|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Home | Research | Research Programs | Artificial Intelligence Technologies |
Artificial Intelligence TechnologiesWheel Impact Load Detector Miner (WILDMiner)The railway industry has undergone a remarkable technological revolution during the past ten years. The companies have adopted a number of new technologies to optimize their productivity and minimize the costs of their operation. A notable example is Wheel Impact Load Detector (WILD) system, a system that makes it possible to detect which wheels will cause impacts that are high enough to damage the railway track. The WILD system consists of sensors installed at the strategic places along the track. When a train passes near the sensors, the latter measure the impact caused by each wheel on the track. The data that is gathered is automatically transmitted to the company's central system. When a wheel causes too high an impact, the train must reduce its speed and then stop at the nearest shop to uncouple the car whose wheel is damaged. This greatly help minimizing the damage to the tracks and bridges. However, reductions in the speed of the trains and the obligation to stop at the shops introduce many delays in the delivery of the goods and reduce the throughput of the company. The objective of the WILDMiner project is to develop the models and systems required to integrate a proactive approach to wheel maintenance. The developed models will predict which wheels are likely to break. The maintenance crew will use this information to plan and perform wheel replacements before the wheels become so damaged that they cause service interruptions. This should reduce operational costs considerably. The WILDMiner project is currently in its first stage, which is to perform a feasibility study. In this first phase researchers study the possibility of using Data Mining techniques, similar to those employed in the ADAM project, to develop the required models. Some preliminary models have been developed and the results appear promising. Research ContactDr. Sylvain Létourneau Business ContactDr. George Forester |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|