PHLAI
Request a DemoExplore the Platform

Solution · Predictive maintenance

Predictive maintenance, earned in stages.

Prediction needs evidence: labelled failures, evaluation and drift monitoring. PHL AI starts where that evidence is built — condition monitoring and transparent anomaly review — and adds trained models only when your data can support them.

PHL MaintainRequest a Demo

01The progression

Four stages. The first two are in the first release.

  1. Stage 1Available in MVP

    Condition monitoring OBS

    Readings, thresholds and freshness for every critical asset.

    Needs
    Asset list, signals, a data path.
    Gives
    You see deterioration that reaches a measured signal.
  2. Stage 2Available in MVP

    Rules and anomaly review STAT

    Sustained-threshold rules, rate-of-change rules and z-score deviation flags against each asset’s own baseline.

    Needs
    Thresholds with owners; 2–4 weeks of history for baselines.
    Gives
    Earlier, explainable flags, ranked by Review Priority.
  3. Stage 3Planned

    Labelled maintenance history NOTE

    Recorded outcomes on every review item and imported CMMS events, linked to the readings before them.

    Needs
    Consistent outcome recording over months.
    Gives
    The training and evaluation data a model would need.
  4. Stage 4Future capability

    Trained predictive models MODEL

    Models per asset class, such as failure-mode classifiers or remaining-useful-life estimates.

    Needs
    Enough labelled failures, holdout evaluation, a model card and your sign-off.
    Gives
    Probabilistic estimates with published precision and recall — only then.

02What stages 1–2 look like

A rising trend, a rule, and a ranked review.

On the demo cell, PMP-001’s vibration has more than doubled in a week. A rule fired, and Review Priority ranks it first. No failure is predicted.

PMP-001 vibration, 30 days (mm/s)4-hour averages (min–max band)
0.02.04.06.08.014 Sep21 Sep28 Sep5 Oct12 OctWarning 4.5Critical 7.1No dataNo data

Vibration velocity RMS over the last 30 days: ranged from 1.8 to 5.0 mm/s, mean 2.8 mm/s. Latest reading 4.9 mm/s at 2026-10-14 14:20 demo time. Thresholds: warning above 4.5 mm/s; critical above 7.1 mm/s. 5 h without data.

Reading warning > 4.5 mm/s critical > 7.1 mm/s No data

Review Priority RULE

80High

  • Active warning rule: Vibration (AL-0007)+35
  • Active attention rule: Bearing temp (AL-0010)+15
  • Rising 7-day trend: Vibration +115 %+20
  • Criticality A+10

Rules-based ranking to help order inspections. Not a probability of failure.

Bring your industrial systems into clearer view.

Start with a focused equipment workflow and expand toward connected digital twins and industrial simulation as your requirements evolve.

Request a DemoExplore TwinCore

Illustrative demo · no signup · not connected to equipment