Novel Adaptive Control Design for Nonlinear System with Extended Partition of Unity Method
Liang Luo,Yin-He Wang,Yong-Qing Fan,Yun Zhang
DOI: https://doi.org/10.1002/asjc.576
IF: 2.4
2012-01-01
Asian Journal of Control
Abstract:One of the amazing abilities of fuzzy logic systems or neural networks is their ability to approximate unknown certainties. When nonlinear systems possess multiple variables, however, the process of the adaptive fuzzy or neural network online control becomes difficult. In this paper, we will introduce the extended partition of unity (EPU), composed of scalars and saturators, to address this problem. The merit of the suggested design scheme is that the construction of the partition of unity and the design of adaptive laws are separate. This means the proposed design method only adjusts the outputs of EPU and one update law, even in nonlinear systems with multiple variables. Therefore, this new method of EPU leads to easier selection of basis functions, reduces the number of adaptive laws, has greater robustness, and is suitable for different kinds of universal approximators. Finally, a numerical example is given to illustrate the effectiveness of the approach.