Multi-agent group combat is an important form in the future battlefield.As a key technology in group operations,formation control has a wide range of application.Focusing on the formation control law,two optimal control methods are applied:linear quadratic regulator (LQR) and model predictive control (MPC).The methods are analyzed in a circular formation scene.Based on the similarity consistency algorithm,utilizes second-order dynamic model to explore the optimal control effect of four agents rotating around the trajectory center in the two-dimension plane.The results show that the model predictive control has larger speed of convergence,higher control stability and smaller side errors than the linear quadratic regulator method.The lateral error of the formation agent following a circular trajectory is greatly reduced by MPC with lateral acceleration compensation.