Research on the Quantitative Method for Performance Dispersion of Turbofan Engines

This study addresses the challenge of performance dispersion in turbofan engines, attributed to deviations from design values due to noise factors during processing and assembly, which can result in a low engine qualification rate and elevated failure rates. A novel performance dispersion analysis method, grounded in a model correction approach, is proposed to quantify the overall and component performance distribution characteristics of 130 engines in batch production. Utilizing a multi-objective genetic algorithm, the Pareto set of correction factors for each engine component is derived from the traditional model identification method. Subsequently, a strategy based on the minimum sum of weighted distances in high-dimensional space is employed to select correction factors using multiple test data sets from a single engine. This method facilitates the modification of personalized component models for multiple engines, yielding performance data sets for various engine components. The unified working condition engine performance is calculated using the modified personalized engine model, obtaining the overall performance data set. Finally, a distribution test quantifies the distribution characteristics of both components and engines. The method is validated with measured data, providing a pivotal foundation for further exploration and control of key uncertain factors affecting engine performance dispersion.