← Platform Overview

AI & Analytics

Automated Work Orders

Physics and ML models detect issues, assess severity, create work orders, assign technicians, and pre-order parts -- automatically. Close the loop from detection to resolution.

AUTOMATION PIPELINESENSORDATA1PHYSICS/MLMODEL2THRESHOLDBREACH3AUTO-GENWORK ORDER4ROUTE TOTECH5RESOLVE6DetectionResolutionAUTO-GENERATED WORK ORDERWO-2400AUTO-GENERATEDAsset:Pump-001 (WTP Building A)Failure:Bearing Inner Race DefectRUL:18 days remainingAction:Replace bearing (SKF 6208-2RS)Assigned:J. Rodriguez (Shift A Lead)Duration:Est. 2.5 hoursParts:SKF 6208-2RS x1, Seal kit x1PRIORITY CLASSIFICATIONP1CRITICALSafety risk or total failure> 90% confidenceP2HIGHRUL < 14 days> 80% confidenceP3MEDIUMEfficiency drop > 10%> 70% confidenceP4LOWMinor drift detected> 60% confidenceSLA TIMER04:00remaining to acknowledgeROUTINGArea: WTP-AJ. RodriguezShift: A (Day)AvailableSkill: Mech.CertifiedTRIGGER SOURCESPhysics Modelη < 65%CorrectiveAnomaly Engine4/5 consensusInspectionRUL ModelRUL < 21dPreventiveThresholdT > 85°CEmergencyDrift DetectCal. offsetCalibrationwo-engine@twinedge:~$ tail -f /var/log/auto-wo.log[00:00] TRIGGER physics_model efficiency_drop Pump-001[00:12] CLASSIFY priority=P2 confidence=87% failure_mode=bearing_defect[00:24] GENERATE WO-2400 asset=Pump-001 type=corrective[00:36] ROUTE assigned=j.rodriguez shift=A sla=4h[00:48] NOTIFY push_sent=true sms_sent=false email_sent=true

Automation Workflow

01Model Detects IssuePhysics model calculates pump efficiency at 62% -- 15 points below baseline. Bearing RUL model estimates 18 days remaining.
02Severity AssessedRule engine evaluates: equipment criticality (high), degradation rate (accelerating), production impact (moderate). Priority: P2.
03WO Auto-CreatedWork order generated with: failure mode, affected sensors, recommended procedure, estimated duration (2.5 hrs), and required parts list.
04Technician NotifiedAssigned technician receives push notification with equipment location, QR code link, and maintenance window recommendation.
05Parts Pre-OrderedIf spare parts inventory is below threshold, purchase requisition is auto-created and routed to procurement for approval.

Trigger Sources

Five different model types can trigger automated work orders.

SourceExample TriggerWO Type
Physics ModelPump wire-to-water efficiency drops below 65%Corrective
Anomaly DetectionMulti-algorithm consensus flags bearing vibration patternInspection
RUL PredictionSeal remaining useful life falls below 21 daysPreventive
Threshold AlarmMotor winding temperature exceeds 85C for >10 minutesEmergency
Drift DetectionSensor calibration drift detected (flow meter reads 8% low)Calibration

Configuration Options

Configurable Thresholds

Set trigger conditions per equipment type: efficiency drop >10%, vibration >4.5 mm/s, RUL <30 days. Combine multiple conditions with AND/OR logic.

Routing Rules

Route by equipment type, location, severity, or time of day. Night shift critical alerts go to on-call lead; daytime P3s go to area technician.

Parts Pre-Ordering

Link failure modes to bill-of-materials. When a bearing failure is predicted, the system checks warehouse stock and creates a PO if needed.

Escalation Chains

If a P1 work order is unacknowledged after 30 minutes, escalate to shift supervisor. After 60 minutes, escalate to maintenance manager.

ML Confidence Gating

Only create work orders when model confidence exceeds your threshold. 90% confidence = auto-create. 70-90% = suggest. Below 70% = log only.

De-Duplication

Suppress duplicate work orders for the same asset and failure mode within a configurable cooldown window. Merge related alerts into a single WO.

Close the Loop Automatically

From anomaly detection to work order completion -- no manual intervention required. Your team focuses on fixing, not dispatching.