A control algorithm for a non-stationary batch service production system using Kalman filter

•In this paper, a lot sizing problem considering the non-stationary nature of order arrivals within a fully automated manufacturing system is considered.•A new control algorithm in the shared domain of CT and OR is presented for the fully automated batch service production system.•The results show that the Kalman Filter based model outperforms its counterparts i.e., the Exponential Filter based model and its Fuzzified version. This paper presents a dynamically optimized control algorithm for a fully automated batch service production process, with non-stationary nature of order arrivals. The problem of interest is a non-stationary SELSP, i.e., stochastic economic lot sizing problem, with erratic order flow, where the objective is to continuously optimize the production lot size according to the real-time detection of demand fluctuations. Such a problem has a dual nature of being a planning problem, which used to be dealt with within the operational research (OR) community and a control problem, often dealt with within the control theory (CT) circle. To tackle the problem a Kalman Filter (KF) based approach has been employed, as one of the most reliable methods in CT, with an edge over its previously studied alternatives, i.e., Exponential Filter (EF) and its fuzzified version. It is illustrated that the adopted KF performs much better both in terms of efficiency metrics of OR, and responsiveness metrics of CT. The results advocate that the new generation of automated production planning and control systems may benefit from the concepts and methods from both communities of OR and CT, in an integrated fashion.