Adaptive neuro-fuzzy inference system combined with genetic algorithm to improve power extraction capability in fuel cell applications

Savrun M. M., Inci M.

JOURNAL OF CLEANER PRODUCTION, vol.299, 2021 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 299
  • Publication Date: 2021
  • Doi Number: 10.1016/j.jclepro.2021.126944
  • Journal Name: JOURNAL OF CLEANER PRODUCTION
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Aerospace Database, Business Source Elite, Business Source Premier, CAB Abstracts, Communication Abstracts, INSPEC, Metadex, Pollution Abstracts, Public Affairs Index, Veterinary Science Database, Civil Engineering Abstracts
  • Bursa Uludag University Affiliated: No

Abstract

This study introduces an improved ANFIS based MPPT method to maximize the power extraction capability of the FC-connected system. The proposed method is tested in a stand-alone system that consists of an FC in the power rating of 1.9 kW, a boost dc-dc converter, local consumer load, and processor unit. The energy transfer between FC and load is handled through the adjustment of a duty cycle of the dc-dc converter. In this context, the output voltage of FC is controlled by the duty cycle to track the MPP. The proposed method called GA-ANFIS computes optimum reference voltages to control the FC output voltage optimally. The GA-ANFIS uses a reduced-size training dataset extracted by GA to train the ANFIS in comparison with conventional ANFIS. Unlike the existing methods, the proposed method tracks the MPP by merely monitoring FC voltage during operation. Besides, it performs precise MPP tracking by considering pressure & temperature variations. Thus, the proposed method provides reduced computational load owing to its current features. The performance of the proposed method compared with the traditional methods like ANFIS and PI. The power extraction ratings and efficiency values validate the viability and effectiveness of the proposed method (>98%). (c) 2021 Elsevier Ltd. All reserved.