HP Labs Technical Reports

Model-Based Reasoning in Analytical Instrumentation

Dickinson, Ian

HPL-92-157

Keyword(s):

Abstract: Developing expert systems to aid users of instruments and instrumentation systems is not a new idea. However, we so far have observed little published success in deploying such systems. This article briefly reviews some of the problems with traditional rule-based architectures, then goes on to suggest a new approach: apply model-based reasoning techniques. Models have various characteristics that address some of the difficulties in traditional expert systems. The article gives a general review of model based reasoning and then focuses on an example of model based diagnosis, with examples from diagnosing faults in gas chromatographs. Development of a prototype of a diagnosis engine using the principles described here is currently underway in a collaborative research project between HP Laboratories and HP Little Falls Division. This article is due for publication in a special edition of Laboratory Robotics and Automation, focusing on work at HP.

Back to Index