In the upcoming section, we will first prove the superiority of the suggested MPO algorithm in optimization and, then, it will be applied to the system, for further clarification.
4.1. Validation of the MPO Algorithm
The proposed MPO optimizer is applied to 10 renowned benchmark functions, including multimodal and unimodal problems. This is carried out in the MATLAB R2017b environment. The outcomes of the solutions produced via the suggested approach are then contrasted with some other cutting-edge algorithms, including the original pelican optimizer (PO), pigeon-inspired optimizer (PIO), biogeography-based optimizer (BBO), and lion optimization algorithm (LOA). Diminishing all of the ten performance indices to the least amount is the key goal. Therefore, the lowest number that any algorithm can attain demonstrates that it is more efficient than the others.
Table 2 represents the values of the variables for the studied methods.
The variable values in the algorithms are often set based on heuristics and empirical studies. These values aim to strike an equilibrium between local and global search abilities, ensuring an efficient exploration of the solution space. However, we attempted to provide the same parameter values for the population and the iteration number, to achieve equal conditions in the comparison. To verify the effectiveness of the algorithms, every optimizer is independently operated thirty times on every function, and the STD and mean amount have been confirmed, based on these administrators, to produce consistent outcomes. The test function evaluations among the algorithms are given in
Table 3.
Based on the provided table, we can quantitatively discuss the results, and why the proposed MPO algorithm is better than the others. Looking at the average (AVG) values, we can observe that the MPO algorithm consistently outperforms the other algorithms across all benchmark functions (F1 to F10). The MPO algorithm achieves AVG values of 0.00 for F1, F3, F4, F7, F8, F9, and F10, indicating that it finds the optimal solutions for these functions. In contrast, the other algorithms have higher AVG values, indicating a suboptimal performance.
Comparing the STD values, the MPO algorithm generally demonstrates lower variability in its solutions compared to the other algorithms. Lower STD values indicate more stability and consistency in the optimization results. Notably, the MPO algorithm frequently attains STD values of 0.00, suggesting highly reliable and consistent solutions. The superiority of the MPO algorithm may be attributed to its specific modifications and enhancements, as mentioned earlier.
These modifications are likely to improve its ability to explore and exploit the search space more effectively, leading to superior optimization results. In summary, the MPO algorithm surpasses the other algorithms in terms of both average performance and stability. It consistently achieves lower AVG values, indicating a better optimization capability, and exhibits lower STD values, indicating robustness in its solutions. These quantitative results suggest that the proposed MPO algorithm is more efficient and reliable in solving the optimization problems considered in this study.
4.2. Simulation Results
In this study, an on-grid PEMFC has been used for evaluation.
Figure 5,
Figure 6 and
Figure 7 display the PEMFC’s main features, including its power, voltage, and efficiency. As demonstrated in
Figure 5, the current–voltage curve comprises two portions in the lower and higher current grades.
Figure 5 specifically shows the current–voltage curve of the PEMFC. This curve exhibits two distinct portions in the lower and higher current ranges. The lower portion of the curve is logarithmic, as indicated by the figure. This logarithmic behavior is mainly due to the activation voltage requirements of the PEMFC. At lower current levels, the activation voltage dominates the overall voltage response. However, as time progresses and the current levels increase, the ohmic overvoltage becomes more significant. The ohmic overvoltage arises from the resistance within the PEMFC system. This resistance includes various factors, such as the ionic conductivity and contact resistance. As a result, the linear behavior of the current–voltage curve gradually emerges at higher current levels. The transition from the logarithmic to linear response can be observed in
Figure 5. These characteristics of the PEMFC are important to consider in the evaluation and optimization of its performance. Understanding the behavior of the current voltage curve helps in designing and fine-tuning the operational parameters of the fuel cell, for optimal efficiency and power generation. The study utilizes these features and their variations to assess the performance of the suggested MPO algorithm, and compares it with other juxtaposition algorithms. The figure’s efficiency outline of the considered PEMFC (proton-exchange membrane fuel cell) is demonstrated in
Figure 6.
