Wind Turbine Modelling

In this study, we’ll use a wind turbine powered by a permanent synchronous generator (PMSG). The wind turbine’s power and rated wind velocity are 300 W and 13 m/s, respectively. The maximum wind velocity is 18 m/s, with a beginning wind velocity of 3 m/s. The maximal power production is 800 W [10].

The output power of a wind turbine can be stated as follows using aerodynamic theory: (1)P=0.5ρπR2V2Cmwp(λ,β)where P is the resultant power of a wind turbine generator. The air density, wind speed and blade radium are represented by, R, and V_w, discretely. The power coefficient (C_p) is defined as: (2)Cp(λ,β)=0.73(151/λ−0.58β−0.002β2.14−13.2)e−18.4/λwhere: (3)λ=11/((λ−0.02β)−(0.003/β3+1))

In Matlab, a fluctuating wind of 13 m/s is generated and shown. The wind turbine’s output power under MPPT control is displayed [11].

PV Array Modelling

A solar array is formed by connecting numerous parallel and series solar cells. The following equation is used to compute the short-circuit current for every solar cell [11]: (4)ISC=ISC0(G/G0)awhere I_sc and I_sc0 are the short-circuit currents, G and G₀ are standard solar radiation, respectively.

The PV cell’s open-circuit voltage is: (5)V0c=Voc01+βln⁡GG0(T0T)αwhere V0c and Voc0 are the open-circuit voltages under average and normal solar energy G and G₀, respectively, and T is the PV cell temperature. β is a technology-specific coefficient for PV cells that depicts temperature nonlinear effects. Using cells arranged in series (N_s) and cells arranged in parallel (N_p), the highest power from the solar array can be written as [12]: (6)Pmax=NpNs(V0c0/nkTq)−ln⁡((V0c/nkTq)+0.72)1+(V0c0/nkTq)×1−RsV0c/IscV0cIsc

Because both the wind turbine and the solar array create power, the collective power produced may be estimated by summing them together. Furthermore, we presume a constant load need of 1 kW. As a result, the power disparity between produced and needed power by load is P [13]. (7)ΔP=PWP+PPV−PLwhere P_WP, P_PV, and P_L are the power output of a wind turbine, power of the solar array, and power demand of the load, discretely [14].

Battery–SC Storage System

An SC, a bidirectional DC-DC converter, control circuitry, and a battery bank are all included in the proposed system’s HESS, as shown in the Simulation of the solar-wind system with Battery-SC system [15]. The Simulink library contains the structure and details of the Battery and Supercapacitor models. The Battery-SC system described in this work uses the benefits of both high power density and energy density storage to accomplish the necessary performance [16]. However, combining the battery-SC as a solitary source of power necessitates a sophisticated conditioning circuitry [17]. Because the SC voltage varies greatly because of its low energy density, the proposed Model’s Battery-SC system is built in a partially active architecture, with a bi-directional DCDC converter located adjacent to the Supercapacitor to isolate the Supercapacitor from the rest of the system [18].

The power electronic component is made up of a DC-DC converter and control circuits [19]. This design provides enough flexibility to implement a variety of control techniques as shown in Fig. 1. Furthermore, with the employment of a single DC-DC converter, this design provides a favourable trade-off between functionality and intricacy of the circuit [20]. It is also necessary to manage the power flow in both directions to properly link the batteries and SCs. The System’s specifications are shown in Table I.

Fig. 1. Hybrid energy storage and renewable energy system.

