Abstract: A production optimization model of nosiheptide fermentation process was built on the basis of the kinetic models of nosiheptide fed-batch fermentation. Decision variables were selected according to the technical flow, and the scope of the variable boundary constraints was decided. PSO (particle swarm optimization) algorithm is easy to fall into local optimum when solving complex optimization problems. To solve this problem, according to the randomicity and ergodicity of chaotic sequences, an improved PSO algorithm was proposed by introducing chaotic migration operator into PSO. The improved PSO algorithm was used to solve the production optimization model of nosiheptide fermentation, and the end time fermentation production was greatly improved. The results showed the effectiveness of the proposed algorithm.