Once, the Energy industry was widely seen as a laggard in embracing AI.
The global pandemic and oil price instability have created intense pressure to cut costs, improve Opex, drive efficiencies, and generate more value. Along with that urgency has come a growing recognition that AI is the technology best-suited to help Energy weather the current crisis, revitalize its operating models, and transform for what lies ahead.
Subsurface data, like seismic data and well logs, are all potentially rich sources for Machine Learning. However according to estimates, only 10% of subsurface data is currently being used. Initiatives like the OSDU are aimed at increasing this number considerably, by liberating data from applications and collecting it in a different way. The core principle of the OSDU is to ‘free’ data from applications; enabling new cloud-native data-driven apps.
But clearly, there is a longer journey ahead. The industry’s ultimate goal must be to formulate a data strategy that harnesses AI to learn from that data — whether it’s neatly structured in rows and columns or completely unstructured — and gain insight. (What’s unstructured data? Examples would include exploration photos included in a presentation or sent to a colleague in an email, or images collected from drone fly-bys.)
Why AI for the Energy Sector?
Unlike conventional computers, artificial intelligence is not programmed. On the one hand, like humans, AI learns from information. Yet on the other, it learns far faster than humanly possible. Machine learning (ML) enables AI to make a correlation between a pattern and an outcome, formulate a hypothesis, take action and then integrate that feedback into its next hypothesis. AI continuously learns, with the goal of predicting future outcomes and events with greater accuracy.
Here are some specific applications every Energy company ought to consider.
Using predictive analytics to drive down maintenance costs
Detecting pipeline cracks and defects at the earliest possible stage can help shrink maintenance costs. Energy companies can leverage AI to anticipate when maintenance of a pipeline is needed or a pump is likely to malfunction.
An AI app would begin by ‘viewing’ thousands of images and videos of pipes. By analysing damage and estimating the degree and probability of fatality, the AI can learn to identify tell-tale signs that predict upcoming failures.
An energy company can use drones to monitor its pipelines, and feed real-time images to the app. By combining these photos and videos with GPS, weather and environmental data, the company can identify damages and defects at an early stage with a high degree of certainty. Such real time data and real time monitoring builds a robust database. The more data AI studies and learns from, the more it can continuously improve its predictive capabilities.
Cognitive assistants and smart expertise locators for a smaller workforce
The continuous aging of the Energy sector workforce has resulted in a loss of core expertise and organisational memory, and has stifled innovation. Slashing overheads in the current crisis is going to painfully exacerbate this trend.
Cognitive assistants and smart expertise locators can help significantly. Through ML (machine learning) a company can capture the expertise of a shrinking labour force, combine it with knowledge datasets and real-time data (e.g. data from connected devices tracking the moves of workers on oil rigs) and use it to train chatbots.
In the field, a worker with an AI-powered wearable assistant can say ‘wake words’ like “I need help” or “I can’t figure this out” and get instant support. This might mean providing step-by-step instructions in video or augmented reality. Or, it might connect the worker with a remote expert somewhere else in the world who can provide direct coaching and guidance.
Avoiding hazardous situations using Image recognition
The Energy sector must now manage facilities that are either somewhat or completely unattended due to staff cuts or Covid-19 containment measures.
Proven technologies, like facial recognition and gesture detection, can help spot behaviours that might create a hazard, like workers showing signs of drowsiness during their shift or moving too close to a dangerous area. When combined with connected worker technology like smart garments with data-collecting sensors, image recognition can power real-time monitoring dashboards and automation that spot signs of danger before it’s too late.
The result is safer employees, avoidance of preventable (and costly) accidents, and increased uptime.
The time for the Energy industry to accelerate its AI journey is now.
The challenges for the industry have never been greater, but its increased interest in AI comes at an ideal moment. A large number of critical applications have already been developed and battle-tested under real field conditions. They are ready to be applied.
With the right partners, the Energy sector can emerge from this crisis stronger and ready for the future. The time to start the journey is now.
AI (Artificial Intelligence) and Machine Learning (ML) glossary
- NLP (natural language processing) or NLP for short, is broadly defined as the automatic manipulation of natural language, like speech and text, by software.
- Sentiment analysis (also known as opinion mining or emotion AI) is defined as the use of natural language processing, text analysis, computational linguistics, and biometrics to systematically identify, extract, quantify, and study affective states and subjective information.
- Cognitive computing describes technology platforms that, broadly speaking, are based on the scientific disciplines of artificial intelligence and signal processing. These platforms encompass machine learning, reasoning, natural language processing, speech recognition and vision (object recognition), human–computer interaction, dialog and narrative generation, among other technologies.
- Machine learning is defined as the study of computer algorithms that improve automatically through experience. It is seen as a subset of artificial intelligence. Machine learning algorithms build a mathematical model based on sample data, known as "training data", in order to make predictions or decisions without being explicitly programmed to do so.
- Deep learning (also known as deep structured learning) is defined as part of a broader family of machine learning methods based on artificial neural networks with representation learning. Learning can be supervised, semi-supervised or unsupervised.
- Predictive analytics is defined as a variety of statistical techniques from data mining, predictive modelling, and machine learning, that analyze current and historical facts to make predictions about future or otherwise unknown events.
- Image recognition or object recognition or classification is defined as when one or several pre-specified or learned objects or object classes can be recognized, usually together with their 2D positions in the image or 3D poses in the scene.
- Voice recognition can be defined as speaker recognition (determining who is speaking), or
- speech recognition, (determining what is being said).
- Intent recognition is defined as Automatic intent recognition, also referred to as plan recognition, is a critical challenge for intelligent user interfaces (to determine what a user is trying to do, and how the UI can offer help), computer security (to determine the objectives of an attacker), sketch understanding, natural language understanding, and other contemporary problems.