24. Automatic Pantograph Drop System Using Real-time Abnormality Detection in the Current Collection System

The next-generation tilt control system developed by the Railway Technical Research Institute detects current running position using a gyroscope, and precisely matches the tilting behavior of the carbody with the geometry of the curved section. This suppresses low-frequency lateral acceleration and reduces the occurrence of motion sickness. Applying this system to higher-speed trains requires technology that allows the carbody to tilt more quickly and stably.

For detecting pantograph part detachments, we developed a detection algorithm combining object detection AI with coordinate processing. This algorithm also supports detection of step-shaped wear on pantograph contact strips and abnormal arcs. For foreign objects such as plastic sheets or futons attached to the overhead contact line, which are difficult to pre-train on, we developed a detection algorithm that combines AI requiring no training data and a system for abstract specification of detection targets within the overhead contact line area setting. Results from in-house testing confirmed that pantograph part detachments can be detected within one second, and foreign objects can be detected up to 40 meters away from the camera, both with no false positives or missed detections.

This system can prevent a chain of equipment damage caused by the current collection system, which is expected to reduce downtime by ensuring train operability and shortening equipment restoration time.