Probabilistic Forecasting of High-Frequency Voltage Distortions in Low-Voltage Networks

Waveform distortion disturbances deteriorate Power Quality (PQ) in power systems and affect electrical installations and equipment. In modern Low-Voltage (LV) distribution systems, electrical signals are characterized by the contemporaneous presence of low-frequency and High-Frequency (HF) spectral components (i.e., up to 150 kHz). This paper focuses on HF voltage distortions, also known as supraharmonics, that can create losses and reduce the efficiency of the equipment. Accurate predictions of future HF voltage distortions help distribution system operators and regulators in taking timely actions to define and withstand PQ limits or to analyze the residual capacity of networks toward the additional installation of power electronic converters. This paper provides a probabilistic approach to forecast HF voltage distortions by applying a methodology based on Quantile Regression Forests (QRFs). Results of numerical experiments based on publicly available actual data collected at a LV distribution system are presented to assess the performance of the proposed methodology.