Adaptive Neural Fuzzy Inference System for Hydrogen Adsorption Prediction

• Main Author: Jufri Afiq Bin Jolan, Jufri Afiq
• Format: Final Year Project
• Language: English
• Published: Universiti Teknologi Petronas 2009
• Subjects: TP Chemical technology
• Online Access: http://utpedia.utp.edu.my/542/1/jufri_ariq_b_jolan.pdf; http://utpedia.utp.edu.my/542/

This report is basically to discuss about the basic concept and implementation of Artificial Neural Fuzzy Inference System (ANFIS) in predicting the hydrogen adsorption isotherm. The objective of this project is to create an ANFIS that is able to predict the hydrogen adsorption isotherm. The challenge in this project is to develop the ANFIS that is able to predict the hydrogen adsorption isotherm at the highest accuracy. ANFIS is developed by using MatLab R2008a. This software which is a mathematicalp ower tool has the ability to developt he ANFIS. This is becauseth e software has the Fuzzy Logic Toolbox which is the basic requirement in building the system. The basic system is developed to receive two inputs which are temperature and pressure from users and gives one output which is the hydrogen adsorption value. Three membership functions are provided for each input which is then used in determining the output of the system. Multiple training data are given to the basic system in order to mature it. Upon completion, the system is then tested with test data and output from the system is analyzed.C alculationso f the output percentagee rror are carried out. From here,t he ANFIS membership functions for each input are fine tuned in order to reduce output percentagee rror in order to increaset he prediction accuracyo f the system. As a result, the ANFIS is able to give prediction data with an error less than 5% which desirable in this project