Clustering Analysis of Electric Vehicle Charging Behavior Based on the Improved Quantum K-Means Algorithm

With the rapid rise of the electric vehicle market and its increasingly significant role in power systems, the clustering analysis of high-dimensional electric vehicle charging data has emerged as an important research direction in load forecasting, demand response, and intelligent dispatch. However, the classical K-Means algorithms often misclassify a large portion of data into “noisy clusters” in high-dimensional settings due to the limitations of the Euclidean distance metric. Thereby the interpretability and practical value of the clustering results are diminished. In this paper, the improved quantum K-Means algorithm is developed upon the principles of normalization and inner product similarity from quantum computing. By converting the Euclidean distance metric into a quantum similarity-based metric, the improved algorithm may capture the subtle differences among the data effectively. Take the electric vehicle charging dataset as an example, the results show that the improved quantum K-Means algorithm surpasses the traditional method in reducing the within-cluster sum of squares and enhancing the silhouette coefficient, which significantly improves the stability and discriminability of the clustering outcomes. This study provides a unique way to identify the electric vehicle charging behaviors accurately for the optimal dispatch of the power system. Additionally, it validates the potential of quantum clustering methods in high-dimensional data processing.