I am trying to learn the fundamentals of the Sugeno-Type Fuzzy Inference system, as it seems to be more favourable to implement than the Mamdani model. For this, I am following the 'Tippersg' example from the MatLab documentation. Theres a few things I have determined such as the Final Output of the system being the weighted average of all rule outputs, but there are a few things I haven't determined as yet:
- When I open the fuzzy GUI and go to the output membership functions, I am perplexed with the 'Params' and 'Range'. The 'tipper' model has 3 output membership functions to represent 5%, 15% and 25%. In the params they are labeled [0 0 5], [0 0 15], [0 0 25] in 'linear' mode, and the range seems to be from [-30 30] when I thought it would range between [0 25]. If I change to 'constant' they are single numbers ,  and , which is what I was expecting in 'linear' mode. What is the difference between 'linear' and 'constant', as I have failed to see a difference in performance within the Rule Viewer so far.
- Moreover, if 'input 1 = x' and 'input 2 = y', then output 'z' is equal to 'ax + by + c' , what do the variables 'a', 'b' and 'c' correlate to? Could these be something to do with each rules weight?
- Finally, how do I know if this tipper demonstration system, 'linear' or 'constant' is a 'zero-order Sugeno model? So far, I have only been able to correlate this with a first-order polynomial ( a linear expression where the highest power to which a variable is raised is 1 ). However, this could be incorrect.