A Control System Based on an Improved Model Auto-Fusion Cerebellar Perceptron and Its Application to the Consensus Problem

作者：Uchiyama Shogo^*; Obayashi Masanao; Kuremoto Takashi; Kobayashi Kunikazu; Mabu Shingo

来源：Electronics and Communications in Japan, 2016, 99(5): 37-46.

DOI：10.1002/ecj.11808

• neural network
• feedback error learning
• multiagent system
• consensus problem

In this paper, we propose a control system based on an improved model auto-fusion cerebellar perceptron using feedback error learning (FEL) which imitates the human cerebellum, and apply it to the consensus problem of a multiagent system (MAS). It is important to control multiple agents because each of them has its own scale and complexity. Therefore, coordinative control of the MAS for each autonomous decision-making instance has been taken as the focus. To control MAS, we consider use of FEL related to biological movement control. We also propose an auto-fusion mechanism for mitigation of neuronal fluctuation. We call the proposed system the Auto-Fusion Cerebellar Perceptron Improved Model-Based Control System (AFCPCS) here. Through a computer simulation of the MAS consensus problem, we demonstrate the effectiveness of the proposed method.

• 出版日期2016-5