Automatic Faults Diagnosis by Application of Neural Network System and Condition-based Monitoring Using Vibration Signals

Companies invest a lot of effort in predictive maintenance due to the relationship with profit and equipment availability. The principal aim is to predict the occurrence of early faults, allowing repairs to be planned. On the other hand, environmentally friendly practices with higher standards of health and safety in industry can be respected, avoiding breaches of the law. With this aim, a special type of spectrum has been developed that uses fixed frequency bands and vibration severity levels. The special spectrum＇s data is used in a neural network, which detects a fault in the early stages and an automatic diagnosis is obtained quickly and reliably. This technique is especially oriented to intranet implementation systems to diagnose the real health condition of any rotating machinery in real-time, optimizing both management maintenance and production. It is used in real measurements of rotating machinery vibration, monitoring signals to obtain the results and conclusions about this technique.