Railway track geometry defect modeling for predicting deterioration, derailment risk, and optimal repair

1. A method, comprising:

logically dividing a railroad network according to spatial and temporal dimensions with respect to historical data collected for the railroad network, the spatial dimensions including line segments of a specified length and the temporal dimensions including inspection run data for inspections performed for each of the line segments over a specified period of time;

creating, via a computer processor, a track deterioration model from the historical data, current track conditions collected from one or more data sources, and traffic data;

identifying geo-defects occurring at each inspection run from the track deterioration model;

analyzing quantified changes in the geo-defects measured at each inspection run;

predicting from the quantified changes in the geo-defects, where at least one of the geo-defects is a Class II geo-defect, with an amplitude below a tolerance level of a safety standard, a probability of the least one Class II geo-defect deteriorating into a Class I defect with an amplitude in violation of the safety standard, within a specified period of time, the Class I defect defined as one that is mandated to be repaired upon discovery;

upon determining the probability reaches a threshold value, scheduling a repair to remedy the at least one Class II geo-defect within the specified period of time;

upon determining the probability does not reach a threshold value, determining a repair decision based on historically determined costs associated with previous comparable repairs; and

initiating the repair decision; wherein

determining the repair decision comprises minimizing a minimum total expected cost based on the historically determined costs.

2. The method of claim 1 , wherein the line segments are defined as connecting two cities, and the spatial dimensions further include at least one of track identifiers and mile post locations.

3. The method of claim 1 , wherein the railroad network is further divided into non-overlapping lots each of a length shorter than a length of the line segments, the method further comprising:

aggregating the geo-defects by inspection run for each geo-defect type.

4. The method of claim 3 , wherein geo-defect types include at least one of:

align;

cant;

dip;

gauge;

harmonic cross-level;

over-elevation;

reverse cross-level;

super cross-level elevation;

surf;

twist;

warp; and

wear.

5. The method of claim 1 , wherein the analyzing the quantified changes in the geo-defects includes analyzing amplitude changes for each of the geo-defects at each inspection run, and wherein a 90 percentile of an amplitude is designated to represent a track condition for a corresponding inspection run.

6. The method of claim 1 , wherein the railroad network is further divided into sections of two miles in length, and the calculating a derailment risk includes:

spatially aggregating the geo-defects at a section level;

temporally aggregating the geo-defects into each inspection level; and

creating a record from results of the spatially aggregating and the temporally aggregating.

7. The method of claim 1 , wherein the determining a repair decision includes, for each of the geo-defects rated as Class II geo-defects:

determining which of the Class II geo-defects are likely to result in a derailment before the defined future point in time; and

determining costs previously associated with repairing the Class II geo-defects.

8. The method of claim 7 , further comprising: calculating an expected cost of derailment.

9. The method of claim 1 , further comprising: calculating a derailment risk by:

spatially aggregating the geo-defects at a section level;

temporally aggregating the geo-defects into each inspection level; and

creating a record from results of the spatially aggregating and the temporally aggregating; wherein the railroad network is further divided into sections of two miles in length.

10. The method of claim 9 , wherein the calculating the derailment risk further includes:

calculating a hazard function representing an instantaneous rate of failure probability given a survival by using a method of partial likelihood.

11. The method of claim 10 , wherein calculating the hazard function comprises fitting a Cox model.

12. A computer program product comprising a computer-readable storage medium having program code embodied thereon, wherein the computer readable storage medium is not a transitory signal per se, which when executed by a computer processor, causes the computer processor to implement a method, the method comprising:

logically dividing a railroad network according to spatial and temporal dimensions with respect to historical data collected for the railroad network, the spatial dimensions including line segments of a specified length and the temporal dimensions including inspection run data for inspections performed for each of the line segments over a specified period of time;

creating, via a computer processor, a track deterioration model from the historical data, current track conditions collected from one or more data sources, and traffic data;

identifying geo-defects occurring at each inspection run from the track deterioration model;

analyzing quantified changes in the geo-defects measured at each inspection run;

predicting from the quantified changes in the geo-defects, where at least one of the geo-defects is a Class II geo-defect, with an amplitude below a tolerance level of a safety standard, a probability of the least one Class II geo-defect deteriorating into a Class I defect with an amplitude in violation of the safety standard, within a specified period of time, the Class I defect defined as one that is mandated to be repaired upon discovery;

upon determining the probability reaches a threshold value, scheduling a repair to remedy the at least one Class II geo-defect within the specified period of time;

upon determining the probability does not reach a threshold value, determining a repair decision based on historically determined costs associated with previous comparable repairs; and

initiating the repair decision; wherein

determining the repair decision comprises minimizing a minimum total expected cost based on the historically determined costs.

13. The computer program product of claim 12 , wherein the line segments are defined as connecting two cities, and the spatial dimensions further include at least one of track identifiers and mile post locations.

14. The computer program product of claim 12 , wherein the railroad network is further divided into non-overlapping lots each of a length shorter than a length of the line segments, the method further comprising:

aggregating the geo-defects by inspection run for each geo-defect type.

15. The computer program product of claim 14 , wherein geo-defect types include at least one of:

align;

cant;

dip;

gauge;

harmonic cross-level;

over-elevation;

reverse cross-level;

super cross-level elevation;

surf;

twist;

warp; and

wear.

16. The computer program product of claim 12 , wherein the analyzing the quantified changes in the geo-defects includes analyzing amplitude changes for each of the geo-defects at each inspection run, and wherein a 90 percentile of an amplitude is designated to represent a track condition for a corresponding inspection run.

17. The computer program product of claim 12 , wherein the railroad network is further divided into sections of two miles in length, and the calculating the derailment risk includes:

spatially aggregating the geo-defects at a section level;

temporally aggregating the geo-defects into each inspection level; and

creating a record from results of the spatially aggregating and the temporally aggregating.

18. The computer program product of claim 12 , wherein the determining a repair decision includes, for each of the geo-defects rated as Class II geo-defects:

determining which of the Class II geo-defects are likely to result in a derailment before the defined future point in time; and

determining costs previously associated with repairing the Class II geo-defects.