Generating flexible convex hyper-polygon validity regions via sigmoid-based membership functions in TS modeling

•A new type of membership function in TS modeling is introduced which can generate flexible convex hyper-polygon validity regions.•Based on the introduced membership function, two types of TS models are proposed whose consequent parts of the rules are linear and quadratic functions.•An incremental learning algorithm is suggested to identify the proposed TS models.•Obtained results demonstrate high accuracy and low redundancy of the suggested TS fuzzy models.