Forecasting of Harmonic Distortions in MV Distribution Network Caused by Illumination Devices

The increase of lighting elements in a power system leads to a corresponding increase in power quality disturbances caused by these devices. When the number reaches hundreds, the distortions begin to affect medium voltage levels rather than low voltage. Outside of specified systems, such as street and outdoor illumination, the use of lighting elements varies widely depending on working conditions and individual or institutional behaviour. Therefore, for an accurate and functional power system planning, it is not enough to only measure the harmonics arising from lighting elements with variable characteristics, but also harmonic estimation is needed, which is also the main motivation of this study. According to this aim, in this study a real-world dataset obtained from a MV distribution system supplying mixed illumination installations used for forecasting of harmonics based on such devices. The analysis of the data set revealed that there is a correlation between current harmonics and voltage variations, so the algorithm was designed based on this relationship, since voltage is an easily measurable power system parameter. Combined AI models were used for forecasting to reach the best results. Forecasting performance is discussed by using numerical values and graphical expressions, where the MSE, RMSE and MAE, and SMAPE values are 1.2834, 1.11328, 0.8095, and 10.7692 respectively.