Linearized modeling of large-scale renewable energy stations with a voltage-source converter-based high-voltage direct current (VSC-HVDC) system faces significant challenges. These arise from the diverse dynamic characteristics of generating units and the high system order. As a result, conventional modal analysis methods are often ineffective. To address this issue, a reduced-order analysis method is proposed for the stability evaluation and oscillation source identification of such systems. Renewable energy plants are equivalenced as multiple identical units. The equivalent systems are then reduced in order, considering the dynamic interactions with the external system. System stability is assessed using modal analysis. Oscillation sources are identified based on unstable modes. Dominant components are determined through participation factor analysis. The proposed method is validated through simulation studies. It is demonstrated that the reduced-order approach significantly lowers system order, reduces computational complexity, and effectively evaluates instability risks. Moreover, system stability is enhanced by tuning parameters of the dominant components.