Using SFA model to decompose and modify the offset of the sample input, and introducing the Malmquist index to improve the operating performance model, then, combined with Tobit model, which is taken into account the combination of environmental effect, random error and dynamic analysis, and the spatial and temporal heterogeneity and influence factors are introduced. Thus, this paper explores the heterogeneity and influence mechanisms of the patented innovations operational performance of knowledge-intensive manufacturing from the method and content. The results show that the patent innovation operational performance which is considering the environmental effect and random error is significantly different, and the environmental factors have a temporal and spatial heterogeneity in the patent innovation operational performance; The adjusted patent innovation operational performance is a double fluctuation, and the difference is obvious in each period; After the adjustment, the growth rate of the patent innovation operational performance is relatively large, and the technical level index is improved in different degrees, which is the main driving force for the improvement of the patent innovation operational performance; Government support has a significant positive effect on the patent innovation operational performance of knowledge-intensive manufacturing, and intellectual property protection have a significant inhibitory effect, and the market competition, enterprise scale and technology digestion and absorption capacity are not obvious to promote the patent innovation operational performance. [ABSTRACT FROM AUTHOR]