Fast real-time SDRE controllers using neural networks. (December 2021)

• Record Type: Journal Article
• Authors: da Costa, Rômulo Fernandes; Saotome, Osamu; Rafikova, Elvira; Machado, Renato
• Is Part Of: ISA transactions. Volume 118(2021)
• Page Start: 133
• Page End: 143
• Publication Date: 2021-12
• Subjects: SDRE control -- Deep learning -- Neural control -- Stacked denoising autoencoders -- Satellite attitude control; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2021.02.019
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: This paper describes the implementation of fast state-dependent Riccati equation (SDRE) control algorithms through the use of shallow and deep artificial neural networks (ANN). Several ANNs are trained to replicate an SDRE controller developed for a satellite attitude dynamics simulator (SADS) to display the technique's efficacy. The neural controllers have reduced computational complexity compared with the original SDRE controller, allowing its execution at a significantly higher rate. One of the neural controllers was validated using the SADS in a practical experiment. The experimental results indicate that the training error is sufficiently small for the neural controller to perform equivalently to the original SDRE controller. Highlights: A real-time SDRE controller is designed using shallow and deep neural networks. A significant reduction in computation is achieved using the neural controllers. Deep denoising autoencoders were trained as deep neural controllers. The neural controller is validated through simulations and a practical experiment. Results show that the neural controller retains its high performance.