Abstract: Aiming at the combinatorial optimization problem of flexible job shop scheduling, an improved genetic algorithm based on global minimum load initialization is proposed, and a mathematical objective model to minimize the maximum completion time is constructed. The proposed algorithm uses an initialization method based on global minimum load selection. The quality of the initial population is improved, the algorithm convergence is speeded up, and the global search efficiency is improved; the selection of the crossover operator is improved by genetic operator and a single-point genetic mutation strategy is proposed that tends to the minimum machine load, and a robust adjustment mechanism is established; combined with the tabu search algorithm and design its neighborhood structure and taboo rules, the algorithm is used to optimize the population after genetic iteration, the shortcomings of the genetic algorithm's are overcame poor local optimization ability, and the quality of the algorithm is improved; the algorithm termination criteria is established, the time complexity is reduced, and solving efficiency of the algorithm is increased. Finally, numerical analysis and comparative experiments are carried out through benchmark test examples to verify the effectiveness of the proposed initialization method and the reliability of the proposed improved algorithm.