This matching operation is hard to do correctly. The quality of this operation is vital to the usability of the results in TRITAPT (or any other system). We have seen really weird problems when matching observed trip data with scheduled data. The table below may be useful for finding faults in similar systems.
Observed problem | Possible causes |
---|---|
Rediculous dates and/or times in recorded events | Failing real-time clock or battery. |
Vehicles (apparently) travel back in time | Failing real-time clock |
Vehicles (apparently) drove forward or backward by some multiple of approximately 32 km (this is the value that we observed in our case) without recording a single event | Memory errors in the on-board computer |
Vehicles that exceed the maximum speed limit for short distances by an impressive value | GPS or odometer errors |
Odometer values do not change | Odometer sensor or cable failure |
Recorded odometer values are off by a few percent | Odometer calibration was not carried out for a long time (and tires have worn), or odometer calibration not carried out after replacing tire(s) |
Recorded odometer values are off by a factor of 2 | Error in odometer catlibration |
Beacons detected that should not be on the route | Wrongly programmed beacon transmitters |
Inconsistencies in trips that run past midnight | Buggy trip matching software |
Impossible dates | Failing battery of real time clock |
Vehicles that never seem to open their doors | Defective door switch |
Vehicles with huge imbalance between boarding and alighting passengers | Failing, or wrongly installed passenger counting equipment on some of the doors |
Vehicles that follow most of the scheduled route but deviate from it shortly before the end of the last trip of a shift | Vehicle was empty and driver / operator decided to not drive the empty vehicle all the way to the terminal stop |
Vehicles that depart early from the initial stop but somewhere before the next stop wait a long time | AVL system mis-identifies departure from initial stop |
Missing stops, route deviations | Detour due to road construction work that was not coded in the schedule |
No trips recognized on a route | Beacon defective, changed, detour due to construction work, trips temporarily carried out with buses hired from another company |
No trips recognized from a particular vehicle | AVL or communication failure with the on-board equipment |
We have developed software that matches trips for a variety of environments. We do not consider this part of the TRITAPT system, but we can certainly help with the development and improvement of trip matching software. Our trip matching software is written in ANSI-C and/or PERL and runs under Linux, MS-DOS, Windows 95, Windows 98, Windows XP, Vista, Windows 7 and probably anything else that is somewhat POSIX compliant.
Besides production of the data files, this phase can also produce daily performance indicators that can be made accessible on web pages. Such web pages have been used to identify recurrent problems in on-board equipment and problems that were related to a particular group of vehicles in a fleet. Equipment in the field fails often and a good monitoring system is required to keep things working satisfactorily.