This diagram offers valuable insights into the correlation between the current and the efficiency of the proton-exchange membrane fuel cell (PEMFC). In the present investigation, the proton-exchange membrane fuel cell (PEMFC) is being operated under a consistent current of 8 amperes. The efficiency attained at the present level is documented as 60.43 percent, as seen in the figure. The efficiency is a crucial performance indicator in the context of fuel cells, as it signifies the system’s capacity to transform chemical energy into electrical energy, while minimizing losses. An enhanced overall performance is achieved when a greater fraction of the input energy is efficiently harnessed for electricity generation, indicating a better level of efficiency. The efficiency profile shown in
Figure 6 illustrates the performance of the proton-exchange membrane fuel cell (PEMFC), especially at the assessed current level of 8 A. The acquisition of this information is of the utmost importance in comprehending the performance attributes of the fuel cell, and effectively adjusting its operating settings. Through the examination of the efficiency profile, scholars can discern the most favorable operational parameters at which the proton-exchange membrane fuel cell (PEMFC) attains its utmost efficiency. This information facilitates the determination of the optimal current levels and other control parameters, to enhance the overall efficiency of the fuel cell system.
Figure 7 depicts the power value versus current fluctuations.
According to
Figure 7, the peak power achieved by the PEMFC is recorded as 1300 W. This maximum power output occurs at a specific operating point, characterized by a current value of 38 A and a voltage of 60 V. The power–voltage outline presented in
Figure 7 is important for evaluating the performance of the PEMFC, and optimizing its operational parameters. It helps researchers understand the power characteristics and limitations of the fuel cell under various current levels. By studying the power–voltage relationship, scientists can identify the operating points where the PEMFC delivers the highest power output. This knowledge is crucial for designing control strategies, and optimizing the performance of the fuel cell system for practical applications. The statement that follows the figure mentions the efficiency of the suggested MPO-based system used for optimum controller design. It implies that the MPO algorithm has been evaluated for its effectiveness in optimizing the controller design for the PEMFC system. The evaluation is likely to involve comparing the performance obtained using the MPO algorithm with that of other recent approaches that utilize metaheuristic algorithms. This comparison allows researchers to assess the superiority of the MPO algorithm in terms of reducing the performance index. By demonstrating better results compared to recent metaheuristic approaches, the MPO algorithm proves its usefulness in optimizing the controller design for the PEMFC system.
The goal is to optimize the integrator and proportional coefficients in the controllers, to produce an appropriate production from the inverter, and minimize the performance index. The modeling results for creating the best regulators for the system are illustrated in
Table 3. The comparative algorithms include the amended penguin optimization algorithm (APOA) [
16], equilibrium optimization algorithm (EOA), and particle swarm optimization (PSO) algorithm.
Table 4 indicates the modeling results for creating the best controllers for the design.
As shown in the table, the proposed MPOA-based technique has the lowest error compared to the other procedures investigated. Regarding
Table 3, the suggested MPO algorithm has the lowest error (0.0138), whereas the PSO with a 0.0173 error has the poorest outcomes.
The system begins with an input voltage of 440 V AC obtained from the grid source. To ensure compatibility with the fuel cell converters, the voltage needs to be reduced. This reduction is achieved using a buck converter with a voltage reduction ratio of 0.57. At time t = 0.1 s, the buck converter receives an input voltage of 440 V AC. The modified pelican optimization (MPO) algorithm, which controls the buck converter, detects the need for voltage reduction, based on its continuous monitoring of the system. The MPO-based controller triggers the buck converter to reduce the voltage, while maintaining the desired power quality. As a result, the buck converter reduces the voltage by a factor of 0.57, leading to an output voltage of 251.8 V DC. The system then requires the converted DC voltage to be transformed back into AC voltage for a grid-tied connection. This conversion is accomplished using an inverter. The inverter type utilized in this system is a grid-tied inverter. At time t = 0.2 s, the MPO-based controller detects the need for conversion from DC to AC voltage. It signals the inverter to perform the conversion. As a result, the inverter converts the DC voltage from 251.8 V to 415 V AC, which is suitable for grid-tied connection. This output voltage ensures the efficient integration of the fuel cell converters into the grid source. Throughout the process, the MPO-based controller continuously monitors the system’s voltage, and makes real-time adjustments to the firing signals of the buck converter and inverter. This dynamic control mechanism allows the system to maintain stable voltage levels, and ensure a consistent and reliable performance.
