A Spacecraft Electrical Characteristics Multi-Label Classification Method Based on Off-Line FCM Clustering and On-Line WPSVM
Fig 3
Flow chart of the FCM used in our article.
The steps of fuzzy c-means clustering have several sections: first, the aim of fuzzy c-means clustering is to divide n samples into a fuzzy set. Second, each group’s clustering center can be calculated by minimizing the value function. Third, fuzzy c-means gives each sample a membership value between 0 and 1 to indicate its degree of membership to each cluster. It is a typical iterative process. Through the clustering algorithm, the off-line system generates the spacecraft electronic load historical data clustering, and gets the clustering center for events.