๐Ÿ”ง Automotive Fitting Gap โ€” Automated Tool

ASSB Predictive Analysis ยท Input data โ†’ analyse failure part/process โ†’ predict future failures
v1 ยท DESIGN
Started: 2026-06-24
TBP Step 2 + Step 6 evidence
AI Firm 2027 candidate

๐ŸŽฏ The 3 Classification Axes

Every fitment gap/levelness issue at ASSB falls into one of three axes. Classify first, then apply the standard fix tree.

1. Part ร— Part

Gap between two assembled components.

๐Ÿ“Œ Example: IP upper โ†” IP lower

Tolerance: Gap โ‰ค 0.5mm ยท Level โ‰ค 0.5mm

Typical root cause: sub-component supplier variation, locator wear, clip fatigue.

2. Part ร— Body

Gap between a panel and a body opening.

๐Ÿ“Œ Example: Headlamp โ†” Fender

Tolerance: Gap โ‰ค 1.0mm ยท Level โ‰ค 1.0mm

Typical root cause: panel locator drift, fixture clamp wear, sensor calibration, paint thickness variation.

3. Body ร— Body

Gap between major body sections.

๐Ÿ“Œ Example: Back door โ†” RR side member

Tolerance: Gap โ‰ค 1.5mm ยท Level โ‰ค 1.5mm

Typical root cause: BIW fixture tolerance stack, weld gun drift, hemming force variation.

๐ŸŽฏ 4-Layer Architecture

Purpose: Input data + requirement table โ†’ analyse failure part/process โ†’ predict future failures via statistics.

Layer 1 โ€” Predetermined Requirement Table

Joint ID | Axis | Class | Spec | Min | Max | Measurement method

IP_UL   | Pร—P | C | Gap | 0.0 | 0.5 | Feeler 0.5mm
HL_FD   | Pร—B | A | Gap | 0.0 | 1.0 | Feeler 1.0mm
BD_RR   | Bร—B | B | Gap | 0.0 | 1.5 | Optical scanner
... (~50+ joints)

Layer 2 โ€” Data Input (line operators)

Web form + CSV upload + (optional) SPC API feed. Each row: Joint ID, value, inspector, station, VIN, timestamp, photo.

Layer 3 โ€” Analysis Engine (Python / SQLite)

Runs on sir's VPS. Computes:

Layer 4 โ€” Output & Alerts

Daily Telegram push + auto-filled 5-Whys template + TBP project auto-feed + weekly trend email.

๐Ÿ› ๏ธ Tech Stack (sir's existing VPS, $0 cost)

Frontend
HTML form

Port 8106, mobile-friendly

Backend
Python Flask

SQLite for measurements

Analysis
pandas + scipy

SPC, Weibull, Pareto, charts

Output
TG + Email

Daily digest, alerts

1. Part ร— Part โ€” fitment miss

Why 1: Gap exceeds 0.5mm โ†’ Why 2: Component dimension out of spec? โ†’ Why 3: Supplier lot variation / locator wear? โ†’ Why 4: Incoming inspection not catching? โ†’ Why 5: Gauge calibration overdue or fixture wear?

2. Part ร— Body โ€” fitment miss

Why 1: Gap exceeds 1.0mm โ†’ Why 2: Panel locator drift or body opening out of position? โ†’ Why 3: BIW fixture wear / hemming force? โ†’ Why 4: Preventive maintenance interval missed? โ†’ Why 5: PM schedule not tied to actual wear data?

3. Body ร— Body โ€” fitment miss

Why 1: Gap exceeds 1.5mm โ†’ Why 2: Weld gun drift or BIW fixture clamp wear? โ†’ Why 3: Weld schedule parameter out of range? โ†’ Why 4: Weld tip dressing interval missed? โ†’ Why 5: SPC data not fed back to PM schedule?

๐Ÿ“‹ Reference to Toyota K-Standard

Toyota K-standard (TMC engineering specification) defines gap/levelness tolerance bands by joint class:

Joint Class Gap Levelness Examples
Class A (visible exterior) โ‰ค 1.0mm โ‰ค 1.0mm Hood-fender, door-body, headlamp-fender
Class B (semi-visible) โ‰ค 1.5mm โ‰ค 1.5mm Back door - side member, IP - console
Class C (hidden / interior) โ‰ค 2.0mm โ‰ค 2.0mm IP upper - lower, console - duct

โš  Sir to confirm K-standard reference + ASSB plant-specific tightening (if any).

๐Ÿ“Š Why This Matters for Plant Director Track

Every fitment miss is a potential 0 COP event under sir's TBP theme:

This 3-axis classification = TBP project Step 2 (Grasp Current Situation) evidence. The fix trees = Step 4 (Establish Countermeasures).

๐ŸŽฏ KPIs (Draft)

1๏ธโƒฃ
3-axis classification complete by 15 Jul 2026
2๏ธโƒฃ
10 real plant cases documented by 31 Jul 2026
3๏ธโƒฃ
Standard fix tree per axis by 15 Aug 2026
4๏ธโƒฃ
Toyota K-standard crosswalk by 31 Aug 2026
5๏ธโƒฃ
Insert into TBP deck (Step 2) by 15 Sep 2026

โ“ Open Questions for Sir

โ“
Toyota K-standard version โ€” K-001 general or K-XXX door-specific?
โ“
Case priority โ€” which 10 COP cases to document first?
โ“
Measurement method โ€” manual feeler only, or include 3D scanner?
โ“
ASSB plant tolerance โ€” own standard tighter than K, or follow K verbatim?