Method and system for data driven machine diagnostics

1. A system for data driven diagnostics of a gearbox, the system comprising:

a gearbox comprising a housing, a first shaft and a second shaft each extending into and protruding outwardly from the housing, the first shaft being supported for rotation by a first bearing pair, the second shaft being supported for rotation by a second bearing pair, a third shaft disposed in the housing and being supported by a third bearing pair, a first gear disposed in the housing and mounted on the first shaft, a second gear disposed in the housing and mounted on the second shaft, a third gear and a fourth gear each mounted on the third shaft, the third gear meshing with the first gear and the fourth gear meshing with the second gear;

a plurality of sensors each positioned in the housing; and

a processor configured to run a machine learning model, the machine learning model being configured to receive operational data from at least one of the plurality of sensors, and process the operational data to determine diagnostic information of the gearbox,

wherein the machine learning model is trained using simulated operational data from a defect simulation from a simulation environment of the gearbox,

wherein the simulation environment comprises a finite element model of structural dynamics of the gear reducer and is configured to generate data pairs which comprise defect types and dynamic response signals, and

wherein the learning model comprises a neural network that comprises an input layer that has a plurality of first nodes that correspond to the dynamic response signals, an intermediate layer configured to group similar sensor signals, and an output layer that has a plurality of second nodes that correspond to the defect types.

2. The system of claim 1 , wherein the diagnostic information comprises a prediction that at least one of a gear defect, a rolling element defect, or a shaft defect exists in the gearbox.

3. The system of claim 1 , wherein the diagnostic information comprises a prediction that a defect of at least one of the following categories exists in the gearbox: plastic deformation, surface fatigue, or cracking.

4. The system of claim 1 , wherein the machine learning model is further configured to receive the operational data from an accelerometer or a microphone.

5. The system of claim 1 , where the neural network is trained using backpropagation.

6. The system of claim 1 , wherein the simulation environment is configured to be updated based on information received about the gearbox to calibrate the simulation environment to the gearbox.

7. The system of claim 1 , wherein the machine learning model is trained using at least one of historical data from other gearboxes that are similar to the gearbox, prototype test data, or historical data of the gearbox.

8. The system of claim 1 , further comprising a data logger configured to store historical data of the gearbox.

9. The system of claim 1 , further comprising a user interface configured to display the diagnostic information to a user.

10. The system of claim 1 , wherein the plurality of sensors comprises at least one gear sensor.

11. The system of claim 1 , wherein the plurality of sensors comprises at least one bearing sensor.

12. The system of claim 1 , wherein the plurality of sensors comprises at least one shaft sensor.

13. The system of claim 1 , wherein the machine learning model is further configured to predict, based on the operational data, an absence of a failure of the gearbox for period of time.

14. A method for data driven diagnostics of a gearbox, the method comprising:

providing a gearbox comprising a housing, a first shaft and a second shaft each extending into and protruding outwardly from the housing, the first shaft being supported for rotation by a first bearing pair, the second shaft being supported for rotation by a second bearing pair, a third shaft disposed in the housing and being supported by a third bearing pair, a first gear disposed in the housing and mounted on the first shaft, a second gear disposed in the housing and mounted on the second shaft, a third gear and a fourth gear each mounted on the third shaft, the third gear meshing with the first gear and the fourth gear meshing with the second gear;

providing a plurality of sensors each positioned in the housing;

training a machine learning model with simulated operational data from a defect simulation from a simulation environment of the gearbox;

receiving, by the machine learning model, operational data from at least one of the plurality of sensors; and

processing, using the machine learning model, the operational data to determine diagnostic information of the gearbox,

wherein the simulation environment comprises a finite element model of structural dynamics of the gear reducer and is configured to generate data pairs which comprise defect types and dynamic response signals, and

wherein the learning model comprises a neural network that comprises an input layer that has a plurality of first nodes that correspond to the dynamic response signals, an intermediate layer configured to group similar sensor signals, and an output layer that has a plurality of second nodes that correspond to the defect types.

15. The method of claim 14 , further comprising:

simulating, by the simulation environment, a plurality of defects to produce a plurality of simulated sensor data to obtain the data pairs; and

sending the data pairs to the machine learning model.

16. The method of claim 14 , further comprising updating the simulation environment based on information received about the gearbox to calibrate the simulation environment to the gearbox.

17. The method of claim 14 , wherein the diagnostic information comprises a prediction that at least one of a gear defect, a rolling element defect, or a shaft defect exists in the gearbox.

18. The method of claim 14 , wherein the training, the receiving, and the processing are executed by a processor.