Efficient Neuro-Fuzzy Identification Model for Electrocardiogram Signal
Abdelaziz Aouiche, Aouiche El Moundher, Dhaouadi Guiza · Journal Européen des Systèmes Automatisés · 2022
This paper addresses the performance of the Artificial Neural Networks (ANNs), Fuzzy inference systems (FISs) and Adaptive Neuro-Fuzzy Inference System (ANFIS) for the identification of some nonlinear systems with certain degree of uncertainty. The efficiency of the suggested methods in modeling and identification the responses were analyzed and compared. The Back-propagation algorithm and Takagi-Sugeno (TS) approach are used to train the ANNs, FISs and ANFIS, respectively. In this study we will show how ANFIS can be put in order to form nets that able to train from external data and information compared to ANNs and FISs. In order, it is proposed forms of inputs that can be used along with ANNs, FISs and ANFIS to modeling nonlinear systems. Two nonlinear systems with an electrocardiogram (ECG) signal in the form of simulation and complexity were used to test the identification of the structure presented. Because ANFIS has an inherent capacity to approximate unknown functions and to adjust the changes in inputs and parameters, it can be used to identify the proposed systems with a very high level of complexity. The results show that the ANFIS technique can provide the most ideal approximation when the right structures are employed.