Estimation of Placement Process Throughput in ElectronicAssemblies Using Neural Networks

摘要

This paper discusses a neural network approach for the prediction of throughput in a placement process. Thernestimation of throughput involved the total number of placements in the process, the number of placements for eachrncomponent, and the time for sub-processes including component picking, vision inspection, and placement. A designrnof experiments based approach was used to determine the network parameters. Multivariate regression models wererndeveloped to compare with the performance of the neural networks. Neural networks can be effectively used tornestimate throughput in a placement process and as part of the objective function for feeder assignment andrnplacement sequence optimization problems.