Abstract: As the key building block of power generation schedules, security-constrained unit commitment (SCUC) is of great significance in power system operation. Consequently, this paper first briefly outlines the SCUC problem from viewpoints of the physical model and the solving methods. Then, the research progress of SCUC driven by the physical model is reviewed from five aspects, namely, multi-objective, diversified decision variables, uncertainty, multi-time scale, and multi-variate constraints, with the knowledge of their underlying challenges. At the same time, the existing research results of SCUC problems based on artificial intelligence and data-driven are generalized, and the characteristics, advantages, and disadvantages of different types of methods are analyzed. Finally, the future research direction of data-driven SCUC is discussed.