The invention relates to an independent microgrid configuration optimization method based on an improved adaptive genetic algorithm. The method includes the following steps: setting the structure of a microgrid system; establishing an optimization configuration model of the microgrid ; on the premise that the power supplying reliability of the microgrid system is met, targeting at minimizing a comprehensive investment cost of the microgrid, using the improved adaptive genetic algorithm to carry out configuration optimization; and searching for a group of optimal power supply configuration so that the value of a cost target function is minimum. The improved adaptive genetic algorithm (IAGA) is used to solve a configuration optimization problem and the IAGA uses a neuron function sigmoid to design adaptive crossover rates and mutation rates, and an crossover rate and mutation rate of a unit with a maximum fitness value are each improved to a specific positive number respectively so that it is prevented that an excellent unit is in a stagnant state and converges too early and thus the IAGA is enabled to jump out of a local optimal solution to look for a global optimal solution.