Discovering system insights using visual interactive meta-simulation

This paper describes the technique of visual interactive meta-simulation with the aid of an example. The method consists of developing a meta-model from an original simulation model, i.e. a model of a model. The meta-model is implemented as a neural network, trained by the user from repeated runs from the simulation model. The paper considers the validation issues for a neural network meta-model. Once a valid meta-model has been produced, its speed can be exploited so that visualisation techniques such as 2D graphs and 3D response or decision surfaces can be quickly generated. This paper describes the visualisation aspect of the meta-modelling technique with reference to a large-scale manufacturing problem.