Universal Temperature Controller using Artificial Intelligence

Completed

Fossil Fuels: Oil Gas and Coal(Oil and Gas, Refining, transport and storage of oil and gas)

Applied Research and Development

PHYSICAL SCIENCES AND MATHEMATICS (Computer Science and Informatics)
ENGINEERING AND TECHNOLOGY (Mechanical, Aeronautical and Manufacturing Engineering)

Not Cross-cutting

Project Contact
SGN

Network Innovation Allowance

Ofgem

01 March 2018

01 March 2019

ENA months

£57,700

Energy

South East

Network Innovation Allowance

Project Contact, SGN

This project will review the feasibility of a universal preheat controller to act as an intermediary between a traditional station exit temperature signal and an installed preheat asset. The controller will be programmed to learn and anticipate site specific dynamics through a desktop exercise learning stage using historical site operating data. The desktop exercise will explore patterns of energy usage and to identify predictable and unpredictable patterns using historic site operation data. This data will be used to identify process optimisation opportunities, bottlenecks and act as a training tool for an Artificial Intelligence (AI) based learning and control algorithm. In parallel a hardware specification of a universal controller which meet safe operating requirements, supporting processing needs of the AI software and offer potential for retrofits across existing network preheat assets.  This project aims to assess the potential application of a universal controller based on AI to act as an intermediary between a traditional station exit temperature signal and an installed preheat asset. By using AI combined with historical data, it is hoped that it will be possible to recognise patterns and learn to predict upcoming events such as variations in demand, and impact of environmental variables such as temperature and humidity. This project will include: Teaching the AI controller: Select a sample of 3 sites with different operating characteristics. Use data mining techniques to identify process bottlenecks, energy saving opportunities and process variability issues. Train the AI to assess performance with the minimum number of inputs for maximum potential application (Inlet Pressure, Outlet Pressure, Station Outlet Temperature, Historic Weather). Development of a virtual system expert to advise on “optimal operation” to help guide process performance from past operating data. Optimise computation for data efficiency. Assess reliability and robustness for suitability on critical infrastructure. Hardware Specification: Determine computation requirement (Central Processing Unit (CPU)/ Graphic Processing Unit (GPU). Develop physical specifications (considering any restrictions). Input / Output specifications and protocols. Review SGN specific requirements / Applicable standards. Determine power requirements. Review connectivity (Global System for Mobile Communications (GSM)/ Wi-Fi etc.). Determine appropriate security & cyber-security protocols. Remote update and control. Implementation: Development of a project implementation roadmap for site integration.

This project aims to assess the potential application of a universal controller based on Artifical Intelligence (AI) to act as an intermediary between a traditional station exit temperature signal and an installed preheat asset.

08/11/22