Abstract:Project management has long faced the problems of a large number of project tasks, complex task relationships, and low parallel efficiency. Traditional project management methods cannot visually express task relationships, and cannot identify and decouple project tasks with high risk coupling, resulting in loss of project management schedule and cost risks, which seriously affects the effect of parallel implementation of tasks. In response to this problem, this paper proposes to use the numerical design structure matrix to express the project task relationship, construct the risk probability matrix and the risk loss matrix to evaluate the overall project risk loss value, and use the genetic algorithm to solve the process of the lowest project risk loss value. This article demonstrates the feasibility of the model's decoupling optimization method through the demonstration of the space science task parallel design project process optimization project.