Abstract: Under the background of major power competition, complex products and systems represented by national heavy equipment are the focus of global technological competition. Breakthroughs in key core technologies face dual challenges across multiple technological levels and fields. It’s very important to enhance the overall efficiency of indigenous innovation. The existing research has identified the dual barriers in complex products and systems innovation, including horizontal integration and vertical collaborative iteration. Case studies of indigenous innovation in complex product systems such as pressurized reactor, heavy-duty gas turbines, and tunnel boring machines reveal that knowledge interaction among innovation actors is the fundamental driving force to overcome above barriers. However, existing research still has several gaps. First, several scholars argue that new knowledge is a critical driver for technological breakthroughs. However, the mechanisms by which new knowledge emerges during the process of breaking through key core technologies remain unclear. Second, existing research tends to regard the innovation consortium as homogeneous entities that uniformly pursue a strategic mission of technological catch-up and indigenous innovation. In reality, the technological bottlenecks faced by different innovation consortia vary significantly in both level and type. This calls for a more nuanced and categorized analysis of the internal mechanisms through which consortia achieve breakthroughs in complex product systems. Third, the emergence of new knowledge results from the joint influence of multiple factors. It is necessary to identify the key influencing factors that shape the evolution of new knowledge during the technological breakthrough process. In response to above research gaps, we try to do a systematic analysis of the knowledge enhancement mechanisms through which the innovation consortium achieve breakthroughs in key core technologies of complex product systems. First, innovation consortia are classified into four distinct types according to their technological breakthrough objectives and knowledge interaction strategies. Second, the model of knowledge interaction is constructed using the Lotka–Volterra framework to uncover the heterogeneous mechanisms of knowledge enhancement across different types of innovation consortia. Finally, simulation analysis is conducted to examine the influence of key factors on knowledge enhancement outcomes. The results of numerical simulations indicate that both mutually beneficial and partially beneficial symbiotic relationships among innovation agents significantly improve the effect of knowledge enhancement. Only temporary and low-level knowledge enhancement can be achieved under the independent symbiotic relationship. Moreover, the architect’s capability for knowledge orchestration is positively correlated with the effectiveness of knowledge enhancement. Even in the context of an independent symbiotic relationship among members, strong knowledge orchestration efforts by the architect can effectively stimulate knowledge interaction dynamics. In addition, the architect's knowledge breadth is also positively associated with the level of knowledge enhancement. Although extensive knowledge interaction may slow down the speed of enhancement, it ultimately results in a higher overall level of knowledge enhancement. In this study, several theoretical and practical contributions are proposed. First, it echoes scholarly discussions on collaborative innovation in complex product systems and proposes knowledge enhancement as the intrinsic driving force behind breakthroughs in key core technologies. In doing so, it advances the mechanism research of innovation consortia from a new perspective, offering a more fundamental understanding of how such breakthroughs occur. Second, based on prior research, this study categorizes the differentiated mechanisms of knowledge enhancement within innovation consortia. It moves beyond the conventional view of consortia as homogeneous entities with unified goals. This new perspective broadens the conceptualization of innovation consortia and deepens our understanding of their heterogeneous dynamics. Furthermore, regarding to practical implications, several recommendations are proposed for policy makers and practitioners. First, it is important to promote the establishment of innovation consortia led by enterprises. Giving full play to the role of enterprises as the principal agents of technological innovation, and fostering enterprise-led consortia, can effectively promote the deep integration between scientific innovation and industrial innovation. Second, in light of the characteristics of complex product systems, it is important to build differentiated types of innovation consortia. On the one hand, coordination among upstream and downstream innovation actors should be supported. On the other hand, participation should also be encouraged among actors operating in the same segment of the value chain, even if potential competitive relationships exist. Third, innovation actors should align their strategies with their positions in the industrial chain and resource endowments. Both vertical collaboration and horizontal integration can be used to form innovation consortia that are well-matched to specific technological breakthrough goals and pathways. Internal relationships within innovation consortia should be effectively managed by improving value distribution mechanisms, strengthening trust-based governance, and stimulating partners’ willingness to share knowledge. These efforts are essential for enhancing knowledge sharing and co-creation efficiency within the consortium.