AI in Action: Mining & Energy4th December 2018
This is the fifth in a series of articles highlighting the many applications of Artificial Intelligence.
The “extractive industries” as they’re sometimes called, Mining and Energy, were the first truly global vertical, and continue to span the globe in the pursuit of access to natural resources. AI can be of great service in innumerable ways in this field. Here’s a few examples:
I was talking to a pipeline company last year about risk management. At the time I was selling a tablet-based inspection system. “Data collection is not our problem” I was told. “we have lots of data automatically collected via SCADA — our need is to interpret this data effectively.” They had a great deal of data being collected along the hundreds of miles of pipeline, but were having trouble processing it all in an effective manner.
This is a great application for AI, specifically predictive modeling. The major risk for a pipeline is a leak or spill. To minimize the risk of such an event, it’s imperative that the pipeline operator is very vigilant when it comes to maintenance. A strong predictive model is a great help in deploying maintenance resources effectively. The model can interpret the data and help prioritize the maintenance schedule. Items can be moved up and pushed down in the schedule according to how critical the maintenance work is deemed to be. Some tasks can wait for a week, some can wait for a month — some have to be fixed as soon as possible. Managing this maintenance schedule effectively can minimize the risk, and maximize the efficiency of the maintenance spend. AI can be invaluable in this process.
A modern refinery plant is packed with production equipment that includes thousands of sensors. These sensors will typically provide a numerical reading as well as a color reading, usually green/yellow/red. For example, a gas pressure reading might be 102.7 kPa, and in the yellow. The automation systems are very effective at managing this river of sensor data. What they aren’t particularly good at, is recognizing more complex patterns of multiple sensor readings that may be a precursor to problems. Identifying problems early are critical to applying predictive maintenance — predictive models based on neural networks are well suited to this purpose.
Over the last century, mining and oil prospecting/exploring has evolved from an art to much more of a science. The image of the individual searching the globe may be much more romantic, but the reality of today is that most of the exploration work is done in office towers by teams of engineers, data scientists, and developers. Poring over satellite data in a boardroom doesn’t compare to James Dean dancing under a gusher, but that’s how things get done in 2018.
Within this science-oriented approach to exploration, there is much service that AI can deliver. There is a neural network type called GANS, which stands for Generative Adversarial Network, which can improve the quality of images, including satellite imagery. Running the satellite imagery through a GAN processor before feeding it into a predictive model can produce superior results when it comes to identifying locations that appear to be resource-rich and are worthy of further study. The cost of in-field testing is very high for all resources at this point — if you can increase the hit rate upstream it can greatly improve overall exploration productivity.
The opposite end of the business to exploration is distributing power through the grid. AI can help make the grid more efficient and more responsive. Matching the demand for power and the supply of power is no different than in other industries. Power management is really a specialized type of sales and operations planning. AI can produce more accurate measures of both commercial and consumer demand, and can also help improve the efficiency of power generation, including solar and wind.
Mining and Energy is one of the largest global industries, and is probably the most “global” of all industries — there is no continent, country, or territory that the industry won’t assess for resource potential. These are just a few ways that AI can benefit Mining and Energy. The true breadth of AI applications within the field is on the same scale as its operations.
Part I: AI in Action: Healthcare
PartII: AI in Action: Financial Services
Part III: AI in Action: Entertainment and Media
Part IV: AI in Action: Manufacturing and Distribution
Ken Tucker is a business consultant specializing in AI and Analytics.
AI in Action: Mining & Energy was originally published in Hacker Noon on Medium, where people are continuing the conversation by highlighting and responding to this story.