Version:
1.0 |
Author:
Stuart G. Mentzer |
Platforms:
Win32, UN*X, DOS, OS/2, OpenVMS |
Function and Methods
VeCor
is a tool for automated assessment and correction of non-oblique VTB
and VTV vehicle crash test data. The impact direction accelerations
and the total barrier force (if VTB test) are input with any available
supplementary information and the set of calibration, bias, and time
shift errors which best explain deviations from physical, kinematic,
and measured relationships are computed and reported. The corrected
UDS files can also be generated if desired.
Default
relationships provided by VeCor consider bias in the pulse tails,
differences in the velocities within each vehicle at the end of the
test, deviations from momentum/impulse balance over the duration of
contact, and the time of initial response for each pulse. The default
relationships can be altered or suppressed by the user.
Optional
relationships include specifying velocities, displacements, or relative
displacements of pulses at specific instant. The weight applied to
each relationship by VeCor to determine a best solution is controlled
by confidence ranges for each relationship, which can be user-specified
or allowed to default to preset values. Relative timing constraints
can also be specified to assure that the sequence of initial responses
agrees with knowledge of the vehicle construction.
Typical
inputs required for an accurate solution would be the velocity and
displacement of a target pulse in each vehicle at a time near the
end of the pulses, relative crush between the accelerometers of each
vehicle, and the approximate time of separation between the vehicles
or vehicle and barrier.
VeCor
uses a weighted quadratic minimization approach to determine the best
solution for the errors in each pulse. The time shifts appear nonlinearly,
requiring a grid/gradient search of the time shift domain. The user
can control the thoroughness of this search, although searching from
many points is time consuming and has been unnecessary in trials.
For any time shift solution, the calibration and bias solutions are
found by solving a linear system of equations for the unique zero
of the gradient of the total deviation error.
Features
and Options
VeCor
allows a wide range of supplementary information to be used in determining
the pulse errors. Although more information assures a better result,
a reasonable solution may be possible with very little information.
By assigning small contributions to the total error for the pulse
errors themselves, VeCor generates determinate (i.e., nonsingular)
minimization problems and allocates reasonable amounts to each pulse
error when the supplementary information alone is insufficient to
uniquely determine all of the pulse errors.
The user
can control the solution by supplying reasonable 95% confidence ranges
for the supplementary information and the pulse errors. This determines
the weighting applied to deviation errors for each relationship.
Setup
and Run Instructions
The command
line syntax is: VeCor [input_file]
VeCor will prompt for a VeCor input file if none is specified on the
command line.
Usage
Recommendations
Barrier force signals with positive-valued impact loads must be given
a "+" sign flag in the VeCor input file. Negative-valued barrier forces
can be given a "-" sign flag or no sign flag.
Relative
displacement (DR) targets can be added for signals from the same or
symmetric locations, such as multiple occupant compartment signals
or the left and right front wheels in a full frontal impact. The DR
confidence bands should usually be fairly loose (50-200 mm), especially
for less consistent motions such as wheel or engine top and bottom
signals.
Set
film displacement constant (DC) targets with a 1 ms time span and
bands of 25-50 mm for high confidence targets and 50-200 mm for low
confidence targets.
Use a velocity scatter (VS) target override specifying the full time-span
for final velocities to be brought together and adjust the confidence
band as needed to obtain sufficiently consistent final velocities.
Set momentum balance (MO) target time span to 10- 20 ms around the
time when the barrier force, if available, and the accelerations drop
to near zero.
Apply top-of- engine film displacement targets to the Engine - Top
signal only.
Use wide/standard target confidence intervals first and then reduce/adjust
as necessary.
Check/correct
any signals yielding negative calibration values for sign-inversion.
Check/correct
any signals yielding large calibration values for units mis-scaling.
If signals with large but unknown mis-scaling will be used an estimated
scaling (1/Calibration) value should be specified for the signal in
the S column and refined over a few VeCor runs to allow the full scaling
correction to be made without distorting the rest of the solution.
A similar approach can be applied for large biases.
Known
Problems and Constraints
VeCor
is currently limited to impact direction pulses from non-oblique VTB
and VTV crash tests.
VeCor
solutions may be degraded by the presence of significant nonsystematic
errors (instrument rotations, signal clips, channel failure, etc.),
or if insufficient or inaccurate supplementary information is provided.
In such cases the VeCor solution will still indicate the presence
of large errors, and the corrected data may still be useful for some
quantitative applications. It is best remove obvious large non-systematic
errors before running VeCor.
VeCor
is limited to one force pulse and up to 20 accelerometer pulses. Search
of the time shift domain from more than a single point may be impractically
time-consuming for interactive runs with more than five pulses.
Additional
Documentation
See The VeCor Program
for complete VeCor documentation.
A template
for the input file can be found in the VeCor
Input Template.
A history
of program changes can be found in the VeCor
Change Log.
|