Skip to content
 
Try for free

ThingsBoard Cloud

Choose your data region

Your data stays in the region you choose, for residency and compliance. No credit card required.

North America Data stored in North America Europe Data stored in the European Union

Rather run it yourself? Install on your own servers

Vibration Analysis for Predictive Maintenance

Vibration analysis catches machine problems from the way a machine shakes, before they turn into a breakdown. Pumps, fans, motors, compressors, and gearboxes all change their vibration when something starts to go wrong, and they do it weeks or months before you can hear it, feel the heat, or see it in the output.

Overview tab of the Conveyor drives vibration model with monitoring enabled: the verdict of Conveyor 4E, the metric trend, the Monitoring card with scheduled scanning, saved metrics, and alarm rules, and the table of machines with their ISO zones and most likely problems

Trendz turns the vibration sensors you already connect to ThingsBoard into an online condition monitoring system. It reads the raw vibration signal of every machine and tells you three things about it: whether it has changed, how serious that is, and what the most likely cause is. Any wired or wireless accelerometer that sends raw waveform samples to ThingsBoard, directly or through a gateway, can be analyzed.

Curve of machine condition over time with vibration change marked first, then audible noise, heat, and failure Curve of machine condition over time with vibration change marked first, then audible noise, heat, and failure

Acting on that early signal instead of waiting for the breakdown is what predictive maintenance means, and vibration analysis in Trendz implements it for your machines.

Predictive maintenance

Section titled “Predictive maintenance”

Predictive maintenance services a machine when its condition calls for it, instead of after a failure or on a fixed calendar. There are three ways to look after the same machine:

  • Reactive: Run it until it fails, then repair it. Every fault becomes unplanned downtime.
  • Preventive: Service it at fixed intervals. Healthy machines are opened for nothing, and worn ones still fail between services.
  • Predictive: Watch its condition and act when it changes. Repairs move into planned downtime, parts are ordered ahead, and healthy machines are left alone.
Three timelines of one machine under reactive, preventive, and predictive maintenance Three timelines of one machine under reactive, preventive, and predictive maintenance

Predictive maintenance is a cycle: measure regularly, learn what is normal for each machine, detect a change, diagnose the cause, plan the repair, and verify the result. Trendz covers every step except the planning and the repair itself, and hands its findings to ThingsBoard alarms, rule chains, and notifications.

Predictive maintenance cycle of six steps, with every step except plan repair marked as covered by Trendz Predictive maintenance cycle of six steps, with every step except plan repair marked as covered by Trendz

How it works in Trendz

Section titled “How it works in Trendz”

A vibration model connects the sensors to the alarms. Once it is set up, every scan runs on its own.

Blueprint of a vibration model across ThingsBoard and Trendz: a sensor sends a 3-second waveform to device telemetry, which ThingsBoard stores in its database; a scheduled scan starts a scan task that reads new samples, analyzes them, saves results to the Trendz database, saves metrics back to device telemetry, and raises alarms, which ThingsBoard stores and passes to dashboards and rule chains Blueprint of a vibration model across ThingsBoard and Trendz: a sensor sends a 3-second waveform to device telemetry, which ThingsBoard stores in its database; a scheduled scan starts a scan task that reads new samples, analyzes them, saves results to the Trendz database, saves metrics back to device telemetry, and raises alarms, which ThingsBoard stores and passes to dashboards and rule chains
  1. Send a sample: The sensor records a 3-second waveform, for example once a day, and sends it to ThingsBoard. The format is described in Getting started.
  2. Store it: ThingsBoard stores the waveform as device telemetry.
  3. Start a scan: On the schedule set on the model, for example every 6 hours, Trendz starts a scan task.
  4. Read new samples: The scan reads the newest sample of each device.
  5. Analyze: Trendz computes the spectra, the metrics, the ISO zone, and the fault scores of each sample.
  6. Save to the database: Trendz stores the spectra and results in its own database, for the Analyzer.
  7. Save metrics: Trendz writes RMS, peak-to-peak, and other metrics back to ThingsBoard as telemetry.
  8. Raise alarms: Trendz checks every alarm rule on the new sample. A breached rule raises an alarm, which ThingsBoard stores. See Monitoring and alarms.
  9. Act on it: Dashboards show the metrics and alarms, and rule chains send them on by email, SMS, or to a ticketing system.

Where to start

Section titled “Where to start”

Features

Section titled “Features”

Trendz gives you the tools for every step of the cycle: it keeps watch over vibration levels and raises alarms on its own, tells you how serious a change is and what is likely behind it, and lets you look into the signal of any machine when you want to see the evidence yourself.

© 2026 The ThingsBoard Authors