Fault Prediction System for Network Operation Status Based on Improved Clustering Algorithm

Cluster analysis algorithm is an essential unsupervised learning algorithm in the field of data mining and machine learning, which can quickly divide large data into different types according to data characteristics. After half a century of accumulation and precipitation, cluster analysis algorithms have achieved quite sufficient research results, including hierarchical clustering, grid-based clustering, and density-based CAs suitable for various application scenarios. The main purpose of this paper is to study the system design of network operating status (OS) fault prediction (FP) based on the improved clustering algorithm (ICA). This paper mainly evaluates the operation state of the distribution network based on the unbalanced data CA. The algorithm in this paper improves the iterative center reduction formula on the basis of the IT2FKM algorithm, and the calculation time required is slightly longer than the classic IT2FKM algorithm. However, with the increase of cluster imbalance, compared with other algorithms, the clustering performance of the proposed algorithm has been significantly improved. It can be seen from the experimental results that the improved IT2FKM algorithm proposed in this paper has strong adaptability when clustering in imbalanced data sets, and does not require too much computational cost.