Accurate classification of electrical equipment types is crucial for enhancing energy efficiency and optimizing power grid operations. This paper presents a method for multiclass classification electrical loads—categorized as heating, cooling, or motor types—based on time series data of electrical measurements, specifically current (I), voltage (U), and power factor ( \(\cos \phi \) ). The proposed method includes an original feature extraction approach for further processing via Random Forest classification models. Feature extraction was conducted using two complementary approaches: statistical characteristics derived from the time domain and harmonic components obtained via Fast Fourier Transform (FFT). Two independent Random Forest classifiers were trained on each respective feature set and evaluated using metrics including accuracy, F1-score, ROC-AUC, and PR-AUC. A synthetic dataset was generated to simulate real-world measurement conditions, capturing operational variability across equipment categories. The results demonstrated high classification performance, with the statistical feature-based model showing slightly more stable outcomes. The proposed dual-model framework facilitates robust and interpretable classification of electrical equipment from limited input data, providing practical utility in grid monitoring.