Artificial Intelligence in Electronic Payment Switches of Banking Systems Using an Enhanced Fuzzy Method Assisted by Metaheuristic Optimization Algorithms
Artificial Intelligence in Electronic Payment Switches of Banking Systems Using an Enhanced Fuzzy Method Assisted by Metaheuristic Optimization Algorithms
safura babajanipoor1 ehsan soleimani2
1) m.sc. Student in Artificial Intelligence, Islamic Azad University, Boroujen Branch
2) Ph.D. Student in Computer Engineering, Islamic Azad University, Meybod Branch
IOI Code
:
ACCD-ACADAD
Publication :
10th International Conference on Applied Researches in Science & Engineering (carseconf.com)
Abstract :
Electronic banking industry, dealing with a massive volume of transactions and an increasing demand for real-time processing, requires intelligent decision-making mechanisms to route transactions toward payment channels. This study designs and evaluates an electronic payment switch system based on fuzzy reasoning, in which membership function parameters and inference rules are automatically tuned utilizing metaheuristic optimization algorithms. In this research, a Mamdani fuzzy inference system model with twenty-four rules and seven input variables—including transaction amount, fee rate, channel response time, historical error rate, network load, customer priority, and gateway reliability—was developed. To optimize the model, four algorithms—Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Grey Wolf Optimizer (GWO)—were compared based on a multi-objective fitness function consisting of total cost, latency, and error rate. Experimental data comprising one hundred thousand simulated transactions were evaluated across four network load scenarios. The results indicate that the fuzzy version optimized with the Whale Optimization Algorithm achieved an 18.6% reduction in cost, a 23.4% reduction in average latency, and a 31.2% reduction in routing error rate compared to the traditional rule-based method. Furthermore, sensitivity analysis demonstrated that the model exhibits favorable stability against network load fluctuations. The findings reveal that the combination of fuzzy logic and metaheuristic optimization provides an efficient and scalable framework for payment switching decision-making, paving the way for operational applications in next-generation banking. Statistical analysis at a significance level of 0.05 showed that the superiority of the Whale algorithm is not random, and the proposed framework offers an interpretability advantage over black-box methods.
Keywords : electronic payment switch fuzzy logic metaheuristic optimization genetic algorithm electronic banking