Master of Science
The complexity of product design in industry has been continuously increasing. More factors are required to be taken into account simultaneously before a decision about the new product could be determined. For this reason, decision-making process costs much more time and it may even be impossible to determine the optimal decision by normal calculations. Therefore, Fuzzy Inference System based on Fuzzy Logic is introduced as a quick decision-making tool to arrive at a good decision within much shorter time.This thesis focuses on studying the features of membership functions in Mamdani-type fuzzy inference process. It is aimed at making the black box of fuzzy inference system to be transparent by adjusting the membership functions to control the relations between input and output variables. Systematic trial and error is implemented based on the Fuzzy Logic Toolbox from MATLAB, and conclusions developed from experiments help eliminate the uncertainties of membership functions, so that the inference process turns to be more precise and reliable. Firstly, Single-Input Single-Output (SISO) Fuzzy Inference System is discussed through the adjustment of membership functions, and the influence on input-output relations are concluded. Next, Two-Input Single-Output (TISO) Fuzzy Inference System is simulated to verify the conclusions from SISO Fuzzy Inference System, and general features of membership functions on affecting input-output relation are developed. Then, an approach using weights on input variables, for practical decision-making process, is derived. Finally, a design problem of timing system of automobile engine is chosen as case study to examine the validity of conclusions on practical decision-making problem.
Wang, Chonghua, "A Study of Membership Functions on Mamdani-Type Fuzzy Inference System for Industrial Decision-Making" (2015). Theses and Dissertations. 1665.