Joint Project to Improve Hydrogen Generation Through XAI
Helping to support operators during the decision making process
Intelligent Plant, along with our partner organisations European Marine Energy Centre (EMEC), and University of Aberdeen, were awarded funding through the Scottish Government's Emerging Energy Technologies Fund. The Scottish Government have a target of 5GW hydrogen capacity by 2030, and with our project we intend to help towards this target.
Our project, 'Feasibility of Explainable AI to support decision making as applied to hydrogen generation' was one of 32 projects to be awarded funding as part of the Hydrogen Innovation Scheme.
Steve Aitken, Founder of Intelligent Plant met up with the Hydrogen Industry Leaders to discuss more about the project and and AI in general - AI Hydrogen Project Interview
Project Outline
This project aims to develop a Decision Support System (DSS) that uses Explainable AI to help the operator when decision making is required.
Traditional DSSs are often 'black boxes', meaning the operator doesn't know the reasoning behind the decisions made by them, introducing distrust towards the machine. With Explainable AI (think ChatGPT), the operator will be able to have a discussion with the AI, helping the operator to understand the reasoning behind the DSS suggestions, ultimately making it a more trustworthy system.
We envisage the end system to answer questions such as:
Operator - "Why are you generating hydrogen?"
System - “Because there are high winds forecast tomorrow, and this will exceed the capacity of our batteries”.
The technical objectives of this project are as follows
-
Undertake knowledge engineering to identify rules which underpin decision-making around hydrogen production and logistics.
-
Encode the rules underpinning the current heuristics in a computer-understandable manner to automate reasoning.
-
Utilise formal argumentation to perform automated reasoning which identifies a course of action; and
-
Introduce a dialogue protocol, allowing users to incrementally query the system to understand why a course of action was recommended. The use of dialogue has been shown to mitigate against information overload and allows for additional information to be injected into the system.
The use of Intelligent Plant's Industrial App Store as an enabler for XAI shall allow operators such as EMEC to better understand their system.
Check out the press releases from our collaborators: