Abstract: Precision RV reducers are the key components in development of the industrial robot technology, but large-scale production is still challenging to achieve domestically. In this paper, a mathematical model is established to address the production imbalance and overall inefficiency in the mixed-flow assembly line of RV reducers. It proposes a swap mutation genetic algorithm considering preceding processes to overcome the traditional genetic algorithm's tendency to get trapped in local optima. The RV reducer mixed-flow assembly production line is modeled and simulated by using Flexsim software, then the effectiveness of the method is also validated. The results show that after optimization, the production cycle time is reduced by 0.8 second, the production balance rate has improved by 18%, and the smoothness index has been reduced from 1.3 to 0.72, which significantly enhances the production efficiency.