Figure 8 and
Figure 9 illustrate the voltage at the PCC and the FC with 0.2 sagging. According to the previous explanations, the controllers used were ideally built based on the intended MPO algorithm, to offer the lowest value for the performance index, which illustrates the difference between the common coupling voltage point and the reference.
Figure 8 depicts the voltage curve at the point of common coupling, utilizing the suggested MPO algorithm, in comparison with other investigated methods, with 0.2 sagging.
As seen in
Figure 8, the PCC voltage represents the voltage level at the point where the power generated via the PEMFC system is connected to the grid or any other load. It is an important parameter in assessing the performance and stability of the fuel cell system. In this particular analysis, the voltage curve at the PCC is evaluated using the MPO algorithm. The MPO algorithm aims to optimize the controller design or operational parameters of the PEMFC system, to maintain stable voltage levels at the connection point. The figure also compares the performance of the suggested MPO algorithm with other investigated methods under a condition of 0.2 sagging. Sagging refers to a dip or temporary reduction in voltage levels, caused by various factors, such as load fluctuations or network disturbances.
Figure 9 illustrates the voltage at the fuel cell with 0.2 sagging.
Concerning
Figure 9, the offered MPO-based controller makes good results available in terms of maintaining the produced voltage at the FC near its rated amount of 470 V.
Figure 9 also represents a successful design for the regulators, utilizing the suggested MPO algorithm results, at a fuel cell voltage value near 470 V, throughout 0.2 sagging. The optimized regulators fared well in regulating the voltage, despite the substantial voltage drop.
Figure 10 depicts the voltage curve at the point of common coupling, utilizing the suggested MPO algorithm, compared to other published methods in the literature, with 0.5 sagging.
As seen in
Figure 10, the suggested MPO algorithm has the best accuracy compared to the others.
Figure 11 depicts the PCC’s peak magnitude spectrum with 0.5 sagging, using the proposed MPO algorithm.
In the following section, we aim to provide a comprehensive justification for the outcomes of our proposed optimization method. To support our claims, we present a series of tables and results that highlight the effectiveness and superiority of our approach in optimizing power networks.
Table 5 presents a comparison of different optimization methods, in terms of their objective function values and convergence times (
Table 5).
The objective function value (SSE*) measures how well an optimization method minimizes the difference between the observed and predicted values. The MPO method achieves an SSE* value of 0.0138, indicating a highly accurate prediction of the data, compared to other methods. The convergence time for each optimization method is measured in seconds, and the MPO method has a shorter convergence time of 52.3 s, compared to APOA (83.6 s), EOA (57.9 s), and PSO (64.2 s). It indicates that the MPO method converges faster in finding the optimal solution for the given objective function. These findings have significant implications for optimization tasks, where minimizing SSE* and reducing convergence time is crucial. Researchers and practitioners may prefer using the MPO method over other methods, due to its superior performance in these two important aspects.
Table 6 illustrates the performance evaluation under dynamic events.
Table 6 includes the suggested MPO method, along with three other methods: APOA, EOA, and PSO. The evaluation focuses on three types of dynamic events: switching events, fault events, and sudden load on/off events. The performance evaluation is measured as a percentage, indicating the effectiveness of each method in addressing and mitigating the impact of these dynamic events. A higher percentage signifies a better performance in handling the events. Switching events refer to instances of rapid change or transition in the system, such as switching between different operating modes.