Table I. Proposed System’s Specifications

Component	Rating	Value
PV	Power	1.2 kW
Wind system	Power	300 W
Li ion battery	Voltage	48 V
	Capacity	300 Ah
Supercapacitor	Voltage	45 V
	Capacitance	500 F

The bidirectional half-bridge DC-DC converter can operate in both buck and boost modes. It comprises of two bi-positional switches constructed in a half bridge arrangement using transistors S1/S2 and diodes D1/D2 [21]. The converter’s high and low voltage sides are interfaced to the DC bus and the Supercapacitor, to enable for constant functioning of the SC [22]. Power is transmitted from the high to the low voltage side when the converter is in buck mode (inductor current, iL > 0). Power is transmitted from the low to the high voltage side (iL < 0) while the converter is in boost mode. The following is a description of the proposed system’s dynamic power balance, based on Fig. 2:

Fig. 2. Simulation inputs: (a) Solar power output, (b) Wind speed, (c) Output from wind turbine, (d) Load profile, (e) Power variation between Power generation and Load demand.

(8)PPV+Pbatt+PSC′+PLoad=0where P_PV is the PV’s power generation, P_batt is the battery’s power, P_load is the load’s power demand, and P_sc0 is the SC power flow after the DC-DC converter’s power conversion. The converter’s efficiency is less than 100% in real-world operation.

As a result, the power shift from the SC to the DC bus is written as expression (9): (9)PSC′=ηDCDC×PSCwhere P_SC denotes SC’s power flow and η_dcdc denotes the converter’s efficiency. The converter’s efficiency is presumed to be 100% in the context of this research.

As a result, (10) defines the amount of power needed by the Battery-SC, ΔP, which is the difference between P_PV and P_load’s output power: (10)ΔP=PPV−PLoad=Pbatt+PSC

Control Strategy

The control strategy, which is based on current system conditions, controls the HESS power flow. In order to accomplish various objectives is frequently complicated and calls for constant operation. Optimizing HESS control is essential for maximizing sustainability and energy efficiency [23]–[28].

There are two different categories of control strategies: traditional control techniques and modern control techniques [29]. Rule-based controllers (RBC) and filtering-based controllers (FBC) are examples of traditional control systems that are simple and easy to implement since they don’t need intensive processing. However, they tend to be rigid and sensitive to parameter changes [30]. Because they develop the dynamic behavior without needing an accurate description of the system, modern control techniques like the FLC are more dependable and effective than traditional control strategies [31]. The MFs of FLC, on the other hand, are normally calculated through trial-and-error, which is ineffective and a lengthy process.

To summarize, REPS with Battery-Supercapacitor HESS frequently use traditional control mechanisms like RBC and FBC. In this study, the proposed control approach is compared to two common traditional control systems (RBC and FBC). All models with SC have a SOCsc operating range of 50% to 100% to enable the use of 75% of the total SC energy [32].

Rule-Based Controller

Based on a set of guidelines, the RBC determines how much power is distributed between the SC and battery. It is easy to implement because it does not necessitate complex processing. RBC, on the other hand, is extremely sensitive to parameter variations due to its pre-defined rules and procedures. An RBC is created and depicted as a dead-zone function using data from [33]. Whenever the current of the battery is between the ib1 and ib2 thresholds, the battery is the sole way to supply the load requirement. when the ib1 or ib2 threshold is reached for battery current demand, the extra current demand is divided between the SC and the battery according to the K1 or K2 ratio [33].

Filtration Based Controller

The FBC splits the active elements of the power demand into low-frequency and high-frequency parts using a filter. This method is easy and requires little computing power. A high pass filter (HPF)-based FBC’s structure was borrowed from [34]. The HPF divides the power requirement into high-frequency (PHF) and low-frequency (PLF) components, with the PHF and PLF fetched by the battery and SC, discretely.

Proposed Control Strategy

Fig. 3 depicts the framework of the suggested control method, which attempts to reduce the battery’s active tension and high current demand. The energy management device in the control strategy is a fuzzy logic controller (FLC). The Grey Wolf Optimization optimisation (GWO) technique is used to optimise the FLC’s membership functions (MFs) in order to obtain optimal performance. The next sections discuss the structure of the suggested control strategy.

Fig. 3. Proposed control strategy.

DC Bus Configuration and Optimization

The voltage on the DC bus is managed using the principle expressed in Fig. 3. The PI device decides the reference current of the DC bus Idc_ref to regulate the bus voltage at V_ref = 400 V.

