Realizing the potential of AI in the oil and gas industry

Artificial Intelligence (AI) is easily one of the hottest areas of technology today, and for good reason— it holds promise in so many areas. Gartner predicts that by 2021 the dominant source of AI business will be new revenue, and by 2021 AI augmentation will generate $2.9 trillion in business value and recover 6.2 billion hours of worker productivity. This means businesses looking at both efficiency and reducing costs should investigate the promise of AI.

AI is already making headlines in industries like retail, banking, hospitality and high tech, and this progress will speed up as other industry factors come in to play. Super-fast computing power with the ability to analyze large datasets, an IoT revolution driven by Internet connectivity, and the on-demand scalability of cloud infrastructure are opening possibilities for AI across many industries.

The oil and gas industry is beginning to experiment cautiously with AI-based solutions. As a whole, the industry is facing challenges with pervasive cheap oil prices, new reserves and alternate fuel choices. These have put pressure on profit margins, driving the industry to explore AI options.

Machine and deep learning can be applied to bring a quantum change in operational efficiency to reduce waste and non-productive time, and prolong asset life and performance. It is an opportunity the industry needs to seize and move forward.

Making better business decisions

Deep learning algorithms can be applied to train neural-network models to analyze vast amounts of data for decision making. Rendering complex multi-layer neural networks on terabytes of data is a reality today. Deep learning can also be used to develop more sophisticated neural networks to handle asset management in the oil and gas industry. For example, time-automated condition monitoring with predictive analytics can assist with operational decision making.

Similarly, unsupervised learning enabled by deep learning can be used to reveal patterns, while drones are being used for visual inspections of pipelines. As a drone flies over a pipeline, it records video footage that may contain signs of leaks or cracks that are undetectable to the human eye. Deep-learning algorithms can automatically detect pixel signatures from drone footage, minimizing asset failure risk at a much-reduced cost and in a timely manner.

Another emerging AI opportunity is the use of conversational bots that apply natural-language processing algorithms enabled by deep learning. These bots can be used by field workers to interact with asset diagnostic applications using speech to text and vice versa.

The AI possibilities are endless and the potential impacts in the oil and gas industry include:

AI capability	Asset-management scenarios
Predictive asset maintenance	Real-time asset operations monitoring	Asset inspection	Asset work order management	Condition-based monitoring
Computer vision: image recognition and classification	Supervised, unsupervised learning and deep learning to identify anomalous behavior and predict failure and corrective actions		Robotic applications: drones using deep learning algorithms like RNN/CNN	Assess the asset condition by automated visual inspection using drones enabled with image recognition/deep learning	Drones enabled with deep learning image recognition features can monitor assets in remote and hard-to-reach locations
Classification	Anomaly detection on time-series data using supervised and unsupervised learning algorithms	Identifying deviations from normal operations using anomaly detection and taking automated corrective actions
Clustering	Unsupervised clustering techniques can predict various possible asset health conditions	Unsupervised cluster analysis can throw light on various symptoms of process or asset degradation
Regression	Failure prediction using regression models
Natural language processing/ query generation	Reinforcement learning algorithms like recurrent neural networks can be used to implement and improve NLP			Use of conversational bots as assistants to the maintenance worker	Workers can use bot assistants to log details about asset health and perform immediate diagnosis
Speech recognition				Voice-based communication to access information from knowledge databases and communicate with experts	The sound from rotating assets can be analyzed for early maintenance issues
Voice to text/text to voice				Bot assistants can help field workers process information in knowledge databases and even facilitate communication with experts
Pattern recognition	Various parameter behaviors can be monitored for specific patterns using unsupervised learning		Robots/drones can identify patterns using visual footage to detect leakages, corrosions, etc., early in assets
Association rule learning	Unsupervised learning algorithms can be used to identify correlations and associations and help decisions in asset health monitoring	Early detection of deteriorating asset health using unsupervised learning algorithms

The promise of AI and transformation

The recent advancements in deep learning and machine learning have brought significant changes to predicting and determining attributes, including insights on anomalous behavior, digital signatures, changes and patterns.

It is true that AI has not always been the most effective solution, and this has led to misconception and hype. AI is still slow in learning new concepts and extending that learning to new applications. However, AI is still in its early stages of development for the oil and gas industry and one can extrapolate recent successes into the future.

For example, AI systems still need a tremendous amount of data to train themselves, while humans do not need to look at 50,000 cat images to identify a cat. In comparison, a human child can look at two cats and distinguish between a cat and a dog. So, in some areas, like speech recognition, AI is more sophisticated than the human mind, while in domains that require reasoning, contextual understanding, and goal seeking, AI cannot even compete with the intelligence of a child.

Even with limited development, companies such as Amazon are using AI to gain a competitive advantage in the market. Soon, every company will need to tap into the potential of AI.

Oil and gas companies must adopt AI in order to compete and grow in the market. We envision a future where AI capabilities will differentiate the winners from the laggards in the increasingly complex energy marketplace, and whoever adopts faster will gain the advantage.

Neeraj Kumar

Digital Solutions Lead, NA Energy Practice