DepotMATE - Multi-sensor Automated Train Examination

点击翻译按钮获取中文摘要

This project and the delivery of the DepotMATE (Multi-sensor Automated Train Examination) solution aims to meet the safety and efficiency challenges associated with depot-based rolling stock inspections. The diverse and congested UK rail network relies on efficient depot operations to ensure passenger, freight, light and heavy rail services run safely and on time, with high-performing assets. With the vast volume of inspection scheduling required to ensure assets are operating effectively, traditional manual examinations require a large amount resource, coming at a cost to both the operator and passenger where services are impacted by depot delays. This project collates a host of technologies incorporating sensor-fusion, advanced Machine Learning, and One Big Circle's industry-leading Intelligent Video expertise to enable operators to remotely examine a multitude of asset conditions in a reduced timeframe.The DepotMATE system will incorporate a multi-sensory lightweight inspection system, deployed to depots or sidings and positioned to capture passing rolling stock vehicles. Combining thermographic, acoustic, and Forwards Facing Video cameras and sensors and additional transmission of train-borne data via edge-processing, the DepotMATE will simultaneously capture a plethora of critical data as rolling stock vehicles pass by. A 'plug & play' design will enable partnering operators to configure sensors for their specific units, effectively targeting each operator's examination requirements and necessary inspection zones. DepotMATE will arm operators with automated monitoring data across a breadth of rolling stock, to assure cost-effective delegation of resources when managing depot operations, and successful delivery of proactive maintenance, helping to reduce costly reactive repairs and optimise the maintenance of rolling stock assets.DepotMATE will assist further automation across rolling stock inspection activities through the application of Machine Learning models, to automatically detect visibly apparent faults/vehicle contamination, exceedingly hot components, and acoustic emissions which may signify defective wheels or braking systems.Data will be accessible in extremely low-latency online, via One Big Circle's Automatic Intelligent Video Review (AIVR) platform, to enable depot operatives, control units, and fleet managers to remotely inspect each vehicle within their depot. Aspects of vehicle cleanliness, asset status, and vehicle allocation within a depot will be presented to the project's partnering operators at the touch of a button, massively reducing requirements for personnel to walk trackside whilst empowering users to inform predictive maintenance decisions.