Train door mechanism failure detection (UK)

Rolling Stock

Train door failure detection (UK)

PROBLEM

Unexpected faults in the train´s pneumatic door components

Neuron Soundware developed a solution to automatically identify failures in the doors. This involved 12 sensors, and IoT device nEdge within the nGuard solution. The customer’s maintenance team collected a number of broken parts to swap with good ones to check if Neuron can identify the failure. After initial 150 rounds of open/close runs to gather nominal sound, an AI model was deployed. Together 16 simulations were performed, 12 with positive detection, 4 not possible to detect by sound analysis. During tests 10 components were swapped, from them 8 were detected as broken, with only 2 breaks didn’t cause change in sound.

SOLUTION

AI Automated Identification of a Failure

12 sensors have been used, IoT device nEdge within the nGuard solution. The customer’s maintenance team collected a number of broken parts to swap with good ones to check if Neuron can identify the failure. After initial 150 rounds of open/close runs to gather nominal sound an AI model was deployed. Together 16 simulations were performed, 12 with positive detection, 4 not possible to detect by sound analysis. During tests 10 components were swapped, from them 8 were detected as broken, 2 breaks didn’t cause change in sound.

Benefits