[Objective] This study proposes a method for identifying technological evolution paths, enabling comprehensive extraction of key technologies and their branches within a specific domain, thus revealing the trajectory of technology evolution. [Methods] Firstly, we devise a graph embedding model to integrate patent characteristics into multidimensional semantic vectors, employed to the technology paths expansion and improve the effect of community division. Secondly, we propose the expansion method of the main path and its derivative path from the perspective of network topology and semantic correlation, and construct the metric of technology convergence point to identify the promising fields. [Results] Taking UAV flight control system technology as a case study, we identify four sub-fields of technological evolution paths and branches, determining that pattern recognition, multiprocessor and data fusion technologies within this field hold promising prospects. [Limitations] Our method has not incorporate the formation mechanism of the technological evolution pattern into the identification framework. [Conclusions] The proposed method, which integrates graph embedding and path analysis, demonstrates significant advantages in terms of evolutionary path identification effectiveness and broad applicability.