Generalized Data–Driven Predictive Control: Merging Subspace and Hankel Predictors

Data–driven predictive control (DPC) is becoming an attractive alternative to model as it requires less system knowledge for implementation and reliable data increasingly available in smart engineering systems. Two main approaches exist within DPC: the subspace approach, which estimates prediction matrices (unbiased large data) behavioral, data-enabled uses Hankel (allows optimizing bias/variance trade–off). In this paper we develop a novel, generalized DPC (GDPC) algorithm by merging predictors. The predicted input sequence defined sum of known, baseline sequence, optimized sequence. corresponding output computed using unbiased, predictor, while matrix predictor. By combining these two types predictors, GDPC can achieve high performance noisy even when small matrix, computationally more efficient. Simulation results benchmark example from literature show that with reduced size match data–enabled larger presence data.