To measure the influence of knowledge diffusion and information exchange between enterprises on high-tech innovation networks’ production and operation, a production model and innovation model based on a network diffusion process are introduced. The knowledge diffusion process treated as an learning-by-observing process in a random network are influenced both by network’s structure and non-structure properties. We analyze the influence of diffusion process theoretically and find that, if given the precondition that initial belief and belief elasticity follow a normal distribution, an increase in mean of initial belief would lead to increase in PV and IV; otherwise, mean of belief elasticity would have a opposite effect on PV and IV under some different conditions. Finally, we give the condition to compare knowledge diffusions in two high-tech networks with same mean degree but different variance.