Effects of operator learning on production output: a Markov chain approach

We develop a Markov chain approach to forecast the production output of a human-machine system, while encompassing the effects of operator learning. This approach captures two possible effects of learning: increased production rate and reduced downtime due to human error. In the proposed Markov chain, three scenarios are possible for the machine at each time interval: survival, failure, and repair. To calculate the state transition probabilities, we use a proportional hazards model to calculate the hazard rate, in terms of operator-related factors and machine working age. Given the operator learning curves and their effect on reducing human error over time, the proposed approach is considered to be a non-homogeneous Markov chain. Its result is the expected machine uptime. This quantity, along with production forecasting at various operator skill levels, provides us with the expected production output.