Precision Makes Perfect: Machine Health Sensing

How many times have you driven in your car and heard something that just didn’t sound right? It’s disconcerting, not knowing if that sound is a minor glitch or a major problem that could have dire safety or financial consequences.

Now imagine if you had the ability to accurately diagnose problems proactively, by sensing and analyzing the slightest vibration and the most miniscule sound difference. Then, after you capture this signal, you quickly diagnose treatment, in real-time, before any potential damage could occur.

Sound impossible? Meet OtoSense technology from Analog Devices.

At a Glance

Companies/Industry:

Machine health sensing

Challenges:

Machine health analysis is often accomplished with skilled engineers (a finite resource); need to increase AI-focused technology in order to create scalable insight on machine performance.

Goal:

Translate raw sensing signals of any asset into insight improving machine uptime and allowing manufacturers to scale diagnostic expertise

Machine health

 

Add Precision, Remove Guesswork from Machine Learning

Today, it’s impossible to pre-emptively maintain machine health at scale, which leads to an inefficient and expensive system of scheduled and reactive maintenance. ADI’s OtoSense brings the possibility of “embedding” a technician in the machine, constantly listening and assessing the machine’s behavior and health.

The key is real-time interpretation of these tough-to-obtain signals. Sound and vibration within a machine are very dense data, emitting hundreds of kilobytes per second, per sensor. With this type of volume, you can’t send all the data from all the sensors to a remote cloud and expect to do accurate, real-time analytics. The interpretation must happen at the edge, continuously, updating at the speed at which data arrives.

OtoSense detects and captures variations in sound and vibration in machines, and synthesizes this data into machine health insights. This ability allows customers to confidently measure tough-to-acquire signals and translate them into actionable information, without requiring the system be connected to a network.


Simplified Block Diagram of the OtoSense System

Machine health

 

Ready for Takeoff with OtoSense

Reliability and uptime are critical to the airline industry. Hundreds of thousands of passengers every day are counting on safe, on-time, fully functioning aircraft to get them to their destinations.

The use of artificial intelligence (AI) is expanding as a diagnostics and decision-making tool for airline maintenance teams at large fleet commercial airlines. Unfortunately for all airlines, there just aren’t enough trained line maintenance technicians to efficiently assess the normal operation of an engine or its starter at the gate, with air traffic volume growing much faster than maintenance capabilities. Starter failure is a common cause for grounding planes, resulting in the disembarkation and rerouting of all passengers. OtoSense can avoid these consequences by detecting early signs of failure, quickly, oftentimes in seconds.

For sensing interpretation leveraging AI, it usually takes hours for a good engineer to browse through sensor data, analyze it, and report on findings. For an experienced technician or engineer to fully understand a machine perfectly, this might take years of training. And then, as so often happens, when these skilled engineers reach proficiency, they retire.

What OtoSense does is effectively “learn” from these domain experts, every day, about any number of machines, and progressively become an expert itself. By developing an automated method of machine health sensing — monitoring the operational performance of machines — ADI’s OtoSense technology is canonizing this painstakingly earned expertise and expanding the scale of engine monitoring to make it a truly data-driven exercise rather than a subjective assessment.


Machine health

 

From How to Wow

As is the case with many new, game-changing technologies, when OtoSense is first presented to a prospect, the reaction is one of skepticism. After a few moments of OtoSense being deployed, the ‘wow’ effect quickly sets in. Prospects soon see how OtoSense can sense the extremely subtle differences in the usual from the unusual, and interact with human experts to name the event, in real time.

Today, OtoSense is having a profound impact across numerous industries: in aerospace, automotive, industrial equipment, healthcare (cough monitoring for COPD patients, sleep apnea detection, chest sounds monitoring, elderly monitoring), building monitoring (increasing the safety of occupants by monitoring emergency areas and places where no camera could be installed), and a long list of upcoming applications.

Hopefully, in the not-too-distant-future, you could be driving down the road and hear an unexplained noise in your car’s engine. Then, with a push of a button you’ll access technology like OtoSense and know whether that strange sound or vibration translates into you needing a simple oil change or complete engine overhaul.


*Gartner, Markets & Markets, Research and Markets, Accenture, Transparency Market Research