The invention discloses a multi-objective optimization array pattern synthesis method based on an improved genetic algorithm, and belongs to the field of antenna array pattern synthesis. According to the method, the genetic algorithm is improved and optimized, and in order to avoid local convergence and prevent close-to-close combination, the Hamming distance is introduced to carry out close-to-close discrimination; in order to improve the convergence speed and ensure that an excellent mode in a parent body can be inherited to the next generation, the size of a crossover operator is adaptively determined according to the fitness. According to the method, a mutual coupling effect between array elements is considered, an active unit directional diagram is combined, an accurate antenna array model is constructed, different fitness function parameters are set for different target array directional diagrams, and an improved genetic algorithm is used for performing multi-target rapid optimization on the antenna array directional diagrams; finally, array synthesis such as beam broadening, low sidelobe, beam null, phased array scanning and the like is realized, and verification is carried out through full-wave electromagnetic simulation software.