Author(s)

R. Strieker

Abstract

Data acquisition and processing using simple sensors, made 'intelligent' by neural nets R. Strieker BMW j4G, D-#m? M%mc/t ^0, German?/ ABSTRACT Efforts in data acquisition and processing to use simpler and cheaper sensors, and to evaluate their unreliable and/or error-prone output by sophisticated algorithms often suffer from the lack of appropriate algorithms, not to mention the difficulties of setting them up to work with sufficient reliability in specific situations. Neural net technology can bring about dramatic changes to this scene: Data registered by the sensors and those anticipated at the output of the data processing system are normally available to serve as exemplary "learning material" to "train" appropriate neural nets. Once designed and trained, these neural nets can often, with minimum effort and complexity, largely replace the currently customary generation and operation of expensive data acquisition and processing systems, leading to simple but powerful so

Keywords