Unit-oriented aeroengine assembly and test-run linkage framework: A data-driven complex network approach

The performance of aeroengines, as complex and precise thermodynamic systems, is highly influenced by assembly quality. However, the lack of integration between assembly and test-run stages hampers the effective use of massive assembly data for closed-loop optimization. Moreover, in mass production characterized by unit-oriented structural design, there is no scalable method to pre-evaluate unit assembly quality. These limitations lead to greater variability in full-machine performance and higher test-run failure rates, which causes rework and significantly increases manufacturing costs. To address this, a unit-oriented aeroengine assembly and test-run linkage framework is proposed. Firstly, assembly data are used to construct network and dynamics models to characterize unit assembly quality. Combined with resilience theory, the unit assembly quality evaluation index and qualified interval are proposed, enabling pre-evaluation and filtering of units before full-machine assembly. Next, using Bayesian theory, the unit-oriented full-machine test-run performance prediction model is constructed, capturing the causal links and probability mappings between unit combinations and full-machine performance. Finally, integrating the above models, the current assembly and test-run process is optimized by building a qualified unit library, predicting the full-machine test-run pass probability of unit combinations, and using feedback from test results to generate actionable assembly recommendations, thereby achieving coordination between assembly and testing. According to the experimental results, the dynamic model achieves fitting errors as low as 2%, underscoring its high accuracy and reliability in characterizing unit assembly quality, while the test-run results and maintenance records provide strong empirical validation. This framework improves coordination between assembly and test-run processes, reduces failure and rework, and supports unit interchangeability and optimal matching. It also underpins the development of digital engineering and intelligent assembly, with significant engineering value and broad application prospects.