Projects
Projects: Custom Search |
||
| Reference Number | NIA2_NGET0059 | |
| Title | Anticipating Gas Insulation Leaks from Electrical assets AGILE | |
| Status | Started | |
| Energy Categories | Other Power and Storage Technologies (Electricity transmission and distribution) 100%; | |
| Research Types | Applied Research and Development 100% | |
| Science and Technology Fields | PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics) 50%; ENGINEERING AND TECHNOLOGY (Electrical and Electronic Engineering) 50%; |
|
| UKERC Cross Cutting Characterisation | Not Cross-cutting 100% | |
| Principal Investigator |
Project Contact No email address given National Grid Electricity Transmission |
|
| Award Type | Network Innovation Allowance | |
| Funding Source | Ofgem | |
| Start Date | 01 March 2024 | |
| End Date | 30 April 2026 | |
| Duration | ENA months | |
| Total Grant Value | £475,000 | |
| Industrial Sectors | Power | |
| Region | London | |
| Programme | Network Innovation Allowance | |
| Investigators | Principal Investigator | Project Contact , National Grid Electricity Transmission (100.000%) |
| Industrial Collaborator | Project Contact , National Grid Electricity Transmission (0.000%) |
|
| Web Site | https://smarter.energynetworks.org/projects/NIA2_NGET0059 |
|
| Objectives | Efforts have already been made by others to employ machine learning (ML) models for predicting when SF6 may be lost from specific assets based on their nameplate and environment with some success but the required level of reliability has not yet been achieved. By adding NGET"s data and making use of data from on-line pressure gauges this project will refine the model with the intention of yielding greater accuracy.The solution will make use of the existing ML model (Python) and using an "ensemble" type regression model to map asset covariates such as age, tank size, corrosion zone, distance from sea etc. onto leak rates calculated from SF6 top up data recorded during asset maintenance. The modelling will be enhanced with live SF6 density data installed specifically for this project. An existing machine learning model written in Python and using an "ensemble" type regression model will be used to map asset covariates such as age, tank size, corrosion zone, distance from sea etc. onto leak rates calculated from SF6 top up data from circuit breakers recorded during asset maintenance. This model has already been benchmarked on other TO circuit breakers and will initially be run on NGET assets to gauge accuracy. An NGET specific model will then be developed. This will entail selecting the optimal inputs for predicting future SF6 loss. A model learned from a combination of all 3 TO"s assets will then be tested against a held out portion of assets from across the GB transmission network. Part of this proposed project will run concurrently with a further iteration of development with SSEN and SPEN (funded separately). Incorporating understandings from online monitored assets will be unique to this project and will help refine the predictive model functionality to take seasonal and diurnal effects into SF6 leakage rates.The project will consider how predictive analytics can change operational decision support through their incorporation into planning tools the form in which this decision support will take. This project will deliver against four objectives:Development of a software model based on gas monitoring data for predicting short term SF6 escapes for inclusion in a longer term general SF6 escape modelApplication of leak prediction model to NGET assets detailing performance against individual families of circuit breakersDevelopment, testing and reporting of a new predictive escape model with combined UK TO dataProof of concept for an automated maintenance planning driven by predictive analytics. | |
| Abstract | This project will use machine learning techniques to improve forecasting capability for SF6 circuit breakers taking nameplate, environmental and operational factors into account. Building on existing work the following will be undertaken:Understand the performance of the existing model when applied to NGET assetsDevelop the model taking NGET assets into account and then test to determine accuracyCollect online density monitoring data and use it to improve the forecastsDemonstrate how forecasting may be used to automate planned interventions. | |
| Data | No related datasets |
|
| Projects | No related projects |
|
| Publications | No related publications |
|
| Added to Database | 02/10/24 | |
