Neural networks are computationally efficient mathematical models that can be used to model quantitative and qualitative data. A neural network can be created through training with known input and output load-deflection data such that it learns to generalize the material characteristics without over-predicting the training data and losing its ability to anticipate behavior outside the training set. The challenge in creating a neural network model of a rubber bushing in a virtual model of a prototype assembly, for instance, is the lack of a physical prototype assembly. This paper describes a method by which data can be measured from a virtual prototype and used to define an appropriate data acquisition for the physical bushing. Training data can then be acquired using these guidelines and used for neural network model development. Subsequently, the enhanced model can then be used in the virtual simulation environment to increase the accuracy of the simulation results.
Multilayer perceptron neural network model development for mechanical ventilator parameters prediction by real time system learning
