An artificial neural network on a field programmable gate array as a virtual sensor

This paper presents a brief description of our achievements and current research on "an artificial neural network on a field programmable gate array as a virtual sensor". One of the characteristics of many industrial processes is the complex interrelations among the variables of the process. This has lead to the development of "software sensors" (soft-sensors or virtual sensor) where, by means of a computer program, variables are estimated from the information gathered by other measurements. Our proposal is based on a modular design of a neural network on using digital nonlinear activation functions. The design methodology, the size and complexity of the synthesised hardware and the results of the tests in a real world problem are presented together with our conclusions on the viability for its wider use on instrumentation applications.