Forward-looking models that drive task scheduling and resource decisions.
Gradient-boosted regression over 32 features (a/c type, load, gate, crew, wx, inbound delay). Outputs P50/P90 TAT per turnaround, refreshed every 5 min.
MILP solver assigning stands, bays, tow-tractors and pushback crews. Honors aircraft compatibility, towing windows, and walk-time penalties for pax connection bank.
Queueing-network model (M/G/k) sizing baggage, fueling and catering teams against rolling 90-min demand curve. Surfaces understaffed windows 60 min ahead.
Detect, score, and classify risks across operational, safety and compliance domains.
Combines likelihood priors (from 5-yr SMS data) with live signal posteriors (wx, NOTAMs, equipment MEL, staffing) into a 0–100 composite risk score per asset, gate, and shift.
Isolation-forest ensemble flagging anomalous taxi-times, fuel-burn deltas, and gate dwell. Routes anomalies to the SMS desk with explainability (top-3 feature contributions).
Maps live ops events to DGCA, ICAO Annex 14 & ISAGO control objectives. Continuously scores residual risk per control and surfaces audit-trail gaps before they age.
Daily composite risk vs alert threshold (70).
Departures within 15 min of STD.
Distribution across SMS risk taxonomy.
Bubble = number of open risks at that cell. Hover for breakdown.
Ranked by composite score = likelihood × impact × controls-gap.
flowchart LR A[Live ops signals
AODB · ATC · Wx · MEL] --> B[TIA-Detect
anomalies] A --> C[TIA-Score
composite risk] B --> D{Severity?} C --> D D -- Low/Med --> E[Auto-log
monitor] D -- High/Crit --> F[Open risk
+ assign owner] F --> G[Mitigation plan
TIA-Allocate suggests] G --> H[Execute & track] H --> I[TIA-Assure
control evidence] I --> J{Residual
acceptable?} J -- Yes --> K[Close · SMS log] J -- No --> F K --> L[Weekly review
Ops · SMS · Engg]