TABLE OF CONTENTS In earlier chapters of the book we have presented various techniques for condition monitoring. It was seen that most faults do have predictable symptoms during their development, from root cause to failure mode, which can be detected by mechanical, electromagnetic, optical or chemical sensors.
Condition monitoring has to establish a map between input signals and output indications of the machine condition. Classifying machine condition and determining the severity of faults from the input signals have never been easy tasks and they are affected by many factors.
Return to our simple analogy in Chapter 4 of an engineer collecting data and acting upon it. It is in this final, diagnostic stage that experienced engineers still outperform most computerised condition-monitoring systems. Experience and intelligence are extremely important in this interpretative stage when information from different sensors is sifted and condition precisely indicated by tracing the probabilities of different root causes. In recent years, there has been considerable effort to develop artificial intelligence systems that can play the roles currently performed by humans. This is important for at least two reasons.
Electrical machines, both motors and generators, are increasingly used as elements in bigger systems where operators are not necessarily experienced in their design.
A human expert can be subject to influences that make quick and consistent judgement impossible, particularly when there are many machines in the plant. Correct judgement may also depend on the knowledge and the experience of many experts who are not all available at the same time.
