Knowledge-based on-line vibration monitoring diagnose

ABB STAL developed some years ago a knowledge-based on-line vibration analysis system (working-name KOVA). KOVA is intended to work together with some type of vibration monitoring system, at present it is adapted to TVM 300. KOVA has no controlling function. It will only diagnose the actual situation and give the user explanations and proposals for actions to be taken. During the developing work, great experience has been gained of the features this type of system demands. This paper will present the outlines of the application and also discuss how to make diagnoses based both on general rules as well as on historical vibration cases for that particular unit (or identical units9. Another subject that this paper will outline, is the representation and evaluation of knowledge. KOVA serves as a decision-support system for the operator. Since KOVA will often give the operator more than one possible diagnosis as the cause of a fault, it is of great importance to give the operator comprehensive explanations and as many facts as possible. It is also important to rank the suggested diagnoses in some way. In KOVA these demands are effectively supported. The models and tools used to realize this functionality will be described in this paper