I am getting the controller output range out of acceptable range, dataset is position error and it's rate of change (cm and cm/s). Therefore, in my opinion, dataset standardization and normalization are not required.
The fuzzy sets that represent the outputs of each rule are combined into a single fuzzy set by the process of aggregation. The input of the aggregation process is the list of truncated output functions evaluated by each rule. Fuzzification is the process of converting a crisp input value to a fuzzy value that is performed by the use of the information in the knowledge base. Although various types of curves can be seen in literature, Gaussian, triangular, and trapezoidal MFs are the most commonly used in the fuzzification process.
try to adjust some parameters like number of cluster and U matrix value but it's not change.
https://kr.mathworks.com/help/fuzzy/fuzzy-inference-process.html
Masoum Mohammadi Gharagoz I tried different types of MF but in vain, I am still getting an output range out of acceptable range, high number of clusters leads to overfitting.
it is right that high number of clusters leads to overfitting
Masoum Mohammadi Gharagoz what can be the max number of Input/Output of the MFs?
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude