Abstract: The rapid and accurate identification of the transient stability status of power systems is a crucial prerequisite for ensuring the safe and stable operation of large power grids. Compared with traditional physical analysis methods, data-driven transient stability assessment technology for power systems has significant advantages in solving complex nonlinear mapping and rapid evaluation, and has become an important direction in current research on transient stability assessment of power systems. This paper establishes the basic architecture of data-driven transient stability assessment technology based on the demand scenarios of power system transient stability assessment and the general intelligent application framework, and analyzes the functions of each process link in a data-driven context from the aspects of offline training, online application and feedback update. Furthermore, focusing on data enhancement, machine learning algorithms, and learning mechanisms, this paper reviews the progress of application research work and key technologies of data-driven technology in power grid transient stability assessment, and analyzes the advantages and disadvantages of different models and methods in solving the fitting and generalization capabilities of power system transient stability assessment models. Lastly, in light of the new characteristics of transient stability assessment for high proportion renewable power systems and the ongoing advancements in artificial intelligence technology, this paper anticipates the future research direction of power system transient stability assessment technology from three perspectives: data, model, and application, aiming to provide technical reference for the digitization and intelligent of power grid transient stability assessment.