Sensor data, ML powers AWS custom predictive maintenance service

Sensor data, ML powers AWS custom predictive maintenance service

Market news |
Amazon Web Services (AWS) has announced the general availability of a machine learning (ML) industrial equipment monitoring service that detects abnormal equipment behavior by analyzing sensor data from a customer's specific equipment.
By Rich Pell

Share:

The new service, Amazon Lookout for Equipment, uses AWS-developed machine learning models to help customers perform predictive maintenance on the equipment in their facilities. The service ingests sensor data from a customer’s industrial equipment (e.g., pressure, flow rate, RPMs, temperature, and power), and then it trains a unique machine learning model to accurately predict early warning signs of machine failure or suboptimal performance using real-time data streams from the customer’s equipment.

With Amazon Lookout for Equipment, says the company, customers can detect equipment abnormalities with speed and precision, quickly diagnose issues, reduce false alerts, and avoid expensive downtime by taking action before machine failures occur. There are no up-front commitments or minimum fees, and customers pay for the amount of data ingested, the compute hours used to train a custom model, and the number of inference-hours used.

“Many industrial and manufacturing companies have heavily invested in physical sensors and other technology with the aim of improving the maintenance of their equipment,” says Swami Sivasubramanian, VP Amazon Machine Learning, AWS. “But even with this gear in place, companies are not in a position to deploy machine learning models on top of the reams of data due to a lack of resources and the scarcity of data scientists. As a result, they miss out on critical insights and actionable findings that would help them better manage their operations.”

Traditionally, when analyzing the data from their equipment, most industrial companies typically use simple rules or modeling approaches to identify issues based on past performance. However, says the company, the rudimentary nature of these approaches often leads customers to identify issues after it is too late to take action, or receive false alarms based on misdiagnosed issues that require unnecessary and timely inspection.

Instead, companies want to detect general operating conditions or failure types (e.g., high temperature due to friction) along with complex equipment failures (e.g., a failing pump indicated by high vibration and RPMs but low flow rates) that can only be derived by modeling the unique relationships between sensors. Today, says the company, advances in machine learning techniques have made it possible to quickly identify anomalies and learn the unique relationships between each piece of equipment’s historical data.

However, most companies lack the expertise to build and scale custom machine learning models across their different industrial equipment. As a result, they often fail to fully leverage their investment in sensors and data infrastructure, causing them to miss out on key actionable insights that could help them better manage their critical equipment’s health and performance.

Amazon Lookout for Equipment is designed to enable industrial and manufacturing customers to quickly and easily build a predictive maintenance solution for an entire facility or across multiple locations. To get started, users upload their sensor data (e.g., pressure, flow rate, RPMs, temperature, and power) to Amazon Simple Storage Service (S3) and provide the relevant S3 bucket location to Amazon Lookout for Equipment. The service will automatically analyze the data, assess normal or healthy patterns, and build a machine learning model that is tailored to the customer’s environment.

It will then use the custom-built machine learning model to analyze incoming sensor data and identify early warning signs of machine failure or malfunction. For each alert, the service will specify which sensors are indicating an issue and measure the magnitude of its impact on the detected event.

For example, if Amazon Lookout for Equipment detected an issue on a pump with 50 sensors, the service could show which five sensors indicate an issue on a specific motor, and relate that issue to the motor power current and temperature, allowing customers to identify the issue, diagnose the problem, prioritize needed actions, and perform precision maintenance before issues happen—saving them money and improving productivity by preventing down time.

Amazon Lookout for Equipment, says the company, allows customers to get more value from their existing sensors, and it helps them make timely decisions that can materially improve operational efficiency. Amazon Lookout for Equipment is available directly via the AWS console as well through supporting partners in the AWS Partner Network. The service is available now in US East (N. Virginia), EU (Ireland), and Asia Pacific (Seoul), with availability in additional regions planned in the coming months.

Amazon Web Services

Related articles:
AWS unveils five industrial machine learning services
NXP, Microsoft demo edge-to-cloud ML predictive maintenance solution
IoT, big data to drive predictive maintenance market to $28B by 2025
AIoT startup uses electrical waveform data to predict machine health


Linked Articles
Smart2.0
10s