In order to solve the problems of large computation amount, long operation time and difficulty in ensuring optima-lity, an improved method combining particle swarm optimization algorithm and Gaussian pseudospectral method was proposed. Firstly, particle swarm optimization algorithm was used for track pre-planning to ensure the fast realization of approximate optimal solution. Secondly, according to the selection of relative positions of collocation points of Gaussian pseudospectral method, the path points of pre-planned particle swarm optimization were fitted, which was used as the initial reference instruction of Gaussian pseudospectral method, so as to solve the problem of initial value sensitivity of pseudospectral method and accelerate the convergence speed of optimization algorithm. Finally, simulation experiments were carried out by taking the performance index of unmanned aerial vehicle formation, flight environment and cooperative flight constraints into consideration. Simulation results verify the importance of the selection of initial value and show that the designed algorithm could improve the optimization and convergence speed of the solution. The research results can be used to quickly plan multi-dimensional and high-precision guidance instructions for coordinated flight control of multiple UAVs, and have certain reference value for the realization of intelligent autonomous flight.