Reliability improvement of airport ground transportation vehicles using neural networks to anticipate system failure

This paper describes a joint industry/university collaboration to develop a prototype system to provide real time monitoring of an airport ground transportation vehicle with the objectives of improving availability and minimizing field failures by estimating the proper timing for condition-based maintenance. Hardware for the vehicle was designed, developed and tested to monitor door characteristics (voltage and current through the motor that opens and closes the doors and door movement time and position), to quickly predict degraded performance, and to anticipate failures. A combined statistical and neural network approach was implemented. The neural network "learns" the differences among door sets and can be tuned quite easily through this learning. Signals are processed in real time and combined with previous monitoring data to estimate, using the neural network, the condition of the door set relative to maintenance needs. The prototype system was installed on several vehicle door sets at the Pittsburgh International Airport and successfully tested for several months under simulated and actual operating conditions. Preliminary results indicate that improved operational reliability and availability can be achieved.