The control object of the PID controller often has the characteristics of highorder nonlinearity. It is easy to make the controller overshoot, oscillate, and deteriorate performance when the parameters are adjusted. To solve this problem, a selfaggregating moth flame optimization（SMFO） algorithm was proposed to optimize the PID parameters. The light intensity attraction characteristic of firefly algorithm was introduced into the conventional moth flame optimization（MFO） algorithm to improve the optimization performance.The Gaussian distribution was used to select suitable individuals in the flame population, and the position of suitable flames were periodically adjusted by using the light intensity attraction characteristic to strengthen the relationship between the flames and enhance the algorithm’s local exploration capability. The improved algorithm was used in PID control parameter optimization experiments. PID parameters were used as individuals in the algorithm, and the ITAE was used as the fitness function to optimize the PID parameters. Simulation results show that the selfaggregating moth flame optimization algorithm has better performance, compared with the moth flame optimization algorithm and Z-N method, the overshot and adjustment time are reduced by at least 26.3% and 69.6% respectively, which ensures the stability of the control system.