Industrial machine operators might want to take a closer look at a new service from Amazon Web Services Inc. that uses machine learning to help them schedule maintenance work on their equipment.

The Amazon Lookout for Equipment service was made generally available today, and works by amassing data from sensors installed on manufacturing equipment and analyzing it to detect any early warning signs that the machine might fail or perform at less than 100%.

The service, which was announced at AWS re:Invent 2020 in December, enables manufacturers to build a predictive maintenance system for single and multiple production facilities, Amazon said. Sensor data related to things such as pressure, flow rates, temperature and power is uploaded to the Amazon Simple Storage Service, and the relevant bucket location is added to Lookout for Equipment.

Then the service uses machine learning algorithms to analyze the data, looking for patterns that might indicate a problem with the machine. If something suspect is found, it will provide detailed alerts that engineers can look at to try to fix the issue before it causes a problem.

With each alert that it provides, the service specifies which sensors have indicated the problem, as well as a measure of the magnitude of the impact on the detected event, Amazon said. For instance, if Lookout for Equipment spots an issue with a pump that’s fitted with 50 sensors, it will identify which of those sensors indicated a problem so engineers know exactly where to look.

AWS Vice President of Machine Learning Swami Sivasubramanian said that though many industrial and manufacturing companies have invested in physical sensors and other equipment that aims to improve maintenance, they lack the expertise and resources to properly assess the data those systems generate.

The service is designed to be used with machinery that is equipped with its own sensors. For machines that don’t have sensors, Amazon offers its Monitron service that comes with a suite of customized sensors that can be attached to machines to monitor them for abnormal vibrations and temperatures that might indicate a problem.

Earlier this year, AWS launched services such as Amazon Lookout for Vision in February, which uses machine learning algorithms to spot defects and anomalies in visual representations using computer vision. Another service, Amazon Lookout for Metrics, debuted in March and helps enterprises detect and respond to important business events such as a sudden jump in user sign-ups.

With these new offerings, Amazon is staking a claim in a predictive factory maintenance market that could be worth as much as $12.3 billion by 2025, according to research from Markets and Markets. It competes against other cloud players such as IBM Corp., as well as startups such as Augury Inc.

Amazon Lookout for Equipment is available now in the US East (N. Virginia), EU (Ireland), and Asia Pacific (Seoul) regions, with availability in more regions to come.

Photo: Medyamen2/pixabay