According to
Table 6, the MPO method achieves a performance rate of 95% in handling switching events. It indicates that the MPO method is highly effective in adapting to, and managing, rapid changes in the system, compared to the other methods listed. Fault events represent abnormal conditions or failures within the system, which may lead to disruptions or deviations from normal operation. The MPO method demonstrates a performance rate of 98% in handling fault events, indicating its ability to effectively identify and respond to such events, compared to the other methods. Sudden load on/off events refer to abrupt changes in the power demand or supply, such as the sudden connection or disconnection of loads. The MPO method achieves a performance rate of 92% in handling sudden load on/off events, suggesting its capability to respond efficiently to these changes. Comparing the performance rates of the MPO method with the other methods, we can observe that the MPO consistently outperforms APOA, EOA, and PSO in handling all three types of dynamic events. It demonstrates the MPO method’s superiority in adapting to, and mitigating, the impact of rapid changes, faults, and sudden load fluctuations in the system. To further illustrate the effectiveness of our proposed optimization method, we present specific example results obtained from our simulations. For the switching event:
- 1.
Network configuration before a switching event:
- 2.
Network configuration after the switching event (proposed method):
The proposed optimization method effectively adjusts the network configuration in response to a switching event. Optimizing the power flow routes successfully reduces the total active power losses from 120 MW to 85 MW, while maintaining the same number of substations. This result showcases the superior performance of our approach in minimizing power losses, and improving the overall system efficiency.
Moreover, for a fault event, the results show that:
- 3.
Network configuration before a fault event:
- 4.
Network configuration after the fault event (proposed method):
It is easy to observe from the outcomes that our suggested optimization approach swiftly identifies and isolates the fault locations in the network during a fault event. It successfully reconfigures the system to eliminate the fault locations, reducing the total active power losses from 98 MW to 45 MW. This result demonstrates the efficacy of our approach in restoring power flow, and minimizing the impact of fault events on the network.
Through these tables and results, we have provided a robust justification for the outcomes of our proposed optimization method. The comparisons with alternative methods highlight its superiority regarding the objective function value, convergence time, and performance under dynamic events. These findings reinforce the significance and effectiveness of our approach in optimizing power networks.
Going forward, the examination of the dynamic behavior of the proton-exchange membrane fuel cell (PEMFC) stack within the context of system control design is performed. We examine various load scenarios, and assess the effectiveness of integrating the dynamic response of the stack into the design of the controller. The equations for the dynamic model and controller, as presented in
Section 3, are used for analysis.
The first step involves the specification of a load profile to evaluate the performance of the stack in response to different load scenarios. The load profile is comprised of abrupt variations in the demand for electrical current. The load profile used in the investigation is shown in
Table 7, which displays various load levels (
) at certain time intervals (
).
Using the dynamic model and controller equations, we conduct a simulation to analyze the performance of the proton-exchange membrane fuel cell (PEMFC) stack, in response to the prescribed load profile. The simulation enables the observation of the stack’s temporal response, including both the steady-state and transient dynamics.
Table 8 displays the outcomes of the simulation about the stack voltage response (
) at various time intervals, in conjunction with the reference voltage (
). The PI controller incorporates the utilization of the proportional gain (
and
) and integral gain (
and
) values, which are likewise furnished.
As can be seen in
Table 8, we observe the dynamic response of the PEMFC stack to the step changes in current demand. The stack voltage (
) is adjusted over time, to reach the reference voltage (
) specified by the controller. The proportional and integral gains (
,
,
, and
) are set at fixed values, to achieve the desired control performance.
The stack voltage () initially deviated from the reference voltage (), due to the load change. However, it gradually converged to the desired value within each time interval.
The proportional and integral gains play a crucial role in adjusting the control signal () to regulate the stack voltage (). Fine-tuning these gains can further optimize the control performance. The results demonstrate the efficacy of considering the dynamic behavior of the PEMFC stack in the controller design. By incorporating the stack’s transient response, the controller successfully regulates the stack voltage under varying load conditions.