The frequency regulation has become one of the major subjects in microgrid power system due to the complexity structure of microgrid. In order to solve this problem, this paper proposes an improved linear active disturbance rejection control algorithm (ILADRC) that can significantly improve system performances through changing feedback control law to reduce the disturbance estimation error of extended state observer. Then, the proposed algorithm is employed in microgrid power system frequency regulation problem, which demonstrates its effectiveness. The parameters of controllers are optimized by particle swarm optimization (PSO) algorithm improved by genetic algorithm (GA). Simulations with different disturbances including sudden and stochastic change of load demand and wind turbine generation are carried out in comparison with previous studies. And robustness testing based on Monte-Carlo approach also shows better performance. So frequency stability of microgrid power system can be well guaranteed by proposed control algorithm.