Applying AI knowledge management
Technology has long been at the heart of capturing and storing knowledge and making it accessible. AI solutions now allow delivery of knowledge to be highly targeted and personalised
An engineer who has worked on an oil rig for 20 years will be an expert in that specific rig. Their ‘been-there-done-that’ experience means they can quickly make good decisions on the best response to a wide range of scenarios.
But it is also knowledge that will be lost when those individuals move on. Companies that fail to leverage the experience of senior engineers risk forever losing valuable insight and industry intelligence, putting their operations at a crucial competitive disadvantage.
Knowledge management—using technology to capture and share knowledge—has long been touted as a potential solution to the problem. Recent advances mean AI is now an important asset to safeguard industry expertise and can make knowledge management much more sophisticated.
Making sense of AI
Over the years, each engineer will have filed thousands of maintenance logs and incident reports where they captured what went wrong and how they solved it.
Each engineer will have filed thousands of maintenance logs and incident reports
Within these reports are the solutions to most engineering issues. In fact, when large oil and gas companies perform audits of errors, they find many came from issues that had previously been resolved on other rigs.
AI can go through those reports, draw out that insight and present it to anyone tackling the same problems. In essence, it is like a virtual brain, with all the knowledge of the company’s past experience, dispensing wisdom to those who need it at the moment they are assigned a task.
Changing the game
If knowledge management were as simple as pulling up past reports, the issue would have been solved years ago. But a system that presents thousands of maintenance reports of varying quality is hardly conducive to a quick response.
The challenge is not finding the documents but spotting and extracting the relevant insight. This is hard for machines because these reports were not necessarily written in a way than can be easily searched using algorithms, which prefer clearly labelled and categorised data. Many documents include unstructured data such as scribbled notes, drawings and informational videos.
Furthermore, task management systems—which allocate engineers to jobs—have not historically been designed to integrate with incident management systems. They speak a different language and are not set up to mine data.
But AI is now sophisticated enough to solve this. It can take complex sets of information from incident reports and make sense of them. Systems can then be set up which map descriptions of work packages in task management systems onto problem reports stored on incident management systems.
Create knowledge management
The first step is to digitise handwritten notes and other information into machine-readable formats, which can be done using off-the-shelf image recognition AI.
Then there is the clever bit. One of the major recent advances in AI has been natural language processing (NLP), where computers learn to capture the linguistic meaning of text. This allows AI to make sense of written language captured in incident reports.
The challenge is not finding the documents but spotting and extracting the relevant insight
An NLP model can be built and fed past incident reports. It then gradually learns to understand the context of the words or phrases and how they relate to others. We worked on a project for Norwegian energy company Equinor to do just that.
The data scientists training the model need to work with experts—maintenance engineers in this case—to understand the context so they can guide the model to recognise what is valuable, and what is not, until it can gradually do this on its own. In the Equinor project, they ran workshops where expert users from their rigs spent hours highlighting what is valuable in different reports.
These models can eventually be combined with other models trained on the design and layout of the asset. Gradually, it will learn to associate human knowledge—captured in text incident reports—with specific problems or components. So, when presented with a new task, it will be able to relate it to a wide range of previous similar tasks and identify the most similar solution.
Making AI useable
AI on its own is no use, though, unless it can integrate appropriately to provide the engineers with information at the right time. This needs new infrastructure to be deployed to link systems together. For example, when a task is allocated, the task management system needs to automatically share it with the central IT system.
Purpose-built software then extracts data from incident management systems, which have APIs set up to allow this. The AI model runs on that data, identifying the most relevant insight for the task and packages it up in an easy-to-understand format.
This is then sent back to the task management system, which has been modified to present that insight to the user via an intuitive interface, e.g. via an app on the tablet they use to manage their workload. Some such systems are now investigating augmented reality to guide engineers through tasks.
Big time and cost savings
AI knowledge systems have many obvious advantages. They allow professionals to benefit from their colleagues’ experience and find solutions quickly, avoid repeating mistakes and quickly solve problems they have not seen before.
They reduce the training needed for new starters of all levels, which in turn allows for a more flexible workforce where experts can be moved around and contractors deployed quickly. It is a protective measure against the loss of expertise and experience.
AI knowledge management is also valuable in the drive towards digital twins. Building up computer-readable knowledge can help improve data management and digitalisation across the board, and identify opportunities to introduce standard approaches that will support other digital projects.
AI knowledge management is a hugely valuable cost-saving exercise in its own right, but it also has the potential to be of huge strategic value.
Warrick Cooke is a senior consultant at Tessella, a data science and AI consultancy.
This article is taken from our forthcoming Digitalisation Review, which will be published in November.