Abstract:     In view of the low degree of technology correlation and lack of effective innovation cooperation among enterprises in China, the 20thNational Congress of China clearly put forward the integration development strategic decision of large, medium and small enterprises. For manufacturing enterprises, as the main force of innovation, it is urgent to establish a boundary-spanning technology innovation network (BSTIN) in which large-scale manufacturing enterprises act as the core, and upstream and downstream supporting enterprises in the supply chain and the cross-field complementary enterprises cooperate with each other. The "dual meta" and "dual degree" technology innovation have become an important strategic choice for enterprises. Therefore, it is of great significance to explore the law of coordinated interaction of manufacturing enterprises to form BSTIN for improving the systematic technology competitive advantage of manufacturing enterprises.Aiming at a series of problems, such as unclear growth elements of BSTIN, lack of the dual meta &dual degree boundary-spanning accurate guidance for the overall network growth research, unclear start-up timing of the dual meta & dual degree, and neglect of research on dynamic transformation rules of different types of networks, this paper analyzed the differences and dynamic transformation rules of different types of BSTIN from the dual meta & dual degree perspectives, so as to effectively facilitate the growth of innovation network and improve the technology innovation competitive advantage. BSTIN is divided into exploratory breadth BSTIN (A), exploratory depth BSTIN (B), incremental breadth BSTIN (C), and incremental depth BSTIN (D). Based on the epistemology and complex network theory, and after comprehensively considering node & network covariates variables and the subject & object, the factors that affect the formation of BSTIN were extracted, namely such four node variables as knowledge siphon(KS), professional matching(PM), ripple extension(RE), and imprinting inheritance(II); and such four network covariates variables as multi knowledge combination network(MKCN), medici network(MN), dynamic network(DN), and inertial network(IN). The influence tendency of the eight variables on BSTIN formation was theoretically demonstrated.In this paper, ERGM model was used to determine the specific elements of different networks, and NCA method was adopted to carry out necessity and robustness tests. Based on this, NK model was used to determine the factors interacting order and mechanism, and the verification was made based on typical cases, thus revealing the network formation transformation rules. The specific research results were obtained as follows:Result 1: The results showed that the common key elements of BSTIN are mainly reflected in the subject attributes, with the dual meta highlighting knowledge base, and the dual degree highlighting dynamic connection while heterogeneous integration, gradient transfer, multiple expansion, and professional deep cultivation elements affecting the formation of exploratory, incremental, wide, and deep BSTIN respectively. Result 2: The optimal order of A, B, C and D are as follows: KS→RE→DN→MN, MN →II→IN→PM→KS, PM→RE→DN→MN→ MKCN, MKCN→II→IN→PM, and the attraction mechanism, centripetal mechanism, combination mechanism and pull mechanism are demonstrated respectively.Result 3: The path transformation rules of different types of manufacturing enterprises′ BSTIN formation were revealed. Knowledge and subject attributes determine the transformation between dual meta BSTIN, while dual degree boundary-spanning transformation emphasizes the adjustment of knowledge, subject attributes and subject behavior, and the incremental-exploration dualmeta adjustment is earlier than the breadth-depth dual degree adjustment. The transformation between A and B, C and D reflects the logical transformation path of end-to-end circular upgrading, so as to promote the technology innovation network to achieve a higher quality through the interlacing of boundary-spanning depth and breadth. A and C, B and D transformation reflects the different choice focus and development direction, so as to achieve exploratory or incremental technology innovation goals, and the network needs to adjust the direction first, following the degree transformation.This paper will provide important enlightenment for the construction of BSTIN for manufacturing enterprises. Manufacturing enterprises can select the necessary factors for network formation according to different network construction types. According to the optimal order and mechanism, a specific network formation law of "first dual meta and then dual degree" is formed, which provides an important decision basis to grasp the overall logic, and this paper will provide effective guidance for manufacturing enterprise to objectively recognize the dynamic transformation rule of network formation and scientifically adjust the key points of network management.

Key words: boundary-spanning technology innovation network(BSTIN), formative element, acting order, dual meta, dual degree