University of Isfahan Computational Intelligence in Electrical Engineering 2821-0689 16 4 2025 12 22 Maximum Power Tracking in Proton Exchange Membrane-Based Fuel Cell Using Neural Network Trained with Metaheuristic Optimization Algorithm Maximum Power Tracking in Proton Exchange Membrane-Based Fuel Cell Using Neural Network Trained with Metaheuristic Optimization Algorithm 1 16 30204 10.22108/isee.2025.145950.1750 FA Mohamad Abedini Associate Professor, Department of Electrical Engineering, Ayatollah Boroujerdi University, Boroujerd, Iran Journal Article 2025 07 13 Considering the limited ability of fuel cells to produce energy, it is necessary to provide solutions in which the efficient power generated by fuel cells can be achieved. Therefore, there is a need for maximum-power-tracking methods that adjust the duty cycle of the boost converter in fuel cells. Thus, in this paper, a new power tracking method has been used that is based on the combination of neural networks and a meta-heuristic algorithm called frog jump optimization to overcome the problems of conventional methods caused by rapid changes in the operating point and power fluctuations. The proposed method has been proposed in order to perform quickly and increase the efficiency achievable from proton exchange membrane-based fuel cells. The modeling results have been presented in the MATLAB environment and compared with several different power tracking methods. The results show that the proposed method shows less than one percent error in the three evaluated temperatures for tracking the maximum power compared to the actual power of the fuel cell, and is also robust to input changes in the fuel cell. Considering the limited ability of fuel cells to produce energy, it is necessary to provide solutions in which the efficient power generated by fuel cells can be achieved. Therefore, there is a need for maximum-power-tracking methods that adjust the duty cycle of the boost converter in fuel cells. Thus, in this paper, a new power tracking method has been used that is based on the combination of neural networks and a meta-heuristic algorithm called frog jump optimization to overcome the problems of conventional methods caused by rapid changes in the operating point and power fluctuations. The proposed method has been proposed in order to perform quickly and increase the efficiency achievable from proton exchange membrane-based fuel cells. The modeling results have been presented in the MATLAB environment and compared with several different power tracking methods. The results show that the proposed method shows less than one percent error in the three evaluated temperatures for tracking the maximum power compared to the actual power of the fuel cell, and is also robust to input changes in the fuel cell. https://isee.ui.ac.ir/article_30204_d215c4b33b8c7015784714f6709e810c.pdf

University of Isfahan Computational Intelligence in Electrical Engineering 2821-0689 16 4 2025 12 22 Design of a DNA Computing–Based Full Adder and Its Mapping onto a Digital Microfluidic Chip Using a Time- and Resource-Efficient Approach Design of a DNA Computing–Based Full Adder and Its Mapping onto a Digital Microfluidic Chip Using a Time- and Resource-Efficient Approach 17 28 30232 10.22108/isee.2025.145235.1738 FA Kosar Hadi Abedini Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran Fatemeh Kazemi HasanAbadi Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran Zohre Beiki Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran Shahram Etemadi Borujeni Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran Mohammad Reza Reshadinezhad Faculty of Computer Engineering, University of Isfahan, Isfahan, Iran Journal Article 2025 05 14 Given the growing importance of biomolecular-scale processing, digital microfluidic technology has emerged as an efficient platform for implementing DNA-based logic circuits. In this work, we propose a novel binary full-adder architecture designed to minimize the number of DNA strands and to reduce fabrication and execution costs. The proposed structure was first modeled and evaluated in the Visual DSD simulation environment, and then mapped onto a custom-designed digital microfluidic biochip (DMFB) with a 15×20 electrode array. By employing time-scheduled modules for droplet splitting, mixing, and detection, the designed chip is capable of performing the complete computational operation within the minimum possible number of operational cycles. Simulation results confirm that the generated outputs fully match the expected logical behavior. Comparisons with existing approaches show that the proposed architecture achieves a significant reduction in the number of required DNA strands while maintaining computational accuracy. Specifically, the proposed full adder operates using only 15 DNA strands, whereas Seesaw- and DNA origami–based methods require 9n+53and 6n+15strands, respectively, for ninputs. In addition, the proposed DMFB mapping completes the addition process in only six operational cycles, resulting in a substantial reduction in latency compared with existing architectures, which typically require 10–14 cycles. These results highlight the potential of the proposed approach as an effective step toward the convergence of biological sciences and computational engineering, and its applicability in the development of intelligent bio-computing systems and lab-on-a-chip technologies. Given the growing importance of biomolecular-scale processing, digital microfluidic technology has emerged as an efficient platform for implementing DNA-based logic circuits. In this work, we propose a novel binary full-adder architecture designed to minimize the number of DNA strands and to reduce fabrication and execution costs. The proposed structure was first modeled and evaluated in the Visual DSD simulation environment, and then mapped onto a custom-designed digital microfluidic biochip (DMFB) with a 15×20 electrode array. By employing time-scheduled modules for droplet splitting, mixing, and detection, the designed chip is capable of performing the complete computational operation within the minimum possible number of operational cycles. Simulation results confirm that the generated outputs fully match the expected logical behavior. Comparisons with existing approaches show that the proposed architecture achieves a significant reduction in the number of required DNA strands while maintaining computational accuracy. Specifically, the proposed full adder operates using only 15 DNA strands, whereas Seesaw- and DNA origami–based methods require 9n+53and 6n+15strands, respectively, for ninputs. In addition, the proposed DMFB mapping completes the addition process in only six operational cycles, resulting in a substantial reduction in latency compared with existing architectures, which typically require 10–14 cycles. These results highlight the potential of the proposed approach as an effective step toward the convergence of biological sciences and computational engineering, and its applicability in the development of intelligent bio-computing systems and lab-on-a-chip technologies. https://isee.ui.ac.ir/article_30232_8e994890148fa972a6d8d5c3789aa5a4.pdf