Deep Hierarchical Interval Type 2 Self-Organizing Fuzzy System for Data-Driven Robot Control

To solve the dimensional explosion problem, this paper proposes a new architecture for fuzzy system, deep hierarchical self-organizing interval type-2 system (DHSOIT2FS). Each sub-fuzzy is constructed online, with rules by rule online update algorithm, consequent parameters updated iterative least squares, and antecedent are using gradient descent algorithm. DHSOIT2FS uses classic serial-layered structure to build overall framework. The first layer two dimensions of data as input. subsequent output previous next input until it built. During training process, each point trained before passing in achieve construction. effectiveness approach illustrated numerical simulation examples. proposed method also applied data-driven control example single-link robot achieves good tracking results.