Bidirectional Traceability and Automated Specification Engineering

Overview

PoC #26, part of the KIMBA Stream 2 (MBSE), explores a visionary "Speedboat" approach. Moving beyond linear workflows, this project demonstrates bidirectional data synthesis: the ability to generate and align Requirements, System Models, and Test Cases in any direction. This lowers entry barriers for complex Model-Based Systems Engineering (MBSE) through intuitive AI chat interfaces.

The "Speedboat" Methodology

As a rapid exploration module, this PoC utilizes the Artificial Engineer tool to evaluate how Large Language Models (LLMs) can master specialized ontologies. By feeding the AI with specific system structures—such as the "Electric Trailer Coupling"—the tool provides:

• Ontological Clarity: Explaining complex MBSE relationships (e.g., how a Logical Cluster relates to a Solution Activity).
• Contextual Analysis: Summarizing inputs, outputs, and central actions within activity diagrams.
• Direct Model Validation: A seamless link between the AI chat and the modeling tool allows engineers to jump directly to a specific action in the diagram for visual verification.

Technical Implementation & Results

The core of this PoC is a highly sophisticated System Prompt combined with compressed JSON model data. The process follows a structured sequence:

1. Ontology Input: Defining the "Mission" and the semantic rules of the system context.
2. Model Processing: Analyzing elements like the Electric Trailer Coupling via JSON-formatted model exports.
3. Automated Specification: Generating structured requirements (Chapters 1-3+) that are ready for import into professional Requirements Management Tools.

Outcome

The experiment successfully extracted 163 individual requirement entries, including headings and classifications, from a single system model snippet. This serves as the foundation for the next stage: a specialized App designed for high-precision synchronization between MBSE models and live Requirement Management databases.