Neural networks for process prediction

Neural networks are powerful tools to model complex nonlinear behavior from process data. An intelligent sensor is a use of a neural network to provide the same information a physical sensor would give. The benefits of using neural networks as intelligent sensors fall into four different categories, which are outlined in this paper. This paper also describes how backpropagation neural networks run, and highlights several important issues that engineers need to keep in mind when building a neural network.