Platform function

Predictive Maintenance: maintenance before something fails

A cooling compressor that runs longer every day, a pump that draws more power for the same work: baiond notices the drift early, tells you which device and why, and helps you plan the repair before it becomes an emergency.

Request a consultation Technical details

AI-generated image: a robot at a desk in front of a city panorama, surrounded by data cards for energy, weather, occupancy, traffic and refrigeration

In one sentence

Predictive Maintenance in baiond is a function that learns how each connected machine normally behaves and warns you when it starts to drift, often well before it fails.

Many breakdowns announce themselves. A compressor runs a little longer each day, a motor draws a little more current, a room takes longer to reach its temperature. The signs are in the data long before the alarm – they are just too small to see on a single reading.

baiond learns what normal looks like for every connected device, from its own history and the weather around it. When the pattern shifts, it tells you in plain words: which device, how far it is off, since when, and what usually causes it. A hot week is far less likely to trigger a false alarm, because baiond takes the heat into account.

In a supermarket, that means the cold room compressor that has run 19.6 hours a day instead of 15 is on the technician's list this week – not on Saturday afternoon, when the fish counter is full.

What it does for you

What changes for you.

Repairs you plan, not emergencies

You see a drift early, often weeks ahead, and book the technician for a quiet morning instead of paying the weekend call-out.
Fewer false alarms

Baselines account for weather and season, so a hot week reads as a hot week, not as a broken machine.
Maintenance by condition

Service the devices that need it and skip the ones that do not – with the readings to show why.

Ask baiond

See it answer.

A question only this function makes answerable, answered the way baiond answers: readings, reasons, a recommendation and the source.

baiond AI Your data. Real answers.

Demo · not connected to live sensorsDemo

Is any of our refrigeration heading for trouble?

You 06:55

Checked Energy meter · Temperature sensor · baiond Gateway · Predictive Maintenance How did baiond know?

Sensor
baiond Gateway
Space System
AI agent

The sensor reports to the baiond Gateway, the Space System puts the reading in context (location, threshold, history), and the AI agent turns it into an answer.

Source: Energy meter · Temperature sensor · baiond Gateway · Predictive Maintenance

Run time / day 19.6h
Normal 14–16h
Outside temp. 18.2°C

Yes, in cold room 2. Its compressor now runs 19.6 h a day against a normal 14–16 h, and the curve has risen for 11 days while the outside temperature stayed around 18 °C. The weather does not explain it – the machine does.

baiond AI 06:55

Recommendation

Check the door seal of cold room 2 today and book the refrigeration technician this week. baiond keeps watching the run time and tells you if it climbs further.

Shall I show you the run time of the last 30 days?

app.baiond.io/predictive-maintenance

Under the hood

For technical readers

Protocols, data model and integration, for the colleague who signs off on the architecture.

Protocols and sources

Modbus
BACnet
M-Bus
MQTT

Inputs

Any time series that describes a device's work: energy and current, run times, temperatures, pressure, humidity, counts. Existing building automation is read through the baiond Gateway; weather data comes in as context.

Sensor and meter data (LoRaWAN, M-Bus, Modbus, BACnet, MQTT)
Local weather: temperature, humidity, rainfall, wind
The device's own history

Detection

baiond builds a baseline per device from its history and compares every new reading against it, adjusted for weather and time of day. It flags gradual drift (rising run time, falling efficiency), sudden changes and unusual cycling, and ranks devices by how far they are off.

Alerts & planning

Thresholds and sensitivity per device or per group in the Space System
Notifications in the app and by e-mail, with the readings attached
Maintenance windows and the history of every intervention in one place

Integration

Predictive Maintenance runs on the data baiond already collects – often without extra hardware where meters and controllers exist. Results are available to the AI agents, so you can ask "Which devices need service this month?" and get a ranked answer.

Where it sits