Machinery Prognostics

Many machinery have rotating components such as a motor, a shaft, bearings and loads. Each of these components can wear out, get damaged, drift out of alignment, become unbalanced, etc. By adding the proper sensors to the machine, these faults may be detected before significant damage occurs. A maintainer can be notified of impending failure before it occurs and repair can be scheduled to minimize downtime and maintenance costs.

The Spectraquest Machine Fault Simulator can be used to simulate these fault conditions. First, a no fault condition is simulated and data recorded. This data is used to train the machine learning algorithm for the baseline normal condition. Then, various faults are introduced gradually. These faults include imbalances in the rotating disks attached to the spinning shaft. These imbalances can be increased stepwise up to the point where the imbalance will nearly cause damage. In the same way, the bearings can be adjusted out of alignment, faulty bearings installed, the shaft can be deformed or damaged, etc. The data from all of these tests are used to train the machine learning algorithms.

Different machine learning algorithms can be trained and the best selected to correctly identify the fault conditions.

The system is connected to either wired Ethernet, Wifi or cellular to provide reports and notifications to maintainers.