Understanding sensor measurements - Amazon Monitron

Amazon Monitron is no longer open to new customers. Existing customers can continue to use the service as normal. For capabilities similar to Amazon Monitron, see our blog post.

Understanding sensor measurements

When a sensor is initially paired to an asset, Amazon Monitron will learn from the vibration and temperature data collected from the equipment, establishing a baseline to determine what is "normal" for that asset. It will use this learning to detect potential failures in the future.

Depending on the situation, operational scenario, use case, and various parameters like the asset’s duty cycle, Amazon Monitron will take between 14 and 21 days to establish this baseline. During this initial learning and training phase, the asset is assumed to be healthy.

After establishing a baseline for the asset, Amazon Monitron monitors the data it collects, looking for an event or trend that indicates a potential failure. It specifically watches for increases in temperature, or vibration levels, or both. Increases in temperature and vibrations are two of the main indicators of a malfunctioning machine. Machine abnormalities often indicate that an asset is starting to fail.

Amazon Monitron uses vibration thresholds established by the International Organization of Standardization (ISO) for your class of machinery. It applies the ISO thresholds in combination with its self-training model to assess actual thresholds to fit your equipment. For example, if your machine runs a little hot or a little cold, or if it vibrates a little bit more than standard, Amazon Monitron adjusts the thresholds slightly so that it can more accurately identify when the machine is acting abnormally.

The only alarms you will recieve during the initial learning and training period will be from the ISO model (which does not require any learning period). You should treat ISO alarms during the training period as you would any alarm—acknowledge the alarm, perform any necessary review of the machine, and then close out the alarm with the suitable action taken code. After that time, Amazon Monitron continues to fine-tune the baseline, building a better picture of "normal" as the sensor collects more data.

If temperature or vibration levels inconsistently rise above the modified threshold, a failure might be possible, but is probably not imminent. In that case, Amazon Monitron sends a Warning notification. If the increase is consistently above the threshold, the conditions are clearly abnormal and a failure is much more likely. Under those circumstances, Amazon Monitron sends an Alarm notification to the mobile or web app.

Notifications screen showing alarms and warnings for various positions with vibration and temperature data.

a mobile app notification

Notifications panel showing alarms and warnings for various positions with detected vibration and temperature issues.

a web app notification

In this example, the Position 3 sensor has detected a persistent increase in temperature and vibration, indicating that a potential failure needs to be investigated.

Pump monitoring interface showing vibration and temperature alarms with graphical data.
Dashboard showing vibration and temperature data for a pump main asset with alarm notifications.