Subspace based wavelet preprocessed data-driven predictive control

Vineet Vajpayee*, S. Mukhopadhyay, A. P. Tiwari

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


This paper introduces a methodology for designing a subspace‐based data‐driven predictive control with wavelet preprocessing. In a data‐driven control, especially when SNR is low, it becomes difficult to obtain reliable predictor coefficients. Therefore, it is imperative to have a processed and informative dataset for stable controller operation. Wavelet being capable of better noise rejection from process dynamics motivates to perform wavelet filtering before designing the control law. Methodology for deriving the predictor from subspace matrices of processed data is presented. A predictive controller, estimated from the dataset, is designed for power control of a nuclear reactor core for a load‐following operation. The efficacy of the proposed technique is demonstrated by Monte Carlo simulations in stationary as well as non‐stationary noise cases.
Original languageEnglish
Pages (from-to)357-371
JournalINCOSE International Symposium
Issue numbers1
Publication statusPublished - 23 Dec 2016
Externally publishedYes
EventAPCOSEC 2016 - Bangalore, India
Duration: 9 Nov 201611 Nov 2016


Dive into the research topics of 'Subspace based wavelet preprocessed data-driven predictive control'. Together they form a unique fingerprint.

Cite this