M. Hadjiski. Model-based ot Model-free Approach in Control

Key Words: Artificial intelligence; automatic control; data driven methods; machine learning; model-based control; model-free control.

Abstract. Classical model-free control methods, which still have extremely large application in industry are considered in historical perspective and in accordance with their advantages and disadvantages. Some of the most popular model-free approaches, adopted last three decades (ultra-local model-based, rule-based, express system-based) are treated tacking into account their pros and cons, depending on the complexity of the control problem and the business aspect. A special attention is paid to contemporary very actual data-driven approaches. The most promising of them, oriented to AI-based technologies (especially machine learning) are discussed. A comparative analysis is carried out in order to clarify the advantages, disadvantages and challenges of model-based and model-free approaches implementation in control in big data era.