The EMS generates the reference currents for batteries and SCs (Ibat_ref and Isc_ref, correspondingly). These reference currents makes the DC bus voltage to remain static irrespective of load conduct or fluctuations in power output. When a fault develops on an element such as SOC, power interruption, or solar irradiation variation, the batteries and/or SCs make sure that the DC bus voltage is regulated. The collective reference currents, Isc_ref and Ibat_ref, should be same as Idc_ref at all times: (11)Idcref=Iscref+Ibatref

The following equation can be used to simulate the function of a DC bus: (12)Cdcdvdc/dt=Iscdc+Ibatdv−Iload

The DC currents of SCs, batteries are represented by ISC_dc, Ibat_dc, respectively. The load current is represented by I_Load. C_dc is the capacity of the central bus that allows a similar DC bus voltage to be imposed on the load and all other inputs. The power oscillations from the chosen stationary converters are filtered by these capacitors.

Low Pass Filter (LPF)

PV power production and load needs are very variable in actual operation. To meet the demands of the traditional system’s highly fluctuating ΔP, the battery is put under stress. The extremely variable battery current results in high internal heat of the battery, which would decrease efficiency and increase internal resistance [4], [6]. LPF is, therefore, used to break down the ΔP into P_HF and P_LF in order to lessen the active load on the battery. The P_HF is the mismatch between ΔP and P_LF, while the P_LF is the resultant of LPF. (13)PLF=lowpassfilter(ΔP) (14)PHF=ΔP−PLF

While the SC should ideally handle the P_HF, a highly variable power demand, the battery should ideally handle the P_LF. This procedure lessens the dynamic stress on the battery by preventing it from delivering the high frequency components. The P_LF is sent to the FLC to achieve battery high current attenuation after the LPF filtration.

Fuzzy Logic Supervisor (FLS)

FLC’s goal is to lower battery high current while continuously taking the SOCsc into account. The Sugeno type fuzzy system, which is utilized in this study, is an effective system for computing that functions best with optimization and adaption [35].

The PLF and the SOCsc are the FLC’s two inputs, as depicted in Fig. 3. The power sharing ratio, which is calculated using the real-time data, is the FLC’s output. The inputs’ MFs have trapezoidal forms. In Fig. 4, the MFs of the FLC are presented.

Fig. 4. Membership functions: (a) Input 1: P_LF, (b) Input 2: SOCsc, (c) Output: α.

The five MFs present in the input variable PLF are positive high (“PH”), positive medium (“PM”), low (“L”), negative low (“NL”), and negative high (“NH”). The HESS must meet the positive P_LF, which is the power demand, and the negative PLF, which is the excess power that must be taken in by the battery-SC system. Contrarily, the input variable SOCsc only has three MFs, denoted by the letters High (H), Medium (M), and Low (Low). The output variable α, meanwhile, has five MFs: PH, PL, zero (“Z”), NL, and NH. The power ratio that will be delivered to and taken-in by the SC is indicated by the positive and negative membership functions, respectively.

Table II includes a list of FLC regulations. Regardless of the SOCsc situation, when the P_LF’s power demand is “L,” the power sharing ratio (α) is “Z,” as the feable power demand places low strain on the battery. To lower the battery’s peak current demand whenever the P_LF is positive, α is fixed according to power demand intensity and SOCsc. To restore the SC’s charge when the P_LF is negative, α is determined with extra power and SOCsc level. Equation (15) can be used to calculate how much electricity the SC and P_H will share:

(15)PH=αPLF

Table II. Rules of FLC

		ΔP
		PH	PL	L	NL	NH
SOCSC	H	PH	PL	Z	Z	Z
	M	PL	PL	Z	NL	NL
	L	Z	Z	Z	NL	NH

PHF and PH are added to determine the overall power that the SC will supply, or PSC∗. (16)PSC∗=PHF+PH

