Process model for a simulation-based early warning system using artificial intelligence
Abstract
In the present paper a procedure model for forecasting component of an early warning system is developed. For this purpose, the simulation tool Siemens - Plant Simulation is combined with Artificial Intelligence (AI) in the form of neural networks. A prototypical production layout is used to calculate the system throughput under different parameter configurations (eg. processing times, buffer sizes). The simulation results are used to train a neural network, which then predicts the system throughput based on the current parameter configuration. The article should show the possibility and the benefits of the combination of simulation and AI. Two approaches have been developed. The first approach describes a general procedure for combining simulation and AI as an early warning system in production and logistics. The second approach describes a possible improvement of the prognosis and solution quality of the system. Here, commercial tools (Matlab, Excel) and open-source tools (TensorFlow, Kreas) were used.
