Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Method for Accurately Detecting Train Stops and Reversals

IP.com Disclosure Number: IPCOM000115396D
Original Publication Date: 1995-Apr-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 4 page(s) / 113K

Publishing Venue

IBM

Related People

Davis, RA: AUTHOR

Abstract

Disclosed is an algorithm for accurately tracking train stops and reversals. The algorithm takes data output from train wheel detect sensors and forms it into a list of train axles. The train axle list can then be used as a footprint for the railroad cars that passed over the wheel detect sensors.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 46% of the total text.

Method for Accurately Detecting Train Stops and Reversals

      Disclosed is an algorithm for accurately tracking train stops
and reversals.  The algorithm takes data output from train wheel
detect sensors and forms it into a list of train axles.  The train
axle list can then be used as a footprint for the railroad cars that
passed over the wheel detect sensors.

      The wheel detector sensors must be spaced close enough together
that a train wheel will activate both sensors at the same time at
some point during the crossing.  The Tieffenbach dual head wheel
detector has 2 sensors spaced 4 inches apart on one device and meets
the criteria for sensor spacing.

      The software algorithm involves monitoring the sensor
detections for patterns that indicate when the train has reversed.
The two sensors are connected to two digital inputs on a computer
system.  Each time an input changes, an event data record is stored
with a value indicating the current condition of the sensors.  If
sensor 1 alone is activated then a data record with a value of 1 is
stored.  If sensor 2 alone is activated then a data record with a
value of 2 is stored.  If both sensors are active then a value of 3
is stored.  It is important the digital inputs be sampled often
enough that both inputs cannot change values in between the samples.
Using the above described wheel detector sensor a polling rate of
about .5 milliseconds should be adequate to handle trains moving at
speeds up to 30 mph.  In typical applications the event data records
would include some form of a time stamp that can be used for
determining the distance between axles.  These distances are used by
algorithms that determine the number and types of cars in the train.

      After a train passes the wheel detect sensors the computer
system should have a list of wheel detect event data records.  These
records include a time stamp and a data value.  The data value should
be a 1, 2 or 3 as described above.  The system must generate a list
of train axles that are the basis of car recognition logic.  If the
train passed straight through the site then all data records should
be ordered 1-3-2 or 2-3-1 depending on which way the train passed by.
If sensor 1 is activated first then the pattern would be 1-3-2.  Any
other sequence of data values indicates a change in direction of the
train movement.

      The wheel detect event data records are used to build a list of
axle data records.  The axle data is used by other algorithms to
build up a listing of the actual cars that make up the train.  The
primary data needed to do this is the distance between axles.
Therefore, each record in the axle data list must have a field
containing the distance to the next axle.

      The following C language structures show an implementation of
the wheel detect data and axle data lists:
  /*
  ** Structure used to hold wheel detect data
  */
  typedef struct wheel_data
    {
    l...