The bidirectional DC-DC converter in the proposed system would control SC power flow in accordance with PSC∗. Therefore, it is anticipated that the battery will deliver the power discrepancy between PSC∗ and ΔP as specified in expression (17): (17)Pbatt=ΔP−PSC∗

Grey Wolf Optimization (GWO) Algorithm

The Grey Wolf Optimization approach is based on the transverse orientation mobility of ants in environment. While exploring their environment, real ants leave pheromone trails that direct each other to resources. Similarly, the simulated ants record their positions and the quality of their solutions in the same way so that more ants find better solutions in subsequent simulation iterations. Therefore, during each flight, Ants function as particles that change their positions and velocities based on their own A_best as well as the collective group’s G_best experience (iteration).

The amount of variables within every problem decides the size of the particles. A fitness function calculated at the particle’s current position determines the quality of the solution for each particle.

The FLC may be optimised based on the projected data since the electrical load, and RES can both be forecasted. In this investigation, the load demand and RES profile presented in Figs. 2d and 2e are taken as the baseline model. The anticipated RES data and load demand are used to optimise the MFs of the FLC’s input variables. Four points together form a complete trapezoidal MF. The first (left) MF and the last (right) MF of a variable with more than two trapezoidal MFs each have just two tuning points. As a result, the equation can be used to determine the number of points that are optimal for a given variable, n (18):

(18)n=(4×mf)−4where mf is the number of MF, in this case. PLF and SOCsc, two input variables, have five and three MFs, respectively, in the suggested control approach. Because of this, the GWO algorithm must optimise 24 points altogether. In light of this, A 24-dimensional particle can be used to represent an FLC or solution. The population size and the number of iterations are user-defined in the GWO procedure, as shown in Fig. 6.

Fig. 6. Flowchart of GWO process.

Fitness Function

The battery’s high discharge current causes extreme damage since it raises the battery’s temperature and the pace at which positive active mass sludges. The major goal is to reduce the battery high current while keeping the SC above 50% of SOCsc. The system’s cost can be decreased while also increasing battery efficiency and lifespan due to the reduction in battery peak current [35]–[37]. The fitness function, f, evaluates the solution’s fitness in GWO. In this study, the minimum battery current, denoted by f(x), is defined in (19):

(19)f(x)=min(Ibatt)where I_batt stands for the battery current. The highest battery current of every function is noted for the GWO algorithm to find the best one in each iteration.

Results and Discussion

GWO Algorithm

Based on the load profiles depicted in Fig. 2, GWO optimises the MFs of FLC. The flowchart for GWO optimization is shown in Fig. 6, where a population size of 20 ants and a total of 80 iterations are specified. Each 24-dimensional particle represents an FLC model. Everywolfin, the population is regulated using the fitness function, which is described in (19). After each evaluation, the ant’s individual best (A_best) and overall best (G_best) are updated.

The fitness value versus iteration count graph is shown in Fig 5. Using a randomly generated solution, the first iteration’s G_best is 6.0389 A. As iterations go longer, the fitness value gets lower. The G_best is maintained at 5.718 A through the 80^th iteration after being decreased to that value at the 60^th iteration. By the time the optimization procedure is complete, the top choice (the particle with G_best of 5.718 A) has been converted into an FLC model, as shown in Fig. 4.

Fig. 5. GWO algorithm’s convergence curve after 80 iterations of optimization.

Simulation

Simulink is used in this study to build the three models stated in the Table III. The solar-wind system with battery-SC model and RBC is referred as Model I. The Solar-wind system with battery-SC model and HPF-based FBC is named as Model II. The hybrid PV-wind system with battery-SC model and the suggested control approach makes up Model III. Table III depicts the configuration of the control strategies. The power source and load profiles are modelled and adapted to all models, as shown in Fig. 2.

