Stochastic mixed model sequencing with multiple stations using reinforcement learning and probability quantiles

Abstract In this study, we propose a reinforcement learning (RL) approach for minimizing the number of work overload situations in mixed model sequencing (MMS) problem with stochastic processing times. The environment simulates times and penalizes overloads negative rewards. To account component problem, implement state representation that specifies whether will occur if are equal to their respective 25%, 50%, 75% probability quantiles. Thereby, RL agent is guided toward while being provided statistical information about how fluctuations affect solution quality. best our knowledge, study first consider variation minimization situations.