Improved K-Prototype Clustering Algorithm for Plant-Side of Power Monitoring System

On the plant side of the power monitoring system, in order to achieve non-destructive and lightweight vulnerability inventory thus proposing cluster analysis, the target assets are clustered, and the impact on the plant-side hosts during vulnerability inventory is greatly reduced through the optimization of the scanning strategy. The asset information collected by the scan is processed to obtain a hybrid dataset. Using the improved k-prototypes clustering algorithm to cluster the target assets, and according to the clusters of hosts with different functions obtained by clustering, an optimized scanning strategy is adopted to change the scanning behavior and reduce the impact caused to the target network. The improvement of the clustering algorithm is mainly reflected in the initial category center selection and distance calculation. For the initial category center selection, the phase dissimilarity measure is adopted for the initial category center selection; in the distance calculation, the entropy weight value is introduced to change the traditional method of randomly determining the attribute weights, which avoids the local optimum and improves the stability of the algorithm.