在测度我国30个省域近10年创新要素集聚水平和创新绩效的基础上构建“创新要素集聚—创新绩效”拟合模型，以Q型聚类法分析为基础证实不同类型区域溢出效应均存在门槛值，运用MATLAB对相关数据进行曲线拟合，得出我国创新要素集聚对创新绩效溢出效应的最优集聚水平。结果表明：我国区域创新要素集聚已形成多层次的分布格局，区域之间的集聚水平差距较大；创新要素集聚对区域创新绩效的溢出效应并非单调递增或递减，而是呈现倒“U”型趋势，说明只有适度的集聚水平才能有效发挥集聚效应；在不同的集聚水平层级下，创新要素集聚对创新绩效的溢出效应门槛值存在差异；当创新要素集聚水平达到门槛值时，创新绩效最大，此时为最优集聚水平。

On the basis of measuring the innovation elements agglomeration level and innovation performance of China's 30 provinces in the past 10 years, the "innovation elements agglomeration-innovation performance" fitting model is constructed. Based on the analysis of Q-type clustering method, it is confirmed that there are thresholds for different types of regional spillover effects. Using MATLAB to curve the relevant data, we can get the optimal agglomeration level of innovation elements agglomeration effect on innovation performance. The results show that China's regional innovation elements agglomeration has formed a multi-level distribution pattern, and the level of agglomeration between regions is relatively large. The spillover effect of the innovation elements agglomeration on regional innovation performance is not monotonously increasing or decreasing, but rather showing an inverted “U” trend. The effect indicates that the agglomeration level can only effectively exert the agglomeration effect under the moderate threshold. Under different levels of agglomeration, there is a difference in the threshold of the spillover effect of innovation elements agglomeration on innovation performance. When the level of innovation elements agglomeration reaches the threshold, the innovation performance reaches the maximum, and the optimal agglomeration level is reached at this time.

教育部人文社会科学规划基金项目“高校创新资源集聚对区域创新绩效的溢出效应研究”（编号：17YJA630111）、国家自然科学基金重点项目“面向产业平台创新生态的协同创新机理与战略选择”（编号：71532003）、中央高校基本科研业务费专项资金资助“企业协同创新风险对协同创新绩效的影响机理研究”（项目编号：PTRW1820）