System, method and apparatus for monitoring the health of railcar wheelsets
A system and method for monitoring the operating condition of a wheelset on a railcar comprising a sealed unit mounted on or near a wheelset of the railcar for collecting data from the wheelset and performing AI analyses on the collected data to determine the operational condition and predict failure modes for the wheelset. Results are communicated off-railcar wirelessly via one or more of several different methods.
1. A method for operating a wheelset health system, comprising:
obtaining, by a processor, sensor data generated by a plurality of sensors coupled to a railway asset;
providing the sensor data as input data to one or more machine learning algorithms;
performing operations by the one or more machine learning algorithms to translate the input data to output data by (i) identifying a pattern in the sensor data that is predictive of a fault condition of a wheelset of the railway asset and (ii) classifying a predicted fault condition into a category for a bearing or wheel of the wheelset of the railway asset; and
causing, by the processor, behavior of the wheelset health system to be modified based on the output data;
wherein the behavior of the wheelset health system is modified by changing a threshold value, changing when a signal is to be wirelessly sent from the wheelset health system, or changing a rate at which the sensor data is sampled.
2. The method according to claim 1 , further comprising:
verifying, by the processor, that the predicted fault condition is a valid fault condition of the bearing based on new sensor data generated by the plurality of sensors; and
causing, by the processor, a remote device to be notified of the predicted fault condition when the predicted fault condition is verified as being a valid fault condition.
3. The method according to claim 1 , wherein the sensor data comprises at least one of vibration data for the wheelset, a temperature measurement for the bearing of the wheelset, a temperature measurement for an ambient environment, a position for a wheel of the wheelset, a speed of the railway asset, and a weight of the railway asset.
4. The method according to claim 1 , wherein at least one sensor of the plurality of sensors is mounted on or adjacent to a bearing adapter of the wheelset or is embedded in the bearing adapter.
5. The method according to claim 1 , further comprising:
determining whether a current speed of the railway asset is within a predetermined range of speeds;
wherein the operations are only performed by the one or more machine learning algorithms when a determination is made that the current speed of the railway asset is within a predetermined range of speeds.
6. The method according to claim 1 , further comprising characterizing a defect in the wheelset by correlating vibration measurements of the sensor data to a load impact on the wheelset, when the fault condition is predicted.
7. The method according to claim 1 , wherein the operations performed by the one or more machine learning algorithms further comprise:
using vibration data to predict failures in at least one of the bearing and wheel of the wheelset; and
using temperature data to identify an imminent defect of the wheelset.
8. The method according to claim 1 , wherein a rate at which the sensor data is sampled is variable based on a type of a previously predicted failure of the wheelset.
9. The method according to claim 8 , wherein the behavior of the wheelset health system is modified by changing the rate at which the sensor data is sampled.
10. The method according to claim 1 , further comprising:
determining a time when the processor should be enabled based on a number of rotations detected for a wheel of the wheelset; and
enabling the processor at the determined time.
11. The method according to claim 1 , further comprising:
determining a time when the processor should be enabled based on a mileage of the railway asset; and
enabling the processor at the determined time.
12. The method according to claim 11 , further comprising disabling the processor when a remote device has been notified of the predicted fault condition.
13. The method according to claim 1 , wherein the fault condition of the bearing is predicted when the sensor data indicates that a temperature of the bearing has increased to a threshold temperature.
14. The method according to claim 1 , further comprising detecting a defect in the bearing when the sensor data indicates that the bearing is experiencing a threshold level of vibration.
15. The method according to claim 14 , further comprising classifying the defect based on at least one of a speed of the railway asset and a weight of the railway asset.
16. The method according to claim 14 , further comprising using a classification of the defect to identify at least one specific component of the bearing.
17. The method according to claim 14 , wherein the operations are performed by the one or more machine learning algorithms to predict the fault condition of the bearing when the defect in the bearing is not detected.
18. A wheelset health system, comprising:
a processor;
a non-transitory computer-readable storage medium comprising programming instructions that are configured to cause the processor to:
obtain sensor data generated by a plurality of sensors coupled to a railway asset;
provide the sensor data as input data to one or more machine learning algorithms;
using the machine learning algorithms to translate the input data to output data by (i) identifying a pattern in the sensor data that is predictive of a fault condition of a wheelset of the railway asset, and (ii) classifying a predicted fault condition into a category for bearing or wheel of the wheelset of the railway asset; and
cause behavior of the wheelset health system to be modified based on the output data;
wherein the behavior of the wheelset health system is modified by changing a threshold value, changing when a signal is to be wirelessly sent from the wheelset health system, or changing a rate at which the sensor data is sampled.
19. The wheelset health system according to claim 18 , wherein the processor is further caused to:
verify that the predicted fault condition is a valid fault condition of the bearing based on new sensor data generated by the plurality of sensors; and
cause a remote device to be notified of the predicted fault condition when the predicted fault condition is verified as being a valid fault condition.
20. The wheelset health system according to claim 18 , wherein the sensor data comprises at least one of vibration data for the wheelset, a temperature measurement for the bearing of the wheelset, a temperature measurement for an ambient environment, a position for a wheel of the wheelset, a speed of the railway asset, and a weight of the railway asset.
21. The wheelset health system according to claim 18 , wherein at least one sensor of the plurality of sensors is mounted on or adjacent to a bearing adapter of the wheelset or is embedded in the bearing adapter.
22. The wheelset health system according to claim 18 , wherein the processor is further caused to:
determine whether a current speed of the railway asset is within a predetermined range of speeds;
wherein the analysis of the sensor data is only performed when a determination is made that the current speed of the railway asset is within a predetermined range of speeds.
23. The wheelset health system according to claim 18 , wherein the processor is further caused to characterize a defect in the wheelset by correlating vibration measurements of the sensor data to a load impact on the wheelset, when the fault condition is predicted.
24. The wheelset health system according to claim 18 , wherein the one or more machine learning algorithms is configured to:
use vibration data to predict failures in at least one of the bearing and wheel of the wheelset; and
use temperature data to identify an imminent defect of the wheelset.
25. The wheelset health system according to claim 18 , wherein a rate at which the sensor data is sampled is variable based on a type of a previously predicted failure of the wheel set.
26. The wheelset health system according to claim 25 , wherein the behavior of the wheelset health system is modified by changing the rate at which the sensor data is sampled.
27. The wheelset health system according to claim 18 , wherein the processor is enabled at a time determined based on a number of rotations detected for a wheel of the wheel set.
28. The wheelset health system according to claim 27 , wherein the processor is enabled at a time determined based on a mileage of the railway asset.
29. The wheelset health system according to claim 28 , wherein the processor is disabled when a remote device has been notified of the predicted fault condition.
30. The method according to claim 18 , wherein the fault condition of the bearing is predicted when the sensor data indicates that a temperature of the bearing has increased to a threshold temperature.
31. The method according to claim 18 , wherein the processor is further caused to detect a defect in the bearing when the sensor data indicates that the bearing is experiencing a threshold level of vibration.
32. The method according to claim 31 , wherein the processor is further caused to classify the defect based on at least one of a speed of the railway asset and a weight of the railway asset.
33. The method according to claim 32 , wherein the processor is further caused to use a classification of the defect to identify at least one specific component of the bearing.
34. The method according to claim 33 , wherein the one or more machine learning algorithms are configured to predict a fault condition of the bearing when the defect in the bearing is not detected.