A Data Driven Approach to Railway Intervention Planning

1
A Data Driven Approach to Railway Intervention
Planning

• Derek Bartram
• Supervisors
• Dr. M. Burrow
• Prof. X. Yao

2
Contents

• Current Technologies
• Structure
• Problems / Issues
• Project
• Aims
• Comparison To Current Technologies
• System Design
• Progress To Date
• Progress Problems

3
Typical Track Deterioration

• Angle 1 lt Angle 2 lt Angle 3

4
Geometry Measurements
5
Geometry Measurements

• Top height
• Web height
• Web thickness
• Ballast thickness
• Ballast SD size
• Corrugation wavelength
• Gauge
• Twist
• Cant

6
Existing Technologies Decision Support Systems
7
Expert System Inference Engine

• If (ballast_type granite)
• then minimum_thickness 50mm
• If (ballast_type sandstone)
• then minimum_thickness 200mm
• If (ballast_thickness lt minimum_thickness)
• then replace_ballast

8
Expert System Inference Engine

• If (ballast_thickness lt 50mm
• ballast_type granite)
• then replace_ballast
• If (ballast_thickness lt 200mm
• ballast_type sandstone)
• then replace_ballast

9
Decision Support Systems Problems / Issues

• Expert system only as good as the rule base
• Simplified models
• Possible rule / intervention flaws
• Large track segments

10
Aims

• Improved deterioration modelling
• Improved intervention planning
• Improved localised fault detection
• Improved total life-cycle costing

11
Static Vs Dynamic Solutions

• Static solution
• Guaranteed good behaviour initially
• Never improves
• Dynamic solution
• Initial behaviour potentially bad
• Requires high quality existing dataset
• Improves with time

12
My Project Assumptions (1)

• The various possible faults for track are
  identifiable by unique combinations of track
  component deterioration

13
My Project Assumptions (2)

• For each type of failure, the solution to the
  problem is not related to other failure types

14
My Project Assumptions (3)

• Once a track sections starts failing with a
  particular failure type, it will continue to fail
  with the same failure type

15
My Solution Tasks

• Classify the various failure types
• Provide a mechanism for classifying unclassified
  track sections
• Produce a deterioration model for each failure
  type
• Determine best intervention for each failure type

16
My Solution Data Processing

• Handle missing data
• Segment data
• Build data runs
• Make absolute values relative

17
My Solution Failure Types

• Plot last data recording of each run in
• n-dimension space

18
My Solution Classification

• We know sets of individual data points and
  associated failure types
• Failure type does not change until intervention
• Decision trees
• Evolutionary algorithms

19
My Solution Classification

• Decision trees

20
My Solution Classification

• Evolutionary Algorithm

21
My Solution Work Determination

• For each run in failure type
• Calculate fitness of subsequent intervention
• Calculate average of fitness's for each
  intervention type
• Choose intervention with best average fitness

22
My Solution Work Determination

• Fitness metric
• Length of time before next intervention
• Next failure type

23
My Solution Deterioration Modelling

• Simple model
• Enhanced simple model
• Evolutionary model building

24
Progress To Date

• Classify the various failure types
• Provide a mechanism for classifying unclassified
  track sections
• Produce a deterioration model for each failure
  type
• Determine best intervention for each failure type

25
My Solution Problems

• Large number of missing values in geometry data
• Inconsistent / missing? work history data
• Data anomalies

26
Conclusions

• Long term improvements over static solutions
• Deterioration models
• Intervention planning
• Costing

27
Thank you for listening
Questions?
28
(No Transcript)