Predictive Dynamic Programming Heuristic Approach for Inventory Control

A common control problem for the design of inventory systems is to respond to demand variability while meeting constraints and performance goals that include a mix of different and often antagonistic objectives. The paper introduces a novel heuristic approach that brings together predictive methods and linear and dynamic programming to develop optimal solutions for controlling the operation of supply–demand systems. The approach is relevant for a broad range of applications such as water supply and distribution, power production and management, or materials storage and supply. The approach is exemplified via a case example of a water supply system comprising a tank filled by a Variable Speed Drive (VSD) pump. The design procedure is structured as a sequence of steps that address component sub-problems. Instances of the modules, together with the simulated controlled process, are implemented within an integrated software tool. The effectiveness of the approach is examined through simulation experiments. Results show that the proposed heuristic approach achieves optimization and efficient operation of supply–demand systems.