An artificial neural network controller based on MPSO-BFGS hybrid optimization for spherical flying robot

4th International Conference on Mechanical, Materials and Manufacturing

材料科学;机械制造

Liu, Xiaolin^1 ; Li, Lanfei^1 ; Sun, Hanxu^1

School of Automations, Beijing University of Posts and Telecommunications, No.10 Xitucheng Road, Beijing, China^1

关键词: Artificial neural network controllers;    Expected values;    Hybrid optimization;    Hybrid optimization algorithm;    Local optimal solution;    Nonlinear coupling;    Strong coupling;    Time-varying disturbance;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/272/1/012002/pdf
DOI  :  10.1088/1757-899X/272/1/012002

学科分类：材料科学（综合）

来源: IOP

【 摘 要 】

Spherical flying robot can perform various tasks in the complex and varied environment to reduce labor costs. However, it is difficult to guarantee the stability of the spherical flying robot in the case of strong coupling and time-varying disturbance. In this paper, an artificial neural network controller (ANNC) based on MPSO-BFGS hybrid optimization algorithm is proposed. The MPSO algorithm is used to optimize the initial weights of the controller to avoid the local optimal solution. The BFGS algorithm is introduced to improve the convergence ability of the network. We use Lyapunov method to analyze the stability of ANNC. The controller is simulated under the condition of nonlinear coupling disturbance. The experimental results show that the proposed controller can obtain the expected value in shoter time compared with the other considered methods.