A Mathematical Approach to Intelligent Control in Artificial Heart Pacemaker Design Using ANFIS with FCM, GA, and PSO

Abstract

This paper presents the design and comparison of an Adaptive Neuro-fuzzy Inference System (ANFIS) based controller of a pacemakerwithFuzzyc-meansclustering, Geneticalgorithm,  and Particles warm optimization learning methods. At the same time, ANFIS is using the benefits of fuzzy logic and neural networks. The input output data for the FCM, GA and PSO based ANFIS heart rate controllers shall be developed after the design of a (proportional integral derivative) PID based heart rate controller so that they can be trained and tested. The results of the Step response in Time domain, have been compared with previous studies and show an optimal pacing rate achieved by ANFIS_GA which is used to automatically set a heart rate for each patient more accurately than other methods, such as Fuzzy, PID or FPID. FCM and PSO have their strengths and can perform well in specific scenarios, GA tends to outperform them due to their  superior exploration capabilities, robustness across problem types, and effective handling of multi-objective optimization. The combination of genetic diversity, population-based search, and adaptability makes GAs a powerful choice for complex optimization tasks, particularly when convergence properties are critical. Compared to other training methods, the  convergence rate was better for FCM. GA’s accuracy was superior to all the other methods. A MATLAB script is used for the controller design. The resulting ’FIS’ (fuzzy inference system) file is imported to SIMULINK and simulation is done by fuzzy controller block that uses ANFIS ’FIS’ file. GA based ANFIS simulation results for determining Membership function’s coefficients show better results than other methods. There is no overshoot and settling time is about 3.13 seconds and rise time 2.28 seconds for step input response (85 bpm input). Beyond this application, this work demonstrates a novel integration of clustering and metaheuristic methods for stability-constrained adaptive fuzzy control, contributing to the theory of nonlinear optimization and function approximation applications in control systems.