TR2017-217

Stochastic Model Predictive Control


    •  Mesbah, A., Di Cairano, S., Kolmanovsky, I.V., Stochastic Model Predictive Control, February 2018.
      BibTeX TR2017-217 PDF
      • @book{Mesbah2018feb,
      • author = {Mesbah, Ali and Di Cairano, Stefano and Kolmanovsky, Ilya V.},
      • title = {Stochastic Model Predictive Control},
      • year = 2018,
      • month = feb,
      • url = {https://www.merl.com/publications/TR2017-217}
      • }
  • MERL Contact:
  • Research Area:

    Control

Abstract:

Stochastic Model Predictive Control (SMPC) accounts for model uncertainties and disturbances based on their statistical description. SMPC is synergistic with the well-established fields of stochastic modeling, stochastic optimization, and estimation. In particular, SMPC benefits from availability of already established stochastic models in many domains, existing stochastic optimization techniques, and wellestablished stochastic estimation techniques. For instance, the effect of wind gusts on an aircraft can be modeled by stochastic von Karman and Dryden's models but no similar deterministic models appear to exist. Loads or failures in electrical power grids, prices of financial assets, weather (temperature, humidity, wind speed and directions), computational loads in data centers, demand for a product in marketing/supply chain management are frequently modeled stochastically thereby facilitating the application of the SMPC framework.
A comprehensive overview of various approaches and applications of SMPC has been given in the article. Another overview article in Encyclopedia of Systems and Control is focused on tube SMPC approaches. This chapter provides a tutorial exposition of several SMPC approaches.