Abstract: An improved invasive weed optimization algorithm is proposed for solving multi-objective flexible job shop scheduling problem (FJSP) with released time and delivery date. The minimum completion time of jobs, the maximum work load of machines and the total work load of all machines are taken as the optimization goals to establish a FJSP model. A random key encoding scheme based on transformed sequences is proposed and a mapping relationship is set up between the continuous space and the discrete space of FJSP. An adaptive Gauss mutation operator is introduced to diversify the population in the process of weed breeding. In spatial diffusion stage, the principle of Levy flight is taken to improve the global search ability, which contributes to escape from local optimal solution. The algorithm is compared with other different algorithms and the statistical results show that the algorithm is effective for solving the multi-objective FJSP.

Key words: flexible job shop scheduling, weed optimization, adaptive Gauss mutation, Levy flight, random key coding