Aluminum fuselage control with AI — riveting, shot peening and coating in a single live dossier.

The problem

A fuselage panel flies for decades — and the dossier should still be readable 20 years later.

The quality manager of a fuselage plant knows that an aerospace aluminum fuselage is designed to fatigue in its own time, not before. A fatigue crack shows up decades later and the NTSB accident report asks: what shot peening did that panel get? What anodizing thickness? Did the riveting follow the recipe? The answer has to be alive, indexed by fuselage serial number, for the whole operational life of the aircraft.

The bottleneck is four realities living in different systems:

1. The sheet — aluminum heat lot (Al 2024, 7075, Al-Li), supplier mill cert. ERP + a supplier PDF.
2. The riveting — correct pattern, head, alignment, no damage to the skin. Today it is validated with a gauge and a veteran's eye.
3. The shot peening — Almen parameter, coverage. In the machine or on a paper report.
4. The coating — anodizing, primer, thickness. On the anodizing line and in thickness-measurement software.

The classic system works 99% of the time. And even so, every time the auditor asks you to rebuild the dossier of a panel made two years ago, the quality team spends weeks assembling a PDF that should have been alive since the very first operation.

How it fits the IRIS system

iLEAN does not add a fifth system — it seals the cracks between the ones you already have.

The problem with fuselage control is not a lack of information: it is information in islands that, at the moment of signing, never quite lands in the same file. iLEAN acts as the putty that fills those gaps, without asking you to change your riveting machine, your shot peening machine or your anodizing line.

Edge sees every panel after every operation. Connect reads the old machine and the bath even when it is isolated. The agent assembles the dossier by serial number and, if something does not add up, it holds the panel. The person signs — never the other way round.

The three iLEAN pieces applied to aluminum fuselage control:

• Edge — terminals with machine vision (CNN) at the critical stations: after riveting (head, alignment, missing rivet, skin damage), after shot peening (visible uniformity), after anodizing/primer (coverage, missing coating). Actuator at 45 ms when a hold is needed. It works with no network.
• Connect — captures the Almen parameter reading from shot peening whether the equipment is modern (direct integration) or old (panel reading by vision, or serial-port reading); captures the anodizing thickness measurements from the vertical software; captures the supplier mill cert whether it arrives by email or as a PDF. Voice assistance for the operator works both ways: the agent talks to them, the operator dictates back.
• Agent — cross-references heat lot, riveting, shot peening, coating, NDT and the human signature. It assembles the dossier by fuselage panel serial number. If any step is missing, it holds the panel before the next assembly step and alerts the person in charge. The person signs; the panel does not move on by itself.

See the full IRIS architecture →

Before and after

Manual control with a checklist vs. cross-referenced control with iLEAN

Aspect	Manual control + checklist	With iLEAN Edge + Connect + Agent
Riveting per panel	Gauge + veteran's eye + paper record	Edge vision per panel, photo filed by serial number
Shot peening (Almen)	Operator reading written on a report	Connect captures it from the equipment (even isolated)
Anodizing thickness	Vertical software, PDF at the end	Cross-referenced with the recipe and filed by serial number
Melt traceability	PDF in a shared folder	Heat lot indexed to the panel dossier
Operation with no network	n/a	Edge keeps running on the panel's own power
Dossier for auditor / customer	Rebuilt by hand, weeks of work	Live dossier by serial number, automatic

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Aerospace Tier 1/2 with fuselage panels in Al 2024 / 7075 / Al-Li. Riveting + shot peening + anodizing + NDT line by eddy current/dye penetrant.
• Edge pilot at the riveting station + Connect over shot peening and anodizing + a dossier agent by serial number. First value expected within a few weeks.
• Indicative payback between 4 and 9 months, depending on the frequency of NCRs for out-of-spec rivets/skin damage and the quality hours spent assembling audit dossiers over recent years.
• Reduction in rework caused by rivet/skin damage detected late: conservative estimate ≥ 30%. The hard lever is the rework avoided on an already assembled panel.

And a side data point from the quality standard: automotive demands levels in the order of 25 PPM ^[2] — aerospace fights with even tighter tolerances and with operational lives measured in decades. The argument is the same: at those levels, the eye alone does not get there.

And the quality data point for the committee

“What if the AI gets it wrong and lets a badly set rivet through?” — hallucination is a problem of free generation, not of anchored tasks. In tasks where the AI merely compares an image with a learned pattern and recontextualizes a shot peening reading into a file, the best models brought error below 1.5% ^[1]. And even so, a critical panel is never signed off alone: iLEAN holds and the person signs. The three safety rings exist precisely for this.

^[1] OpenAI paper “Why Language Models Hallucinate”, 2025 — on the reliability of AI in anchored tasks.

^[2] The 25 PPM quality standard in automotive — Symestic.

Cross-links: composite wing control · aerospace material traceability by melt · ultrasonic NDT in aerospace