Table III. Model Configurations

Model	Energy source	Energy storage system	Control strategy
I	PV/Wind	Battery/SC	RBC
II	PV/Wind	Battery/SC	FBC
III	PV/Wind	Battery/SC	FLC/GWO (Proposed control strategy)

A number of battery metrics are assessed, including average battery SOC (SOC_{batt average}), peak battery current (I_{batt peak}), peak battery power (P_{batt peak}), and final battery SOC (SOC_{batt final}). Lower battery stress, greater battery efficiency, and a decrease in internal voltage are all effects of reducing I_{batt peak} and P_{batt peak} [4], [5]. In this study, the SOC_{batt average} and SOC_{batt final} are examined. Higher SOC_{batt average} and SOC_{batt final} would increase the life of battery and decrease system LPSP.

|ΔP| stands for the absolute rate of change of power in a time step of dt, expressed in Watt per second (W s1). Equation (20) can be utilised to calculate |ΔP|:

(20)|ΔP|=|(P(t)−P(t−Δt))/Δt|where P(t) denotes battery power at time t, P(t−Δt) denotes battery power at time t−Δt, and Δt is the study’s 1 s time step. In other words, |ΔP| can be used to estimate the battery power’s amount of fluctuation, with a greater value indicating a higher level of variation. Low levels of battery power fluctuation can boost the battery’s efficiency and life expectancy.

This study computes the battery power’s maximum |ΔP| |ΔP_max| and mean |ΔP| |ΔP_mean| to assess the battery's level of dynamic stress.

The simulation model of all battery profiles simulation are shown in Figs. 7a–7c. Table IV compares and summarises each model’s battery performance.

Fig. 7. Battery power (a) Battery power—Model I, (b) Battery power—Model II, (c) Battery power—Model III.

Table IV. Summary of Battery performance

Battery parameters	Value	Model I	Model II	Model III
I batt_peak	Current (A)	6.053	6.152	5.718
	Attenuation (%)	–	−1.02	5.9
P batt_peak	Power (W)	293.6450	296.5788	275.4768
	Attenuation (%)	–	−1.01	6.19
SOC batt_average	SOC (%)	67.3008	66.7895	67.2767
	Increase (%)	–	0.76	0.4
SOC batt_final	SOC (%)	47.8976	47.7657	47.8734
	Increase (%)	–	0.01	1.15
|ΔP| max	Rate (ws-1)	24.1243	8.7685	7.4899
	Attenuation (%)	–	74.17	77.01
|ΔP| mean	Rate (ws-1)	0.346	0.017869	0.01567
	Attenuation (%)	–	95.01	95.59

The I_{batt peak} and P_{batt peak} are decreased for Model II, according to Fig. 7b and Table IV, but the battery still endures a significantly variable power demand. As the SC expends most of its energy to deliver the load, the SOC_{batt average} and SOC_{batt final} are increased by 0.76 percent and 0.01 percent, discretely. These improvements are the largest of all the models. Table IV and Fig. 7b for Model II demonstrate a notable decrease in the active strain of the battery but only a marginal improvement in I_batt peak and P_batt peak. This is due to the FBC, which was created to lessen the dynamic load on the battery without taking peak demand into account. However, the SOC_{batt average} and SOC_{batt final} are not significantly developed (0%), as the SC only absorbs the highly fluctuating low power components.

As shown in Fig. 7, the battery power profile for Model III is noticeably smoother than that of Models I and II. SC compensates for the difference between ΔP and battery power.

As previously stated, one of the suggested model’s objectives is to attenuate the battery’s peak demand. The maximum battery current in Model III in the simulation is 5.718 A, which is identical to the best because the GWO optimization and simulation use the same energy source and load profiles.

Due to the FLC’s inclusion in the suggested control plan, the SC delivers to satisfy the high demand while continuously taking its SOC level into account. As a result, when compared to Model I, the I_{batt peak} and P_{batt peak} are decreased by 5.9% and 6.19%, discretely.

