Harmonic Source Modeling Method Based on Random Forest Algorithm

Abstract

In recent years, with the rapid development of power electronics technology, power electronic devices are widely used in life, making nonlinear loads a large class of harmonic sources injected into the power grid. Due to the small capacity and low power of loads for home users, the harmonic problem can be ignored, while industrial users, due to their large capacity and many nonlinear devices, cause serious harmonic problems that cannot be ignored. If the harmonics are not reasonably and effectively intervened, they will flow into a higher level power system and cause a series of related power quality problems, which makes the design of power quality management scheme difficult. In the process of harmonic analysis and management, it is necessary to establish an accurate model of harmonic sources. By applying the power quality monitoring data of industrial users, this paper proposes a harmonic source modeling method. Firstly, preprocess the power quality monitoring data, then the characteristic matrix of harmonic voltage and current is constructed, and the decision tree is built by training the harmonic data, and the unknown harmonic voltage and current values are regressed and analyzed, and finally a harmonic source model based on random forest algorithm is established, and the proposed method is verified by applying the power quality monitoring data of industrial users at 110 kV voltage level in a certain place. The results prove that the proposed method has higher accuracy for harmonic current prediction compared with the traditional harmonic source model, and can accurately evaluate and predict the impact of each harmonic voltage difference on harmonic current, which is beneficial to provide theoretical basis for preventing harmonic pollution and analyzing the influencing factors of power quality characteristics of power system.