Aiming at the problem about initial clustering center was randomly assigned in K-means clustering algorithm, the improved K-means clustering algorithm based on hierarchical clustering algorithm and K-means clustering algorithm was proposed in this paper. In the improved algorithm, first of all K was calculated by hierarchical clustering. When K was determined, K-means clustering was implemented. The results of the aero-engine vibration data clustering shown that not only the k value was to quickly and accurately determined, but also the number of clusters can be reduced and higher computing efficiency can be attained by the improved K-means clustering algorithm.

A improved genetic algorithm is proposed based on a new fitness function in allusion to the problem that the traditional genetic algorithm is not fully consider the knowledge of the problem itself.The improved genetic algorithm is used to analyze the fault feature , to extract the fault and remove redundant characteristic parameters for the fault classification and calculation.The diagnosis example shows that the method has faster convergence speed and can be effective for fault identification.

No abstract