By
The Future of AI in Asset Management What’s Real and What’s Fiction
Let’s start with the “What is Fiction” question first, Skynet from the Terminator is not real, at least not in Facilities and Asset Management. The machines are not taking over your facility operation, yet.
What is real is the fact that AI, or Artificial Intelligence, is the field of machine learning and decision making. AI systems can learn about what you do, how you do it and what you should be doing. With the right level of access, AI systems can issue not only alerts, but create work orders for problems that it has detected. Given the amount of knowledge and processing power today’s AI systems have, you need to learn how to train your AI.
The truth is that we are in the early stages of AI. Variations of it are present in our everyday lives but is more commonly used as “Expert Systems” that interpret parameters and input into a set of activities that can be executed by a computer system at a speed much faster than a human is capable of. Where AI departs from these “Expert Systems” is in its ability to adapt, learn, and consume information that seemingly is not related, and draw conclusions from the data.
In Facilities / Asset Management, we see AI systems being deployed and given time to “learn” about the facility. The AI System will begin exploring the data and start looking for event correlations. For example, predictive equipment failure is often a task for AI systems. AI systems are used, for example, with vibrational analysis. An acoustic sensor will “listen” to the equipment and as the equipment is maintained it will note the changes in the acoustic signature of the machine. Should the piece of equipment fail, it will take note of the sound of the equipment (acoustic signature) prior to failure. The AI system will then match the acoustic signature to the maintenance and failure records and “predict” when that equipment will either fail or requires maintenance. If there are multiple pieces of equipment, it will apply what it learns from one piece of equipment to the others.
The other use case is where an equipment fails with no apparent cause. After review by the AI system it is found that the specific crew install the equipment. The AI system can identify what other equipment was installed by the crew and is at risk for failure.
These are great examples of the future but there are a couple of pitfalls that need to be considered. First, the AI system needs access to all the information about your operations. This could be sensor data, work orders, manufacture specification sheets, etc. Without a complete data set it cannot make accurate predictions. The second pitfall is a need for all the information to be digital and readable, which sounds obvious but in most cases information still exists in paper format Even the simple case where scanned manufacturers cut sheets are scanned as an image results in data that cannot be read by the AI system.
These are real challenges for most organizations. While some organizations are trying out AI, it will take most organizations considerable effort and cost to implement. However, the upside is that when sensors and work order data is connected to AI, it will learn very quickly. It can be used to give insights into operations. This results in improved equipment uptime and operational efficiencies.
