Robust Fuzzy Regression Clustering Model Based on Maximum Entropy and L1 Norm and its Application to Power Load Analysis
摘要
The traditional fuzzy C-regression model (FCRM) is highly sensitive to noise and outliers. Although using L1 norm can enhance robustness, it lacks an analytical solution. This paper introduces a robust fuzzy regression clustering model by integrating the entropy regularization term based on the principle of maximum entropy and the alternating direction method of multipliers (ADMM). The proposed model, FCRE_L1, incorporates both the L1 norm and entropy penalty into the objective function and solves it through a block coordinate descent strategy. Simulation and real-world power load data demonstrate its superior performance in accuracy and robustness compared to traditional models.