Supervised and Reinforcement Group-Based Hybrid Learning Algorithms for TSK-type Fuzzy Cerebellar Model Articulation Controller

Abstract

In this study, a Takagi-Sugeno-Kang (TSK)-type fuzzy cerebellar model articulation controller (T-FCMAC) based on a group-based hybrid learning algorithm (GHLA) was proposed for solving various problems. The proposed T-FCMAC model was mainly derived from a traditional cerebellar model articulation controller and the TSK-type fuzzy model. For supervised learning, the proposed GHLA was developed by combining an improved quantum particle swarm optimization algorithm and the Nelder-Mead method for adjusting the parameters of a T-FCMAC. The fuzzy C-mean clustering technique was adopted to improve the performance of quantum particle swarm optimization. A fitness threshold was used to determine the number of clusters in fuzzy C-mean clustering. The grouping concept was also used to improve the search ability and increase the convergence rate. Moreover, because exact training data may be expensive or even impossible to obtain in some real-world applications, a reinforcement GHLA (R-GHLA) was proposed. Experimental results revealed the performance and applicability of the proposed GHLA and R-GHLA.