Skip to main content
This project ended in Feb 2019 and is now closed.Dismiss

LCT Detection

Funding mechanismNetwork Innovation Allowance (NIA)
DurationOct 2018 - Feb 2019
Project expenditure346k
Research areaTransition to Low Carbon Future
Regions
  • South West
  • South Wales
  • West Midlands
  • East Midlands
  • January 2019

    The LCT Detection project has proven the concept that a model can be developed, using cognitive analytics and AI, to spot unregistered LCTs connected at hous…

Objectives

By using Electralink’s DTS dataset, combining this with a range of other structured and unstructured data and then applying IBM’s Cognitive analytics, the objective is to identify patterns in the data that indicate the presence of EV, PV or other LCTs that had not previously been identified. IBM will apply its Watson technology to perform advanced analytics on the ElectraLink, combined with other datasets. IBM will use a progressive and iterative methodology to detect patterns in the data that was not detected hitherto. By improving detection of LCT on the network, the project will also build the foundation for improving forecasting capabilities and, ultimately, garner an understanding the effectiveness and costs for the various options would allow for the validation process to be optimised.

Problems

The energy market is complex and evolving – particularly with growing smart technologies and embedded, renewable generation. For DNOs, the increasing number of ‘invisible’ changes (growth of Electric Vehicles (EV), photovoltaic (PV) and other Low Carbon Technologies (LCTs)) challenge existing network practices. At present, technology change is outpacing changes in modelling and forecasting of consumer uptake of ‘smart’, Distributed Energy Resources (DER) or Electric Vehicle (EV) technology; therefore, it is difficult to monitor or understand the change in requirements on the LV network under existing arrangements, without monitoring EV and DER impacts directly at source (or substation level). While smart meters will improve the visibility of network load and generation in the longer term, there is a need for a solution that can identify unregistered equipment. The problem this project addresses is how to improve WPD’s ability to identify EVs, DERs and other LTCs connected to its network so that future operational and future investment decisions can be improved. It will also support some of the informational requirements needed in its transition to a DSO.

Methods

The project will use IBM’s cutting-edge AI and cognitive analytics capability to extract key information, related to EV/DER proliferation from the DTS dataset. The dataset has been constructed by ElectraLink using over six years of market interactions. This data contains 100s of millions of transactions, structured and unstructured market messages related to WPD’s area, across over 100 market processes. The project is looking to improve modelling at LV in particular, though having more realistic distribution substation data/profiles will potentially benefit HV planning. 

ElectraLink will extract data sent across the DTS regarding consumption and export relating to WPD’s network. This data will be analysed by IBM’s cognitive analytics and where appropriate combined with third party datasets, to develop candidate locations for validation. Once validated, this improved an output that can be overlaid onto WPD substation information can be used to develop a reporting framework to enable WPD to forecast future requirements for network monitoring and potential sites for active network management.