Vibration-based predictive maintenance · by aQb Solutions
Vibration-Based Predictive Maintenance for Rotating Equipment.
Vibera is predictive maintenance software for rotating equipment that diagnoses faults from vibration data and estimates remaining useful life.
Video transcript
Meet Vikram, a maintenance engineer at a plant in India. Six rotating machines, each producing vibration data. Which one will fail next? Siemens found that unplanned downtime costs the world's 500 biggest companies 11 percent of their revenues. Waiting for a failure is costly, and guessing which machine to service is no better. Then he opens Vibera by aQb Solutions, an AI predictive maintenance copilot for rotating equipment. Every asset is ranked by maintenance priority, with its vibration zone, health score and remaining life. Open an asset to see its vibration trend, spectrum and bearing envelope, with the diagnosis. Ask the copilot which assets need attention, and it ranks them, with evidence. Ask for a diagnosis, and it explains the fault and the remaining life range. Compare an asset before and after maintenance, to see whether the fix worked. Then it raises work orders for the priority assets, ready to connect to your maintenance system. Vibera finds the fault. Vikram plans the repair. The fleet shown is simulated. Predict the failure. Plan the repair. Protect uptime.
Vibera by aQb Solutions is an AI predictive maintenance copilot for rotating equipment that diagnoses bearing and shaft faults from vibration data, estimates remaining life and raises work orders. Fix it before it fails: it reads vibration and condition data, ranks every asset by maintenance priority, diagnoses the fault, estimates the remaining useful life and raises the work order. The AI explains, while the physics comes from deterministic analysis, so every number traces back to a tool call you can see. The fleet shown is simulated.
Impact you can measure
Illustrative targets, to be confirmed in a pilot on your own data.
30 min → 2 minTarget time to diagnose an asset
30%+Target reduction in unplanned downtime, to be confirmed in a pilot
100%Assets ranked by maintenance priority
Days → Same shiftTarget time to raise a work order
20%+Target reduction in unnecessary maintenance, to be confirmed in a pilot
100%Findings traced to the tool calls behind them
What it does
Fleet health ranking
Every asset gets a vibration zone, health score, remaining useful life and priority, so the fleet is ranked by what needs maintenance first.
Sensor and condition data intake
Load asset details, daily condition history and vibration waveforms exported from your sensors or condition-monitoring system. Live feeds from sensors and historians are scoped for each deployment.
Vibration analysis and anomaly detection
Trend, velocity spectrum and envelope spectrum show the fault frequencies behind a rising reading, and machine-learning anomaly detection compares each asset with its own healthy baseline.
Fault diagnosis
Rules separate imbalance, misalignment and bearing race, ball and cage defects, and show the evidence behind each call.
Remaining useful life
A degradation forecast extrapolates to the failure limit and gives the remaining life as a range, and it says so when there is no trend.
Work orders and maintenance integration
Priority assets get corrective actions and work orders in the built-in register, ready to connect to your maintenance system, such as SAP PM or IBM Maximo, as part of integration.
Key Questions Answered
01 / 04
Which machinery, sensors and data formats are supported?
Rotating equipment such as pumps, gearboxes and compressors, with faults including imbalance, misalignment and bearing race, ball and cage defects. Input is asset details, daily condition history and vibration waveforms exported from your sensors or condition-monitoring system. Sampling and operating-condition needs are confirmed on your own data.
02 / 04
Is it imported-data analysis or live monitoring?
Today it analyses data you import. Live feeds from sensors and historians are scoped for each deployment.
03 / 04
How is a diagnosis evidenced, and how reliable is remaining life?
Each diagnosis shows its evidence, such as the fault frequencies in the spectrum, and every number traces to a tool call. Remaining life is a range from a degradation forecast extrapolated to the failure limit, and it says so when there is no trend.
04 / 04
What has been validated, and what does it connect to?
It is shown on a simulated fleet, which shows the workflow and is not production validation or proof of downtime reduction. Performance on known faults, false alarms and remaining-life error is evaluated in a pilot on your equipment. Work orders are raised in the built-in register, and connecting to SAP PM or IBM Maximo is integration work.
How it works
01Analyse Vibration
02Detect Anomalies
03Diagnose Faults
04Predict Remaining Life
05Prioritise & Advise
06Raise Work Order
Technology stack
LLM
Agent framework
Python & FastAPI
scikit-learn & SciPy
SQLite
Use cases
01Bearing failures caught early
01Process plants and utilities
Bearing failures caught early
Detect early bearing defects in pumps and motors from vibration data, and plan the repair inside the remaining-life range.
02The right machine serviced first
02Maintenance planning
The right machine serviced first
Rank every rotating asset by maintenance priority, with the evidence behind each ranking, so crews go where the risk is highest.
03Proof that the repair worked
03Reliability engineering
Proof that the repair worked
Compare vibration before and after maintenance, and close the work order only when the readings confirm the fault is fixed.