Using neural networks to select a control strategy for automated storage and retrieval systems

The development of an automated control system for automated storage and retrieval systems (AS/RS) is described. The control system is able to deal with changes in system configuration and multiple performance requirements. The approach utilizes artificial neural networks to learn from the simulation results of three sets of experimental designs: complete design, fraction design, and orthogonal array. The inputs of the neural network include system configuration and required performance levels, while the outputs are the control strategy for four decision points: storage location assignment, retrieval location selection, queue selection, and job sequencing. Different topologies and training parameters are examined to obtain a properly trained neural network for AS/RS control strategy. The experimental results show that the trained network is able to identify 84 novel data correctly and thus the feasibility of the proposed approach is demonstrated.