Adaptive neural network control of hexapod for aerospace application

Zhukov, Yu. A.; Korotkov, E. B.; Moroz, Alexander; Zhukova, V. V.; Abramov, A. M.

doi:10.1088/1757-899x/441/1/012001

Cited by 3 publications

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The nonredundant limbs are controlled by the BP neural network PID controller. Figure 3 shows the block diagram of the BP neural network model which is employed in the intelligent gain tuning by online learning [37][38][39][40][41]. In this model, the NN has four input layers, eight hidden layers, and three output layers, and w o and w i are the weight factors of input layers and output layers that can be continuously updated by machine learning.…”

Section: Control Designmentioning

confidence: 99%

Trajectory Tracking Control Study of a New Parallel Mechanism with Redundant Actuation

Zhang

Fang

Zhang

et al. 2020

International Journal of Aerospace Engineering

View full text Add to dashboard Cite

Parallel mechanisms with redundant actuation are attracting numerous scholars’ research interest due to their inherent advantages. In this paper, an efficient trajectory tracking control scheme for the new redundantly actuated parallel mechanism by integrating force/position hybrid control with the combination of inertia feed-forward control and back propagation (BP) neural network PID control is proposed. The dynamic models including the joint space and task space are formulated explicitly in efficient and compact form by means of the principle of virtual work and d’Alembert formulations. The force/position hybrid control is implemented to perform trajectory tracking and optimize the driving force configuration in MATLAB/Simulink environment, before being applied to an actual parallel mechanism. The illustrative simulation results demonstrate that the force/position hybrid control scheme is available to provide good trajectory tracking performance. Simultaneously, the feasibility of the proposed control scheme is verified by comparison analysis with the aforementioned conventional control method.

show abstract

Section: Control Designmentioning

confidence: 99%

Trajectory Tracking Control Study of a New Parallel Mechanism with Redundant Actuation

Zhang

Fang

Zhang

et al. 2020

International Journal of Aerospace Engineering

View full text Add to dashboard Cite

show abstract

Adaptive Neural Network Control of Space-Mission Universal Mechatronic Module

Furtat,

Zhukov,

Korotkov

et al. 2023

Russ. Aeronaut.

View full text Add to dashboard Cite

Reinforcement Learning-Based Intelligent Control for Stewart Platform

Duyun,

Kabalyants,

Duyun

et al. 2024

Bulletin of Belgorod State Technological University Named After. V. G. Shukhov

View full text Add to dashboard Cite

This article presents an intelligent control system for the Stewart platform, a parallel kinematic mechanism with six degrees of freedom. One of the main challenges of such mechanisms is the presence of singular positions within the workspace, which can lead to instability. Standard control algorithms often prove ineffective when navigating these potentially unstable zones. Therefore, the proposed control system utilizes Reinforcement Learning, a type of machine learning, as its core component. The choice of this method is motivated by its effectiveness in continuous action spaces, which is crucial for ensuring smooth movement within a workspace with variable coordinates. In conditions involving operational loads, vibrations, and temperature fluctuations, real-time correction is necessary. The application of an artificial neural network enhances accuracy and flexibility, enabling the system to adapt to changing operational conditions without compromising performance by modeling complex nonlinear dependencies and learning from accumulated experience. The paper discusses the following aspects of the control system: the working scheme and overall architecture of the reinforcement learning method, the learning algorithm scheme, UML diagrams of the environment and agent classes, the architecture of the actor-critic network, the network training process, and the results of its testing, which demonstrate high efficiency in its application.

show abstract

Adaptive neural network control of hexapod for aerospace application

Cited by 3 publications

References 4 publications

Trajectory Tracking Control Study of a New Parallel Mechanism with Redundant Actuation

Trajectory Tracking Control Study of a New Parallel Mechanism with Redundant Actuation

Adaptive Neural Network Control of Space-Mission Universal Mechatronic Module

Reinforcement Learning-Based Intelligent Control for Stewart Platform

Contact Info

Product

Resources

About