利用随机微分博弈理论研究了产学研协同创新主体间的知识共享问题，建立了一个随机微分博弈模型。运用哈密尔顿-雅可比-贝尔曼方程分别求得了Stackelberg主从博弈和协同合作博弈下均衡的知识共享策略、知识创新量的期望值和方差以及Stackelberg主从博弈下均衡的创新补贴比例，并对此两种博弈结果进行了比较。研究发现：(1)无论是Stackelberg主从博弈还是协同合作博弈，当知识共享的成本越高，共享的知识量越少；知识创新能力越大，知识共享的边际收益越高，共享的知识量越多；知识折旧率越高，共享的知识量越少。(2)协同合作博弈下创新主体共享的知识量分别高于Stackelberg主从博弈下的知识共享量，且协同合作博弈下产学研知识协同创新系统的总收益大于Stackelberg主从博弈下系统的总收益。(3)协同合作博弈下知识创新量的期望值和方差大于Stackelberg主从博弈下知识创新量的期望值和方差。

In this paper, knowledge sharing between enterprise and university in industry-university-research synergetic innovation is studied by constructing a stochastic differential game model involving stochastic disturbance. Dynamic programming method is employed to derive the equilibrium knowledge-sharing strategies and subsidy ratio both in Stackelberg game and cooperative game. By comparing the obtained results, we find that: (1) both in Stackelberg game and cooperative game, the more knowledge sharing cost and knowledge decay rate, the less knowledge is shared. The higher knowledge innovation ability and marginal profit, the more knowledge is shared; and (2) the optimal knowledge sharing, the total profit, the expected innovative knowledge and its variance of the cooperative game are higher than that of the Stackelberg game respectively.