The innovation network is the frontier of economic geography, although the drivers and the underlying mechanism of the evolution of this innovation network are not clear. This study employs the Exponential Random Graph Model (ERGM) and Stochastic Actor-Oriented Model (SAOM) to analyse the evolution of the drivers and the formation of the innovation network in China's photovoltaic industry, as a case study. Results show that, organizational proximity plays a direct and important role, and innovative actors tend to cooperate with actors in the same group. However, geographic proximity is not the most important influencing factor but indirectly affects the evolution of the innovation network through other proximity (organizational and cognizant). When a distance threshold of 100 km is used, organizational proximity plays a more important role as the geographic distance increases.