The current market economy has stepped into the stage of high-quality development led by innovation. How to effectively improve the innovation capacity of enterprises has become a widespread concern in the academic community. The article first analyzes the formation factors and consequences of transgressive innovation behavior and then combs through the dynamic collaboration and system performance generation in multi-team network structures. Based on the network centrality of social network analysis, two models based on sub-team type and team structure balance are constructed by combining structural balance theory. Regression analysis was conducted for Model 1 to clarify the effects of transgressive innovation behavior on network centrality and subteam type, and social network analysis was used to investigate the influencing factors of team structure balance in Model 2. Network centrality increases by 0.121 percentage points when the subteam type variable increases by 1 percentage point. The regression coefficient of subteam type on network centrality decreases from 0.138 to 0.114 after adding the variable of team members’ transgressive innovation behavior to Model 1. When the number of subteams is 3 and 4, the team network structure can reach a strong balance state at this time, while more than 5 will cause the growth of negative team relationships because of the transgressive innovation behaviors of team members, which will affect the balance of the team network structure. Therefore, there is a dynamic relationship between the transgressive innovative behavior of high-tech enterprises and the team network structure, and it is necessary to set up a reasonable multi-team structure in order to effectively ensure that the performance of high-tech enterprises can be improved.