Issue link: https://fhpublishing.uberflip.com/i/782355
NETWORK / 26 / FEBRUARY 2017 2 Artifi cial intelligence for pipe coating inspection NGGT – launched NGGT is looking to apply arti cial intelli- gence to the process of inspecting the deg- radation of plant coatings, paintings and cladding on above-ground assets across the system. NGGT has approximately 450 sites that are currently inspected by tech- nicians, which results in inconsistent cat- egorisation and variable visibility of asset health condition. Such inspections provide key information which supports invest- ment decisions across the coating systems – the primary form of corrosion preven- tion – on the national transmission system. NGGT believes that there is an opportunity to improve data collection, categorisa- tion accuracy, maintenance strategies and investment e• ciency through the use of arti- cial intelligence. This project will assess the feasibility of applying patented machine learning tech- nology that is currently being used to con- duct inspections in the motor industry to the key process speci ed in the speci cation for the assessment and reporting of plant coat- ings, painting and cladding inspections for national transmission system assets. Harmonics caused by wind farms con- necting to the transmission network result in voltage rise through increased reactive power. This is o€ en mitigated through recon- ductoring and reinforcement, or the use of other generation or power devices to absorb the reactive power. On method involves indi- vidual wind farm developers implementing their own solutions, but such an approach may lead to duplication of investment. SPEN is currently developing an exten- sive network in south-west Scotland for the connection of a number of windfarms, in particular the double-circuit between Coyl- ton 275kV and New Cumnock 275kV and extend to Tongland 132kV. SPT will evaluate the harmonic per- B E T T E R C O N N E C T E D 1 2 Innovative approach for transmission harmonics issues SPT – launched WI-POD control approaches NGET – launched The increased displacement of conventional carbon-intensive synchronous generation is expected in future years, particularly dur- ing system minimum demand conditions. At these times certain boundary transfers of power Œ ow (the Anglo-Scottish transfer, for example) are expected to remain high. The stability of these boundary Œ ows will be largely dependent on the control response performance of connected renewables both on- and o' shore. The inter-area characteris- tic frequency of the Anglo-Scottish transfer oscillation is known to be around 0.5Hz, which runs the risk of signi cant increased system operating costs for manage it. As system operator, National Grid has the right to specify that any generator be tted with a power system stabiliser to dampen any oscillations. This is known as power oscillation damping (POD). In its discussions with developers sur- rounding the delivery of such arrangements, National Grid has noted signi cant concern by the o' shore wind community of the prac- tical impact of power oscillation on the wind turbines themselves. This project will investigate if there is a plausible risk to wind turbines – their opera- tion life, performance or structural integrity – from interaction with PSS requirements that may be imposed by National Grid, and the level of that risk. NGET will develop an o' shore wind tur- bine model based on the widely accepted NREL (National Renewable Energy Labora- tory) o' shore 5MW baseline wind turbine model, which represents current typical o' shore wind turbines, to investigate the impact of POD control actions on o' shore wind turbine control performance. This will involve developing an appropriate model for the interactions between the o' shore wind turbine and grid using the Matlab Simulink- based simulation package. If› the› project› outcome› helps› to› clar- ify› the› POD› issues› with› o' shore› wind› tur- bine› developers,› it› will enable the incorporation of POD controls on future wind turbines and windfarms. Incorpo- The algorithm will be trained through the analysis of the library of digital photo- graphs to identify the seven di' erent catego- ries of assets to 80% accuracy. It will then be trained to recognise the di' erent levels of corrosion present before work starts on the development of an app that provides an interface for users in the eld to take photos, organise photos, conduct assessments and develop a report in a format can be emailed to engineers. As a conservative estimate, reducing expenditure by 10% on above-ground coat- ing maintenance activities would result in savings of more than £1 million in the RIIO T1 and RII T2 periods. formance of this network with the aim of developing a coordinated, e• cient and cost- e' ective harmonic ltering solution. There will be a resulting cost reduction to con- sumers through the reduced reinforcement required. The level of that cost reduction is di• cult to estimate at this stage. on subjects that are associated with techni- cal standards, resulting in an annual cost of between £2›million and £4 million. A key element in achieving a successful algorithm is to identify suitable methods to interact with cognitive technologies, in par- ticular when capturing from individuals in the business who provide the information verbally based on their own knowledge and experience. This project will develop a pilot using the IBM Watson core engine, which will capture knowledge from 630 technical standards documents and use cognitive techniques to provide intelligent and reasoned responses to queries.