An Artificial Neural Network Model for Maintenance Planning of Metro Trains

In urban transportation, trains have an increasingly important place due to the increase in number of passengers. Meeting passengers is directly related operated on a line. Thus, frequency operation affects level wear equipment. This makes train maintenance more important. Equipment faults are basis for maintenance. However, fault times equipment which unknown causes uncertainty activities and plans. results from many factors that affect train. If historical data, affecting known, effective use resources (time, cost personnel, etc.) provided eliminated. this study, firstly, data Ankara Metro between 2017 2018 examined evaluated with expert opinion. Artificial Neural Network (ANN) model created set along each according type ANN model, 5 (Equipment Type, Preventive Maintenance Frequency, Material Quality, Life Cycle, Line Status) determined as inputs failures outputs. The mean absolute percent error (MAPE) value found 11%, square (MSE) 0.0028229 training test stages ANN. Then, 10-week planning applied. compared current planning. As result applied planning, average decreases by 27%, uninterrupted service rate increases 40% heavy errors also prevented. Fault removal resulted 10% improvement. showed models could be used effectively prediction rail system multiple types literature, there no study implements where all equipments failure together. first field systems literature will reference future studies.