Existing work on knowledge management in aircraft assembly provides limited insight into the optimal knowledge content, proactive decision frameworks and the concrete role of knowledge push in improving decision accuracy. To address these gaps, this study aims to propose a flow-driven knowledge push method that predicts and delivers task-relevant information in line with task progress and designer preferences, thereby enhancing decision-making precision in aircraft assembly processes (AAPs).
The study first analyzes the knowledge push requirements of AAP and the theoretical foundations of active knowledge delivery. On this basis, a multilayer knowledge push model driven by knowledge flow is developed. The model integrates a document- and topic-layer representation, term frequency-inverse document frequency and cosine similarity for content modeling, particle swarm optimization for clustering and a dynamic programming algorithm for computing knowledge flow similarity. A time-weighted push rating mechanism is then formulated to rank and recommend knowledge documents. The approach is implemented and validated in a decision-making system for tooling design of aircraft wing spar assembly.
Experimental results show that the proposed mechanism achieves high performance, with Precision = 0.91, Recall = 0.93 and F1-score = 0.92, while also reducing runtime compared with an existing static model. By jointly considering user preferences, task progress, knowledge flow similarity and time-weighted document engagement, the method significantly improves the accuracy and relevance of knowledge push, thereby increasing tooling design efficiency and shortening the design cycle.
This research introduces a dynamic and flow-driven active knowledge push framework that integrates explicit and tacit knowledge for engineering decision support in aircraft assembly. The two-layer knowledge flow model and time-aware push rating mechanism provide a more personalized, precise and context-aware alternative to conventional static or one-dimensional recommendation methods.
