A hybrid feature selection model for identifying groups of critical elements in aero-engine assembly

Aero-engine assembly has a wide range of indicators, and finding critical assembly elements to guide assembly and troubleshooting is essential in improving the reliability and safety of aero-engine. In order to identify critical elements in aero-engine assembly components, this study aims to establish a two-stage hybrid feature selection model, namely, FSBP approach, which integrated filter method and particle swarm algorithm with Bayesian optimization. Specifically, individual filter feature selection methods are compared to select a relatively effective method to reduce the data dimensions and ensure the quality of the initial subset. Then, the particle swarm algorithm combined with Bayesian optimization obtains a subset of features in the second stage that are more suitable for predictive models with more robust classification prediction capability. The algorithm is successfully applied to real aero-engine assembly and trial test datasets, and the experimental results show that our proposed two-stage hybrid feature selection model outperforms other benchmark methods.