Computer Vision For Cable Tunnels

Completed

Other Power and Storage Technologies(Electricity transmission and distribution)

Applied Research and Development

ENGINEERING AND TECHNOLOGY (Electrical and Electronic Engineering)

Not Cross-cutting

Project Contact
National Grid Electricity Transmission

Network Innovation Allowance

Ofgem

01 June 2013

01 November 2014

17 months

£95,000

Power

London

Network Innovation Allowance

Project Contact, National Grid Electricity Transmission

To produce a system that is able to perform condition monitoring on tunnels, without the need for personnel to be in the tunnel. This project will be successful if a solution is developed that enables personnel to be removed from the tunnel system whilst condition monitoring is being performed.

Cable tunnels have to be monitored for structural integrity. Currently condition monitoring of the cable tunnels are carried out by personnel operating in a confined space using visual inspections which are subjective. In addition enclosed spaces are a potentially hazardous area which National Grid tries to avoid sending personnel into wherever possible. National Grid has to perform this condition assessment, and has no other option but to use people to complete this assessment. Development The method proposed is a computer vision based system for the automated mapping of tunnels for inspection and monitoring, building upon experiments conducted during a site visit with Costain and National Grid to the Channel Gate Road site, London, in July 2012. This is a technological transfer with modification from a device that is utilised on condition monitoring of bridges, but will require modifications to make functional in the cable tunnel. The proposed system receives as input a collection of image data taken by a camera array system moving along the tunnel length (e.g. hung from the existing I-beam). The data is processed automatically and the system provides a number of useful outputs/capabilities: 1. A 3D model of the tunnel, formed using image data from digital cameras 2. Mosaic images for long sections of tunnel, to aid in 2D inspection reporting 3. The ability to image behind occluding objects such as wires or pipes 4. The ability to automatically localise new images precisely within the 3D model. Such images might include a second set of images taken by the rig one week later, or a random assortment of images taken from different viewpoints by inspection engineers5. The ability to automatically detect and potentially classify anomalous visual changes in these new images versus previous data6. The ability to collect thermal data using a infra-red camera and register it to the 3D modelNote : Project Documents may be available via the ENA Smarter Networks Portal using the Website link above

10/07/18