Dynamic modelling of proton exchange membrane fuel cell system for electric bicycle / Azadeh Kheirandish

Azadeh, Kheirandish (2016) Dynamic modelling of proton exchange membrane fuel cell system for electric bicycle / Azadeh Kheirandish. PhD thesis, University of Malaya.

Preview

PDF (Thesis PhD) Download (5Mb) | Preview

Abstract

Fuel cell systems with high-energy efficiency provides clean energy with lower noise and emissions that have attracted significant attention of energy. Proton exchange membrane (PEM) fuel cell has high power density; long stack life and low-temperature operation condition, which makes it a prime candidate for the vehicles. Performance optimization of PEM fuel cell has been a topic of research in the last decade. The efficiency of fuel cells is not specific; it is a subordinate to the power density where the system operates. The fuel cell performance is least efficient when functioning under maximum output power conditions. Modelling the PEM fuel cell is the fundamental step in designing efficient systems for achieving higher performance. In spite of affecting factors in PEM fuel cell functionality, providing a reliable model for PEM fuel cell is the key of performance optimization challenge. There have been two approaches for modelling and prediction of commercial PEM fuel cell namely, theoretical and empirical models. Since theoretical modeling is not achievable in experimental conditions, the empirical modeling has attracted significant attention in researches. Various types of algorithms have been utilized for modelling these systems to achieve a high accuracy for predicting the efficiency and controlling the system. Recent models provide high accuracies using complex systems and complicated calculations using advanced optimization algorithms. However, designing an accurate dynamic model for prediction and controlling the system in a real time condition is a challenge in this field. By utilizing the state of the art soft computing algorithms in modeling the technical systems to reduce the complexity of the models artificial neural networks have had a great impact in this field. This study has multifold objectives and aim to design models for a 250W proton exchange membrane fuel cell system that is used as the power plant in electric bicycle. Classical linear regression and artificial neural networks as the most popular and accurate algorithms have been optimized and used for modeling this system. In addition, for the first time fuzzy cognitive map has been utilized in modeling PEM fuel cell system and targeted to provide a dynamic cognitive map from the affective factors of the system. Controlling and modification of the system performance in various conditions is more practical by correlations among the performance factors of the PEM fuel cell resulted from fuzzy cognitive map. On the other hand, the information of fuzzy cognitive map modeling is applicable for modification of neural networks structure for providing more accurate results based on the extracted knowledge from the cognitive map and visualization of the system’s performance.

Actions (For repository staff only : Login required)