There are many health status parameters for aeroengine, leading to partly information overlap. The accuracy of commonly used evaluation methods is seriously restricted, due to the reliance on human subjective experience. The PCA-Kmeans combination algorithm for aeroengine health status evaluation is constructed, the specific steps are proposed, and the result is evaluated and verified by RBF neural network. Taking NASA public dataset as an example, the experimental results suggest that the PCA-Kmeans combined algorithm is well suited to health status clustering based on PCA dimension reduction, and is basically consistent with the evaluation results of RBF neural network. It provides a reference for large scale objective evaluation of aeroengine health status, comprehensively mastering the overall performance degradation of engine and scientifically making maintenance decisions.
|