Comparison of Modified Karnik-Mendel Algorithm Based Interval Type-2 ANFIS and Type-1 ANFIS Controllers

This study aims to test Interval Type-2 Adaptive Neuro-Fuzzy Inference System (IT2-ANFIS) based Neuro-Fuzzy Control (NFC) by using the Modified Karnik Mendel Algorithm (M-KMA). Determining the FLS antecedent and consequent parameters is the main problem in engineering issues. The Adaptive Neuro-Fuzzy Inference System (ANFIS) is the most used training method in the literature. The ANFIS model works very well for Type-1 FLS (T1-FLS) but is inconvenient for Interval Type-2 FLS (IT2-FLS) due to uncertainties. In 2021, a new IT2-ANFIS method is proposed to tune the IT2-FLS parameters by designing a reduction method named M-KMA. The M-KMA based IT2-ANFIS is superior to T1-ANFIS, so it is used in this paper. The M-KMA based IT2-NFC that consists of IT2-ANFIS structure is implemented to a DC motor control problem and the results are compared to Type-1 NFC. The results show that IT2-NFC performs better compared to T1-NFC.