AI based decision-making system for tooling design of aircraft product assembly with developed knowledge retrieval algorithm

This research develops a decision-making system for aircraft wing-spar tooling design that employs ontology-based Knowledge Retrieval Practices (KRP) to reduce search effort, improve traceability, and deliver decision-ready guidance. This research formalized a domain ontology, encode rule-based constraints, and structure case libraries, then introduce a query-information model that maps natural-language questions to machine-interpretable intents. The system orchestrates three retrieval modes, including ontology based semantic (OBS), rule-based inference (RBI), and case-based reasoning instances (CBRI), within a five-layer browser and server platform integrated with Product Development Management (PDM)/ Computer Aided Design (CAD). Evaluation spans twenty decision-like tasks, graded tooling corpora (100-420 documents), and a domain-agnostic stress test. The system achieved mean task-level accuracy of 93.1%, with the hybrid OBS+RBI+CBRI configuration reaching 96.9% confidence. Document-level accuracy was 98-99% on the tooling corpora, and the stress test showed small but consistent gains over traditional retrieval. OBS excelled for conceptual and attribute queries; RBI for computable dimensioning and sequencing; and CBRI for structural analogies and fine adjustments. Precision declined for legacy materials lacking ontology tags or using obsolete terminology, motivating retro-tagging and synonym expansion. Practically, it delivers a deployable KRP platform with role-based governance and enterprise interfaces, offering explainable, decision-ready guidance for complex assembly tasks. The study provides actionable guidance on when to use each retrieval strategy and establishes a reproducible evaluation protocol, laying the groundwork for longitudinal impact studies and public benchmarks in aerospace knowledge management. Limitations include the domain focus and reliance on curated corpora; these inform the roadmap for broader validation.