NEURAL SIMULATION OF A PROTOTYPE KEYBOARD ASSEMBLY CELL WITH ADAPTIVE-CONTROL

This paper describes the integration of a feedforward neural network w ith a real-time adaptive control technique to model an assembly cell f or computer keyboards. The simulation model is able to undertake real- time forecasting of the expected behaviour of the cell, and also able to display in real time the current status of the assembly cell. The p otential of integrating artificial neural networks into simulation tog ether with an on-line real-time interface between the simulation model and the real-world system is investigated in this study. A simulation model was built in SIMAN with a neural network written in NeuralWork Professional II Plus embedded in the model to allow training and testi ng of the network. The control of the cell was achieved through a thir d-party program written in C that interfaces directly to the cell.