Nonlinear Data-Driven Predictive Control Using Deep Subspace Prediction Networks

Mircea Lazar, Mihai Serban Popescu, Maarten Schoukens

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

1 Downloads (Pure)


Indirect data-driven predictive control (DPC) algorithms for nonlinear systems typically employ multi-step predictors, which are identified from input-output data using neural networks. In this paper we put forward a unifying multi-step prediction network architecture, i.e., the deep subspace prediction network (DSPN). We then prove that the DSPN architecture specialized to multi-layer-perceptron neural networks recovers the linear predictor corresponding to subspace predictive control for a sufficient number of hidden layer neurons. Hence, we establish a well-posed generalization of subspace predictive control for nonlinear systems. Moreover, we develop a regularized DSPN architecture that embeds a linear subspace predictor to improve extrapolation properties for non-training data. Simulation results on a benchmark inverted pendulum show that nonlinear DPC based on DSPN achieves high control performance for both noiseless and noisy data.

Originele taal-2Engels
Titel2023 62nd IEEE Conference on Decision and Control, CDC 2023
UitgeverijInstitute of Electrical and Electronics Engineers
Aantal pagina's6
ISBN van elektronische versie979-8-3503-0124-3
StatusGepubliceerd - 19 jan. 2024
Evenement2023 62nd IEEE Conference on Decision and Control (CDC) - Singapore, Singapore, Singapore, Singapore
Duur: 13 dec. 202315 dec. 2023
Congresnummer: 62


Congres2023 62nd IEEE Conference on Decision and Control (CDC)
Verkorte titelCDC 2023


Duik in de onderzoeksthema's van 'Nonlinear Data-Driven Predictive Control Using Deep Subspace Prediction Networks'. Samen vormen ze een unieke vingerafdruk.

Citeer dit