Since the SC expends the majority of its energy achieving the control strategy’s objectives, the SOC_{batt average} and SOC_{batt final} are enhanced by 0.4% and 1.15%, discretely which is significantly less than Model I.

By determining the absolute value of the total ampere-hours (amount of charge) travelling to and from the SC, |Ah|SC, it is possible to estimate the SC utilisation in the HESS. The SC current is integrated over time to determine the SC ampere-hours.

The SOCsc and |Ah|SC of Models I, II, and III are shown in Fig. 8 during the simulation. The level of battery use decreases as SC utilisation increases. As a result, by raising the SC utilisation level, the system’s internal losses can be decreased. By boosting SC use, the system can be made smaller. This is because a significant portion of the resultant current passes via the SC, which has a lower internal resistance and, as a result, causes the battery to heat up less and last longer.

Fig. 8. Supercapacitor rating: (a) Accumulated ampere hour of supercapacitor, (b) State of charge of capacitor.

According to Fig. 8b, Model I’s final SOCsc is same as the required minimum SOCsc of 50%. On comparison, Model I performs the poorest in terms of SC utilisation and battery active strain reduction, meaning that it does not properly use the SC. It also consumes the majority of the SC energy. The result is that Model II’s total |Ah|SC is 32.65% less than Model III but 487.86% more than Model I. Only 0.56% net SOCsc is used during the simulation, resulting in a final SOCsc of Model II of 92.69%.

In other words, Model II’s SC is not fully utilised. In Model III, the final SOCsc is kept at a greater level than in Model I while being kept 7.45% over the required minimum SOC of 50%. In compared to the other models, Model III has the greatest total |Ah|SC, with Model I and Model II’s respective SC utilisation levels being 687.122% and 32.65% lower.

In other words, the suggested technique can run the SC under the specified SOC range and effectively utilise the SC’s constricted energy limit to produce promising performance.

The suggested system (Model III) is tested using the load profile depicted in Fig. 2b and various weather conditions. As would be expected, the scenario’s most important changes are the decrease in Peak demand and strain level of the battery.

Regardless of the weather and load profile, the simulation results demonstrate that the suggested method greatly attenuates the battery’s active strain ΔP_mean by more than 80% when compared to Model I. In the meantime, the decrease in battery’s high need depends on the stress level of ΔP, which is established by the power source and the load demand. With the high ΔP and low PV output or peak load demand, the battery peak demand can be significantly reduced [38].

Conclusion

In conclusion, this research underscores the pivotal role of innovative control strategies in enhancing the longevity and efficiency of energy storage systems, particularly in the context of modern urban energy infrastructures reliant on renewable sources. By addressing the challenges posed by the unpredictable behaviour of renewable energy sources and varying load demands, the study presents a significant step forward in sustainable energy solutions.

The proposed control approach, integrating a Low-Pass Filter (LPF), Fuzzy Logic Controller (FLC), and Grey Wolf Optimization (GWO), stands out as a pioneering method to tackle the detrimental effects of deep cycles and irregular charging patterns on batteries. Through the meticulous optimization of FLC’s membership functions using GWO, the system achieves optimal peak current attenuation, reducing the strain on batteries substantially. The simultaneous monitoring and adjustment of the supercapacitor’s charge levels further contribute to maximizing the operational life of the entire hybrid energy storage system.

Comparative analyses against traditional control techniques, such as Rule-Based Controller and Filtration-Based Controller, demonstrate the superiority of the proposed method. Substantial reductions in peak current and high power requirements not only alleviate strain on the battery but also significantly enhance the utilization of the supercapacitor. This improvement not only augments the system’s efficiency but also mitigates replacement costs, contributing to the long-term sustainability of energy storage solutions.

In essence, this study exemplifies the potential of integrating advanced control methodologies and optimization techniques in shaping the future of energy storage systems. By extending the operational life of batteries, this research paves the way for more reliable, cost-effective, and environmentally friendly energy infrastructures, fostering a greener and more sustainable future.