A Neural Network Implementation on Embedded Systems

Abstract

This dissertation presents a solution for embedded neural networks across many types of hardware and for many applications. The software package presented here allows the user to develop a neural network for a desired application, train the network, embed it on most platforms, and verify its functionality. This software supports advanced and very powerful types of neural networks including cascade, fully, and arbitrarily connected networks. It also supports several different training algorithms both first and second order. This system automates the process of transforming the trained neural network to an embedded neural network on most microcontrollers with a C compiler. There is also an assembly language neural network highly optimized for speed based on an inexpensive 8-bit PIC microcontroller. Software for testing and verifying functionality of the embedded neural networks is also included. Several neural network examples are also shown being calculated on the embedded system.