A clustering method for power time series curves based on improved self-organizing mapping algorithm
Jiahao Yan, Chao Zhang, Yaping Li · 2023
To solve the clustering task of power time series curves, we propose an improved self-organizing mapping algorithm (improve SOM). We optimized the process of feature extraction before SOM algorithm considering the massive and noisy nature of power load data: firstly, we used wavelet transform to denoise and reduce the interference of noise on the clustering process; then extracting power features from denoised power time series curves; then use the improved SOM clustering algorithm to cluster based on the extracted power features. The clustering algorithm proposed in this paper can deal with the clustering task of a large amount of power data and has good accuracy and robustness.