Abstract

Condition Monitoring usually analyzes each tag individually using limit values provided by human experts. This results in false alarms and unhealthy conditions that are not alarmed. Using machine learning techniques, all tags on a piece of equipment, such as gas and steam turbines, compressors, pumps, valves, heat exchangers and so on, can be analyzed as a single coherent whole to draw conclusions about its current state of health. A mathematical model of the relevant tag is learned using the other tags of the same equipment. This model represents the equipment as a whole and its behavior when it is operating as it should as only data from known healthy operations are used for learning. The computed value is compared to the measured value. If they agree, the current state is healthy. If they do not, an alarm is released, and a maintenance activity must follow. This method is proven to be far more successful than standard condition monitoring in preventing false alarms and alarming unhealthy states.

You can access this article if you purchase or spend a download.